The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
4 T h e R o le o f M o n e y and M o n e ta ry P o lic y 23 T h e Causes and C on seq u en ces o f L e v e r a g e d Buyouts 35 T h e L in k B e tw e e n M l and th e M o n e ta r y Base in th e 1980s 53 M o n e ta r y T a r g e tin g w ith E xch an ge Rate C onstraints: T h e B un desbank in th e 1980s t iie FEDERAL A K1SIR M HAN K of ST.IXH IS 1 Federal R eserve B ank o f St. Lou is R eview September/October 1989 In This Issue . . . W e begin this issue o f the Review with a brief eulogy for Karl Brun ner who passed away on May 9, 1989. All of us who are concerned with monetary theory and policy have lost an inspired teacher, an en joyable colleague and a good friend. In his memory, we have reprinted his article on "The Role of Money and Monetary Policy,” which was originally published in this Review in 1968. * * * The wave of leveraged buyouts over the past decade has led many to question the social value of this type o f activity in the market for cor porate control. Some observers, arguing that leveraged buyouts merely redistribute wealth with a possible negative net effect or deteriorate the stability o f the financial system, have proposed that they be restricted. In the second article of this Review, "The Causes and Consequences of Leveraged Buyouts,” Michelle R. Garfinkel argues that the expressed skepticism about leveraged buyouts is implicitly premised on the notion that these transactions have no real impact on the target firms’ produc tive capacity. In discussing the theory and existing empirical evidence that suggests leveraged buyouts are productive, at least in an expected sense, she points out that arguments to restrict this activity are less compelling; if leveraged buyouts actually enhance the target firms’ per formance, then legal restrictions could prove to be harmful. * * * During the 1980s, several changes in the Federal Reserve System’s reserve requirements have altered the relationship between the money stock, M l, and the monetary base. The Monetary Control Act of 1980 brought all depository institutions under a uniform set o f reserve re quirements and removed reserve requirements on a broad category of time and savings deposits. In 1984, the Federal Reserve switched to con temporaneous reserve accounting. In the third article o f this Review, "The Link Between M l and the Monetary Base in the 1980s,” Michelle R. Garfinkel and Daniel L. Thornton show that, under fairly general condi tions, these changes should have made the link between M l and the monetary base tighter in this decade, by eliminating or, at least, diminishing the importance of some sources o f variability in that link. Garfinkel and Thornton present evidence that is consistent with their arguments. * * * SEPTEMBER/OCTOBER 1989 2 Coordinating economic policies among countries, particularly to con trol exchange rate movements, has been advocated by an increasing number of analysts in the 1980s. While such coordination seems desirable in theory, however, its goals can often be at odds with an in dividual nation’s domestic monetary policies. In the final article in this Review, “Monetary Targeting with Exchange Rate Constraints: The Bundesbank in the 1980s,” Jurgen von Hagen ex amines the German central bank’s experience with international ex change rate policies and domestic monetary control. The author reviews recent research, which finds that the Bundesbank’s ability to achieve its monetary targets has been unhindered by its coordinated exchange rate policies. Von Hagen’s analysis, however, shows that the Bundesbank’s participation in coordinated interventions since 1985 have contributed significantly to the growth of the German money supply in excess o f its monetary targets in the second half of the 1980s. http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis 3 In M em oriam Karl Brunner 1916— 1989 When Karl Brunner died on May 9, 1989, the Federal Reserve Bank of St. Louis lost a teacher, a supporter, a critic, and above all, a very dear friend. His association with this Bank began in 1967, when he began policy discussions with the staff. During this span of 22 years, he read the ar ticles in our Review, both criticizing and com plimenting them, never allowing us to forget that improving monetary policy was our first and foremost task. His participation at our an nual economic policy conferences provoked the kind of heated discussions that make such meetings worthwhile. Karl was born on February 16, 1916, in Zurich, Switzerland. He completed his formal education in 1942 and came to the United States in 1950. Over the years, he was a member of the faculties of the University of California, Los Angeles, Ohio State University and the Univer sities of Konstanz and Bern in Switzerland. He joined the University of Rochester in 1971, where he remained until his death. Throughout his career, Karl was one of those rare individuals who neither stopped learning nor stopped teaching. He is best known as one of the major advocates of monetarist theory— indeed, I believe he coined the word “monetar ism" in our Review—but he never merely passed on what he had learned from others. Like most students o f his generation, he had a thorough grounding of Keynesian economics; unlike most others, he questioned its logical and empirical implications, which led him to abandon the doc trine in favor o f the monetarist approach to which he added so much. Despite his prodigious contributions to the body of economic knowledge, advanced in more than 200 books and articles, his chief concern was always with the passing on o f his discoveries to others. I think that he was hap piest either in a classroom, giving advice to young economists, or at the various central banks, giving advice to old institutions. The con ferences that he helped to establish made an enormous contribution to the education o f a couple o f generations of American and Euro pean economists. His impact on policy was substantial as well. His teachings directly or indirectly touched two central banks in particular: the Swiss National Bank and the German Bundesbank. Not surpris ingly, these two institutions today have the most stable price levels among the developed coun tries. Even the Federal Reserve System, which steadfastly downplayed the usefulness of monetary discipline, has become more sensitive to the causes and effects of fluctuations in the nation's money stock than it was 20 years ago. This heightened sensitivity is due, in large measure, to Karl’s research and teachings. Karl Brunner was a totally honest individual: he didn’t boast about his own achievements, nor did he suffer fools lightly. This straightforward ness, though it made some uncomfortable, gave his words credibility. It was this credibility that made him influential among monetary policymakers. He has left an indelible imprint on monetary economics and monetary policy; his influence will continue in the future through the work of his students and colleagues who learaned so much from him. Like all of them, we shall miss him. —A. B. Balbach Director of Research SEPTEMBER/OCTOBER 1989 4 The follow ing article is reprinted fr o m the July 1968 issue o f the Federal Reserve Bank o f St. Louis Review. Karl Brunner The Role of Money and Monetary Policy TSL HE DEVELOPMENT of monetary analysis in the past decade has intensified the debate con cerning the role of money and monetary policy. Extensive research fostered critical examinations o f the Federal Reserve’s traditional descriptions o f policy and of the arrangements governing policymaking. Some academic economists and others attribute the cyclical fluctuations of monetary growth and the persistent problem concerning the proper interpretation of monetary policy to the established procedures of monetary policy and the conceptions tradi tionally guiding policymakers. The critique o f established policy procedures, which evolved from this research into questions concerning the monetary mechanism, is de rived from a body of monetary theory referred to in this paper as the Monetarist position. Three major conclusions have emerged from the hypotheses put forth. First, monetary im pulses are a major factor accounting for varia tions in output, employment and prices. Second, movements in the money stock are the most reliable measure o f the thrust o f monetary im pulses. Third, the behavior of the monetary authorities dominates movements in the money stock over business cycles. 1Lyle Gramley and Samuel Chase, “ Time Deposits in Monetary Analysis,” Federal Reserve Bulletin, October 1965. John H. Kareken, “ Commercial Banks and the Supply of Money: A Market Determined Demand Deposit Rate,” Federal Reserve Bulletin, October 1967. J. A. Cacy, http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis A response to the criticisms of existing mone tary policy methods was naturally to be ex pected and is welcomed. Four articles which de fend present policy procedures have appeared during the past few years in various Federal Reserve publications.1 These articles comprise a countercritique which argues that monetary im pulses are neither properly measured nor ac tually transmitted by the money stock. The authors reject the Monetarist thesis that mone tary impulses are a chief factor determining variations in economic activity, and they con tend that cyclical fluctuations of monetary growth cannot be attributed to the behavior of the Federal Reserve authorities. These fluctua tion are claimed to result primarily from the behavior o f commercial banks and the public. The ideas and arguments put forth in these articles deserve close attention. The controversy defined by the critique of policy in professional studies and the countercritique appearing in Federal Reserve publications bears on issues of fundamental importance to public policy. Under lying all the fashionable words and phrases is the fundamental question: What is the role of monetary policy and what are the requirements of rational policymaking? “ Alternative Approaches to the Analysis of the Financial Structure,” Monthly Review, Federal Reserve Bank of Kan sas City, March 1968. Richard G. Davis, “ The Role of the Money Supply in Business Cycles,” Monthly Review, Federal Reserve Bank of New York, April 1968. 5 The following sections discuss the major as pects of the countercritique. These rejoinders may contribute to a better understanding of the issues, and the resulting clarification may re move some unnecessary disputes. Even though the central contentions of the controversy will remain, the continuous articulation of opposing points of view plays a vital role in the search for greater understanding of the monetary process. A SUM M AR Y OF THE CO UNTERCR ITIQUE The four articles relied on two radically dif ferent groups o f arguments. Gramley-Chase, Kareken and Cacy exploit the juxtaposition “New View versus Traditional View" as the cen tral idea guiding their counter critique. The ana lytical framework developed by the critique is naturally subsumed for this purpose under the "Traditional View” label. On the other hand, Davis uses the analytical framework developed by the critique in order to organize his arguments. Gramley-Chase describe their general argu ment in the following words: "(New) developments have reaffirmed the bankers' point of view that deposits are attracted, not created, as textbooks suggest. In this new environ ment, growth rates of deposits have become more suspect than ever as indicators of the conduct of monetary policy. . . . A framework of analysis [is required] from which the significance of time deposits and of changing time deposits can be deduced. Traditional methods of monetary analysis are not well suited to this task. The 'New View’ in monetary economics provides a more useful analytical framework. In the new view, bankslike other financial institutions—are considered as suppliers of financial claims for the public to hold, and the public is given a significant role in deter mining the total amount of bank liabilities. . . . Traditional analysis. . . fails to recognize that substitution between time deposits and securities may be an important source of pro-cyclical varia tions in the stock of money even in the face of countercyclical central bank policy.”2 This general argument guided the construction of an explicit model designed to emphasize the 2Gramley-Chase, pp. 1380, 1381, 1393. 3Cacy, pp. 5 & 7. role of the public’s and the banks’ behavior in the determination of the money stock, bank credit and interest rates. Kareken’s paper supplements the GramleyChase arguments. He finds “the received money supply theory” quite inadequate. His paper is designed to improve monetary analysis by con structing a theory o f an individual bank as a firm. This theory is offered as an explanation of a bank’s desired balance sheet position. It also appears to form the basis of a model describing the interaction of the public's and the banks' behavior in the joint determination of the money stock, bank credit and interest rates. The whole development emphasizes somewhat suggestively the importance of the public’s and banks’ behavior in explanations of monetary growth. It is also designed to undermine the empirical hypotheses advanced by the Monetar ist position. This is achieved by means of ex plicit references to specific and “obviously desirable" features of the model presented. Cacy’s article develops neither an explicit framework nor a direct critique of the basic propositions advanced by the Monetarist thesis. However, he provides a useful summary of the general position of the countercritique. The Monetarist analysis is conveniently subsumed by Cacy under a "Traditional View” which is jux taposed to a "New View” of monetary mecha nisms: "The new approach argues. . . that there is no essential difference between the manner in which the liabilities of banks and nonbank financial institutions are determined. Both types of institutions are subject in the same way to the portfolio decisions of the public.”3 The new approach is contrasted with the Traditional View, which "obscures the important role played by the public and overstates the role played by the central bank in the determination of the volume o f money balances."4 The general comparison developed by Cacy suggests quite clearly to the reader that the Traditional View allegedly espoused by the Monetarist position cannot match the "realistic sense” of the New View advocated by the countercritique. In the context o f the framework developed by the critique, Davis questions some basic proposi tions of the Monetarist position: 4lbid., p. 7. SEPTEMBER/OCTOBER 1989 6 "In the past five to ten years, however, there has come into increasing prominence a group of economists who would like to go considerably beyond the simple assertion that the behavior of money is a significant factor influencing the behavior of the economy. . . . In order to bring a few of the issues into sharper focus, this article will take a look at some evidence for the ‘money supply' view. . . . It confines itself to examining the historical rela tionship between monetary cycles and cycles in general business. The article concludes that the relationship between these two kinds of cycles does not, in fact, provide any real support for the view that the behavior of money is the predomi nant determinant of fluctuations in business activi ty. Moreover, the historical relationship between cycles in money and in business cannot be used to demonstrate that monetary policy is, in its effects, so long delayed and so uncertain as to be an un satisfactory countercyclical weapon.”5 A N EX A M IN A TIO N OF THE ISSUES A careful survey o f the countercritique yield ed the following results. The Gramley-Chase, Kareken, and Cacy papers parade the New View in order to question the status of empirical theories used by the Monetarist critique in its examination o f monetary policy. The Davis paper questions quite directly, on the other hand, the existence and relevance of the evi dence in support o f the Monetarist position, and constitutes a direct assault on the Monetarist critique. The others constitute an indirect assault which attempts to devalue the critique's analysis, and thus to destroy its central proposi tions concerning the role of money and mone tary policy. The indirect assault on the Monetarist position by Gramley-Chase, Kareken and Cacy requires a clarification concerning the nature o f the New View. A program o f analysis must be clearly distinguished from a research strategy and an array of specific conjectures.6 All three aspects are usually mixed together in a general descrip tion. It is important to understand, however, that neither research strategy nor specific em 5Davis, pp. 63-64. 6These three aspects of the New View will subsequently be elaborated more fully. Their program of analysis refers to the application of relative price theory to analysis of finan cial markets and financial institutions. Their research strategy refers to a decision to initiate analysis in the con text of a most general framework. Their specific conjec tures refer to propositions concerning the causes of fluc http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis pirical conjectures are logical implications o f the general program. The explicit separation o f the three aspects is crucial for a proper assessment of the New View. Section A examines some general character istics of the countercritique’s reliance on the New View. It shows the New View to consist of a program acceptable to all economists, a re search strategy rejected by the Monetarist posi tion, and an array o f specific conjectures ad vanced without analytical or empirical substanti ation. Also, not a single paper of the countercritique developed a relevant assessment o f the Monetarist’s empirical theories or central prop ositions. In sections B and C detailed examinations of specific conjectures centered on rival explana tions of cyclical fluctuations o f monetary growth are presented. The direct assault on the Monetarist position by Davis is discussed in some detail in section D. This section also states the crucial propositions of the Monetarist thesis in order to clarify some aspects of this position. This reformulation reveals that the reservations assembled by Davis are quite innocuous. They provide no analytical or empirical case against the Monetarist thesis. Conjectures associated with the interpretation of monetary policy (the “indicator problem") are presented in section E. A. The New View The countercritique has apparently been decisively influenced by programmatic elabora tions originally published by Gurley-Shaw and James Tobin.7 The program is most faithfully reproduced by Cacy, and it also shaped the arguments guiding the model construction by Kareken and Gramley-Chase. The New View, as a program, is a sensible response to a highly unsatisfactory state o f monetary analysis in herited in the late 1950’s. A money and banking syndrome perpetuated by textbooks obstructed the application of economic analysis to the financial sector. At most, this inherited litera ture contained only suggestive pieces of an alysis. It lacked a meaningful theory capable of tuation of monetary growth and propositions about proper interpretation of policy. 7John G. Gurley and Edward F. Shaw, Money in a Theory of Finance, (Washington: Brookings Institute, 1960). James Tobin, “ Commercial Banks as Creators of Money,” Bank ing and Monetary Studies, ed. Deane Carson (R. D. Irwin, 1963). 7 explaining the responses o f the monetary sys tem to policy actions or to influences emanating from the real sector. The New View proposed a systematic application of economic analysis, in particular an application of relative price theory, to the array of financial intermediaries, their assets and liabilities. This program is most admirable and in contestable, but it cannot explain the conflict revealed by critique and countercritique. The Monetarist approach accepted the general prin ciple of applying relative price theory to the analysis of monetary processes. In addition, this approach used the suggestions and analytical pieces inherited from past efforts in order to develop some specific hypotheses which do ex plain portions o f our observable environment. The New Viewers’ obvious failure to recognize the limited content o f their programmatic state ments only contributes to maintenance of the conflict. A subtle difference appears, however, in the research strategy. The New View was introduc ed essentially as a generalized approach, in cluding a quite formal exposition, but with little attempt at specific structuring and empirical content. The most impressive statements pro pagated by the New View w ere crucially in fluenced by the sheer formalism of its exposi tion. In the context of the New View’s almost empty form, little remains to differentiate one object from another. For instance, in case one only admits the occurrence of marginal costs and marginal yields associated with the actions of every household, firm, and financial interme diary, one will necessarily conclude that banks and non-bank financial intermediaries are restricted in size by the same economic forces and circumstances. In such a context there is truly no essential difference between the deter mination of bank and non-bank intermediary liabilities, or between banks and non-bank in adequate analysis of the medium of exchange function of money, or of the conditions under which inside money becomes a component of wealth, was obstructed by the programmatic state of the New View. The useful analysis of the medium-of-exchange function depends on a decisive rejection of the assertion that “ everything is almost like everything else.” This analysis requires proper recognition that the marginal cost of information concerning qualities and properties of assets differs substantially between assets, and that the marginal cost of readjusting asset positions depends on the assets involved. The analysis of the wealth position of inside money requires recognition of the marginal productivity of inside money to the holder. Adequate attention to the relevant differences between termediaries, or between money and other fi nancial assets. The strong impressions conveyed by the New View thus result from the relative emptiness of the formulation which has been used to elabo rate their position. In the context of the formal world of the New View, "almost everything is almost like everything else.” This undifferentia ted state of affairs is not, however, a property of our observable world. It is only a property of the highly formal discussion designed by the New View to overcome the unsatisfactory state o f monetary analysis still prevailing in the late 1950’s or early 1960’s.8 Tw o sources o f the conflict have been recog nized thus far. The Monetarists’ research strat egy was concerned quite directly with the construction of empirical theories about the monetary system, whereas the New View in dulged, for a lengthy interval, in very general programmatic excursions. Moreover, the New Viewers apparently misconstrued their program as being a meaningful theory about our obser vable environment. This logical error con tributed to a third source o f the persistent conflict. The latter source arises from the criticism ad dressed by the New Viewers to the Monetarists’ theories o f money supply processes. Three of the papers exploit the logically dubious but psychologically effective juxtaposition between a "New View” and a “Traditional View." In doing this they fail to distinguish between the inheri ted state of monetary system analysis typically reflected by the money and banking textbook syndrome and the research output of economists advocating the Monetarist thesis. This distinction is quite fundamental. Some for mal analogies misled the New Viewers and they did not recognize the logical difference between detailed formulations o f empirical theories on various cost or yield functions associated with different assets or positions is required by both problems. The blandness of the New View’s standard program cannot cope with these issues. The reader may consult a preliminary approach to the analysis of the medium of ex change function in the paper by Karl Brunner and Allan H. Meltzer, in the Journal of Finance, 1964, listed in footnote 9. He should also consult for both issues the important book by Boris Pesek and Thomas Saving, Money, Wealth and Economic Theory, The Macmillan Company, New York, 1967, or the paper by Harry Johnson, “ Inside Money, Outside Money, Income, Wealth and Welfare in Monetary Theory,” to be published in the Journal of Money, Credit and Banking, December 1968. SEPTEMBER/OCTOBER 1989 8 the one side and haphazard pieces o f unfinished analysis on the other side.9 A related failure accompanies this logical er ror. There is not the slightest attempt to assess alternative hypotheses or theories by systematic exposure to observations from the real world. It follows, therefore, that the countercritique scarcely analyzed the empirical theories advanc ed by the Monetarist critique and consequently failed to understand the major implications of these theories. For instance, they failed to recognize the role assigned by the Monetarist view to banks’ be havior and the public’s preferences in the monetary process. The objection raised by the New View that "the formula [expressing a basic framework used to formulate the hypothesis] obscures the important role played by the public” has neither analytical basis nor meaning. In fact, the place of the public’s behavior was discussed in the Monetarist hypotheses in some detail. Moreover, the same analysis discussed the conditions under which the public’s behavior dominates movements o f the money stock and bank credit.10 It also yielded informa- 9As examples of the empirical work performed by the Monetarists, the reader should consult the following works: Milton Friedman and Anna Jacobson-Schwartz. A Monetary History of the United States, 1867-1960, (Princeton: Princeton University Press, 1963). Philip Cagan, Determinants and Effects of Changes in the Stock of Money, (Columbia: Columbia University Press, 1965). Karl Brunner and Allan H. Meltzer, “ Some Further In vestigations of Demand and Supply Functions for Money,” Journal of Finance, Volume XIX, May 1964. Karl Brunner and Allan H. Meltzer, “ A Credit-Market Theory of the Money Supply and an Explanation of Two Puzzles in U.S. Monetary Policy,” Essays in Honor of Marco Fanno, 1966, Padova, Italy. Karl Brunner and Robert Crouch, ‘‘Money Supply Theory and British Monetary Experience,” Methods of Operations Research III—Essays in Honor of Wilhelm Krelle, ed. Rudolf Henn (Published in Meisenheim, Ger many, by Anton Hain, 1966). Karl Brunner, “ A Schema for the Supply Theory of Money,” International Economic Review, 1961. Karl Brunner and Allan H. Meltzer, “ An Alternative Approach to the Monetary Mechanism,” Sub committee on Domestic Finance, Committee on Banking and Currency, House of Representatives, August 17, 1964. 10The reader will find this analysis in the following papers: Karl Brunner and Allan H. Meltzer, “ Liquidity Traps for Money, Bank Credit, and Interest Rates,” Journal of Political Economy, April 1968. Karl Brunner and Allan H. Meltzer, “ A Credit-Market Theory of the Money Supply and an Explanation of Two Puzzles in U.S. Monetary Policy,” Essays in Honor of Marco Fanno, Padova, Italy, 1966. ” The reader is, of course, aware that these assertions re quire analytic substantiation. Such substantiation cannot be supplied within the confines of this article. But the reader could check for himself. If he finds, in the context of the countercritique, an analysis of the monetarists’ ma jor hypotheses, an examination of implication, and ex http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis tion about the response o f bank credit, money stock and time deposits to changes in ceiling rates, or to changes in the speed with which banks adjust their deposit-supply conditions to evolving market situations. Every single aspect of the banks’ or the public’s behavior emphasiz ed by the countercritique has been analyzed by the Monetarist’s hypotheses in terms which render the results empirically assessable. Little remains, consequently, of the suggestive countercritique assembled in the papers by Gramley-Chase, Kareken and Cacy.11 B. A Monetarist Examination o f the N ew View ’s M o n ey Supply Theory Three sources of the conflict have been dis cussed thus far. Tw o sources w ere revealed as logical misconstruals, involving inadequate con struction and assessment of empirical theories. A third source pertains to legitimate differences in research strategy. These three sources do not explain all major aspects of the conflict. Beyond the differences in research strategy and logical misconceptions, genuinely substantive issues re main. Some comments of protagonists advocating posure to observations, I would have to withdraw my statements. A detailed analysis of the banks’ and the public’s role in the money supply, based on two different hypotheses previously reported in our papers will be developed in our forthcoming books. This analysis, by its very existence, falsifies some major objections made by Cacy or Gramley-Chase. Much of their criticism is either innocuous or fatuous. Gramley-Chase indulge, for in stance, in modality statements, i.e. statements obtained from other statements by prefixing a modality qualifier like “ maybe” or “ possibly.” The result of qualifying an em pirical statement always yields a statement which is necessarily true but also quite uninformative. The modality game thus yields logically pointless but psychologically ef fective sentences. Cacy manages, on the other hand, some astonishing assertions. The New View is credited with the discovery that excess reserves vary over time. He totally disregards the major contributions to the analysis of excess reserves emanating from the Monetarists’ research. A detailed analysis of excess reserves was developed by Milton Friedman and Anna Schwartz in the book mentioned in footnote 9. The reader should also note the work by George Morrison, Liquidity Preferences of Commercial Banks, (Chicago: University of Chicago Press, 1966), and the study by Peter Frost, “ Banks' Demand for Excess Reserves,” an unpublished dissertation submitted to the University of California at Los Angeles, 1966. The classic example of an innocuous achievement was sup plied by Cacy with the assertion: “ . . . the actual volume of money balances determined by competitive market forces may or may not be equal to the upper limit established by the central bank” (p. 8). Indeed, we knew this before the New View or Any View, just as we always knew that “ it may or may not rain tomorrow.” The reader should note that similar statements were produced by other authors with all the appearances of meaningful elaborations. 9 the New View should probably be interpreted as conjectures about hypotheses to be expected from their research strategy. It should be clear ly understood that such conjectures are not logi cal implications o f the guiding framework. In stead, they are pragmatic responses to the gen eral emphasis associated with this approach. A first conjecture suggests that the money stock and bank credit are dominated by the public’s and the Banks’ behavior. It is suggested, therefore, that cyclical fluctuations of monetary growth result primarily from the responses of banks and the public to changing business con ditions. A second conjecture naturally supple ments the above assertions. It is contended that the money stock is a thoroughly "untrustworthy guide to monetary policy.” Articles by Gramley-Chase and Kareken at tempt to support these conjectures with the aid o f more explicit analytical formulations allegedly expressing the general program o f the New View. The paper contributed by Gramley-Chase has been critically examined in detail on another occasion, and only some crucial aspects relevant for our present purposes will be con sidered at this point.12 Various aspects of the first conjecture are examined in this and the next section. The second conjecture is examined in sections D and E. A detailed analysis of the Gramley-Chase model demonstrates that it implies the following reduced form equations: M = g(B', Y, c) E = h(Be, Y, c) g, > 0 < g 2i hi > 0 > h 2, and \ > g 113 explaining the money stock (M) and bank credit (E) in terms of the extended monetary base (B'), the level of economic activity expressed by na tional income at current prices (Y), and the ceil ing rate on time deposits (c).14 The Gramley-Chase model implies that mone tary policy does affect the money stock and bank credit. It also implies that the money stock responds positively and bank credit negatively to economic activity. The model thus differs from the Monetarist hypotheses which imply that 12The reader may consult my chapter “ Federal Reserve Policy and Monetary Analysis” in Indicators and Targets of Monetary Policy, ed., by Karl Brunner, to be published by Chandler House Publishing Co., San Francisco. This book also contains the original article by Gramley-Chase. Fur ther contributions by Patric H. Hendershott and Robert Weintraub survey critically the issues raised by the Gramley-Chase paper. both bank credit and the money stock respond positively to economic activity. The GramleyChase model also implies that the responses of both the money stock and bank credit to monetary actions are independent o f the general scale o f the public's and the banks’ in terest elasticities. Uniformly large or small in terest elasticities yield the same response in the money stock or bank credit to a change in the monetary base. A detailed discussion of the implications derivable from a meaningfully supplemented Gramley-Chase model is not necessary at this point. W e are foremost interested in the rela tion between this model and the propositions mentioned in the previous paragraph. The first proposition can be interpreted in two different ways. According to one interpretation, it could mean that the marginal multipliers g, and h; (i = 1, 2) are functions o f the banks’ and the public’s response patterns expressing various types o f substitution relationships between dif ferent assets. This interpretation is, however, quite innocuous and yields no differentiation relative to the questioned hypotheses o f the Monetarist position. A second interpretation suggests that the growth rate of the money stock is dominated by the second component (changes in income) of the differential expression: AM = ga AB' + g2 AY This result is not actually implied by the Gramley-Chase model, but it is certainly consis tent with the model. However, in order to derive the desired result, their model must be supplemented with special assumptions about the relative magnitude of gi and g2, and also about the comparative cyclical variability of AB' and AY. This information has not been provided by the authors. Most interesting is another aspect of the model which was not clarified by the authors. Their model implies that policymakers could easily avoid procyclical movements in AM. This model exemplifying the New View thus yields 13ln the Gramley-Chase model, g3 and h3 are indeterminant. 14This implication was demonstrated in my paper listed in footnote 12. The monetary base is adjusted for the ac cumulated sum of reserves liberated from or impounded in required reserves by changes in requirement ratios. SEPTEMBER/OCTOBER 1989 10 little justification for the conjectures of its proponents. analysis o f firm behavior developed on the first level of his investigation. A central property of the Gramley-Chase model must be considered in the light of the programmatic statements characterizing the New View. Gramley-Chase do not differentiate between the public's asset supply to banks and the public’s demand for money. This procedure violates the basic program o f the New View, namely, to apply economic analysis to an array of financial assets and financial institutions. Economic analysis implies that the public’s asset supply and money demand are distinct, and not identical behavior patterns. This difference in behavior patterns is clearly revealed by dif ferent responses o f desired money balances and desired asset supply to specific stimuli in the en vironment. For instance, an increase in the ex pected real yield on real capital raises the public’s asset supply but lowers the public’s money demand. It follows thus that a central analytical feature of the Gramley-Chase model violates the basic and quite relevant program of the New View. This disregard for the construction o f an economic theory relevant for the real world is carried into the second level o f analysis where the author formulates a system o f relations describing the joint determination of interest rates, bank credit, and money stock. A remark able feature of the Karenken model is that it yields no implications whatsoever about the response of the monetary system to actions of the Federal Reserve. It can say nothing, as it stands, about either open market operations or about discount rate and reserve requirement ac tions. This model literally implies, for instance, that the money stock and the banking system's deposit liabilities do not change as a result of any change in reserve requirements ratios. Karenken’s construction shares this fundamen tal analytical flaw with the Gramley-Chase model, but this is not the only problem faced by his analysis. The Karenken analysis proceeds on tw o levels. First, he derives a representative bank's desired balance sheet position. For this purpose he postulates wealth maximization sub ject to the bank’s balance sheet relation be tween assets and liabilities, and subject to reserve requirements on deposits. On closer ex amination, this analysis is only applicable to a monopoly bank with no conversion of deposits into currency or reserve flows to other banks. In order to render the analysis relevant for a representative bank in the world of reality, ad ditional constraints would have to be introduced which modify the results quite substantially. It is also noteworthy that the structural properties assigned by Karenken to the system of market relations are logically inconsistent with the im plications one can derive from the author’s 15Two direct objections made to the Brunner-Meltzer analysis by Kareken should be noted. He finds that the questioned hypotheses do not contain “ a genuine supply function” of deposits. Accepting Karenken’s terminology, this is true, but neither does the Gramley-Chase model contain such a supply function. But the objection has no evidential value anyway. If a hypothesis were judged un satisfactory because some aspects are omitted, all hypotheses are “ unsatisfactory.” Moreover, the cognitive status of a empirical hypothesis does not improve simply http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis None of the conjectures advanced by the countercritique concerning the behavior o f the money stock and the role o f monetary policy find analytical support in Karenken’s analysis. To the extent that anything is implied, it would imply that monetary policy operating directly on bank reserves or a mysterious rate o f return on reserves dominates the volume o f deposits—a practically subversive position for a follower of the New View.15 C. Alternative Explanations o f Cyclical Fluctuations in M onetary Growth The examination thus far in this article has shown that even the most explicit formulation (Gramley-Chase) of the countercritique, allegedly representing the New View with respect to monetary system analysis, does assign a signifi cant role to monetary policy. This examination also argued that the general emphasis given by the New View to the public’s and the banks’ behavior in determination of the money stock and bank credit does not differentiate its pro duct from analytical developments arising from the Monetarist approach. It was also shown that because an ‘‘analytical underpinning” has been provided. Karenken also finds fault with our use of the term ‘‘money supply function.” Whether or not one agrees with this ter minological preferences surely does not affect the relation between observations and statements supplied by the hypothesis. And it should be clear that the status of hypothesis depends only on this relation, and not on names attached to statements. 11 the only explicit formulation advanced by the New Viewers does not provide a sufficient basis for their central conjectures. It is impossible to derive the proposition from the Gramley-Chase model that the behavior of the public and banks, rather than Federal Reserve actions, dominated movements in the money supply. But the declaration of innocence by the countercriti que on behalf of the monetary authorities with respect to cyclical fluctuations of monetary growth still requires further assessment. The various conjectures advanced by GramleyChase, Cacy, and Davis in regard to causes of movements in money and bank credit can be classified into two groups. One set o f conjec tures traces the mechanism generating cyclical fluctuations o f monetary growth to the re sponses of banks and the public; the behavior of monetary authorities is assigned a com paratively minor role. The other group o f con jectures recognizes the predominant role of the behavior of monetary authorities. The detailed arguments advanced to explain the observed cyclical fluctuations of monetary growth differ substantially among the con tributors to the countercritique. Gramley-Chase maintain that changing business conditions modify relative interest rates, and thus induce countercyclical movements in the time deposit ratio. These movements in demand and time deposits generate cyclical fluctuations in monetary growth. On the other hand, Cacy develops an argument used many years ago by Wicksell and Keynes, but attributes it to the New View. He recognizes a pronounced sen sitivity of the money stock to variations in the public’s money demand or asset supply. These variations induce changes in credit market con ditions. Banks, in turn, respond with suitable adjustments in the reserve and borrowing ratios. The money stock and bank credit conse quently change in response to this mechanism. In the following analysis the framework pro vided by the Monetarist view will be used to assess these conflicting conjectures. The em phasis concerning the nature of the causal mechanisms may differ between the various conjectures regarding sources o f variations in money, but the following examination will be applied to an aspect common to all conjectures emphasizing the role of public and bank behavior. Davis actually advances two radically different conjectures about causes of cyclical fluctuations of monetary growth. The first conjecture at tributes fluctuations o f monetary growth to the public’s and banks' responses. Changing busi ness conditions modify the currency ratio, the banks’ borrowing ratio, and the reserve ratio. The resulting changes generate the observed movements in money. His other conjecture at tributes fluctuations in monetary growth to Federal Reserve actions: "the state of business influences decisions by the monetary authorities to supply reserves and to take other actions likely to affect the money supply.”16 16Davis, p. 66. One argument about monetary policy in the same paper requires clarification. Davis asserts on p. 68 that the money supply need not be the objective of policy, and “ given this fact, the behavior of the rate of growth of the money supply during the period cannot be assumed to be simply and directly the result of monetary policy deci sions alone.” This quote asserts that the money supply is "simply and directly the result of policy alone” whenever policy uses the money supply as a target. This is in a In the context o f the Monetarist framework, the money stock (M) is exhibited as a product of a multiplier (m) and the monetary base (B), (such that M = mB). This framework, without the supplementary set of hypotheses and theo ries bearing on the proximate determinants of money summarized by the multiplier and the base, is completely neutral with respect to the rival conjectures; it is compatible with any set of observations. This neutrality assures us that its use does not prejudge the issue under con sideration. The Monetarist framework operates in the manner of a language system, able to ex press the implications of the competing conjec tures in a uniform manner. The first group of conjectures advanced by the countercritique (behavior of the public and banks dominates movements in money) implies that variations in monetary growth between upswings and downswings in business activity are dominated by the variations in the mone tary multiplier. The second group (behavior of monetary authorities dominates movements in sense correct. But the quote could easily be misinter preted due to the ambiguity of the term “ policy.” This term is frequently used to designate a strategy guiding the adjustment of policy variables. Is is also frequently used to refer to the behavior of the policy variables or directly to the variables as such. The quote is quite acceptable in the first sense of “ policy,” but thoroughly unacceptable in the second sense. SEPTEMBER/OCTOBER 1989 12 money) implies that, in periods with unchanged reserve requirement ratios and ceiling rates on time deposits, variations in the monetary base dominate cyclical changes in monetary growth. The movements of the monetary multiplier which are strictly attributable to the changing of requirement ratios can be separated from the total contribution of the multiplier and com bined with the monetary base. With this adjust ment, the second group of conjectures implies that the monetary base, supplemented by the contribution of reserve requirement changes to the multiplier, dominates variations in the money stock. In this examination o f contrasting explanations of monetary fluctuations, values of the money stock (M), the multiplier (m), and the monetary base adjusted for member bank borrowing (B) are measured at the initial and terminal month of each half business cycle (i.e., expansions and contractions) located by the National Bureau of Economic Research. W e form the ratios of these values and write: The subscript 1 refers to values of the terminal month and the subscript 0 to values of the in itial month. These ratios were measured for each half cycle in the period March 1919 to De cember 1966. They were computed for two def initions o f the money stock, inclusive and exclusive of time deposits, with corresponding monetary multipliers. Kendall’s rank correlation coefficients between the money stock ratios (pi) and the multiplier ratios (a), and between (/u) and the monetary base ratio (ft) were computed. W e denote these correlation coefficients with q l/u, a) and q l/u, ft). The implications of the two rival conjectures can now be restated in terms of the two coeffi cients. The first group of conjectures implies that g(fx, a) > g(fu, ft); while the second group implies that in periods of unchanged reserve re quirements ratios and ceiling rates on time deposits, the coefficient q(^, ft) exceeds the coef ficient q(/u, a). The second group implies nothing about the relation of the two coefficients in periods of changing reserve requirements and ceiling rates on time deposits. It follows, therefore, that observations yielding the ine quality g(f.i, ft > q(h, a) disconfirm the first group and confirm the second group. The correlations obtained are quite unam biguous. The value of q (/u, ft) is .537 for the http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis Table 1 A Comparison of Alternative Contribu tions to the Average Annual Growth Rate of the Money Stock and Bank Credit Rank Correlations Contribution made by: Public’s currency and authorities’ behavior Time deposit substitution mechanism Wickseli-Keynes mechanism Money Bank Credit .905 .333 .048 .143 .381 -.333 Remarks: The figures listed state the rank correlation between the average growth rate of the money stock and bank credit with three different contributing sources. whole sample period, whereas q tju, a) is only .084. The half-cycle from 1929 to 1933 was omitted in the computations, because move ments in the money stock and the multiplier were dominated by forces which do not dis criminate between the rival conjectures under consideration. The sample period, including 1929 to 1933, still yields a substantially larger value for g(pi, ft). The same pattern also holds for subperiods. In particular, computations bas ed on observations for 1949 to 1966 confirm the pattern observed for the whole sample period. The results thus support the second group of conjectures but not the first group. These results also suggest, however, that forces operating through the multiplier are not quite negligible. The surprisingly small correlation q()x, a) does not adequately reveal the operation of these forces. Their effective operation is revealed by the correlation q(/u, ft), which is far from perfect, even in subperiods with constant re serve requirement ratios. This circumstance suggests that the behavior o f the public and banks contributes to the cyclical movements of monetary growth. The main result at this stage, however, the clear discrimination between the two groups of conjectures. The results are quite unambiguous on this score. Additional information is supplied by table I. For each postwar cycle beginning with the downswing of 1948-49, the average annual growth rate o f the money stock was computed. The expression M = mb was then used to com pute the contribution to the average growth 13 Table II Regressions of the Money Supply On the Monetary Base and Gross National Product* Regression Coefficients For: Monetary Base Cycle First Differences Gross National Product Log First Differences First Differences Log First Differences IV/48 to II/53 2.03 (9.80) .92 .77 (10.02) .93 .04 (3.12) .62 .11 (3.39) .65 II/53 to III/57 1.75 (1.89) .44 .63 (1.96) .45 .02 (1.02) .26 .07 (1.23) .30 III/57 to II/60 4.59 (11.76) .97 1.66 (11.81) .97 .06 (5.10) .86 .19 (5.34) .67 II/60 to III/65 2.76 (7.56) .87 1.08 (8.54) .89 -.01 (-.3 3 ) -.0 8 -.0 3 (-.2 7 ) -.0 7 *The monetary base was adjusted for reserve requirement changes and shifts in deposits. All data are quarterly averages of seasonally adjusted figures. The first entry in a column for each cycle is the regression coefficient, t-statistics are in parentheses, and partial correlation coefficients are below the t-statistics. rate of money from three distinct sources: (i) the behavior of monetary authorities (i.e., the monetary base and reserve requirement ratios), and the public’s currency behavior, (ii) the time deposits substitution process, and, (iii) the varia tions in the excess reserve and borrowing ratios o f commercial banks (Wicksell-Keynes mecha nism). The rank correlations between each contribu tion, and the average growth rate o f the money stock over all postwar half-cycles clearly sup port the conclusion o f the previous analysis that cyclical movements in the money stock are dominated by Federal Reserve actions. Table I also presents the results of a similar examination bearing on causes of movements in bank credit. The reader should note the radical difference in the observed patterns o f correla tion coefficients. The behavior o f monetary authorities, supplemented by the public’s cur rency behavior, does not appear to dominate the behavior o f bank credit. The three sources contributing to the growth rate of money all ex erted influences of similar order on bank credit. It appears that bank credit is comparatively less exposed to the push of Federal Reserve actions than was the money stock. On the other hand, the money stock is less sensitive than bank credit to the time-deposit substitution mechanism emphasized by Gramley-Chase, and the Wicksell-Keynes mechanism suggested by Cacy. Most astonishing, however, is the negative association between the average growth rate of bank credit and the Wicksell-Keynes mechanism emphasized by Cacy. It should also be noted that the average growth rate of money conforms very clearly to the business cycle. Such conformity does not hold for bank credit over the postwar half cycles. This blurring occurred particularly in periods when the ceiling rate on time deposits was increased. These periods exhibit relatively large contributions to the growth rate o f bank credit emanating from the time deposit substitu tion mechanism. A regression analysis (table II) of the reduced form equations derived from the Gramley-Chase model confirms the central role of the monetary SEPTEMBER/OCTOBER 1989 14 base in the money supply process. Estimates of the regression coefficient relating money to in come are highly unstable among different sam ple periods, relative to the coefficient relating money to the monetary base. Furthermore, estimates of regression coefficients relating money to income occur in some periods with signs which contradict the proposition of Gramley-Chase and Cacy, or exhibit a very small statistical significance. These diverse patterns of coefficients do not occur for the estimates of coefficients relating money and the monetary base. It is also noteworthy that the average growth rate of the monetary base (adjusted for changes in reserve requirement ratios), over the upswings, exceeds without exception the average growth rate of adjacent downswings. This observation is not compatible with the con tention made by Gramley-Chase that policy is countercyclical. Additional information is supplied by table III, which presents some results o f a spectral analy sis bearing on the monetary base and its sources. Spectral analysis is a statistical procedure for decomposing a time series into seasonal, cyclical, and trend movements. After such an analysis was conducted on the monetary base and its sources, a form of correlation analysis was run between movements in the monetary base and movements in its various sources. The results of this procedure (table III) indicate that movements in Federal Reserve credit dominate seasonal and cyclical movements in the mone tary base. In summary, preliminary investigations yield no support for the contention that the behavior of banks and the public dominates cyclical move ments in the money stock. The conjectures ad vanced by Gramley-Chase or Cacy are thus disconfirmed, whereas Davis' second conjecture that fluctuations in monetary growth may be at tributed to Federal Reserve actions seems sub stantially more appropriate. However, further investigations are certainly useful. D. Relevance o f M o n ey and M onetary Actions with Respect to Econom ic Activity At present, a broad consensus accepts the relevance o f money and monetary policy with respect to economic activity. But this consensus concerning the relevance of money emerges from two substantially different views about the nature of the transmission mechanism. One http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis Table III Spectral Correlation Between the Monetary Base, Federal Reserve Credit and Other Sources of the Base _____ Special Correlation Between_____ Period in Months OO 120 60 40 30 24 20 17.14 15 13.33 12 6 4 3 Monetary Base and Federal Reserve Credit Monetary Base and Other Sources of the Base .65 .69 .74 .74 .73 .71 .60 .43 .51 .82 .94 .91 .92 .90 .24 .61 .71 .45 .25 .18 .11 .11 .07 .48 .71 .21 — — Remarks: The monetary base equals Federal Reserve Credit plus other sources of the base. The spectral analysis is bas ed on first differences between adjacent months. The data used were not seasonally adjusted. view is the Keynesian conception (not to be con fused with Keynes' view), enshrined in standard formulations o f the income-expenditure frame work. In the view, the interest rate is the main link between money and economic activity. The other view rejects the traditional separation of economic theory into parts: national income analysis (macro economics) and price theory (micro economics). According to this other view, output and employment are explained by a suit able application o f relative price theory. With regard to discussions o f the impact of money and monetary actions on economic activity, this latter view has been termed the Monetarist position. This position may be divided into the weak Monetarist thesis and the strong Monetarist thesis. In a sense, both the New View and the Monetarist extension o f the "tradi tional view ” are represented in the weak Monetarist position. The following discussions develop the weak and the strong Monetarist thesis. The weak thesis is compared with some aspects of the income-expenditure approach to the determina tion of national economic activity. The strong 15 thesis supplements the weak thesis with special assumptions about our environment, in order to establish the role of monetary forces in the business cycle. 1. The Weak Monetarist Thesis According to the weak Monetarist thesis, monetary impulses are transmitted to the economy by a relative price process which operates on money, financial assets (and liabili ties), real assets, yields on assets and the produc tion of new assets, liabilities and consumables. The general nature o f this process has been described on numerous occasions and may be interpreted as evolving from ideas developed by Knut, Wicksell, Irving Fisher and John Maynard Keynes.17 The operation of relative prices between money, financial assets and real assets may be equivalently interpreted as the working of an interest rate mechanism (prices and yields of assets are inversely related). Monetary impulses are thus transmitted by the play o f interest rates over a vast array of assets. Variations in interest rates change relative prices o f existing assets, relative to both yields and the supply prices of new production. Acceleration or deceleration of monetary impulses are thus con verted by the variation o f relative prices, or in terest rates, into increased or reduced produc tion, and subsequent revisions in the supply prices o f current output. This general conception of the transmission mechanism has important implications which conflict sharply with the Keynesian interpreta tion of monetary mechanisms expressed by standard income-expenditure formulations.18 In the context of standard income-expenditure analysis, fiscal actions are considered to have a "direct effect” on economic activity, whereas monetary actions are considered to have only and "indirect effect.” Furthermore, a constant budget deficit has no effect on interest rates in 17The reader may consult the following studies on this aspect: Milton Friedman and David Meiselman, “ The Relative Stability of Monetary Velocity and the Investment Multiplier in the United States, 1897-1958,” in Stabilization Policies, prepared by the Commission on Money and Credit, Englewood Cliffs, 1963. The paper listed in foot note 21 by James Tobin should also be consulted. Harry Johnson, “ Monetary Theory and Policy,” American Economic Review, June 1962. Karl Brunner “ The Report of the Commission on Money and Credit,” The Journal of Political Economy, December 1961. Karl Brunner, “ Some Major Problems of Monetary Theory,” Proceedings of the American Economic Association, May 1961. Karl Brunner and Allan H. Meltzer, “ The Role of Financial Institutions in a Keynesian framework, in spite of substantial accumulation of outstanding government debt when a budget deficit continually occurs. And lastly, the operation of interest rates on invest ment decisions has usually been rationalized with the aid of considerations based on the ef fects o f borrowing costs. These aspects of the income-expenditure ap proach may be evaluated within the framework of the weak Monetarist thesis. The effects of fiscal actions are also transmitted by the relative price mechanism. Fiscal impulses, i.e., govern ment spending, taxing, and borrowing, operate just as “indirectly” as monetary impulses, and there is no a priori reason for believing that their speed of transmission is substantially greater than that of monetary impulses. The relative price conception of the transmission mechanism also implies that a constant budget deficit exerts a continuous influences on eco nomic activity through persistent modifications in relative prices of financial and real assets. Lastly, the transmission of monetary impulses is not dominated by the relative importance of borrowing costs. In the process, marginal costs of liability extension interact with marginal re turns from acquisitions o f financial and real assets. But interest rates on financial assets not only affect the marginal cost o f liability exten sion, but also influence the substitution between financial and real assets. This substitution modi fies prices of real assets relative to their supply prices and forms a crucial linkage of the mone tary mechanisms; this linkage is usually omitted in standard income-expenditure analysis. The description of monetary mechanisms in Davis’ article approaches quite closely the no tion developed by the weak Monetarist thesis. This approximation permits a useful clarification o f pending issues. However, the criticisms and objections advanced by Davis do not apply to the the Transmission Mechanism,” Proceedings of the American Economic Association, May 1963. Karl Brunner “ The Relative Price Theory of Money, Output, and Employment,” unpublished manuscript based on a paper presented at the Midwestern Economic Association Meetings, April 1967. 18The paper on “ The Effect of Monetary Policy on Expen ditures in Specific Sectors of the Economy,” presented by Dr. Sherman Maisel at the meetings organized by the American Bankers Association in September 1967, ex emplifies very clearly the inherited Keynesian position. The paper will be published in a special issue of the Journal of Political Economy. SEPTEMBER/OCTOBER 1989 16 weak Monetarist position. They are addressed to another thesis, which might be usefully labeled the strong Monetarist thesis. 2. The Strong Monetarist Thesis If the theoretical framework of the weak Monetarist thesis is supplemented with addi tional and special hypotheses, the strong Mone tarist thesis is obtained. An outline of the strong thesis may be formulated in terms of three sets of forces operating simultaneously on the pace of economic activity. For convenience, they may be grouped into monetary forces, fiscal forces and other forces. The latter include technological and organizational innovation, revisions in supply prices induced by accruing information and expectation adjustments, capital accumulation, population changes and other related factors or processes. All three sets o f forces are acknowledged by the strong thesis to affect the pace of economic activity via the relative price process previously outlined. Moreover, the strong Monetarist point of view advances the crucial thesis that the vari ability of monetary forces (properly weighted with respect to their effect on economic activi ty) exceeds the variability of fiscal forces and other forces (properly weighted). It is argued further that major variabilities occurring in a subset of the other forces (e.g., expectations and revisions o f supply prices induced by informa tion arrival) are conditioned by the observed variability o f monetary forces. The conjecture thus involves a comparison of monetary vari ability with the variability of fiscal forces and independent "other forces.” According to the thesis under consideration, the variability of monetary o f impulses is also large relative to the speed at which the economy absorbs the impact of environmental changes. This predomi nance of variability in monetary impulses im plies that pronounced accelerations in monetary forces are followed subsequently by accelera tions in the pace of economic activity, and that pronounced decelerations in monetary forces are followed later by retardations in economic activity. The analysis of the monetary dynamics, using the relative price process, is accepted by both the weak and the strong Monetarist theses. This analysis implies that the regularity of the ob 19U.S. Financial Data, Federal Reserve Bank of St. Louis, week ending February 14, 1968. Also see “ Money Supply http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis served association between accelerations and decelerations of monetary forces and economic activity depends on the relative magnitude of monetary accelerations (or decelerations). The same analysis also reveals the crucial role of changes in the rate of change (second differ ences) o f the money stock in explanations of fluctuations in output and employment. It im plies that any pronounced deceleration, occurr ing at any rate of monetary growth, retards total spending. It is thus impossible to state whether any particular monetary growth, say a 10 percent annual rate, is expansionary with respect to economic activity, until one knows the previous growth rate. The monetary dynam ics of the Monetarist thesis also explains the simultaneous occurrence o f permanent priceinflation and fluctuations in output and employ ment observable in some countries. The nature and the variability o f the “Fried man lag” may also be analyzed within the framework of the Monetarist thesis. This lag measures the interval between a change in sign of the second difference in the money stock and the subsequent turning point located by the Na tional Bureau. In general, the lag at an upper turning point will be shorter, the greater the absorption speed of the economy, and the sharper the deceleration of monetary impulses relative to the movement o f fiscal forces and other forces. Variability in the relative accelera tion or deceleration o f monetary forces neces sarily generates the variability observed in the Friedman lag. What evidence may be cited on behalf of the strong Monetarist thesis? Every major inflation provides support for the thesis, particularly in cases of substantial variations of monetary growth. The attempt at stabilization in the Con federacy during the Civil W ar forms an im pressive piece of evidence in this respect. The association between monetary and economic ac celerations or decelerations has also been observed by the Federal Reserve Bank o f St. Louis.19 Observations from periods with diver gent movements o f monetary and fiscal forces provide further evidence. For instance, such periods occurred immediately after termination o f W orld W ar II, from the end of 1947 to the fall o f 1948, and again in the second half of 1966. In all three cases, monetary forces pre- and Time Deposits, 1914-1964” in the September 1964 issue of this Review. 17 vailed over fiscal forces. The evidence adduced here and on other occasions does not "prove” the strong Monetarist thesis, but does establish its merit for serious consideration. Davis’ examination is therefore welcomed. His objections are summarized by the following points: (a) observations of the persistent associa tion between money and income do not permit an inference of causal direction from money to income; (b) the timing relation between money and economic activity expressed by the Fried man lag yields no evidence in support of the contention that variations in monetary growth cause fluctuations in economic activity; (c) the correlation found in cycles o f moderate ampli tude between magnitudes o f monetary and eco nomic changes was quite unimpressive; (d) the length of the Friedman lag does not measure the interval between emission of monetary im pulse and its ultimate impact on economic ac tivity. Furthermore, the variability o f this lag is due to the simultaneous operation and interac tion of monetary and non-monetary forces. Davis' first comment (a) is of course quite true and well known in the logic of science. It is im possible to derive (logically) causal statements or any general hypotheses from observations. But w e can use such observations to confirm or disconfirm such statements and hypotheses. Davis particularly emphasizes that the persistent association between money and income could be attributed to a causal influence running from economic activity to money. Indeed it could, but our present state of knowledge rejects the notion that the observed association is essentially due to a causal influ ence from income on money. Evidence refuting such a notion was presented in section C. The existence of a mutual interaction over the shorter run between money and economic ac tivity, however, must be fully acknowledged. Yet, this interaction results from the conception guiding policymakers which induces them to ac celerate the monetary base whenever pressures on interest rates mount, and to decelerate the monetary base when these pressures wane. Ad mission of a mutual interaction does not dispose of the strong Monetarist thesis. This interaction, inherent in the weak thesis, is quite consistent with the strong position and has no disconfirming value. To the contrary, it offers an explana tion for the occurrence of the predominant variability of monetary forces. The same logical property applies to Davis’ second argument (b). The timing relation ex pressed by the Friedman lag, in particular the chronological precedence of turning points in monetary growth over turning points in eco nomic activity, can probably be explained by the influence of business conditions on the money supply. Studies in money supply theory strongly suggest this thesis and yield evidence on its behalf. The cyclical pattern of the curren cy ratio, and the strategy typically pursued by monetary policymakers explain this lead of monetary growth. And again, such explanation of the timing relation does not bear negatively on the strong conjecture. The objection noted under Davis’ point (c) is similarly irrelevant. His observations actually confirm the strong thesis. The latter implies that the correlation between amplitudes of monetary and income changes is itself cor related with the magnitude of monetary ac celerations or decelerations. A poor correlation in cycles of moderate amplitude, therefore, yields no discriminating evidence on the validity discriminating evidence on the validity of the strong thesis. Moreover, observations describing occurrences are more appropriate relative to the formulated thesis than correlation measures. For instance, observations tending to disconfirm the strong Monetarist thesis would consist o f oc currences of pronounced monetary accelera tions or decelerations which are not followed by accelerated or retarded movements of economic activity. Point (d) still remains to be considered. Once again, his observation does not bear on the strong Monetarist thesis. Davis properly cau tions readers about the interpretation of the Friedman lag. The variability of this lag is pro bably due to the interaction of monetary and non-monetary forces, or to changes from cycle to cycle in the relative variability of monetary growth. But again, this does not affect the strong thesis. The proper interpretation of the Friedman lag, as the interval between reversals in the rate of monetary impulses and their prevalence over all other factors simultaneously operating on economic activity, usefully clarifies a concept introduced into our discussions. This clarification provides, however, no relevant evidence bearing on the questioned hypotheses. In summary, the arguments developed by Davis do not yield any substantive evidence against the strong Monetarist thesis. Moreover, SEPTEMBER/OCTOBER 1989 18 the discussion omits major portions of the evidence assembled in support o f this position.20 E. Countercyclical Policy and the Interpretation o f M onetary P olicy The usual assertion of the New View, at tributing fluctuations of monetary growth to the public’s and the banks' behavior, assumed a strategic role in the countercritique. The coun tercritique denied, furthermore, that monetary actions have a major impact on economic activi ty. With the crumbling of these two bastions, the monetary policymakers’ interpretation of their own behavior becomes quite vulnerable. In a previous section, the substantial contribu tion of the monetary base to the fluctuations of monetary growth has been demonstrated. These facts, combined with repeated assertions that monetary policy has been largely counter cyclical, suggest the existence of a pronounced discrepancy between actual behavior of the monetary authorities and their interpretation of this behavior. A crucial question bearing on this issue per tains to the proper measure summarizing actual behavior of the monetary authorities. Tw o ma jor facts should be clearly recognized. First, the monetary base consists of "money” directly issued by the authorities, and every issue of base money involves an action of the monetary authorities. This holds irrespective of their knowledge about it, or their motivation and aims. Second, variations in the base, extended by suitable adjustments to incorporate changing reserve requirement ratios, are the single most important factor influencing the behavior of the money stock. And this second point applies ir respective of whether Federal Reserve author ities are aware of it or wish it to be, or whatever their motivations or aims are. Their actual behavior, and not their motivations or 2()Milton Friedman’s summary of the evidence in the FortyFourth Annual Report of the National Bureau of Economic Research is important in the respect. Davis overlooks in particular the evidence accumulated in studies of the money supply mechanism which bears on the issue raised by point (a) in the text. A persistent and uniform associa tion between money and economic activity, in spite of large changes in the structure of money supply processes, yields evidence in support of the Monetarist theses. The reader should also consult Chapter 13 of the book by Milton Friedman and Anna Schwartz listed in footnote 9; Studies in the Quantity Theory of Money, edited by Milton Friedman, University of Chicago Press, 1956; and a doc toral dissertation by Michael W. Keran, “ Monetary Policy and the Business Cycle in Postwar Japan,” Ph.D. thesis at the University of Minnesota, March 1966, to be published http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis aims, influences the monetary system and the pace of economic activity. Thus, actual changes in the monetary base are quite meaningful and appropriate measures of actual behavior of monetary authorities.21 The information presented in table IV sup ports the conjecture that monetary policy makers’ interpretation of their own behavior has no systematic positive association with their actual behavior. Table IV was constructed on the basis of the scores assigned to changes in policies, according to the interpretation o f the Federal Open Market Committee.22 Positive scores w ere associated with each session of the FOMC which decided to make policy easier, more expansionary, less restrictive, less tight, etc., and negative scores indicate decisions to follow a tighter, less expansionary, more restric tive course. The scores varied between plus and minus one, and expressed some broad ordering of the revealed magnitude o f the changes. An examination of the sequences o f scores easily shows that the period covered can be naturally partitioned into subperiods exhibiting an overwhelming occurrence o f scores with a uniform sign. These subperiods are listed in the first column o f table IV. The second column cumulated the scores over the subperiods listed in order to yield a very rough ranking of the policymakers’ posture according to their own interpretation. Table IV reveals that the FOMC interpreted the subperiods from August 1957 to July 1958, and from July 1959 to December I960 as among the most expansionary policy periods. The period from November 1949 to May 1953 appears in this account as a phase o f persistent ly tight or restrictive policy. The next two col umns list the changes of two important vari ables during each subperiod. The third column as a chapter of a book edited by David Meiselman. 2,The reader may also be assured by the following statement: “ . . . monetary policy refers particularly to determination of the supply of (the government’s) demand debt . . .” This demand debt coincides with the monetary base. The quote is by James Tobin, a leading architect of the New View, on p. 148 of his contribution to the Commission on Money and Credit, “ An Essay on Principles of Debt Manage ment,” in Fiscal and Debt Management Policies, Prentice Hall, Englewood Cliffs, 1963. 22The scores were published as Appendix II to “ An Alter native Approach to the Monetary Mechanism.” See foot note 9. 19 Table IV The Association between Policymakers’ Interpretation of Policy, Changes in the Monetary Base and Changes in Free Reserves Periods Cumulative Scores of Policymakers’ Interpretation over the Period Changes in Free Reserves over the Period in $ Million Changes in the Monetary Base over the Period in $ Million 11/49- 5/53 6/53 - 11/54 12/54 - 10/55 11/55- 7/56 8/56- 7/57 8/57 - 7/58 7/58 - 6/59 7/59 - 12/60 1/61 - 12/62 -4 .7 5 + 2.63 -3 .3 7 + 1.12 -1 .0 0 + 3.50 -2 .1 2 + 2.62 -.6 3 -1030 + 286 -8 18 + 352 -4 4 + 1017 -1059 + 1239 -4 2 8 + 5216 + 1321 + 345 + 399 + 657 + 1203 + 531 -5 3 + 3288 describes changes in free reserves, and the fourth column notes changes in the monetary base. A cursory examination of the columns im mediately shows substantial differences in their broad association. The rank correlation between the various columns is most informative for our purposes. These rank correlations are listed in table V. The results expose the absence o f any positive association between the policymakers’ own in terpretation or judgment of their stance and their actual behavior, as indicated by move ments in the monetary base. The correlation coefficient between the monetary base and cumulated scores has a negative value, sug gesting that a systematic divergence between stated and actual policy (as measured by the monetary base) is probable. On the other hand, Table V Rank Correlation Between Changes in the Monetary Base, Changes in Free Reserves and the Cumulated Scores of Policym akers’ Interpretations Cumulated scores and base Cumulated scores and free reserves Free reserves and base -.0 9 + .70 -.2 6 the correlation between the policymakers’ de scriptions of their posture, and the movement of free reserves, is impressively close. This cor relation confirms once again that the Federal Reserve authorities have traditionally used the volume o f free reserves as an indicator to gauge and interpret prevailing monetary policies. Yet little evidence has been developed which estab lishes a causal chain leading from changes in free reserves to the pace of economic activity. Another observation contained in table IV bears on the issue o f policymakers interpreta tion of their own behavior. Changes in the cumulated scores and free reserves between the periods listed always move together and are perfect in terms of direction. By comparison, the co-movement between cumulated scores and changes in the monetary base is quite haphaz ard; only three out of eight changes between periods move together. This degree of co-movement between cumulated scores and the monetary base could have occurred by pure chance with a probability greater than .2, whereas the probability of the perfect co movement between cumulated scores and free reserves occurring as a matter o f pure chance is less than .004. The traditional selection of free reserves or money market conditions as an indicator to interpret prevailing monetary policy and to gauge the relative thrust applied by policy, forms the major reason for the negative association (or at least random association) be tween stated and actual policy. SEPTEMBER/OCTOBER 1989 20 Attempts at rebuttal to the above analysis often emphasize that policymakers are neither interested in the monetary base, nor do they at tach any significance to it. This argument is ad vanced to support the claim that the behavior o f the monetary base is irrelevant for a proper examination o f policymakers’ intended behavior. This argument disregards, however, the facts stated earlier, namely, movements in the mone tary base are under the direct control and are the sole responsibility of the monetary author ities. It also disregards the fact that actions may yield consequences which are independent of motivations shaping the actions. These considerations are sufficient to acknowledge the relevance of the monetary base as a measure summarizing the actual behavior of monetary authorities. However, they alone are not sufficient to determine whether the base is the most reliable indicator o f monetary policy. Other magnitudes such as interest rates, bank credit and free reserves have been advanced with plausible arguments to serve as indicators. A rational procedure must be designed to determine which of the possible entities frequently used for scaling policy yields the most reliable results. This indicator problem is still very poorly understood, mainly because of ambiguous use of economic language in most discussions of monetary policy. The term "indicator” occurs with a variety of meanings in discussions, and so do the terms "target” and "guide.” The in dicator problem, understood in its technical sense, is the determination of an optimal scale justifying interpretations of the authorities’ ac tual behavior by means of comparative state ments. A typical statement is that policy X is more expansionary than policy Y, or that cur rent policy has become more (or less) expan sionary. W henever we use a comparative con cept, we implicitly rely on an ordering scale. The indicator problem has not been given adequate treatment in the literature, and the recognition o f its logical structure is often obstructed by inadequate analysis. It is, for in stance, not sufficient to emphasize the proposi tion that the money supply can be a "misleading guide to the proper interpretation of monetary policy.” This proposition can be easily demon strated for a wide variety of models and hy potheses. However, it establishes very little. The same theories usually demonstrate that the rate of interest, free reserves, or bank credit can FEDERAL RESERVE BANK OF ST. LOUIS also be very misleading guides to monetary policy. Thus, we can obtain a series of proposi tions about a vast array of entities, asserting that each one can be a very misleading guide to the interpretation of policy. W e only reach a useless stalemate in this situation. The usual solution to the indicator problem at the present time is a decision based on mystical insight supplemented by some impressionistic arguments. The most frequently advanced argu ments emphasize that central banks operate directly on credit markets where interest rates are formed, or that the interest mechanism forms the centerpiece of the transmission pro cess. Accordingly, in both cases market interest rates should "obviously” emerge as the relevant indicator of monetary policy. These arguments on behalf of market interest rates are mostly supplied by economists. The monetary authorities’ choice of money market conditions as an indicator evolved from a dif ferent background. But in recent years a subtle change has occurred. One frequently encoun ters arguments which essentially deny either the existence of the indicator problem or its ra tional solution. A favorite line asserts that "the world is very complex” and consequently it is impossible or inadmissible to use a single scale to interpret policy. According to this view, one has to consider and weigh many things in order to obtain a “realistic” assessment in a compli cated world. This position has little merit. The objection to a "single scale” misconstrues the very nature of the problem. Once we decide to discuss mone tary policy in term of comparative statements, an ordinal scale is required in order to provide a logical basis for such statements. A multiplicity of scales effectively eliminates the use of com parative statements. Of course, a single scale may be a function of multiple arguments, but such multiplicity o f arguments should not be confused with a multiplicity o f scales. Policy makers and economists should therefore realize that one either provides a rational procedure which justifies interpretations of monetary policy by means of comparative statements, or that one abandons any pretense of meaningful or intellectually honest discussion of such policy. Solution of the indicator problem in the technical sense appears obstructed on occasion by a prevalent confusion with an entirely dif ferent problem confronting the central banker— 21 the target problem. This problem results from the prevailing uncertainty concerning the na ture of the transmission mechanism and the substantial lags in the dynamics of monetary processes. In the context of perfect information, the in dicator problem becomes trivial and the target problem vanishes. But perfect information is the privilege of economists’ discourse on policy; cen tral bankers cannot afford this luxury. The im pact of their actions are both delayed and uncertain. Moreover, the ultimate goals of monetary policy (targets in the Tinbergen-Theil sense) appear remote to the manager executing general policy directives. Policymakers will be inclined under these circumstances to insert a more immediate target between their ultimate goals and their actions. These targets should be reliably observable with a minimal lag. It is quite understandable that central bankers traditionally use various measures of money market conditions, with somewhat shifting weights, as a target guiding the continuous ad justment of their policy variables. This response to the uncertainties and lags in the dynamics of the monetary mechanism is very rational in deed. However, once we recognize the rationali ty of such behavior, we should also consider the rationality of using a particular target. The choice o f a target still remains a problem, and the very nature of this problem is inadequately understood at this state. This is not the place to examine the indicator and target problem in detail. A possible solution to both problems has been developed on an other occasion.23 The solutions apply decision theoretic procedures and concepts from control theory to the determination of an optimal choice of both indicator and target. Both pro blems are in principle solvable, in spite of the "complexity o f the world.” Consequently, there is little excuse for failing to develop rational monetary policy procedures. CO NCLUSION A program for applying economic analysis to financial markets and financial institutions is certainly acceptable and worth pursuing. This program suggests that the public and banks in 23The reader may consult the chapter by Karl Brunner and Allan H. Meltzer on “ Targets and Indicators of Monetary Policy,” in the book of the same title, edited by Karl Brun teract in the determination o f bank credit, in terest rates, and the money stock, in response to the behavior of monetary authorities. But the recognition of such interaction implies nothing with respect to the relative importance of the causal forces generating cyclical fluctuations of monetary growth. Neither does it bear on the quality of alternative empirical hypotheses, or the relative usefulness of various magnitudes or conditions which might be proposed as an in dicator to judge the actual thrust applied by monetary policy to the pace of economic activity. The Monetarist thesis has been put forth in the form of well structured hypotheses which are supported by empirical evidence. This ex tensive research in the area o f monetary policy has established that: (i) Federal Reserve actions dominate the movement of the monetary base over time; (ii) movements o f the monetary base dominate movements o f the money supply over the business cycle; and, (iii) accelerations or decelerations of the money supply are closely followed by accelerations or decelerations in economic activity. Therefore, the Monetarist thesis puts forth the proposition that actions of the Federal Reserve are transmitted to economic activity via the resulting movements in the monetary base and money supply, which initiate the adjustments in relative prices of assets, liabilities, and the production of new assets. The New View, as put forth by the counter critique, has offered thus far neither analysis nor evidence pertaining relevantly to an ex planation of variations in monetary growth. Moreover, the countercritique has not devel oped, on acceptable logical grounds, a system atic justification for the abundant supply of statements characterizing policy in terms of its effects on the economy. Nor has it developed a systematic justification for the choice o f money market conditions as an optimal target guiding the execution of open market oper ations. But rational policy procedures require both a reliable interpretation and an adequate deter mination of the course of policy. The necessary conditions for rational policy are certainly not satisfied if policies actually retarding economic activity are viewed to be expansionary, as in the ner. The book will be published by Chandler House Publishing Co., Belmont, California. SEPTEMBER/OCTOBER 1989 22 case of the 1960-61 recession, or, if inflationary actions are viewed as being restrictive, as in the first half of 1966. The major questions addressed to our mone tary policymakers, their advisors and consul tants remain: How do you justify your inter FEDERAL RESERVE BANK OF ST. LOUIS pretation of policy, and how do you actually ex plain the fluctuations of monetary growth? The major contentions o f the academic critics of the past performance o f monetary authorities could possibly be quite false, but this should be dem onstrated by appropriate analysis and relevant evidence. 23 Michelle R. Gurfinkel Michelle Ft. Garfinkel is an economist at the Federal Reserve Bank of St. Louis. Thomas A. Pollmann provided research assistance The Causes and Consequences of Leveraged Buyouts I n THE MARKET for corporate control during the past decade, leveraged buyouts have become increasingly popular. Many observers, speculat ing about the causes o f this recent trend, have expressed concern about the potential problems arising from such activity.1 Implicit in many casual discussions is the assumption that leveraged buyouts—hereafter LBOs—are merely some type of cosmetic surgery. That is, an LBO has no impact on the productive capacity of the target firm, while unjustifiably inflating the value of the stock. Under this assumption, any observed gains to the existing shareholders of the target firm are likely to be matched, if not dominated, by losses to others; since there is no net gain and possibly a loss to society, LBO activity should be re stricted. Some analysts also argue that LBOs have contributed to the unprecedented growth o f outstanding debt in recent years. If, as many contend, the large growth in debt is associated with increased instability in the financial sys tem, public policy might aim to reverse or at least curb debt growth. In addition, tax reform might be an appropriate way to reduce this 1For example, see Lowenstein (1986), “ When Industry Bor rows Itself” (1988), Friedman (1989) and Kaufman (1989). 2When the target of an LBO is a division of a public com pany, the transaction is typically called a “ management buyout.” Stancill (1988), p. 18, who points out that LBO activity targeting smaller (private) firms is not a new debt growth, if it stems chiefly from tax incen tives. This article examines whether LBOs have had a productive impact on the target firm. If eco nomic theory and evidence suggest that LBOs generally are productive, then arguments for legal restrictions on LBO activity are less per suasive. Alternatively, if there are few, if any, gains from LBOs, the idea that LBOs pose a problem for the economy might be valid. W H A T ARE LBOs? Despite the ever-expanding literature on LBOs, there does not appear to be a single, clear defi nition of what an LBO really is. Loosely speak ing, an LBO is simply the purchase of a firm by an outside individual, another firm or the in cumbent management with the purchase being financed by large amounts o f debt; the resulting firm is said to be "highly leveraged.” The target firm can be a free-standing entity or a division o f a public corporation.2 Although the target of the LBO can be a private firm, recent discus sions about LBO activity have focused primarily phenomenon, provides a very general definition of an LBO: “ whenever a buyer lacks the requisite cash and bor rows part of the purchase price against the target com pany's assets (receivables, equipment, inventory, real estate) or cash flow (future cash), that’s an LBO.” SEPTEMBER/OCTOBER 1989 24 on instances in which a public firm is taken private.3 Upon this type o f transaction, the target firm’s stock shares are no longer traded publicly in equity markets. The greatest ambiguity about what constitutes an LBO concerns the degree to which the pur chase is financed with debt.4 Typically, debt finance provides about 80 percent to 90 percent o f the funds for the purchase. Equity finance, in which the resulting shares are held by the purchasers o f the target firm and, often, an out side investment group, provides the remaining funds.5 D ebt Finance in an LBO Tw o types of debt are usually employed in an LBO transaction: senior debt and subordinated debt. Senior debt typically accounts for the greatest proportion, usually 50 percent to 60 percent, o f financing for the LBO. Sometimes called secured debt, senior debt specifies a lien on a particular piece o f property. In the case that the firm ’s earnings are insufficient to ser vice the firm ’s debt obligations fully, the holders of senior debt can have the pledged property sold to recover the unpaid interest and prin cipal. Funds through senior debt are often pro vided by commercial banks, insurance com panies, leasing companies and limited partner ships specializing in LBOs and venture capital investments.6 Subordinated debt, or "mezzanine” debt, is considered to be more speculative than senior debt because it is issued without a lien against specified property. Although the holders of sub ordinated debt are protected in the case of de fault, only assets not pledged explicitly and any cash remaining after paying other creditors are available to satisfy these unsecured claims. Ac counting for about 30 percent o f the financing for the transaction, subordinated debt is usually provided by pension funds, insurance com panies and limited partnerships.7 3See Lehn and Poulsen (1988, 1989) and “ Corporate America Snuggles Up to the Buy-Out Wolves” (1988), for example. DeAngelo, DeAngelo and Rice (1984), p. 370, use a narrower definition by making a distinction between pure going-private transactions, where “ incumbent management seeks complete equity ownership of the sur viving corporation,” and leveraged buyouts, where “ management proposes to share equity ownership in the subsequent private firm with third-party investors.” 4The Federal Reserve Board recently established a set of guidelines for banks involved in a broader class of leverag ed financing, called “ highly leveraged financing.” This class of leveraging includes all borrowers having debt-to http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis In many LBOs, after the purchase, the new owners sell some of the firm’s assets and use the proceeds to retire some of the debt. Cash flows from continued operations are used to service the remaining debt obligations. K ey Features o f Recent LBOs The typical LBO in recent years has two in teresting characteristics that distinguish it from other takeover and merger activities. First, the equity of the target firm usually is held by few er individuals following the financial reorganiza tion. This increased concentration of ownership is especially typical of a "going-private” transac tion in which the stock is no longer publicly traded.8 Second, although alternative sources o f funds are available to obtain corporate ownership, going-private transactions usually are financed heavily with debt, leaving the target firm in a highly leveraged position. In essence, the tran saction involves a substitution of debt for equi ty. For example, in a sample o f 58 LBOs be tween 1980 to 1984, the average debt-to-equity ratio rose from 0.457 to 5.524, a percentage change exceeding 1100 percent.9 The higher degree of leveraging means that a larger proportion of claims against the target firm’s assets and operations are fixed obliga tions. Because holders of these claims can push the firm into bankruptcy if these obligations are not met fully, the greater leveraging, holding all else constant, erodes the target firm’s insulation from unexpected declines in earnings and, hence, increases the firm ’s risk o f bankruptcy. RECENT TRENDS IN LHO A C T IV IT Y The following discussion defines an LBO as a highly leveraged, going-private transaction. This total-asset ratios exceeding 75 percent. See “ Board Issues Guidelines for LBO, Other Highly Leveraged Loans ...” (1989). Although LBOs are included in this class, they have not been specifically defined by the Federal Reserve System. 5Thomson (1989) and Lehn and Poulsen (1988, 1989). 6lbid. 7lbid. 8Many of these firms, however, subsequently go public. 9Lehn and Poulsen (1988), table 2, p. 48. 25 Table 1 Table 2 LBO Activity, 1979-88: Going Private Transactions Premium Paid Over Market Price in LBOs Year Number of transactions Median purchase price 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 16 13 17 31 36 57 76 76 47 125 $ 7.9 25.3 41.1 29.6 77.8 66.9 72.6 84.5 123.3 79.8 Total dollar value paid $ 636.0 967.4 2,338.5 2,836.7 7,145.4 10,805.9 24,139.8 20,232.4 22,057.1 60,920.6 SOURCE: Merrill Lynch Business Brokerage and Valuation, Inc. Mergerstat Review (1988), p. 92. NOTE: These numbers do not include management buyouts. See Merrill Lynch Business Brokerage and Valuation, Inc. (1988), p. 82. narrow focus permits the discussion to address recent concerns about LBO activity that appear to revolve around those transactions in which public firms are taken private primarily through debt financing. Year Average Median 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 106.4% 49.2 31.3 41.4 36.7 36.3 30.9 31.9 34.8 33.8 65.6% 36.2 26.7 38.6 31.3 33.7 25.7 26.1 30.9 26.3 SOURCE: Merrill Lynch Business Brokerage and Valuation, Inc., Mergerstat Review (1988), p. 92. NOTE: Premiums over the market price were calculated on the basis of the market value of the firm's closing stock price five days before the initial announcement. These data are for going-private transactions only. firms included in the New York Stock Exchange. Even accounting for inflation, the increase in the average purchase price was substantial— from $50.6 million to $400.5 million in 1982 prices, a real annual growth rate of 25.8 percent. As shown in table 1, the number of goingprivate transactions in 1988 was nearly eight times that in 1979.10 Just in the past year, the incidence of these transactions has more than doubled. Furthermore, the table indicates that the average as well as the median purchase price rose dramatically over the same period. In 1979, the average purchase price was $39.8 million, whereas in 1988 it was $487.4 million. The average purchase price rose at an annual rate o f 32.1 percent, nearly three times the 11.1 percent annual rate of increase in the value of While the average purchase price generally rose during the 1980s, the “premium” or the price paid for these firms above their initial market value (the value of their stock shares before the initial announcement) as a percent age of the market value has been relatively stable. As table 2 shows, average and median premiums paid over the prior market price of the target firms from 1979 to 1988 have been quite large. Even excluding the extremely large 1979 values, the average and median premiums averaged about 36.3 percent and 30.6 percent, respectively.11 These large premiums indicate ' “Merrill Lynch Business Brokerage and Valuation, Inc. (1988) reports, “ Like the majority of unit management buyouts, most, if not all, of the ‘going private’ transactions also are leveraged buyouts, i.e., transactions in which the buyers put up only a small part of the purchase price and borrow the rest.” (p. 91) Management buyouts have also increased less markedly, from 59 in 1979 to 89 in 1988. See Merrill Lynch Business Brokerage and Valuation, Inc. (1988), p. 82. 11Lehn and Poulsen (1988), table 5, p. 52, report the premiums, as determined in the market, for the target firms of LBOs included in the COMPUSTAT data tape be tween 1980 and 1984. The “ market-valued” premium was measured as the percentage increase in the stock price from 20 days before the LBO announcement until the day of the announcement for LBOs between 1980 and 1984. They find that market-valued premium as a percentage of the market price before the announcement averaged 39.5 percent, ranging from 1.7 percent to 120 percent. During the same period, the “ cash-offer” premium (the cash offer above the market price 20 days before the announcement) as a fraction of the market price ranged from 2 percent to 120 percent, averaging 41 percent. The sample standard deviation of both these premiums was 23.2 percent. SEPTEMBER/OCTOBER 1989 26 that the target firm’s stockholders have cap tured significant capital gains upon the LBO transaction.12 AR E L B O s PR ODUCTIVE?— SOME FINANCE TH E O R Y A N D EVIDENCE The growing incidence of LBO activity in the market for corporate control has sparked many to question the social value of this activity. Many expressed concerns are predicated im plicitly on the notion that the changes in the firm ’s financial structure associated with the LBO transaction have no positive real effects on that firm ’s output. If the transaction were mere ly a device to realize some short-term gain, at the expense of long-term growth and a reduc tion in social wealth, then these concerns would be justified. Finance theory, however, suggests that LBOs can be productive. The gains derive from two key features o f LBOs in recent years—namely, going private and highly leveraged financing. These related features permit a reorganization o f the firm to alter its incentive structure and produce an increase in its earnings potential. The Advantages o f Going Private The theory of corporate finance shows how the distinction between ownership and control, or equivalently the differences between the in centives and constraints of the firm’s stockholders and those of the firm ’s managers, can have important implications for the perfor mance of the firm. Specifically, this distinction can create a situation in which the firm does not achieve its maximum earnings potential— 12Also, see DeAngelo, DeAngelo and Rice (1984), Torabzadeh and Bertin (1987) and Lehn and Poulsen (1988, 1989), who find that announcements of LBOs have signifi cant positive effects on the target firm’s stock price. For example, Lehn and Poulsen (1989, p. 776) calculate the average daily return from holding the stock of the target firm of the LBO for various holding periods, abstracting from movements in the firm’s stock price due to economywide factors. They find that the “ cumulative average daily abnormal return” (CAR) from 20 days before to 20 days after the LBO announcement averaged 20.54 percent across the firms included in the sample during the period 1980-87. This means that an individual buying a stock of an LBO target 20 days before the announcement and then selling it 20 days after the announcement could have made a 20.5 percent return on average above a normal (the market) return over the same period. Even holding the stock from one day before the announcement until the end of the announcement day yielded, on average, a CAR of 16.3 percent, a return too high to be attributed solely to http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis that is, the firm is not being run efficiently from the stockholder’s perspective.13 By going private, the distinction is removed and earnings can increase.14 If the manager’s actions were monitored easily and costlessly, going-private transactions would have no implications for the performance o f the firm. A contract for compensating the manager could be designed by the owners to encourage the manager to act entirely on their behalf. The ideal contract would specify the appropriate ac tions to be taken by the manager to maximize the firm ’s value under all possible contingencies; the contract would penalize the manager if he failed to act in accordance with its specifica tions, thereby ensuring that the manager always acted in the interests of the owners. The efficacy o f such contracts, however, hinges on the ability and costs o f monitoring. Typically, the firm ’s owners do not observe the actions of the managers directly, nor are they fully aware o f the economic environment (specif ic to the firm) in which a manager’s decisions are made. For example, owners do not have complete information about the firm’s oppor tunities for investment and growth or about the daily events that influence a manager’s deci sions. Holding all else constant, as the number o f the holders of the firm’s stock increases,— that is, as the firm’s ownership becomes more dispersed—the potential gains realized by one owner monitoring the manager’s actions decline, because the potential net gains that the individual can capture become smaller relative to the costs he incurs. In this case, monitoring activity de clines and contracts designed to align the man- chance. Similarly, for the period 1973-80, DeAngelo, DeAngelo and Rice (1984), pp. 394-95, estimate a signifi cant CAR of 16.99 percent for the same holding period. 13For example, see Manne (1965) and Jensen and Meckling (1976). 14Another gain from going private, which is more obvious, involves circumventing the explicit costs that are otherwise incurred with outside ownership, such as registration and listing fees and other stockholder service costs. Relative to the market value of the public firm, these explicit costs can be significant. For example, in the early 1980s, estimates of the costs of public ownership incurred annual ly ranged from $30,000 to $200,000. The value of the stream of this annual cost (for an indefinite time) dis counted at a rate of 10 percent, ranges from $300,000 to $2,000,000, whereas the median value of a sample of 72 firms attempting to go private between 1973 and 1980 was $2,838,000. See DeAngelo, DeAngelo and Rice (1984) and references cited therein. 27 ager’s incentives with those of the owners can not be enforced completely. To see why the distinction between owner ship and managerial control can be important when monitoring incentives are weaker, con sider the following extreme example in which a firm has such a large number o f owners that no individual finds it worthwhile to monitor the manager at all. As is typical in any publicly owned firm, the owners have voting rights, but do not participate directly in the daily opera tions and decision-making of the firm. Suppose that the firm ’s manager, who exercises full con trol over these operations, has the opportunity to undertake a new project whereby the present value of cash flows (that is, revenues net of operating costs) can increase by $100. If the manager had a fixed salary and no owner ship claims in the firm, he would be completely indifferent between exploiting this opportunity and not doing so, as long as the expansion re quired no additional time by the manager. If the expansion actually required any additional time, however, he might well choose to forgo the op portunity; after all, what’s in it for him? In this example, the distinction between ownership and control is meaningful because the manager does not fully bear the wealth con sequences o f his actions. In the absence of ef fective monitoring by the owners, the decisions of the manager, acting on his own behalf, are not likely to maximize the owners’ wealth; in stead, they will maximize the manager’s utility. As the distinction between ownership and control becomes less clear, the conflict of in terests between owners and managers becomes less severe. In the example above, if the manager owned a fraction of the firms’ stock, say 5 percent, he would be less reluctant to in itiate the new project; the additional cash flow created by the new project would increase the total value o f his stock and wealth by $5. Never theless, the manager would not act entirely on 15Note that a manager who dislikes risk would not willingly enter into a wage contract specifying that his compensa tion be a function only of the market value of the firm's stock. Doing so would involve taking on a large amount of risk—i.e., possible, large fluctuations in income that are not entirely under his control. Provided that there is com petition in the market for managers, owners of the firm must bear some of the risks and offer a compensation schedule such that risks are shared by owners and managers. Bennett (1989), however, reports that ex ecutives increasingly are taking on some of the risks, in the sense that the link between their salaries and the market value of the firm, through long-term incentive behalf of all the owners unless the marginal gain from doing so, $5 in this example, exceeded the marginal value of his time used in other ways, including leisure. The problems that potentially arise from the distinction between ownership and control, called "agency problems,” explain why we observe managerial contracts that are more complicated than those that simply specify a fixed income. The problem o f "incomplete monitoring” ex plains why the observed managerial contracts are less complicated than those that could perfectly remove the conflict of interests be tween owners and managers. A contract that partially links the manager’s income to the firm’s characteristics observed easily by stockholders—for example, sales, profits or the firm ’s stock performance—could help alleviate the conflict.15 A change in the organizational structure of the firm, such as that engendered by an LBO, however, is another and potentially more effective method to circumvent the firm ’s organizational inefficiencies attributable to the meaningful separation o f control and ownership. In a going-private transaction, the interests of owners and the manager generally are closely, if not fully, reconciled. Once the manager be comes the owner, there is no conflict; the wealth consequences of the manager’s actions are entirely internalized by the firm’s reorganization. Even when a third party (an other company or an individual) finances the purchase, monitoring possibilities improve, sim ply because the transaction decreases the number o f owners—or, equivalently, concen trates the ownership o f the firm—thereby rais ing the level of monitoring and the possibility that enforceable contracts can be designed to resolve the conflict o f interests more effectively. By improving the organizational efficiency of the firm through a change of ownership, the LBO can increase the firm ’s earnings.16 schemes (such as stock options and restricted stock), has become substantial over the past decade. In the absence of complete monitoring, the problems that typically arise from the distinction between ownership and control are be ing partly mitigated by tying executive compensation to the performance of the firm. 16An inefficient organization of a firm provides a motivation for others to take over that firm. Note that such a takeover need not involve taking that firm private. Rather, the takeover is necessary to reorganize the firm to effect a higher concentration of ownership. SEPTEMBER/OCTOBER 1989 28 The Advantages o f Highly Leveraged Financing That most going-private transactions are financed with a large proportion o f debt sug gests that leveraging itself must augment the potential gains from the buyout. That is, the high degree of leveraging in the buyout need not indicate that the buyers do not have the re quisite cash for the transaction. One widely mentioned source o f gain from ex tensive leveraging is based on the incentive structure o f the tax system. Because interest payments on debt are tax deductible, debt financing is relatively more attractive (ceteris paribus) than other methods of finance. The double taxation of dividends, first as corporate income and then as shareholder income, further increases the incentive to issue or sell debt to finance the purchase o f the firm. The gain from leveraged financing, however, need not be restricted to reducing the tax liabil ity of the target firm. Another motive for the use of debt finance stems from the misalign ment of the manager’s incentives with those of the owners in cases where the firm faces low growth prospects and a large “free cash flow .”17 When the firm's cash flow exceeds what is necessary to finance its own projects that are expected to yield positive (discounted) net revenues, the firm is said to have a positive free cash flow. That is, the firm has reached its op timal size; additional projects to expand its operations would not maximize its profits. There are cases, however, in which the manager of a firm that has reached its optimal size might choose not to maximize the share holders’ wealth by paying out the free cash flow in the form of dividends. For example, if the manager’s compensation w ere linked to the firm ’s growth in sales, he would have a greater incentive to invest the free cash in any project that increases the firm's sales, even if the pro 17Jensen (1986, 1988). Also see “ Management Brief: The Way the Money Goes” (1989) for a brief discussion of this hypothesis as well as others to explain the increasing degree of leveraging by corporations in recent years and Laderman (1989a) for a discussion of the concept of free cash flow and its relation to cash flow and operating cash flow. 18Of course, free cash flow could also explain the growing acquisition activity that has generated losses to stockholders. See Jensen (1986, 1988) for details. 19See the evidence cited in footnotes 11 and 12. FEDERAL RESERVE BANK OF ST. LOUIS ject’s net return would be insufficient to main tain the firm’s value. The incentive to use the free cash inefficiently (from the stockholders' and society’s perspective) to increase the firm ’s size is greater if the manager values his power as measured by the amount of resources under his control.18 In this case, the market value of the stock and the wealth o f existing shareholders will not be maximized. The problem of free cash flow, a particular type of agency problem, can be mitigated in a buyout that is financed with debt. Issuing debt and using the entire proceeds to purchase equi ty in an LBO enables the stockholders to cap ture the present value of the future free cash flow that otherwise would be used inefficiently. The firm’s increased leveraged position after the transaction, in effect, imposes a binding commit ment on the manager to not waste future cash flow; specifically, the manager cannot repudiate the firm’s debt obligation to pay out the future free cash flow as interest payments because the bondholders could then push the firm into bankruptcy. By circumventing or reducing the agency problem associated with free cash flow, the use of debt essentially improves the produc tive efficiency o f the firm. Evidence The empirical observation that the purchase price in an LBO is, on average, considerably higher than the market price before the LBO announcement suggests that these transactions have increased the value o f the target firm and, hence, the wealth of the shareholders.19 The observed gain to shareholders is consistent with the notion that market participants at least ex pect the changes brought about by the LBO ac tivity to be productive.20 The basic idea here is that by increasing the efficiency with which the firm's resources are used, the LBO transaction is expected to in- 20The issue of whether merger and acquisition activity in general is productive has also received attention by researchers in finance as well as the news media. See Jarrell, Brickley and Netter (1988) and Jensen and Ruback (1983) for recent reviews of the empirical studies on the effects of merger and takeover activity. These studies generally indicate that stockholders gain, on average, from this activity in the market for corporate control. Also, see Ott and Santoni (1985) who present a useful theoretical discussion of the productiveness of mergers and acquisi tions and place this activity into an historical perspective. 29 crease economic earnings, which would even tually be paid out as dividends. Because the price of a firm ’s stock is equal, in theory, to the expected present discounted value o f future dividends, the transaction also raises the price of the stock. In equilibrium, the gains to stock holders or the premium paid over the market price before the transaction should be identical to the expected increase in the present dis counted value of economic earnings to the target firm.21 firm's tax liability do not add statistically signifi cant information for predicting the marketvalued premium above the information provided by the cash flow measure.23 Hence, the ex pected gains from the LBO transactions appear to be over-and-above the tax advantages of debt finance. In an attempt to identify the sources o f the increase in value from LBOs, one recent study found that the increase in the market price of the target firm’s stock is largely explained by its cash flow as a fraction of the market value of its equity before the transaction.22 This evidence suggests that, with greater cash flow and the greater agency costs potentially associated with that flow, there is more room to improve the firm’s productive efficiency and, accordingly, to increase the firm’s value. Indeed, although dif ferences in the firm's tax liabilities are associ ated with significant differences in the observed magnitudes of the premiums, measures of the Despite the gains typically realized by a target firm ’s shareholders, some observers have ex pressed doubt about the benefits o f LBOs. These doubts stem from two types o f potential “bad” effects of LBOs: wealth redistributions and increased instability o f the economy. 21For example, in the simple case where expected future dividends, d, for t> 0 , grow at a constant rate, g, the price di of the firm’s stock can be written a s ____r is the conr-9 stant discount rate appropriately adjusted for risk, and d, is next period's dividend payment. Hence, by increasing expected dividends (d, or g)—or, equivalently, expected economic earnings—the transaction can increase the market value of the firm’s stock. Assuming that market participants correctly value the firm’s stock, the observed increases in the stock price cast some doubt on the general criticism of activity in the market for corporate control, that managers are exploiting opportunities for short-term gains at the expense of long term performance. Rather, this activity effectively removes myopic incentives so as to increase long-term economic earnings. Of course, the claim that observed unusual in creases in the stock price supports the hypothesis that mergers and acquisitions are productive presumes that capital markets are efficient. In particular, firms are not systematically undervalued (given public information) and daily changes in the price of the firm’s stock reflect new information that is made available to the public and is rele vant for determining the firm’s value. Otherwise, the observed increase in the stock price could merely reflect a re-evaluation of the firm’s productiveness, without any fun damental change expected to arise from this activity in the market for corporate control. 22Lehn and Poulsen (1988), table 6, p. 54. The measure of cash flow used in their empirical analysis, however, does not control for the firm’s growth prospects and so only crudely captures the firm’s “ free cash flow.” But in a subsequent analysis, Lehn and Poulsen (1989), using un distributed cash flow (that is, the firm’s after-tax cash flow net of interest and dividend payments) and attempting to control for the firm’s growth prospects, get similar results for LBOs between 1984 and 1987 (table V, p. 782). Also, SKEPTICISM A B O U T THE SOCIAL V A LU E OF LB O s LBOs and Wealth Redistributions One version of the redistribution criticism is the claim that LBOs generate gains for the stockholders at the expense of those holding the target firm’s original bonds; the redistribution presumably results from a reduction in the Lehn and Poulsen (1989), table III, p. 778, find that firms going private have a significantly higher flow of un distributed cash flow as a fraction of their equity value and possibly lower growth prospects than a control group of firms. Recently, Mitchell and Lehn (1988), who attempt to iden tify the source of gains to shareholders in takeover activi ty, present some preliminary evidence to support the hypothesis that the growth in productive takeover activity is partly an attempt to prevent the target firm from using free cash flow in an unprofitable way or to reverse the earlier unprofitable takeover activity due to the free cash flow problems. 23Lehn and Poulsen (1988, 1989). Lehn and Poulsen (1988), table 9, p. 60, divide their sample into two equal sub samples according to the magnitude of the firm’s tax liability as a fraction of the market value of the firm’s outstanding equity before the transaction. They find that the mean market-valued premium for those firms with the higher tax liability measure was 47.7 percent, whereas that for firms with the lower measure of tax liability was 32.1 percent. The difference in the premiums for the two sub samples cannot be due to chance alone. (See footnote 11 for their definition of the market-valued premium.) However, the firm’s tax liability does not explain variation in the premium not already explained by variation in the firm’s undistributed cash flow. See Lehn and Poulsen (1989), table V, p. 782. Also, they do not find a significant difference between the mean tax liability for firms that went private and that for a control group of firms (table III, p. 778). SEPTEMBER/OCTOBER 1989 30 market value of the firm’s outstanding debt.24 The value o f debt allegedly falls because the target firm's increased leveraged position, typically in the form of low-quality, highyielding (junk) bonds, increases the probability that its future revenues will be insufficient to cover its higher interest payments. That is, the value o f the firm’s bonds outstanding before the announcement of the LBO drops because market participants believe that the probability of default has increased as a result of the LBO transaction.25 Even if LBOs w ere to redistribute wealth in this way, however, whether or not public policy should aim to discourage LBO activity is not ob vious.26 Economics has nothing meaningful to say about the “fairness” of wealth redistribu tions that leave social wealth unchanged. The key economic issue is whether LBOs reduce the market value of the firm’s outstanding debt by more or less than the increase in the value of its outstanding stock. If the net change in the value o f stockholders’ and bondholders’ claims on the firm is negative, then LBOs reduce social wealth. In this case, LBOs would be socially in efficient and public policy to limit such activity could be justified. The evidence discussed above, however, casts some doubt on the validity of the claim that LBOs merely redistribute wealth among those having claims in the firm with no net gain to society. Specifically, the alleged positive effect of the increase in leveraging on the firm's default probability should not emerge. If such an effect were to emerge, it would first be reflected in the price of the stock. Because the new owners of the firm will be the residual claimants of the firm’s earnings, they take on the greatest amount of risk in the transaction. The bidders must ex pect that, while future debt-servicing increases, the LBO will improve the firm’s productivity so 24For example, see “ A Big Event for American Bonds” (1988) and, “ When Industry Borrows Itself” (1988). 25The value of preferred stock is also said to fall. Specified payments or dividends, distributed to holders of these stock shares unless earnings are insufficient to cover in terest payments on outstanding debt, are fixed like interest payments on debt. 26The forms of protection, offered in financial markets, against such losses weakens the role for public interven tion. See, for example, “ The Debt Deduction” (1988) and Lehn and Poulsen (1988). 27Lehn and Poulsen (1988), table 8, p. 57. Also, see Marais, Schipper and Smith (1989) who similarly find that bond values did not significantly decline following 290 proposed management buyouts between 1974 and 1985. Further http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis as to augment the future cash flow available for servicing that increased debt obligation; other wise, they would not be willing to pay such a premium to purchase the firm. Confirming this line o f reasoning, empirical studies indicate that LBO announcements have an insignificant effect on the market value of the firm’s outstanding debt. One study found that, for a sample o f 13 target firms between 1980 and 1984, the average percentage change in the bond price from 10 days before to 10 days after the announcement was -1.42 percent, much smaller than the average 7.21 percent decline in the Wall Street Journal’s 20-bond in dex over the same period.27 Another study of 20 LBOs between 1984 to 1988 found that the like lihood of the bond price falling was virtually equal to the likelihood of the price increasing upon the LBO announcement.28 However, a re cent study found that, for 33 successful buyouts between 1974 and 1985, the default risk o f the target firms’ bonds (as measured by Moody’s) typically increased.29 Another version o f the redistribution hypothesis is based on the widely cited reason for the recent growth of LBOs—that is, the tax system produces a bias for debt finance. By reducing the firm ’s tax liability, the LBO in creases the firm’s after-tax earnings and, conse quently, the market value of the firm’s stock. According to some observers, the observed in crease in stock value takes place at the expense o f taxpayers. Because these transactions permit the target firms to reduce their tax liability, tax gains to the target firms realized by the share holders are said to be offset indirectly by in creasing the tax liabilities of all taxpayers.30 Begardless o f the issues related to the fairness o f the tax system, the critical economic issue for public policy toward LBOs is whether the more, preferred stock values do not appear to be significantly affected by the announcement. 28Fortier (1989). Out of a sample of 20 LBOs, the bond prices of only eight target firms fell. The average change in price as a percentage of the bond’s face value, abstrac ting from general market interest rate movements was only -0.50 percent, too small to be attributed to the LBO an nouncement. However, she finds that after January 1987, when the elimination of preferential tax treatment of capital gains made debt finance even more attractive, bond holders, on average, experienced significant losses (5.1 percent). 29Marais, Schipper and Smith (1989), tables 8 and 9, pp. 184-85. 30For example, see Lowenstein (1986). 31 would increase later; increased future tax revenues would offset partially, if not fully, the loss in tax revenues now due to the use of debt finance in the LBO transaction. Table 3 Growth of GNP and Debt___________ Nominal GNP Total credit market debt owed by domestic nonfinancial sectors U.S. government State and local governments Households Corporate Farm corporate Nonfarm, noncorporate 1960-69 1970-79 1980-88 6.89% 10.24% 7.57% 6.80 1.95 10.34 8.67 10.85 13.76 7.56 8.54 8.47 8.95 7.27 11.31 9.09 12.56 8.69 9.96 10.41 -0.31 13.51 16.07 13.13 SOURCE: Federal Reserve Board, “ Flow of Funds.” NOTE: All data are annual percentage changes. net effect on social wealth is negative. But, for example, even if LBOs had no effect on the firm’s performance, the net effect of LBO activi ty on tax revenues is unlikely to be negative. While the tax liability of the target firm falls with increased leveraging, that of the share holders realizing capital gains and new bond holders increases. Moreover, the evidence that the tax benefits do not fully explain the observed gains to shareholders suggests that the gains to shareholders do not simply come at the expense of taxpayers.31 Thus, the argument that the gains to shareholders are offset by losses to tax payers ignores the future increased tax base resulting from the LBO’s predicted effect on the firm’s productivity. If LBOs enhance the firm’s performance, then income subject to taxation 31See evidence cited in footnote 23. 32See Friedman (1989) and Kaufman (1989), for example. Gilbert and Ott (1985) found that the increase in corporate merger activity financed with debt (including LBOs) ac counted for a substantial amount of the unusually large growth of business loans in the first half of 1984. 33During the 1980s, corporate debt growth has exceeded nominal GNP growth in all but two years and by as much as 9.96 percentage points. See also Bernanke and Camp bell (1988), who provide a detailed analysis of the recent trends in corporate debt. They look at disaggregated data in an attempt to determine the financial stability or solven M a croecon om ic Instability Some individuals have argued that the recent activity in the market for corporate control has contributed to an excessive growth o f debt by nonfinancial borrowers in this decade.32 As table 3 shows, the growth of nominal GNP ex ceeded that o f total debt o f nonfinancial bor rowers slightly during the 1960s and was mar ginally smaller in the 1970s. In the 1980s, however, the growth of total outstanding debt for nonfinancial borrowers exceeded that of nominal GNP by more than 3 percentage points. Table 3 indicates that all borrowers contributed to this recent trend except for the farm, and nonfarm, noncorporate sectors. But the primary contributors appear to be the U.S. government and the corporate sector.33 Some observers have suggested that the growth rates of corporate and public debt, which appear high relative to GNP growth in the 1980s, especially by post-World W ar II stan dards, reflect a greater instability in financial markets and, hence, the economy. According to this view, for any given slowdown in economic activity, the higher degree of leveraging by firms implies a greater likelihood that these firms will be forced to default on their debt obligations; if the affected creditors who suffer from deficient cash flows, in turn, are unable to service their own debt, then the severity o f a slowdown in economic activity will be aggravated as the inci dence of default is transmitted throughout the financial system.34 Despite the fact that the recent growth in cor porate debt and LBO activity appear to be strik ing, whether or not these new trends indicate a threat to the stability o f the financial system or cy of those firms most likely to default on their debt obligations. 34See, for example, “ Taking the Strain of America’s Leverage” (1988) and Ferguson (1989), Kaufman (1986, 1989), Friedman (1986, 1989) and Greenspan (1989, especially p. 269). Friedman (1989) also argues that “ because of the increased likelihood of debtors’ distress in the event of an economic downturn, the Federal Reserve system is likely to be less willing either to seek or to per mit a business recession in the United States.” According to Friedman, a consequence of the higher degree of leveraging is the prospect for greater inflation. SEPTEMBER/OCTOBER 1989 32 the economy is not obvious. If LBOs or, more generally, merger and acquisition activity had no other benefit than providing a channel through which tax advantages of debt finance could be realized, then the growth of debt that only recently has significantly exceeded the growth of nominal output might seem alarming. The existing empirical evidence briefly discussed above, however, suggests that LBOs provide anticipated gains over and above the tax gains to the target firm. Since these an ticipated benefits include enhancing the earn ings potential of the firm, simply comparing debt growth with nominal GNP growth does not provide a complete picture from which to iden tify the effects of debt growth on the stability of financial markets. Specifically, the increased debt as a fraction o f nominal output could re flect an increase in expected future cash flows relative to the prior post-World W ar II trends. In this case, the increased debt would be associated with a rise in the market value of firms’ assets. Indeed, aggregate debt-to-asset ra tios, which more accurately indicate financial stability, hardly changed on net from 1969 to 1986. For example, one measure of this ratio us ing "flow of funds” data, rose from 34 percent in 1969 to 42 percent in 1986, peaking in 1974 at 51 percent.35 Aggregate debt-to-asset ratios, however, can be misleading, because they mask the financial condition of those firms with especially high debt-to-asset ratios. In fact, such firms have ex hibited only a slightly higher increase in debt-toasset ratios than would be suggested by the ag gregate data. Specifically, a recent study found 36Bernanke and Campbell (1988), table 3, p. 98. 36lbid., table 5, p. 104. As predicted by the “ free cash flow” theory, the study found a dramatic increase in real and nominal interest expenses as a percentage of cash flows over this same period (see tables 6 and 7, pp. 106-07). Because expectations about increased future cash flows (as reflected in the increased market value of the firms’ outstanding assets that has left debt-to-asset ratios virtual ly unchanged on net from 1969 to 1986) might not be fulfilled, however, concerns about recent trends in debt growth are not entirely unwarranted. Another recent study found that the default rate on junk bonds, commonly used to finance transactions in the market for corporate control, could be as high as 34 percent, much higher than the average 2.5 percent reported by an earlier study. See Laderman (1989b) for a brief discussion of these two studies and Mitchell (1989) and Fidler and Cohen (1989) for discussions of a more recent study by Moody’s In vestors Services, Inc. Also see Passell (1989) who sum marizes two other studies’ findings that the greater risk of default has been compensated by higher realized returns on average. http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis that while, for a full sample o f firms, the debtto-asset ratio fell from 31 percent in 1969 to 27 percent in 1986, for those firms in the 99th per centile (that is, having a higher debt-to-asset ratio than 99 percent of the sample), debt-toasset ratios rose from about 74 percent to 82 percent.36 SOME U N A N SW ER ED QUESTIONS A N D P O LIC Y IM PLIC AT IO N S The existing evidence cannot rule out the validity o f all critical concerns about LBOs. In particular, most research on LBOs has examined the impact of the transaction on pre-buyout stockholders and bondholders of the firm. As such, these studies provide evidence on finan cial market participants’ expectations about the impact of LBOs on the target firms perfor mance. Although these studies generally indicate that these transactions on average are expected to generate gains beyond tax liability reductions, w e will have to wait to see if these gains are ac tually realized. Several recent studies on post buyout performance of LBO firms provide evi dence suggesting that those transactions, on average, have actually improved the firm's per formance; however, evidence is preliminary and particularly subject to many methodological prob lems due to data limitations.37 Nevertheless, with out evidence that LBOs are harmful or are like ly to be harmful to the economy, policy actions to restrict LBO activity seem to be premature; indeed, such restrictions could themselves be harmful, especially if LBO activity actually enhances the productiveness of the target firms. 37For example, see Deveny (1989), who discusses a recent study indicating that companies involved in the market for corporate control have not, on average, exhibited a decrease in expenditures on research and development, as predicted by some critics. Also, see Yago (1989), who reports one study’s finding that target firms of manage ment buyouts are less likely to close plants than are other firms. Francis (1989) discusses evidence from another study indicating that, upon a change in ownership of a firm, the ratio of the administrative employees to plant employees fell 11 percent on average. Indeed, one study found that for LBO firms between 1984 and 1986, average annual growth of the firm’s productivity (measured by sales per employee) increased from an average of 3.6 per cent before the transaction to 17.4 percent after the tran saction. See Yago (1989). Also, Palmeri (1989) recently found that the stocks of 70 LBO target firms that subse quently went public performed significantly better than the market since going public. But see Long and Ravenscraft (1989) for a brief summary of a few other existing studies providing mixed evidence on post-LBO performance and a critical assessment of the validity of these studies. 33 “A Big Event for American Bonds,” The Economist (October 29, 1988), p. 81. Bennett, Amanda. "A Great Leap Forward for Executive Pay,” Wall Street Journal, April 24, 1989. Bernanke, Ben S., and John Y. Campbell. “ Is There a Cor porate Debt Crisis?” Brookings Papers on Economic Activity (1:1988), pp. 83-125. “ Board Issues Guidelines for LBO, Other Highly Leveraged Loans ...” The Fed Letter, Federal Reserve Bank of Kansas City (April 1989), p. 1. “ Corporate America Snuggles Up to the Buy-Out Wolves,” The Economist (October 29, 1988), pp. 69-72. DeAngelo, Harry, Linda DeAngelo, and Edward M. Rice. “ Going Private: Minority Freezeouts and Stockholder Wealth,” Journal of Law and Economics (October 1984), pp. 367-401. “The Debt Deduction,” New York Journal of Commerce, November 29, 1988. Deveny, Kathleen. “ Progress Isn’t Drowning in Debt—Vet,” Business Week: Innovation in America (Special 1989 Bonus Issue), p. 110. Dowd, Ann Reilly. "Washington’s War Against LBO Debt,” Fortune (February 13, 1989), pp. 91-92. Ferguson, Douglas E. “ Solving the Leverage Problem,” New York Journal of Commerce, January 9, 1989. Fidler, Stephen and Norma Cohen. “Widening the Junk Default Debate,” Financial Times, July 20, 1989. Fortier, Diana L. “ Buyouts and Bondholders,” Chicago Fed Letter (January 1989). Francis, David R. “Takeovers Cut Central-Office Costs,” The NBER Digest, June 1989. Friedman, Benjamin M. “ Increasing Indebtedness and Finan cial Stability in the United States” in Federal Reserve Bank of Kansas City, Debt, Financial Stability, and Public Policy (August 1986), pp. 27-53. _______ . “Tread Carefully on Takeovers,” New Ybrk Jour nal of Commerce, April 27, 1989. Gilbert, R. Alton, and Mack Ott. “ Why the Big Rise in Business Loans at Banks Last Year?” this Review (March 1985), pp. 5-13. Greenspan, Alan. “ Statement Before the Committee on Ways and Means, United States House of Represen tatives,” Federal Reserve Bulletin (April 1989), pp. 267-72. Jarrell, Gregg A., James A. Brickley, and Jeffry M. Netter. “ The Market for Corporate Control: The Empirical Evidence Since 1980,” Journal of Economic Perspectives (Winter 1988), pp. 49-68. Jensen, Michael C. “Takeovers: Their Causes and Conse quences,” Journal of Economic Perspectives (Winter 1988), pp. 21-48. _______ . “Agency Costs of Free Cash Flow, Corporate Finance, and Takeovers,” American Economic Review (May 1986), pp. 323-29. Jensen, Michael C., and William H. Meckling. “ Theory of the Firm: Managerial Behavior, Agency Costs and Ownership Structure,” Journal of Financial Economics (October 1976), pp. 305-60. Jensen, Michael C., and Richard S. Ruback. “The Market for Corporate Control: The Scientific Evidence,” Journal of Financial Economics (April 1983), pp. 5-50. Kaufman, Henry. “ Halting the Leverage Binge,” Institutional Investor (April 1989), p. 23. _______ . “ Debt: The Threat to Economic and Financial Stability,” in Federal Reserve Bank of Kansas City, Debt, Financial Stability, and Public Policy (August 1986), pp. 15-26. Laderman, Jeffrey M. ’’Earnings, Schmernings—Look at the Cash,” Business Week (July 24, 1989a) pp. 56-57. _______ . “ Does Junk Have Lasting Value? Probably,” Business Week (May 1, 1989b), pp. 118-19. Lehn, Kenneth, and Annette Poulsen. "Free Cash Flow and Stockholder Gains in Going Private Transactions,” Journal of Finance (July 1989), pp. 771-87. 38For example, see “ The Debt Deduction” (1988) and Fried man (1986, 1989) and Dowd (1989). Also see U.S. Con- gress, Joint Committee on Taxation (1989) for a more detailed and exhaustive list of policy proposals. Although the recent behavior o f various debtto-asset ratios does not indicate a drastic deterio ration of corporate solvency, the higher debt-toincome ratios do suggest some increased risk of financial stress. That is, the recent behavior of these latter ratios indicate a higher degree of pressure on cash flows exerted by interest ex penses (a reduction in liquidity), which could ex acerbate the severity of any given slowdown in economic activity. To the extent the tax advan tages o f debt finance are not necessary to realize the gains from LBO activity, as well as from other highly leveraged transactions in the market for corporate control, a change in public policy might be warranted. A widely discussed policy recommendation in tended to slow the growth o f all corporate debt involves eliminating the tax advantages of debt finance, in particular, by eliminating the tax de ductibility o f interest payments on debt.38 Another policy recommendation would involve removing the double-taxation of dividends by relieving the tax burden on dividends at the corporate level or stockholder level. Whether the latter approach to curb debt growth is polit ically feasible, given the wide concern about the unprecedented growth in public debt along with explicit commitments made by the administra tion to reduce the budget deficit, remains un clear. In any case, if, as suggested by the em pirical evidence, LBO activity has benefits in addition to the tax advantages, these tax reforms should be considered on their own merits, not chiefly as a way to reduce LBO activity. REFERENCES SEPTEMBER/OCTOBER 1989 34 _______ . “ Leveraged Buyouts: Wealth Created or Wealth Redistributed?” in Murray L. Weidenbaum and Kenneth W. Chilton, eds., Public Policy Toward Corporate Takeovers (Transaction Inc., 1988), pp. 46-62. Long, William F., and David J. Ravenscraft. “The Record of LBO Performance,” mimeo (May 17, 1989). Lowenstein, Louis. “ No More Cozy Management Buyouts,” Harvard Business Review (January/February 1986), pp. 147-56. “ Management Brief: The Way the Money Goes,” The Economist (July 15, 1989), pp. 70-71. Manne, Henry G. “ Mergers and the Market for Corporate Control,” Journal of Political Economy (April 1965), pp. 110-20. Marais, Laurentius, Katherine Schipper and Abbie Smith. “ Wealth Effects of Going Private For Senior Securities,” Journal of Financial Economics (June 1989), pp. 155-91. Merrill Lynch Business Brokerage and Valuation, Inc., Mergerstat Review (1988). Mitchell, Constance. “ JunkMssuer Rate of Default is Put at Average 3.3%,” Wall Street Journal, July 20, 1989. Mitchell, Mark L., and Kenneth Lehn. “ Do Bad Bidders Become Good Targets?” mimeo (August 1988). http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis Ott, Mack, and G.J. Santoni. “ Mergers and Takeovers—The Value of Predators’ Information,” this Review (December 1985), pp. 16-28. Palmeri, Christopher. “ Born-Again Stocks,” Forbes (March 20, 1989), pp. 210-11. Passell, Peter. "Economic Scene: The $12 Billion Misunder standing,” New York Times, July 17, 1989. Stancill, James McNeill. “ LBOs for Smaller Companies,” Harvard Business Review (January/February 1988), pp. 18-26. “ Taking the Strain of America’s Leverage,” The Economist (November 5, 1988), pp. 87-88. Thomson, James B. “ Bank Lending to LBOs: Risks and Supervisory Response,” Economic Commentary, Federal Reserve Bank of Cleveland (February 15, 1989). Torabzadeh, Khalil M., and William J. Bertin. “ Leveraged Buyouts and Shareholder Returns,” Journal of Financial Research (Winter 1987), pp. 313-19. U.S. Congress. Joint Committee on Taxation. Federal Income Tax Aspects of Corporate Financial Structures. (GPO, 1989). “When Industry Borrows Itself,” The Economist (October 29, 1988), pp. 17-18. Yago, Glenn. “ LBOs, UFOs and Corporate Perestroika,” Wall Street Journal (July 19, 1989). 35 Michelle R. Garfinkel and Daniel L. Thornton Michelle R. Garfinkel is an economist and Daniel L. Thornton is an assistant vice president at the Federal Reserve Bank of St. Louis. David H. Kelly and Thomas A. Pollmann provided research assistance. The Link Between M l and the Monetary Base in the 1980s G L^7lNCE 1980, there have been several changes in the Federal Reserve System’s reserve require ments that have altered the relationship be tween the money stock, M l, and the monetary base. The Monetary Control Act o f 1980 (hence forth, MCA) brought all depository institutions— member and nonmember commercial banks, saving and loan associations, mutual savings banks and credit unions—under a uniform set of reserve requirements and removed reserve requirements on a broad category o f savings time deposits that are close substitutes for checkable deposits. In February 1984, the Fed eral Reserve switched from lagged reserve ac counting to contemporaneous reserve accounting.1 This article shows how these changes affected the relationship between the money stock and the monetary base, arguing that, under fairly general conditions, the relationship should have become less variable since 1980. Evidence con sistent with this argument is then presented. CHANGES IN THE M ONEY S U P P L Y PROCESS SINCE 1980 A simple model of the money supply process provides a useful framework to illustrate how the link between M l and the monetary base has changed in the 1980s. This model is summar ized by the following equation: (1) M l = m-MB, where M l denotes the stock o f money consis ting o f checkable deposits and currency held by the non-bank public; MB denotes the stock of the monetary base consisting of total reserves and currency; and m represents the money multiplier. The money multiplier, which translates fluc tuations in the monetary base into fluctuations in M l, depends on the reserve requirements that the Federal Reserve imposes on depository institutions and a number o f ratios that reflect portfolio decisions of both depository institu- 1The Fed moved from contemporaneous to lagged reserve accounting in 1968. SEPTEMBER/OCTOBER 1989 36 tions and the public (see the appendix for de tails and verification o f the claims made in the text). For a given set of portfolio preferences and reserve requirements, equation 1 shows what level of M l will result from any given level of the monetary base. The ratios that reflect portfolio preferences of depository institutions and the public generally are not constant. As a result, even if reserve re quirements were unchanged, variation in these ratios would produce variability in the money multiplier. The MCA was intended to strengthen the link between Federal Reserve actions and changes in the money stock by reducing or eliminating specific sources of variability in the multiplier. Uniform R eserve Requirem ents f o r M e m b e r and State-Chartered N on m em ber Ranks The MCA imposed uniform reserve require ments on all depository institutions. Before 1980, reserve requirements on deposits of statechartered nonmember banks were established by the state in which they were domiciled. These requirements were generally lower than those imposed by the Federal Reserve. More im portantly, while only vault cash held by these institutions was part of the monetary base, checkable deposits held by these institutions were included in M l.2 Without uniform reserve requirements on checkable deposits, the multiplier would change as deposits shifted between member and non member banks. For example, as checkable de posits flowed from member to nonmember banks, reserves would be released so that a larger money stock could be supported by the same level o f the monetary base. That is, the multiplier would increase. The opposite would 2For a discussion of state reserve requirements, see Gilbert and Lovati (1978) and Gilbert (1978). 3See Hafer (1980) for a detailed discussion of the redefini tion of the monetary aggregates. While the redefinition did not change the aggregate level of the monetary base, vault cash holdings of thrifts were moved from currency to total reserves. There were two other important definitional changes in the aggregates in the 1980s. Starting in February 1980, demand deposits of foreign commercial banks and official institutions were excluded from M1. In July 1981, non bank traveler’s checks were included in M1. In both cases, the M1 series was revised historically. The latter revision introduces an additional source of variability in the M1-base relationship because non-bank traveler’s checks http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis occur when deposits flowed from nonmember to member banks. With uniform reserve re quirements, such shifts in checkable deposits are no longer a source o f variation in the multiplier. The same reasoning applies to shifts of time and savings deposits between nonmember and member banks. Before the MCA, as these deposits flowed from member to nonmember banks, reserves were released that could sup port a larger volume of checkable deposits. Other things the same, the multiplier would fluctuate as time and savings deposits shifted between member and nonmember banks. Again, uniform reserve requirements established by the MCA removed this source o f variation. Extending R eserve Requirem ents to Thrifts The MCA also extended the same set of re serve requirements to deposits at thrift institu tions, thereby removing another source of varia tion from the multiplier. Before 1980, these institutions were not subject to the Fed's reserve requirements and checkable deposits held at these institutions, called NOW accounts, were not included in M l. In February 1980, however, M l statistics were revised to include interestbearing checking accounts held at thrifts and the historical data w ere revised to reflect this change.3 Consequently, shifts of checkable deposits between thrifts and banks influenced the money multiplier prior to the adoption of the MCA. Now, deposit shifts between thrifts and banks can no longer be a source of varia tion in the money multiplier. Before the nationwide introduction o f interestbearing checking accounts on January 1, 1981, however, thrifts did not hold a large amount of NOW accounts.4 Accordingly, deposit shifts be- are not reservable. Further, since foreign deposits are still subject to reserve requirements, they absorb the base even though they are no longer included in M1. 4For example, as of December 31, 1979, the non-bank depository institutions held only $4.2 billion in NOW accounts. 37 tween thrifts and banks may not have been an important source of variation in the multiplier prior to 1981.® Elimination o f R eserve R e quirements on Savings and Time Deposits Before the MCA, the Federal Reserve imposed reserve requirements o f 3 percent on commer cial bank savings and time deposits.6 The MCA eliminated reserve requirements on a broad class o f savings and time deposits (hereafter, S&Ts).7 As a consequence, shifts between formerly reservable deposits and non-reservable time deposits no longer affect the multiplier. Of course, shifts between currently reservable time deposits and checkable deposits or nonreservable time deposits remain a source of variability in the multiplier. Hence, the elimina tion of reserve requirements on a broad class of S&Ts does not guarantee that the variability of the multiplier will be reduced. The Gradual Implementation o f the M CA Reserve requirement changes under the MCA were phased in over several years. The ad justments for most nonmember banks and 5Conversations with Board staff suggest that thrifts may have held vault cash in excess of what would have been required on NOW accounts. If, in effect, thrifts were holding vault cash in the form of reserves against these accounts as if they were member banks, a shift from de mand deposits in a member bank to a NOW account at a thrift would have no effect on the multiplier using current data. Prior to the revision of the monetary aggregates, however, such a shift would have caused the money supp ly to decline with no corresponding change in the monetary base. Consequently, it would have affected the multiplier. Furthermore, prior to January 1, 1981, member banks in Connecticut, Maine, Massachusetts, New Hamp shire, New Jersey, New York, Rhode Island and Vermont also issued interest-bearing NOW accounts that were not included in the money stock at that time. Member banks, however, were required to hold reserves against these deposits. Nevertheless, a shift from a member bank NOW account to a thrift NOW account would have left both the money stock and the monetary base unchanged if thrifts were holding vault cash as reserves against these deposits. It should be noted, however, that in both cases above, the measured ratio of currency to either M1 or checkable deposits would have changed before the redefinition of money and the adoption of MCA. Nonetheless, the varia tion in the currency ratio would have been reflected in the measured multiplier only in the first case. 6The actual system of reserve requirements was slightly more complicated, as there were different marginal reserve requirements on time deposits by total size of outstanding deposits and by term to maturity. thrifts occurred gradually over an eight-year period. Beginning November 1980, these institu tions had to maintain only one-eighth o f the re quired reserves they would eventually hold when the act was fully implemented. Each suc cessive September until 1987 (when the phase-in was completed), these institutions had to hold an additional one-eighth of the target level of required reserves. Member bank reserve requirements generally w ere reduced by the MCA.8 Starting November 1980, a seven-step phase-down began, with onefourth of the new reserve requirements being implemented in November 1980 and one-sixth of the remainder being implemented in six steps. The full phase-down was completed on March 1, 1984. For member banks whose reserve re quirements w ere raised, the phase-in was im plemented in four steps, with one-fourth of the increase being required in November 1980 and one-fourth being met in each of the next three Septembers. Although member banks had completely ad justed to the new reserve requirements by March 1, 1984, the full impact o f the MCA on the multiplier could not have emerged until the phase-in was completed for all depository insti tutions—unless the effect of extending reserve Specifically, the MCA imposed reserve requirements on time deposits except some of those that have an original maturity shorter than 31/2 years; shorter-maturity time deposits that are transferable, or that are non-transferable and owned by anybody excluding an individual person or a sole proprietorship, are still subject to a 3 percent reserve requirement. 8Prior to the MCA, a system of marginal reserve re quirements on transaction deposits varied with the deposit size of the institution. For example, just before the im plementation of the MCA, the marginal reserve require ment on demand deposits more than $400 million was 1 6 1/4 percent, while that on deposits less than $2 million was 7 percent. Hence, the money supply could change relative to the base as transaction deposits shifted bet ween institutions of different size. By reducing the number of tiers in the marginal system from five to two,—that is, by partially removing the marginal reserve requirement system—the MCA reduced the importance of this source of variability in the multiplier. Moreover, the new system generally lowered reserve requirements to be maintained against transaction accounts. Starting in November 1980, the marginal reserve requirement was only 3 percent for accounts less than $25 million and was 12 percent for ac counts in excess of $25 million. With the exception of member banks holding balances of checkable deposits between $25 million and $100 million, member banks were subject to lower marginal reserve requirements on checkable deposits. SEPTEMBER/OCTOBER 1989 38 requirements to thrifts is quantitatively unim portant. Because of the nature o f the phase-in, the variability of the multiplier might not have dropped sharply at any particular time during the 1980s. Instead, MCA’s impact on the vari ability of the multiplier could have occurred gradually throughout the transition period. The Impact o f the M CA on the Level o f the Multiplier In addition to reducing the variability of the multiplier, the MCA's changes in reserve re quirements had divergent effects on the level of the multiplier.9 While higher reserve require ments for nonmember banks, thrifts and some member banks reduced the multiplier, the elimination of reserve requirements on a broad class of S&Ts for member banks and lower reserve requirements for most member banks increased it. The net effect of the MCA on the size of the multiplier depends on the relative magnitude of these effects.10 9For a given level of the monetary base, an expected in crease (decrease) in the money multiplier would imply an increase (decrease) in the money supply. If the Fed removes reserves from (injects reserves into) the system, however, the money supply need not be affected by the expected increase in the multiplier. Typically, changes in the money supply produced by changes in the multiplier as a result of reserve-requirement changes are largely off set through open market operations. See Burger (1979). 10See footnote 8. As of September 30, 1978, large banks held more than 48 percent of the total demand deposits outstanding. The net effect of the MCA on reserve re quirements for all depository institutions on the phase-in dates is shown in table 2 which lists all reserverequirement changes from 1973 to 1988. "Strictly speaking, reserve requirements under CRA are not completely contemporaneous. There is a two-day lag on reserve requirements on transaction accounts and a 14-day lag on liabilities other than transaction deposits. See Gilbert and Trebing (1982) for details. 12One of the main concerns about the effect of LRA on monetary control was that LRA encouraged the Fed to validate deposit creation of depository institutions. Specifically, some observers argued that under LRA depository institutions were free to create any desired amount of the checkable deposits. The Fed would be forc ed to supply the necessary reserves two weeks later; otherwise, there would be a sharp increase in the federal funds rate. At one level, this argument reflects a view that the Fed might be more concerned with movements in the federal funds rate than with its money supply objective. At another level, however, it was frequently suggested,—e.g., Laufenberg (1976)—that LRA severed the contemporaneous link between the money stock and the monetary base. Thornton (1983), however, has shown that the link need not be affected by the accounting procedure for reserve requirements; a contemporaneous link between the money stock and the monetary base could be maintained either through depository institutions’ holdings of excess reserves http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis The Change to Contem poraneous R eserve Requirem ents In February 1984, the Fed changed its reserve accounting procedures from lagged reserve ac counting (LRA) to contemporaneous reserve ac counting (CRA). Under LRA, depository institu tions were required to hold reserves during the current reserve-maintenance period based on the average of reservable deposits (both check able and time deposits) held during the prior two weeks. Under CRA, required reserves for the current reserve-maintenance period are based more heavily on the amount of reservable deposits in that period.11 The major reason for adopting CRA was to in crease the Federal Reserve’s control over the money stock.12 It could have increased or had no effect on the variability of the multiplier.13 If it affected the multiplier’s variability, the level of the multiplier also would have declined; or through the currency-deposit ratio under LRA. von Hagen (1987) arrives at a similar conclusion, but empha sizes the role of interest rate expectations under LRA. Thornton (1984) provides some early evidence on the effect of the move to CRA on the variability of money and in terest rates. Also, see Thornton (1982) for an analysis of money stock control under LRA and CRA. 13While the contribution of the variance of the currencydeposit ratio and the ratio of excess reserves to checkable deposits to the variability of the multiplier both decline with the adoption of CRA, the contribution of the variance of the other ratios could get larger. The net effect on the variance of the multiplier depends on the relative magni tude of these effects. Given the importance of the cur rency-deposit ratio, in particular, the variance of the multiplier should decline with the adoption of CRA. This conjecture depends on modeling depository institutions’ holdings of excess reserves as a proportion of their check able and time deposits. If this specification is inappropri ate, the only link between M1 and the base would be through the currency-deposit ratio. In this instance, the variance of the multiplier would increase with the move to CRA. Also, if depository institutions hold excess reserves as a buffer stock under CRA, there might be no change in the variability of the multiplier. See Thornton (1983) for details. Although it is not immediately obvious why depository in stitutions would behave that way, Tarhan and Spindt (1983) provide some evidence that banks maintain excess reserves as a buffer stock. 39 otherwise, it would have no effect on the level of the multiplier. EM PIR ICAL EVIDENCE The multiplier is measured as the ratio o f M l to the monetary base. The effects o f reserve re quirement changes are reflected in the adjusted monetary base, so that they are not reflected in its multiplier. The effect o f such changes are not reflected in the source base, so they are re flected in the multiplier obtained using it. Be cause the above analysis abstracts from reserve requirement changes, normally it would be pre ferable to use the adjusted monetary base to construct the multiplier. The adjusted monetary base can also yield misleading results, however, because the ratio o f reservable time deposits to total checkable deposits appears in the adjusted monetary base and not its multiplier after November 1980.14 Removing this component from the multiplier only after November 1980 biases the results toward finding a reduction in the multiplier’s variance.15 Although the multiplier derived from the source base does not suffer from this limitation, it reflects reserve-requirement changes. The ef fect of such changes on the variability of the multiplier before and after the MCA depends on the frequency and magnitude of reserve-requirement changes during the two periods. If reserve-requirement changes were more fre quent or larger before the MCA, failure to abstract from such changes produces results that are biased in favor of seeing a reduction in variability after the MCA. If they are less fre quent or smaller, the bias would be in the op posite direction. The analysis presented here is carried out for multipliers based on the adjusted monetary base (mA) and the source base (ms) to see if the results are affected by these factors. The data are monthly and cover the period from January 1973 through December 1988. The Level o f the Multiplier As noted previously, the net effect o f the MCA on the level o f the multiplier is analytically indeterminate. On one hand, extending reserve requirements to nonmember banks and non bank depository institutions and increasing reserve requirements for some member banks cause the multiplier to fall. On the other hand, the elimination o f reserve requirements on a broad class of S&Ts and the reduction in re serve requirements for most member banks cause the multiplier to rise. The effect of the move to CRA is somewhat less indeterminate analytically. If it had any affect at all, the multiplier would decline. Figure 1 shows the levels of the two multi pliers over the period. The vertical lines corres pond to the initiation of the MCA and the adop tion o f CRA. Both multipliers generally decline from January 1973 through early 1980. Follow ing a sharp decline in early 1980 and a sharp rise in mid-1980, the multipliers generally rose until mid-1986 and declined thereafter. Al though both multipliers declined slightly during 1984, the beginning of the decline, especially for the adjusted monetary base multiplier, predates the adoption of CRA by several months. Relating the behavior of the level o f the multipliers to the adoption o f the MCA and CRA by direct in spection is complicated by the fact that the mul tipliers are influenced greatly by the "k-ratio,” the ratio of currency to checkable deposits, which changed markedly during this period.16 The Effect o f the M CA and the Adoption o f CRA on the N onCurrency Ratio Com ponents o f the Multiplier One can abstract from movements in the kratio by obtaining a joint representation for the other components. Each multiplier can be w rit ten in the general form, 14See Gilbert (1987), especially the appendix, for a discus sion of the revised adjusted monetary base. 15The strength of this conclusion is based on an implicit assumption that the covariance between this and other multiplier components is zero. If the covariance is nonzero, the direction of the bias could be the opposite of that stated in the text. 16For a discussion of the importance of the k-ratio and its behavior during the 1980s, see Burger (1988). SEPTEMBER/OCTOBER 1989 40 Figure 1 Adjusted Monetary Base and Source Base Multipliers (2) m = (1 + k)/(z + k), where k is the k-ratio and z is a composite of the required reserve ratios and the other ratios that reflect the portfolio preferences o f deposi tory institutions and the public. Equation 2 can be solved for z to yield (3) z = (1 + k - mk)/m. This calculation of z is done for both mA and ms; the results are denoted respectively as zA and zs. Figures 2 and 3 show the behavior of the multiplier, the k-ratio and z for the adjusted monetary base and source base, respectively, over the full sample period. In both cases, z declines following the adoption of the MCA, 17The credit controls imposed new reserve requirements on increases in credit card lending, on large-denomination time deposits and on money market mutual funds. The credit controls were imposed in March 1980 and removed in July 1980. 18Actually, this observation is not too surprising. The removal of reserve requirements on a large class of time and savings deposits should have caused the multiplier to increase significantly. Moreover, the Board’s estimates in http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis although the timing o f this descent does not match precisely the implementation of the MCA. The behavior of zs in the early to mid-1980s was influenced greatly by the Federal Reserve's imposition and subsequent elimination of credit controls.17 Nonetheless, its decline through February 1984 suggests that the elimination of reserve requirements on many savings and time deposits and/or the lowering o f reserve re quirements for most member banks are the dominant factors influencing the level o f the multiplier.18 zs declines markedly through February 1984 and increases slightly thereafter. The increase following the move to CRA is con sistent with the hypothesized effect of CRA, but is so small that the move to CRA might not have had an important impact on the level of the multiplier. zA behaves similarly, except that dicate that the largest effect of reserve-requirement changes for member and nonmember institutions on reserves was through institutions that had their reserve re quirements decreased (see table 2). 41 Figure 2 Adjusted Monetary Base Multiplier and Its Components mA zA k 3 .2 0.42 3.1 0.38 3.0 0.34 2 .9 0.30 2.8 2 .7 0.26 2 .6 0.22 2 .5 0.18 2 .4 0.14 2.3 2.2 0.10 1973 74 75 76 77 78 79 80 it continues to fall following the move to CRA.19 Figures 2 and 3 reveal that much o f the move ment in the multipliers is associated with move ments in the k-ratio. The dominant effect of the k-ratio on the multipliers is particularly evident for the source base multiplier after 1984, when zs hardly changed. Indeed, the decline in both multipliers since mid-1986 is associated with a rise in the k-ratio; it appears to be unrelated to movements in z. The effect o f reserve-requirement changes on the level of zs is seen more clearly when the 19The divergent behavior of zA and zs, especially after the move to CRA, is difficult to explain. With the exception of the ratio of reservable time deposits to checkable deposits, changes in all other ratios should be reflected in the same way in both measures of z. An increase in the ratio of reservable time deposits to checkable deposits would cause zs to rise; because of the way that the adjusted monetary base has been calculated since November 1980, however, such an increase would have no effect on zA. In any event, the disparate movements in the z’s had a very small effect on the multipliers; both multipliers have moved together after February 1984. It should be noted, that because zs reflects the actual level of reserve requirements while zA reflects the average 81 82 83 84 85 86 87 1988 data are differenced. The differences of zA and zs, denoted Aza and Azs, respectively, are pre sented in figure 4. Beginning in 1980, there are several pronounced spikes in Azs. The first two are the large positive and negative spikes associated with the introduction and subsequent elimination of the credit controls. The next seven large negative spikes are associated with the important phase-in dates for the MCA for member banks. The presence o f spikes in Azs related to reserve-requirement changes and their absence in Aza attests both to the importance of the ef- level over some base period before November 1980 and the marginal reserve requirement on transaction deposits (12 percent) thereafter, zs is larger than zA until mid-1982 and is smaller thereafter. The level of zs in recent years is somewhat puzzling, however, because it is substantially less than the marginal reserve requirement on transaction deposits. Moreover, both measures suggest that the pro portion of the z’s not accounted for by reserve require ments is very small. Indeed, the excess reserve ratio and the ratios of government and foreign deposits to total checkable deposits averaged .0018, .0397 and .0231, respectively, from February 1984 through December 1988. SEPTEMBER/OCTOBER 1989 42 Figure 3 Source Base Multiplier and Its Components Zs k September 1980 February 1984 3 . 2 --------------------------------------------------------------------------------------------------------------------------- 0.42 3.1 0.38 0.34 0.30 0.26 0.22 0.18 0.14 0.10 1973 74 75 76 77 78 79 80 81 82 83 84 85 86 87 1988 Figure 4 Changes in the Non-k Components of the Multipliers 1973 74 75 76 77 http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis 78 79 80 81 82 83 84 85 86 87 1988 43 Table 1 Variable 1980.12-1988.12 1980.12-1984.1 1984.2-1988.12 R AmA 1.731 1.576 1.641 1.554 1.1 Ams 2.987 1.762 1.945 1.630 1.7* .022 .015 .020 .012 1.5 Azs .072 .023 .031 .013 3.1* Ak .063 .081 .085 .079 .8 N 1973.1-1980.11 > > Variances of Am, Az and Ak1 Observations Deleted3 Ams 2.046 1.414 .866 1.624 1.4 Azs .047 .010 .006 .008 4.7* Ak .055 .078 .062 .085 .7 .4 A ctual variance is 10 times the reported variance. 2An F-statistic with an * indicates the hypothesis that the variances of the variable are equal across the pre- and postMCA periods can be rejected at the 5 percent significance level. Observations were deleted based on the following rule: if the reserve requirement change took place on or before the 15th of the month, that month was deleted; if it took place after the 15th, the next month was deleted. A close examina tion of deleted observations of Ams and Azs with the corresponding dates indicates that this rule performs reasonably well. feet o f reserve-requirement changes on ms and to the usefulness of the monetary base adjust ment in capturing their effect. Furthermore, the fact that these spikes are completely eliminated after the phase-in of member banks confirms the conclusion reached by the analysis of zA and zs that extending reserve requirements to nonmember banks and thrifts had a relatively unimportant effect on the multiplier. With so little variation in reserve ratios evident in zs after February 1984, assessing the effect of the move to CRA is difficult. Furthermore, the final phase-in of the MCA for member banks coin cides closely with the adoption of CRA. The Variability o f the Multiplier Because the levels o f the multipliers, the cor responding z ’s and the k-ratio have definite trends, the variances o f the levels are not very useful as measures of variation. More appropri “ Diagnostic tests indicate that mA, ms, zA, zs and k are non-stationary in levels but stationary in first differences. Moreover, in most cases, the hypothesis that, in levels, ate measures are the variances o f the first dif ferences (A) of these variables.20 The variances o f the first differences of the multipliers, the z ’s and the k-ratio for various periods are presented in the upper part o f table 1. This table also presents the F-statistic for a test o f the null hypothesis that the variances of each series for the periods 1973.1-1980.11 and 1980.12-1988.12 are equal against the alternative that the variance is larger during the earlier period. These data show that the variance of both AmA and Ams declined following the adop tion of the MCA; however, only the decline for Ams is statistically significant at the 5 percent level. There is also a decrease in the variances of Aza and Azs following the adoption of the MCA. Caution must be exercised in interpreting the decline in the variance of Aza; it is biased downward because of the elimination of the ratio o f reservable time deposits to checkable these series follow a random walk cannot be rejected at the 5 percent level. SEPTEMBER/OCTOBER 1989 44 Table 2 Reserve Requirement Changes, 1973-88 Effective Date Reserve requirement change June 21, 1973 The Board amended its Regulation D to establish a marginal reserve requirement of 8 per cent against certain time deposits and to subject to the 8 percent reserve requirement certain deposits exempt from the rate limitations of the Board’s Regulation Q. In addition, reserves against certain foreign branch deposits were reduced from 10 percent to 8 percent. These changes had little effect on required reserves. July 12, 1973 Reserve requirements were imposed against finance bills. This action increased required reserves approximately $90 million. July 19, 1973 The reserve requirement against all net demand deposits, except the first $2 million was in creased 1/2 percentage point. This action increased required reserves approximately $760 million. October 4, 1973 The marginal reserve requirement against certain time deposits was increased from 8 percent to 11 percent. This action increased required reserves approximately $465 million. December 27, 1973 The marginal reserve requirement against certain time deposits was reduced from 11 percent to 8 percent. This action reduced required reserves approximately $360 million. September 19, 1974 The marginal reserve requirement against time deposits in denomination greater than $100,000 and more than four-month maturity was eliminated. This action reduced required reserves approximately $510 million. December 12, 1974 The reserve requirement against all time deposits with an original maturity of six months or longer was reduced from 5 percent to 3 percent; the reserve requirement against all time deposits with an original maturity of less than six months was increased from 5 percent to 6 percent; and the reserve requirement against net demand deposits more than $400 million was reduced from 18 percent to 17-1/2 percent. In addition, the 3 percent marginal reserve requirement on large certificates of deposit with an initial maturity of less than four months was removed. These actions reduced required reserves approximately $710 million. February 13, 1975 The reserve requirements against all categories of net demand deposits up to $400 million were reduced by one-half of 1 percentage point, and the reserve requirement against net de mand deposits of more than $400 million was reduced 1 percentage point. This action re duced required reserves approximately $1,065 million. May 22, 1975 The reserve requirement against foreign borrowings of member banks, primarily Eurodollars, was reduced from 8 percent to 4 percent. This action reduced required reserves approximate ly $80 million. October 30, 1975 The reserve requirement against member bank time deposits with an original maturity of four years or more was reduced from 3 percent to 1 percent. This action reduced required reserves approximately $360 million. January 8, 1976 The reserve requirement on time deposits maturing in 180 days to 4 years was reduced from 3 percent to 2-1/2 percent. This action reduced required reserves by approximately $500 million. December 30, 1976 The reserve requirement against net demand deposits up to $10 million was reduced by 1/2 percentage point, and the reserve requirement against net demand deposits more than $10 million was reduced by 1/4 percentage point. This action reduced required reserves by ap proximately $550 million. November 2, 1978 A supplementary reserve requirement of 2 percentage points was imposed on time deposits of $100,000 or more. This action increased required reserves approximately $3.0 billion. http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis 45 Table 2 (Continued) Reserve Requirement Changes, 1973-88 Effective Date Reserve requirement change October 11, 1979 A marginal reserve requirement of 8 percent was imposed on “ managed liabilities” of member banks, Edge Act corporations, and U.S. agencies and branches of foreign banks above a base average for the two weeks ending September 26, 1979. Managed liabilities in cluded large time deposits ($100,000 and more with maturities of less than one year), repur chase agreements against U.S. government and federal agency securities, Eurodollar borrow ings, and federal funds borrowings from a nonmember institution. On October 25, required reserves and reserves held by Edge Act corporations were included in member bank reserves. (Previously reserves held by these institutions were recorded as “ other deposits” by Federal Reserve Banks.) These actions raised required reserves approximately $355 million and $320 million, respectively. March 12, 1980 The 8 percentage point marginal reserve requirement was raised to 10 percent. In addition, the base upon which the marginal reserve requirement was calculated was reduced. This ac tion increased required reserves about $1.7 billion. May 29, 1980 The marginal reserve requirement was reduced from 10 percentage points to 5 percentage points and the base upon which the marginal reserve requirement was calculated was raised. This action reduced required reserves about $980 million. July 24, 1980 The 5 percent marginal reserve requirement on managed liabilities and the 2 percent sup plementary reserve requirement against large time deposits were removed. These actions reduced required reserves about $3.2 billion. November 13, 1980 Required reserves of member banks and Edge Act corporations were reduced about $4.3 billion and required reserves of other depository institutions were increased about $1.4 billion due to the implementation of the Monetary Control Act of 1980. February 12, 1981 In conjunction with the transitional phase-in program under the Monetary Control Act, re quired reserves of certain nonmember banks and foreign-related institutions increased by ap proximately $245 million. March 12, 1981 In conjunction with the transitional phase-in program under the Monetary Control Act, re quired reserves of small nonmember “ quarterly reporters” increased about $75 million. May 14, 1981 In conjunction with the transitional phase-in program under the Monetary Control Act, re quired reserves of certain nonmember banks and foreign-related institutions increased by ap proximately $245 million. August 13, 1981 In conjunction with the transitional phase-in program under the Monetary Control Act, re quired reserves of certain nonmember banks and foreign-related institutions increased ap proximately $230 million. September 3, 1981 In conjunction with the transitional phase-in program under the Monetary Control Act, re quired reserves of member banks were reduced about $2.0 billion, and required reserves of other depository institutions were increased about $0.9 billion. November 12, 1981 In conjunction with the transitional phase-in program under the Monetary Control Act, re quired reserves of certain nonmember banks and foreign-related institutions increased about $210 million. January 14, 1982 The low reserve tranche for transaction accounts at depository institutions was raised from $25 million to $26 million. This action reduced required reserves approximately $60 million. February 11, 1982 In conjunction with the transitional phase-in program under the Monetary Control Act, re quired reserves of certain nonmember banks and foreign-related institutions increased about $170 million. SEPTEMBER/OCTOBER 1989 46 Table 2 (Continued) Reserve Requirement Changes, 1973-88 Effective Date Reserve requirement change March 4, 1982 In conjunction with the transitional phase-in program under the Monetary Control Act, re quired reserves of member banks decreased by about $2.0 billion. May 13, 1982 In conjunction with the transitional phase-in program under the Monetary Control Act, re quired reserves of certain nonmember banks and foreign-related institutions increased about $150 million. August 12, 1982 In conjunction with the transitional phase-in program under the Monetary Control Act, re quired reserves of certain nonmember banks and foreign-related institutions increased about $140 million. September 2, 1982 In conjunction with the transitional phase-in program under the Monetary Control Act, re quired reserves of member banks were reduced about $2.1 billion, and required reserves of other depository institutions were increased about $0.9 billion. October 28, 1982 In accordance with provisions of the Depository Institutions Act of 1982, required reserves of certain former member banks were reduced by approximately $100 million. December 23, 1982 In accordance with provisions of the Depository Institutions Act of 1982 that exempted the first $2.1 million of reservable liabilities at all depository institutions from reserve re quirements, required reserves were reduced by an estimated $800 million. March 3, 1983 In conjunction with the transitional phase-in program under the Monetary Control Act, re quired reserves of member banks were reduced by approximately $1.9 billion. April 14, 1983 Required reserves were reduced an estimated $80 million as a result of the elimination of reserve requirements on nonpersonal time deposits with maturities of 2-1/2 to 3-1/2 years. September 1, 1983 In conjunction with the transitional phase-in program under the Monetary Control Act, re quired reserves of member banks were reduced about $2.0 billion, and required reserves of other depository institutions were increased about $0.9 billion. October 20, 1983 Required reserves were reduced an estimated $100 million as a result of the elimination of reserve requirements on nonpersonal time deposits with maturities of 1-1/2 to 2-1/2 years. January 12, 1984 The low reserve tranche for transaction accounts at depository institutions was raised from $26.3 million to $28.9 million. Also, in accordance with the provisions of the Depository In stitutions Act of 1982, the reserve requirement exemption applied to total reservable liabilities was raised from $2.1 million to $2.2 million. These actions reduced required reserves by about $350 million. February 2, 1984 In conjunction with the transitional phase-in program under the Monetary Control Act, re quired reserves of member banks were reduced about $2.0 billion. September 13, 1984 In conjunction with the transitional phase-in program under the Monetary Control Act, re quired reserves of certain nonmember depository institutions increased about $1.08 billion. January 3, 1985 The low-reserve tranche for transaction accounts was raised from $28.9 million to $29.8 million. The exemption applied to reservable liabilities was also raised from $2.2 million to $2.4 million. These actions reduced required reserves by about $190 million. September 12, 1985 According to the transitional phase-in program under the Monetary Control Act, required reserves of certain nonmember depository institutions were increased about $1.23 billion. January 2, 1986 The low-reserve tranche for transaction accounts was raised from $29.8 million to $31.7 million. The exemption applied to reservable liabilities was also raised from $2.4 million to $2.6 million. These actions reduced required reserves by about $340 million. http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis 47 Table 2 (Continued) Reserve Requirement Changes, 1973-88 Effective Date Reserve requirement change April 24, 1986 Money market deposit accounts (MMDA), which had previously been subject to full reserve requirements, were made subject to the transitional phase-in program of the Monetary Con trol Act. In addition, the order of application of the exemption applied to reservable liabilities was changed. These actions reduced required reserves by about $260 million. September 11, 1986 According to the transitional phase-in program under the Monetary Control Act, required reserves of certain nonmember depository institutions were increased about $1.58 billion. January 1, 1987 The low-reserve tranche for transaction accounts was raised from $31.7 million to $36.7 million. The exemption applied to reservable liabilities was also raised from $2.6 million to $2.9 million. These actions reduced required reserves by about $970 million. September 10, 1987 According to the transitional phase-in program under the Monetary Control Act, required reserves of certain nonmember depository institutions were increased about $1.70 billion. December 31, 1987 The low-reserve tranche for transaction accounts was raised from $36.7 million to $40.5 million. The exemption applied to reservable liabilities was also raised from $2.9 million to $3.2 million. The actions reduced required reserves by about $740 million. December 29, 1988 The low-reserve tranche for transaction accounts was raised from $40.5 million to $41.5 million. The exemption applied to reservable liabilities was also raised from $3.2 million to $3.4 million. The actions reduced required reserves by an estimated $210 million. deposits from mA. Care must also be taken in interpreting the decline in the variance o f Azs because reserve-requirement changes affect that variance in an indeterminate way. To remove the effect of reserve-requirement changes, the variances of Ams, Azs and Ak were recalculated from smaller samples in which observations for months affected by the reserve-requirement changes were deleted. These variances are presented in the bottom portion of table 1. The list of reserve requirement changes from Jan uary 1973 through December 1988 is presented in table 2.21 The results show a large, though not statistically significant, decline in the vari ance of Ams and a large and statistically signifi cant decline in the variance o f Azs.22 Hence, while it is clear that reserve-requirement changes had a substantial effect on the vari ances of Ams and Azs, these changes do not seem to qualitatively affect the observed impact of the MCA. 21Reserve-requirement changes from 1960 to 1973 can be found in Burger (1979), pp. 6-7. 22As expected, the variance of Azs from the sample in which observations affected by reserve-requirement changes The variance of Ak increased slightly, but not significantly so over these periods. Hence, it would appear that the observed reduction in the variances of AmA and Ams can be attributed to the predicted reduction in the variances of Aza and Azs. This is not necessarily the case, however. The variance of Am is given by an ex pression like (4) Var(Am) = a2Var(Ak) + b2Var(Az) 2abCov(Ak,Az), where Var and Cov denote the variance and covariance of the variables in parentheses, respectively. Because the coefficients, a and b, change with the MCA and the adoption of CRA, it is impossible to say that the observed decline in the variance o f Am is due solely to the de cline in the variance o f Az. A clearer picture of the effects of the MCA and the adoption of CRA on the variance o f Am can be obtained by calcu lating the proportion of the variance o f Am ac- were deleted is substantially smaller than that from the full sample. The same is generally true for the variance of Ams. SEPTEMBER/OCTOBER 1989 48 Table 3 Relative Contributions of the Components to the Variance of Am Variable1 1973.1-1980.11 1980.12-1988.12 1980.12-1984.1 1984.2-1988.12 a^arfAkJ/VarfAmA) .389 .599 .496 .683 b2Var(AzA)A/ar(AmA) .339 .272 .298 .248 2abCov(AzA,Ak)/Var(AmA) .272 .129 .206 .068 a2Var(Ak)/Var(Ams) .170 .591 .431 .797 b2Var(Azs)/Var(Ams) .516 .400 .406 .300 2abCov(Azs,Ak)/Var(Ams) .315 .009 .162 -.0 97 a2Var(Ak)/Var(Ams) .215 .854 .733 .875 b2Var(Azs)/Var(Ams) .490 .249 .187 .192 2abCov(Azs,Ak)/Var(Ams) .296 -.094 .080 -.067 Observations Deleted2 1The coefficients a and b are (z-1)/(z + k)2 and (1 +k)/(z + k)2, respectively. These coefficients are evaluated at the means of z and k during the periods, and the proportions are based on the Var(Am) calculated from the approximations rather than the actual sample variance for Am. 2See footnote 3 from table 1. counted for by each component on the righthand side of equation 4. These proportions for the relevant periods are presented in table 3. For both multipliers, the proportion of the variance of Am explained by the Az component declines after November 1980, while the proportion of the variance explained by the Ak component rises. Furthermore, the decline in the proportion of the variance of the change in the multiplier explained by the Az component continues after the adoption of CRA. This latter observation is not necessarily evi dence that the move to CRA reduced the vari ability of the multiplier; however, the comple tion of the MCA phase-in for member banks coincides closely with the adoption of CRA. CONCLUSION The changes in reserve requirements specified by the Monetary Control Act of 1980 and the 23See Johannes and Rasche (1979) and Hafer and Hein (1983) for a discussion of the control problem for M1 and how it is related to the Fed’s ability to forecast the money multiplier. Johannes and Rasche (1987) argue that their money stock control model performs well during the 1980s. http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis switch to contemporaneous reserve accounting in February 1984 imply that the link between M l and the monetary base should have become tighter in the 1980s. The empirical evidence presented in this article suggests that, in fact, the money-base relationship has strengthened in the sense that, for a given k-ratio, the multiplier has become less variable. By eliminating or, at least, diminishing the im portance o f some sources of variability in the multiplier, these changes have potentially enhanced the Federal Reserve’s control over M l. The degree to which control over M l has improved, however, hinges on how these changes, among others, have affected the predictability of the multiplier.23 The slight reduction in the variance o f the change in the multiplier in the 1980s does not necessarily imp ly that the multiplier itself is easier to predict. The evidence presented in this article suggests that the predictability o f the multiplier, especial- 49 ly that for the source base, depends crucially on the predictability o f the k-ratio in the 1980s. In any case, further research is necessary to deter mine whether predicting the multiplier has become more or less difficult. REFERENCES Burger, Albert E. “Alternative Measures of the Monetary Base,” this Review (June 1979), pp. 3-8. ________“The Puzzling Growth of the Monetary Aggregates in the 1980s,” this Review (September/October 1988), pp. 46-60. Gilbert, Ft. Alton.“A Revision in the Monetary Base,” this Review (August/September 1987), pp. 24-29. _______ . “ Effectiveness of State Reserve Requirements,” this Review (September 1978). Gilbert, R. Alton, and Jean M. Lovati. “ Bank Reserve Re quirements and their Enforcement: A Comparison Across States,” this Review (March 1978), pp. 22-32. Gilbert, R. Alton, and Michael E. Trebing. “ The New System of Contemporaneous Reserve Requirements,” this Review (December 1982), pp. 3-7. Hafer, R. W. “ The New Monetary Aggregates,” this Review (February 1980), pp. 25-32. Appendix (7) CD, = CD“ + CDn The following equations form a simple static model of the money supply process. This model is intended solely to illustrate the main effects o f the passage o f the Monetary Control Act on the money supply process. It is not intended to represent completely the money supply process.1 (1) M l, = CD, + C, (2) SB, = RR, + ER, + C, (3) RR, = rc [CD“ y + CD,(l-y)] + rT TR f + r cGD, (4) ER, = a[CDM + <5,CD, + GD, + /JCD* + £ (TD, - <5,CD,)]y + a(CD, + d2 CD, + GDt) (1 - y) (5) T D f = d,CD,y + <52CD, (1 - y ) , d, > <52 (6) TD, = TDf* + TDNR 'For example, the model does not account explicitly for the fact that required reserves under both LRA and CRA con sist of deposits at the Federal Reserve plus vault cash held during the two weeks prior to the current reserve maintenance period. Since the impact of the variability of changes in vault cash is the same under all regimes, ac counting for this fact would merely add another random Hafer, R. W., and Scott E. Hein. “The Wayward Money Supply: A Post-Mortem of 1982,” this Review (March 1983), pp. 17-25. Johannes, James M., and Robert H. Rasche. “ Predicting the Money Multiplier,” Journal of Monetary Economics (July 1979), pp. 301-25. _______ . Controlling the Growth of Monetary Aggregates (Kluwer Academic, 1987). Laufenberg, Daniel E. “ Contemporaneous Versus Lagged Reserve Accounting,” Journal of Money, Credit and Banking (May 1976), pp. 239-45. Tarhan, Vefa, and Paul A. Spindt. “ Bank Earning Asset Be havior and the Causality Between Reserves and Money: Lagged Versus Contemporaneous Reserve Accounting,” Journal of Monetary Economics (August 1983), pp. 333-41. Thornton, Daniel L. “ Simple Analytics of the Money Supply Process and Monetary Control,” this Review (October 1982), pp. 32-39. _______ . “ Lagged and Contemporaneous Reserve Account ing: An Alternative View,” this Review (November 1983), pp. 26-33. _______ . “An Early Look at the Volatility of Money and Inter est Rates Under CRR,” this Review (October 1984), pp. 26-32. von Hagen, Jurgen. “ Money Stock Targeting with Alternative Reserve Requirement Systems,” Zeitschrift fur Wirtschaftsund Sozialwissenschaften (1987), pp. 379-95. (8) CDm = 0CD, (9) C, = kCD, (10) TD, = ACD, (11) GD, = gCD, The superscripts, M and N, distinguish depos its held at member depository institutions from those held at nonmember institutions. (Nonmem ber institutions include both banks and thrifts). The superscripts R and NR distinguish reser vable from nonreservable time and savings deposits. The variable names are: C = the currency component of the money stock CD = checkable deposits TD = time and savings deposits component to all of the reduced-form expressions; so it does not qualitatively affect the conclusions. The same conclusion holds for Eurodollar deposits and traveler’s checks, which are also not explicitly treated in the model. SEPTEMBER/OCTOBER 1989 50 GD = government deposits SB = source base, currency plus total reserves RR = required reserves ER = excess reserves M l = the M l definition of the money stock rc = reserve requirement ratio on reservable checkable deposits rT = reserve requirement ratio on reservable savings and time deposits y is a shift parameter with the characteristic, _ ^ ( 1 before November 1980 \ 0 after November 1980 The reserve requirement ratios, rc and rT, and the coefficients, ft and £, are assumed to be fixed parameters, whereas the ratios, 0, k, A, d,, d2, g and a, are treated as independent random variables with time-invariant (stationary) distri butions. These equations establish the relationship be tween the monetary base and M l and the fea tures of the MCA that have altered that rela tionship. In the context o f this static model, there are two distinct regimes: before the MCA, y = l; and, after the MCA, y = 0. Equation 1 is simply the current definition of M l, currency plus checkable deposits, including demand de posits and NOW accounts. Equation 2 defines the uses of source base as the sum of required and excess reserves and currency held by the nonbank public. Equation 3 specifies required reserves in the pre-MCA period (y = l) and under MCA (y = 0). Before the MCA, reserves w ere required to be held against checkable deposits and savings and time deposits at member banks, as well as total government deposits. Equation 5 says that, prior to the MCA, reservable time and savings deposits w ere a fraction, d,, of checkable deposits. Equation 5 defines that class, under the MCA, to be a smaller fraction of checkable deposits, d2, where d, > d2. Equation 4 de scribes excess reserve holdings by all depository institutions under both regimes and is general enough to capture the possibility that nonmember banks acted as if.they were subject to reserve requirements. Specifically, if r> « + rc a + rT ft = _______ and £ = ______I , a a http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis then member and nonmember banks acted identically before and after the MCA. Equation 6 is an identity for total time and savings deposits. Similarly, equation 7 is an identity for total checkable deposits. Equations 8 through 11 establish proportional links of checkable deposits at member banks, currency, total time and savings deposits and government deposits, respectively, to total checkable deposits. Equations 1 through 11 are general enough to show the potential impact o f the MCA on the multiplier. A dynamic specification, however, is necessary to illustrate the possible effects o f the switch from LRA to CRA. Introducing a dynam ic element into the model can be accomplished easily by substituting the following equations for equations 3 and 5: (3') RR, = rcCDM, y + rcCD,., ( l - y ) ( l - i f i ) + rcCD, ifj + rx TD*1 + rcGD,., (1 - V )+ rcGD,ifi (5’) T D ?= d,CDt., y + d2CDtM ( l - y ) ( l - y j ) + d2CD,tyi where ip = 0 under LRA and ip = l under CRA. The Effects o f the M CA The multipliers (denoted by m), linking M l to the source base in the static framework for the pre- and post-MCA regimes, are given respec tively by: (12a) m = 1 + k ; A where A = a(0 + d, + g + /?(l-0) + £ (A - d,)) + rc(0 + g) + rTd, + k (12b) m = 1 + k ; B where B = a( 1 + d2 + g) + rc( l + g) + rxd2 + k The impact of the MCA on the multiplier can be seen partly by comparing the expressions in 12. First, notice that the ratio of checkable de posits at member banks to total checkable de posits, dj, does not influence the multiplier under the MCA. Second, the parameters that capture the nonmember banks’ preferences for holding excess reserves disappear from the 51 multiplier under the MCA. If, however, nonmember banks held reserves as if they were member banks, deposit shifts between member and nonmember banks would not have been a source of variability in the multiplier. Never theless, even in this case, the MCA would have affected the multiplier because o f the MCA’s reclassification of reservable time deposits. This new classification means that d2 replaces d, and A in the multiplier under the MCA regime. To see how the MCA affected the variability of the multiplier, we can compare the variances of m expressed in equations 12a and 12b. By us ing a Taylor’s series expansion to approximate the variances of m before the MCA, one can verify the following: Similarly, we can approximate the variance of the multiplier under MCA: (14) Var (m) = 3m + ( M 2 o2 + 3k/ 3d2 where 3m B - (1 + k) 3k B2 3m 3m dd2 + dm da \ag 3m 3m\ 2 3d. 3 A, where _ A - (1 + k) A2 3k 3m = 30 aP - (rc + tt) (1 + k) A2 3m (1+k) A2 dS 3m 0 + g + dj + /J(l —0) + £ (A - d ,) _ 3a 3m 3d, 3m 3A A2 _ at, - (rT + a) (1 + k) A2 (1 + k) A2 and a2 denotes the constant variance of the ran dom variable j. This approximation assumes that the covariances between the random variables is 0. = rc ag dm / 3m\ 2 „ 2 da 3m + a (1 + k) B2 1 + g + ^2 n da (13) Var (m) = 3m \ 3g / B2 = rT + a (1+k) B2 Comparing equations 13 and 14 reveals that some sources of variation present before the MCA are no longer relevant—namely, a2, o2<j and o2x• Variability of the ratios of time deposits at member banks to total checkable deposits and total time deposits to total checkable depos its does not contribute to the variance of the multiplier under the MCA. The MCA, however, does maintain reserve requirements on some time deposits, represented here by d2CDt. Ac cordingly, variability in the ratio of these deposits to total checkable deposits essentially represents a new source o f variation in the multiplier under MCA. The MCA had another important effect on the variance of the multiplier. In particular, by changing the level of the multiplier, it changed the coefficients on each o f the individual vari ances. The multiplier in the MCA regime will be unambiguously larger than in the pre-MCA regime if, before the adoption of the MCA, non member banks acted identically to member banks—that is, if () = a + rc and £ = a + rT. a a As fi and £ approach 0, the difference in the multipliers for the two regimes gets smaller. But, provided that (d j-d J (rx + a) > (1 —0) (rc + «), B < A and the multiplier is larger under MCA. That is, if the impact of eliminating reserve requirements on a large class of time and savings deposits is greater than the effect of extending reserve requirements to all depos itory institutions, then the multiplier is SEPTEMBER/OCTOBER 1989 52 larger in the current regime than in the preMCA regime.2 That the multiplier can be larger under MCA implies that the sources of variability remaining under MCA can make a greater contribution to the variability o f the multiplier. Even under the simplifying assumption that the magnitudes of the remaining sources of variability do not change across regimes, the variance of the multiplier could be larger under the current regime. Although it is highly unlikely that the variance would increase, the variability in k is likely to have a greater impact on the variance of the multiplier under MCA than in the preMCA regime. The Effects o f the M o v e to CRA f r o m LRA To investigate the possible effects o f the switch to CRA from LRA, we employ the dy namic version of the model. In the dynamic model, there are three regimes o f interest: preMCA, LRA (y = 1, ty» = 0); MCA, LRA (y = 0, if* = 0); and, MCA, CRA (y = 0, xp = 1). The contemporane ous multipliers in the dynamic model for these three regimes are given, respectively, by: r \ (15a) m = w h ere A ' s 1 + k ____ , A' a[9 + 6 ] + g + p (1-0) + £ (A - (*,)] + k , *. 1 + k (15b) m = ____ , B' where B' = a( 1 + d2 + g) + k r- \ 1 + k (15c) m = ____ , C' w h ere C ' = a(l+ c52 + g) + r c(l + g) + r xc52 + k 2As discussed in the main text but not captured in this sim ple model, if the effect of reducing reserve requirements on checkable deposits held at many member banks is large, the adoption of the MCA would tend to increase the multiplier. 3As Thornton (1983) shows, the isolated impact of the move to CRA on the multiplier is diminished if depository institu tions hold excess reserves as a buffer stock to absorb changes in required reserves under CRA, a possibility not http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis Before discussing the effects of the move to CRA, we can see how the MCA influenced the multiplier under LRA by comparing 15a with 15b. As in the static model, the random ratio of checkable deposits at member banks to total checkable deposits and the fixed parameters, describing the behavior of nonmember banks’ holdings of excess reserves prior to MCA, are no longer relevant for the multiplier in the MCA/LRA regime. Also, the MCA influences the level of the multiplier in an analytically indeter minant way. The move to CRA unambiguously decreased the average level of the multiplier, however, as can be seen by inspecting equations 15b and 15c. Nevertheless, the net effect from the first to the third regimes predicted by the dynamic model is identical to that predicted by the static model. That is, holding all else constant, the level o f the multiplier is most likely higher now than before the MCA if the net effect of the MCA was to decrease average reserve requirements. In fact, the dynamic version of the model of the money supply process has similar predic tions about the impact of the MCA on the variability of the multiplier to those from the static model. The similarities of the predictions of both models can be verified by approxima ting the variance of the multipliers expressed in 15 with a Taylor’s series expansion. Since the multiplier declines from the second to the third regimes, variation in k and a provide smaller contributions to the variability of the multiplier upon the move to CRA. The change in the im portance of the variability of d2 and g for the variability o f the multiplier could be smaller, but is likely to be larger. Nonetheless, the pre dicted effect o f the MCA on the variability of the multipliers and its components from the first to the third regimes in the dynamic model is qualitatively identical to the effect predicted by the static model.3 Specifically, the variance of the multiplier should fall with the implementa tion of the MCA and the switch to CRA. captured by the dynamic model. To the extent that these institutions hold excess reserves as a buffer stock, the switch from LRA to CRA has a smaller effect on the dynamic structure of the money supply process. Further, one can verify, by setting a = 0, that the move to CRA could have increased the variability of the multiplier if, under LRA, the only contemporaneous link between the monetary base and M1 were through currency holdings. 53 Jurgen von Hagen Jurgen von Hagen, an assistant professor of business economics and public policy at Indiana University, was a visiting scholar at the Federal Reserve Bank of St. Louis. Kevin L. Kiiesen provided research assistance. Monetary Targeting with Exchange Rate Constraints: The Bundesbank in the 1980s p JL S-ECENT programs for international coordi nation of economic policies have focused on the control of exchange rate movements among the major industrial countries. Such efforts gained visibility in the 1985 Plaza Agreement among the G5 nations (the United States, Canada, France, Germany and the United Kingdom) to curb the rising dollar, and in the subsequent joint efforts to prevent it from falling too low.1 Discussions of exchange rate coordination often neglect the potential conflict between exchange rate targets and domestic monetary policy objec tives. Exchange rate policies may be costly, because a central bank may lose the ability to control domestic money growth and, hence, the domestic rate of inflation, in the effort to con trol exchange rates. The purpose of this paper is to explore the impact o f exchange rate policies on domestic monetary control during the 1980s for one of the main players in the international arena, the German Bundesbank. The Bundesbank presents a particularly interesting case. On the one hand, it maintains a formal, explicit commitment to monetary targeting. On the other hand, it en gages in exchange rate stabilization policies both inside the European Monetary System (EMS) and vis-a-vis the U.S. dollar. Recently, a number of authors have concluded that these joint commit ments do not lead to significant conflict among the Bundesbank’s policy objectives.2 Specifically, they argue that its participation in the EMS and in coordinated exchange rate policies in the G5 does not affect the Bundesbank’s ability to achieve its monetary targets. The analysis in this paper suggests that this conclusion is too optimistic for two reasons. First, it neglects important institutional aspects of the Bank’s operating procedure; second, it neglects the fact that its exchange rate policies are geared to two different markets, the EMS and the dollar. When these aspects are taken in to account, the evidence shows that German domestic money growth has been significantly affected by the Bundesbank’s exchange rate pol icies in at least five years over the decade from 1979 to 1988. 1See Funabashi (1988). 2Bofinger (1988), Camen (1986), Mastropasqua et al. (1988), Obstfeld (1983), Rieke (1984), Roubini (1988). SEPTEMBER/OCTOBER 1989 54 M O N E TA R Y TAR G ETIN G IN GERM ANY Shortly after the 1973 breakdown of the Bretton Woods system of fixed exchange rates, which freed the Bundesbank from the obliga tion to intervene in the deutsche mark-U.S. dollar market to maintain the fixed dollar pari ty, the Bundesbank established monetary tar geting as its monetary policy regime. A mone tary target was first announced in late 1974. Monetary targeting has remained the basic policy regime in Germany, although, occasional ly, changes have occurred in implementation procedures. The Bundesbank announces annual monetary targets for a broad monetary aggregate. During 1975 to 1987, the targeted aggregate was the "central bank money stock,” a weighted M3 monetary aggregate, where M3 is the sum of currency in the non-bank sector, demand de posits and reserveable time and savings depos its. This aggregate is similar to the Federal Reserve’s money stock definition for M2. In 1988, the Bank adopted the simple sum M3 as its target aggregate. Between 1979 and 1987, the targets w ere expressed as ranges of growth rates from the fourth quarter to the fourth quarter; only in 1989, the Bundesbank returned to its pre-1979 practice of announcing a precise fourth-quarter-to-fourth-quarter target growth rate. Reasons f o r M onetary Targeting Annual monetary targets impose limits on monetary policy activism and discretion. They imply that, over a year’s time horizon, the cen tral bank’s actions have to be reconciled with its targeted growth rate of the money supply. The Bundesbank adopted monetary targeting to in fluence the public’s expectations of future infla tion and to provide the public with a standard of monetary policy that can be easily monitored to assess the credibility of the Bank's commit ment to price stability. Targeting a monetary ag gregate helps to reduce the economic cost o f ex pectation errors about inflation that cause fluc 3See e.g. Bundesbank, Monthly Report February 1975, An nual Report for 1975, Schlesinger (1979, 1983). 4For a recent review of the European Monetary System, see Fratianni and von Hagen (1990). German law places the authority to participate in international exchange rate arrangements with the Ministry of Finance, not the http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis tuations in output and employment, and allows the central bank to establish a reputation for commitment to price stability. This reputation helps to further reduce inflation expecta tions. Monetary targeting also permits the Bundes bank to deny responsibility for labor and output market disequilibria, and to signal that mone tary policy will not be available as an instru ment of discretionary aggregate demand man agement. Both arguments arise from the basic view—laid down in the Bank's legal constitu tion—that the principal goal of German mone tary policy is price stability.3 Reasons f o r Exchange Rate Management Exchange rate management has been the se cond important determinant of Bundesbank monetary policy during the 1980s. Exchange rate considerations arise from two grounds. First, Germany’s membership in the EMS obli gates the Bundesbank to intervene in foreign ex change markets to maintain stable parities of the DM with the other participating currencies; these currencies are the French franc, the Belgian franc, the Dutch guilder, the Italian lira, the Irish punt, and the Danish korner.4 Under the EMS arrangement, member central banks are required to intervene without limits if neces sary, to keep exchange rates within target zones of ±2.25 percent (± 6 percent for the lira) around predetermined central parities. Second, the Bundesbank has often argued that intervention in the DM-dollar market is neces sary at times to maintain "orderly” market con ditions; that is, to dampen exchange rate fluctua tions perceived to be unwarranted by the prevailing basic economic conditions or "funda mentals.” While the Bundesbank has repeatedly denied having specific dollar targets for such in terventions, the concept of unwarranted ex change rate movements implies that some funda mental value of the exchange rate is being con sidered and serves as a target for interventions. Bundesbank. One may, therefore, argue that the obliga tions implied by the EMS were imposed on the Bank. In deed, the Bundesbank strongly opposed the formation of the EMS. 55 Table 1 Monetary Targets and Exchange Rate Performance Year Monetary target (percent) Realized money growth (percent) 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 6-9 5-8 4-7 4-7 4-7 4-6 3-5 3.5-5.5 3-6 3-6 6.3 4.9* 3.5* 6.0 7.0 4.6 4.5 7.7* 8.1* 6.7* Relative position of the DM in the previous year Against US$ In EMS ____ ____ mixed weak weak weak weak weak strong strong strong mixed weak mixed strong mixed mixed strong strong strong 'Indicates violation of monetary target. SOURCES: Deutsche Bundesbank, Annual Reports for 1979-87, Monthly Reports, various issues. The Relationship Retween M onetary Control and Exchange Rate Management The link between exchange rate policies and monetary control arises from the central bank's balance sheet. Domestic monetary control is achieved primarily by controlling the growth rate of the monetary base—total bank reserves plus currency in circulation—at an appropriate rate. The monetary base, from the sources' side, consists of various domestic assets and the cen tral bank's stock o f international reserves. Ex change rate control requires intervention in foreign exchange markets: purchases and sales of foreign assets which change the international reserves component of the base. Consequently, to assess the extent to which exchange rate policies affect domestic monetary control, one must answer the question of how much foreign exchange interventions affect the growth of the monetary base. This is called the "sterilization” issue. Only if foreign exchange interventions are "sterilized” completely, that is, have no effect on monetary base growth, will exchange rate poli cies not impede domestic monetary control. Otherwise, foreign exchange market interven tions have some impact on the growth of the monetary base, and, consequently, on domestic monetary control. Sterilization of foreign ex change market interventions requires offsetting sales or purchases of domestic assets, such that the total base remains unchanged. M onetary Targeting with Exchange Rate Constraints The monetary policy regime prevailing in Ger many in the 1980s can best be characterized as one of monetary targeting with EMS and U.S. dollar exchange rate constraints. Did these con straints prevent the Bundesbank from reaching its monetary targets? Table 1, which reports the monetary targets and realized growth rates in Germany since 1979, provides a preliminary look at the answer to this question. During these years, the Bundesbank met its target on five occasions, and missed its target on five oc casions. Money growth was too slow in 1980 and 1981, and too fast in 1986 to 1988. The table also presents some qualitative information about the performance o f the mark vis-a-vis the dollar and in the EMS during the previous years. Here, a “ strong” position of the mark is typified by an appreciation o f the mark and sales o f DM reserves against foreign currency in the relevant market; the opposite is true for a "weak” position. A "mixed” position indicates that the mark switched between "strong” and "weak” during the year. SEPTEMBER/OCTOBER 1989 56 Figure 1 illustrates these qualitative characteri zations. The upper part of the figure shows an index of the DM-dollar exchange rate (red-line) and a weighted index of the DM exchange rates in the EMS (black-line). The lower part o f figure 1 shows an index of the Bundesbank's net foreign asset position (red line) excluding net claims on the European Monetary Cooperation Fund (EMCF), and an index of its net claims on the EMCF (black line). Changes in these two foreign asset positions reflect the Bank’s inter ventions in the dollar market and in EMS cur rencies, respectively. During 1980 and most of 1981, the dollar was rising and net foreign assets fell due to intervention supporting the mark. The mark’s exchange rate in the EMS remained flat, but net claims on the EMCF fell, too. From 1982 to the end of 1984, the dollar kept rising, while the mark appreciated steadily in the EMS. The two net foreign asset positions oscillated with little apparent trend. With the dollar’s decline from early 1985 to the end of 1987, a period o f intervention to support the dollar began, reflected in the increase in net foreign assets during that period. As the mark remained strong in the EMS, net claims on the EMCF began to rise, too. Finally, in 1988, the mark was flat in the EMS, while the dollar was stronger again and the Bank heavily sold dollar assets. Table 1 reveals a clear pattern in the connec tion of exchange rate performance and monetary targeting. There are four years, 1980, and 1985-87, when the mark’s movements against both the dollar and the EMS currencies were in the same direction. Each of these years is followed by one in which the monetary target was violated, 1981 and 1986-88. Conversely, each year in which the mark’s performance against the dollar was different from its perfor mance in the EMS was followed by a year in which the monetary target was achieved. Since the mark’s 1980 weakness against both already began in late 1979, the slight undershooting of the monetary target in 1980 fits into the same pattern. Note that the failures to meet monetary targets in each case were consistent in direction 5ln practice, IRE consists of net claims of the Bundesbank against the European Monetary Cooperation Fund (EMCF), which pools reserves available for exchange market in tervention in the EMS contributed by the participating cen tral banks. Participating central banks can obtain strongcurrency reserves for obligatory interventions supporting their own currency from the EMCF, which leads to an in crease in the net foreign asset position of the strong http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis with the previous exchange market operations reflected in Figure 1: undershooting the targets in 1980 and 1981 followed a period of net sales o f foreign assets, overshooting the targets in 1986-88 followed periods of net purchases of foreign assets by the Bundesbank. Table 1 suggests that the pursuit of exchange rate objectives was associated with a failure to achieve the monetary targets in Germany. How ever, the importance of the Bundesbank’s ex change rate constraints seems to depend on whether the mark is moving in a similar or dissimilar direction in the dollar market and the EMS. Moreover, the link between the foreign exchange market operations and the growth rate o f the money supply is not immediate; it in volves time lags, although the lag is not neces sarily as long as one year. The following section explains these qualifications in more detail. M O N E T A R Y BASE C O N T R O L IN G ERM ANY Equation 1 below describes the monetary base, B, by three net positions on the sources’ side: assets in EMS currencies (IRE), dollar de nominated assets (IR$), which, for simplicity, in clude gold, reserves in the International Monetary Fund (IMF) plus Special Drawing Rights (SDRs) and other foreign assets, and domestic assets (D).5 (1) B = IR$ + IRE + D A characteristic o f the German financial sys tem is that the domestic component consists almost entirely o f Bundesbank loans to domestic commercial banks. In contrast to other major central banks, the Bundesbank's portfolio of open market paper—government securities and Treasury bills—makes up only a very small frac tion of its assets. The structure of the sources of base money is illustrated by table 2, which shows the base and the percentage shares o f its main components for the period of 1978 to 1988. At the end of 1978, international reserves currency central bank. Conversely, if the latter intervenes to support the weak currency, it will transfer the acquired weak-currency reserves to the EMCF, again resulting in an increased net claim position. These rules of the EMS were designed to assure “ symmetry” of interventions, i.e. that it does not matter which bank intervenes. 57 Figure 1 Indexes of DM Exchange Rates and Bundesbank Net Foreign Assets M o n th ly D ata, A pril 1979 = 100 D M /U SS D M /E M S C U R R E N C IE S 90 70 N ET D O L LA R R E S E R V E S 1 110 50 N E T RESER VES IN T H E EM CF 200 150 1Total international reserves less net claims on the EMCF Vertical lines indicate realignments in the EMS. SEPTEMBER/OCTOBER 1989 58 Table 2 Main Sources of the Monetary Base (end of year) International Domestic Base IRS IRE DISC LOMB REPO Year (DMbill) (%) (%) (%) (%) (%) 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 140.7 151.2 145.0 142.2 148.6 157.4 163.8 170.5 179.6 196.4 216.1 68.2 46.0 37.4 34.3 33.3 33.0 31.8 28.2 28.8 36.2 21.2 _ 12.8 6.3 11.7 13.2 9.8 8.8 10.1 9.1 14.7 10.0 12.9 21.9 30.1 35.3 36.2 35.6 38.2 36.0 33.9 27.5 25.7 4.4 2.1 5.3 4.2 8.1 8.4 4.9 1.4 1.3 0.4 5.2 0 0 4.3 8.4 6.1 10.3 15.7 24.4 18.5 14.1 36.1 MOB (%) SEC <%) -3 .2 1.3 3.1 2.7 2.7 1.5 1.3 0.0 2.2 1.8 1.8 3.0 1.4 2.7 2.6 3.6 5.0 2.6 2.4 2.9 2.3 2.3 NOTE: Base - currency in circulation plus central bank deposits of commercial banks; IR$ - net monetary reserves less net claims on the European Monetary Cooperation Fund, IRE; DISC - dis count window borrowing (domestic and foreign bills); LOMB - Lombard borrowing; REPO loans to domestic commercial banks through repurchase agreements; MOB - net stock of money market paper (Mobilisierungs-und Liquiditatspapiere); SEC - stock of government securities. SOURCE: Bundesbank, Monthly Reports. accounted for nearly two thirds of the base. That share declined to about 45 percent at the end of 1981. It climbed again from 38 to 51 per cent during 1987 and fell sharply in 1988. Table 2 distinguishes the three most impor tant types of loan operations in the domestic component: discount credit (DISC), Lombard credit (LOMB) and repurchase agreements (REPOs). In Germany, discount credit is rationed and the discount rate is kept consistently below money market rates, and commercial banks have additional incentives to fully utilize their discount quota.6 Consequently, the Bundesbank can tightly control the quantity of discount credit. Discount credit is collateralized with trade bills and has a fixed maturity of up to 90 days. Lombard credit, in contrast, is freely accessi ble to banks under normal circumstances at a 6For example, an individual bank’s quota may be reduced if it has not been exhausted for some time. This practice im plies that variations in the discount rate have no direct im pact on the money supply. The important variable deter mining the quantity of discount credit is the total size of discount quota. http://fraser.stlouisfed.org/ FEDERAL BANK OF ST. LOUIS Federal Reserve Bank of St. RESERVE Louis rate that is kept higher than money market rates. The Bundesbank may, however, impose quantitative restrictions on Lombard credit also. Lombard credit, which is collateralized with trade bills, securities and Treasury bills and can be repaid at any time, is a more flexible refi nancing instrument than discount credit for commercial banks. Repurchase agreements are loans to commer cial banks collateralized with securities, tradebills or foreign assets.7 Typically, they have fixed maturities between three and 30 days and are available only at the Bundesbank's discretion. The table shows that REPOs gained importance during the 1980s. The two remaining components of the domestic sources o f the German monetary base shown in table 2 are the Bundesbank’s stock of open market paper, securities issued by the Fed eral or State Governments, the Federal Railroad ’The official terms for these operations, “ open market operations with repurchase agreements,” is misleading in that their economic nature is a loan to a commercial bank, using a bank’s asset as collateral. That is, the Bundes bank does not purchase or sell securities or trade bills in these operations. See Bundesbank (1985). 59 or the Federal Post Office (SEC), and the net Treasury bill position (MOB).8 The Bank can issue Treasury bills, “Mobilisierungs-undLiquiditatspapiere” on its own initiative, but on ly up to the amount of DM 16.5 billion. During the period under consideration, the combined share o f MOB and SEC never exceeded 7 per cent of the base. An important consequence of the Bundes bank’s asset structure is that the domestic com ponent, D, is controlled mainly via loans to domestic banks. For this purpose, the Bundes bank has developed a two-stage strategy.9 It is based on the decomposition of the domestic component into a permanent part (P) and a transitory part (T): (2) D = P + T. The permanent part (P) is used to achieve the desired trend growth o f the domestic compo nent over a time horizon of several months. The main policy instruments for its control are purchases and sales o f securities (SEC), reserve requirements, the discount rate and discount quota.10 The transitory component (T) is used to con trol the short-term growth of the monetary base with regard to current money market con ditions. Here, the main policy instruments are the issue and redemption of Treasury bills (MOB), repurchase agreements, including loans collateralized with foreign assets (REPO) and Lombard credit (LOMB).11 Variations in T are geared primarily at reduc ing short-run fluctuations in the interbank rate for overnight central bank funds, the Bundes bank’s principal operating target for monetary control.12 The Bundesbank stresses the shortrun character of these operations by calling them "reversible money market operations.” Reversible money market operations have typical maturities between two and thirty days; 8MOB is the difference between “ Equalization Claims” on the Federal Government and “ Liabilities to Banks from Is suing Mobilization and Liquidity Paper” in the Bundesbank balance sheet. 9For detailed discussions, see Dudler (1988), Neumann (1988). 10Note that SEC and discount credit policies operate on the supply side of base money; reserve requirements impact on its demand. "Loans collateralized with foreign assets, foreign exchange swaps—the combination of a foreign exchange operation only recently has the Bundesbank introduced repurchase agreements that extend over two months. Under the current two-stage strategy, increases or decreases of the transitory compo nent are generally reversed after some time. That is, they are neither intended nor permitted to have a lasting impact on the total domestic component or on the base. The notable excep tion to this policy design was the increase in REPO during 1984/85, when the Bundesbank decided to increase the stock of REPO to gain more flexibility in its use to control money market conditions.13 STERILIZING FOREIGN EX CHANGE INTERVENTIONS: THE CASE OF THE BU N D ESBAN K In view of equations 1 and 2, we can now write the change in the monetary base during a period t as: (3) AB, = AIR, + AP, + AT,, where AIR, (= AIRS, + AIRE,) is the change in the total international reserves component. Con sider now the impact of an exchange market in tervention, say, to support the dollar. This re quires an increase in foreign assets, AIR, > 0. Under the two-stage control procedure de scribed above, the Bundesbank may simultane ously reduce the transitory component if it wants to neutralize the immediate effect of the intervention on the monetary base. Thus, we may assume: (4) AT, = £AIR, + v „ where v, is a random variable that represents all other changes in the transitory component. Inserting (4) into (3) yields: (5) AB, = (1 + £)AIR, + AP, + v,. in the spot market with the opposite operation in the for ward market—and “ deposit policy” —temporary transfers of central bank deposits of the Federal Government into and out of the banking sector—are additional, but less im portant, components of T. Loans collateralized with foreign assets and foreign exchange swaps do not change the Bundesbank’s net foreign asset position. See Bundesbank (1985). 12See von Hagen (1988). 13See Bundesbank, Monthly Report, October 1985. SEPTEMBER/OCTOBER 1989 60 If £ = -1 , the intervention is fully sterilized in the current period: There is no current effect on the base. over time. Only long-run sterilization, however, can make monetary base growth independent of the consequences of exchange rate policies. This The two-stage monetary control strategy im plies, however, that things do not end here. The initial decrease in the transitory component is reversed during subsequent periods. As this happens, the effect of the initial intervention on the growth of the monetary base is gradually realized. Therefore, even if foreign exchange in terventions are sterilized completely in the cur rent period, the growth of the base need not be independent of foreign exchange interventions in the longer run. Independence of base growth from foreign exchange interventions in the long er run can only be achieved by counteracting their effects with appropriate changes in the permanent component. requires 1. 6S = -1 , and is independent of the m As a consequence of its control strategy for the monetary base, the Bundesbank has devel oped a two-stage sterilization procedure.14 The essential point is to distinguish between shortrun and long-run sterilization. The former is brought about by variations of the transitory component as discussed above; the latter re quires changes in the permanent component. To illustrate the two-stage procedure, let the change in the permanent component be: (6) AP, = ut + I <Jj AIR,^ j = 1 where ut denotes changes in the permanent component, independent of exchange rate m policies. Here, the sum 1. 6i determines the de- i =i gree of sterilization in the long run, and the in dividual dj parameters show how the steriliza tion with the permanent component is distri buted over time. The effect of a one-time in tervention in period t on base money growth in the current period can be represented as: (7) Bt — Bt_1 = ut + (l + £)AlRt + v„ while the long-run effect is apparent from: (8) Bt+m- B t_, = £ (ut+J + vt+J) + (1 + £ djJAlR,. j = 0 J= 1 Equations (7) and (8) highlight the different roles o f short-run and long-run sterilization in influencing the growth of the monetary base. Short-run sterilization reduces the immediate impact of an intervention on the monetary base and serves to distribute its effect more smoothly 14See Dudler (1988). http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis j = 1 degree of short-run sterilization. For the conflict between monetary targeting and exchange rate targeting, the degree of long-run sterilization is the relevant issue because it determines the consequences of exchange market interventions on money growth over time. Tw o conclusions can be drawn at this point. First, the strong negative correlations between contemporaneous or short-lagged changes in the domestic component and the international re serves component o f the base pointed out by Camen (1986), Obstfeld (1983), Roubini (1988), Mastropasqua et al. (1988) and Bofinger (1988) only tell us about a high degree of short-run sterilization. Given the control strategy o f the Bundesbank, these results per se have no im plications for the independence of the Bundes bank's domestic monetary policy goals from its exchange rate policies in the EMS or in the coordination efforts of the G5. Second, even if the degree of short-run sterilization is high, foreign exchange interven tions can produce accelerations or decelerations in base and money over time simply because changes in the transitory component are revers ed only gradually and with a lag of several periods. Therefore, the lag pattern found in table 1 is consistent with this distinction bet ween short-run and long-run sterilization. Offsetting Interventions In practice interventions may not seriously en danger monetary control even if they are not sterilized in the long run. For example, different interventions can offset each other and thus neutralize their individual effects on monetary base growth. Such “offsetting" interventions can occur either over time or across markets. Offsetting interventions that occur over time are possible if, in the absence of interventions, the exchange rate is subject to purely tran sitory, random fluctuations of mean zero, while the underlying "fundamental” exchange rate is constant. If the central bank decides to dampen exchange rate movements by intervening in the 61 foreign exchange market, the resulting interven tions would produce purely transitory fluctua tions in the international reserves component. Since these fluctuations average out to zero over time, they contribute nothing to the growth o f the monetary base over time, even without sterilization.15 Unfortunately, the fundamental exchange rate is not constant, as illustrated by the large swings in the DM-dollar rate and the persistent appreciation o f the mark in the EMS since 1981 shown in figure 1. Worse than that, the fun damental rate is not known. Therefore, every change in the exchange rate can represent a fundamental change, a transitory change, or the sum of both. Thus, interventions to smooth only transitory fluctuations require knowledge of what is transitory and what is not, knowledge which is not available to policymakers. Misinter preting fundamental for transitory movements will lead to interventions that do not average out over time, and therefore cannot be steriliz ed by the transitory base component alone.16 The dual exchange rate constraints in the EMS and the dollar market create an opportuni ty for the second kind of offsetting interven tions. This could occur if, for example, interven tion in the EMS is required to bring the mark’s value in other EMS currencies down, while the mark is weak against the dollar at the same time. Intervention in the EMS then requires purchases of reserves, AIRE > 0, while stabiliz ing the mark against the dollar requires sales of dollar assets, AIR$ < 0. Such offsetting opera tions across the two markets seem to have oc curred frequently between 1982 and 1985. For example, the Bundesbank reports that "the con siderable inflows o f foreign exchange from the EMS area did not pose any problem for Ger many on balance because the Bundesbank simultaneously sold heavily in the dollar market in the form o f smoothing interventions in favor of the Deutsche Mark."17 In the figure, this is most clearly visible during 1982. Since the 15They do, however, increase the variability of base growth. Statements by the Bundesbank indicate that this is indeed its paradigm for exchange rate policies with regard to the dollar: Interventions are geared to dampen “ erratic” fluc tuations of the DM-dollar rate around the fundamental rate. See e.g. Dudler (1988), p. 69, Scholl (1983), p. 120. 16The Bundesbank has admitted that such interpretation er rors have led to undesirable monetary developments on numerous occasions. See Bundesbank, “ Vierzig Jahre Deutsche Mark,” Monthly Report, May 1988. monetary base is affected only by the net change of the total international reserves com ponent, the sterilization problem is considerably alleviated in this situation. Thus, with two exchange rate constraints, the consequences o f exchange rate policies for monetary targeting depend on the relative movement o f the mark in the two markets. Specifically, the Bundesbank’s exchange rate constraints are more likely to jeopardize mone tary control if the mark is moving in the same direction in both markets, because this limits the possibility of offsetting interventions across markets. This suggests that empirical studies of the degree of sterilization by the Bundesbank should distinguish between periods o f equal or opposite movements of the mark against the dollar and the EMS. EM PIR IC AL ESTIMATES OF SH O R T-R UN A N D LONG-RUN ST E R ILIZA T IO N In this section I estimate the degree of steril ization of foreign exchange interventions for the Bundesbank over the period of 1979-88. In con trast to earlier studies, I distinguish between short- and long-run sterilization and look at subperiods according to the mark's relative strength. The Data The analysis is based on monthly data for changes in the monetary base and its sources from the Bundesbank's Monthly Reports. The data are calculated from averages of daily figures and are not seasonally adjusted. Changes in the international reserves component, AIR, are measured as “foreign exchange inflows to or outflows from the Bundesbank." These foreign exchange flows are reported in actual, effective transaction values and measure precisely their effect on the base. This is the 17Annual Report for 1981, p. 76; see also Annual Report for 1982, p. 72, Scholl (1983), p. 124. SEPTEMBER/OCTOBER 1989 62 main advantage of using these data rather than balance sheet data, where, in accordance with German accounting laws, international reserves are reported at constant exchange rates throughout each year.18 One problem arising with these data—as well as with balance sheet data—is that they do not distinguish between changes in international reserves due to in tervention and changes due to other sources. Even without intervention, the net foreign asset position of the Bundesbank would change over time as interest income on foreign assets is col lected and exchanges o f DM for dollars takes place in regular business with U.S. armed forces stationed in Germany.19 However, in the present context o f testing for short-run and long-run sterilization, this does not pose a serious problem.20 For interventions in EMS currencies, balance sheet data must be used. IRE is measured by the Bundesbank’s net claims on the EMCF in connection with the EMS. Dollar market in terventions are then obtained from AIR - AIRE = AIR$. Finally, to obtain an empirical counter part of the transitory base component (T), we use the bank’s "balance of short-term assistance measures on the money market” and add Lom bard loans outstanding.21 The monthly changes in these variables are normalized by the lagged monetary base to obtain scale-free variables. Regression Equations I estimate the following regression model for the transitory component: (9) AT, = 7i, £ P„i 1=1 + i aui A lR $. + ,-i + ^ 1=1 A T t_j + aini y „ j A lR E .+ 1-i A T , _ 12 + + v,. 18See Roubini (1988) for a discussion of valuation problems with balance sheet data. 19See e.g. Monthly Report, November 1988, p. 32. 20We may interpret AIR as the sum of the true effect of in terventions plus a measurement error from other changes in international reserves. It is well-known that such errorsin-variables bias regression coefficients toward zero, unless the covariance of the measurement error and the equation error is positive and large relative to the magni tude of the coefficient being estimated. In tests of a shortrun sterilization coefficient of negative one, and of no longrun effect of AIR on the base, the test is biased against full short-run and in favor of full long-run sterilization. Since the test results accept the former and reject the lat ter, they are actually stronger than indicated by the nominal significance levels used. 21All data except IRE are obtained from the Bundesbank’s table III.3. IRE is taken from the Bank’s table IX.6a. http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis The model has an autoregressive part and two sets of regressors, the two intervention vari ables AIR$ and AIRE. In equation 9, the paramn m eter sums 1 /3, . and 1 y s estimate the total j= 1 * J = 1 ’ degree of short-run sterilization achieved with the transitory component, while the individual coefficients and yui, show how this is distributed over the current and the following months. The model for the monetary base is: K (10) ABt = n2 + I a j AB,., + a ABt_ 12 + J= 1 N M z p Jtl a i r s , ., + j=i I j=i y2li A ire , , + w,. N Here, the sums of parameters Z p2ij and J=1 9 M Z y2j measure the total impact of a given inter im i ' vention on the base; /?2;j and y2)j reflect its dis tribution over time. I estimate equations (9) and (10) over the total period, 1979 to 1988, and three subperiods.22 These subperiods, while necessarily somewhat arbitrary, were chosen in accordance with the characterizations shown in table 1: 1980-81 was a period in which the mark was generally weak against both the dollar and the EMS currencies, 1985-87 was a period in which the mark was mostly strong against both currency groups; 1982 to 1984 was a period in which the mark was generally weak against the dollar and strong against the EMS currencies. Preliminary estimates of the two equations were run to select the appropriate lag lengths for the regressors. The preferred specifications have k = K = 3 lags for the autoregressive part 22We estimate equations 6 and 8 together with an auxiliary equation for total interventions, AIR, or dollar market in terventions, AIR$, using a three-stage least-squares (3SLS) estimator. The equation for AIR$ is a regression of these interventions on a set of instruments, namely lagged in terventions, lagged changes in the DM-dollar rate and lagged German-U.S. interest rate differentials for call money and government securities. The use of these in struments allows us to solve the simultaneity problem that arises in equations 7 and 9, as exchange rates and hence interventions may be contemporaneously affected by changes in the base. Furthermore, the use of the 3SLS estimator is appropriate to account for the likely correlation of the error terms ut and w,. For the EMS interventions, an instrumental estimator was not used because there were no appropriate instruments. 63 Table 3 Results for Transitory Sterilization (dependent variable AT) Summary of Estimates Sample 1/79-12/88 1/80-12/81 1/82-12/84 1/85-12/87 own F„ 4.3” 6.0** 2.8* 4.6** AIRS lags -0.21 12.1** -1.33 6.1** 0.32 10.4** -0.31 11.4** AIRE F, t f i, -0.74 10.9** -1 .1 7 8.9** -1.25 6.2** -0.80 6.5** * 1. --0.82 --1.87 --1.37 --0.30 SE(%) F R2 AR(6 2.70 4.72 2.35 2.72 16.0** 4.0** 5.6** 7.8** 0.54 0.68 0.62 0.70 3.8 2.3 7.8 3.3 F-Tests for Short-Run Sterilization (degrees of freedom in parentheses) 1/79-12/88 1/80-12/81 1/82-12/84 1/85-12/87 1/3,,= -1 2.3 (1,325) 0.3 (1,40) 0.8 (1,76) 0.6 (1,76) = -1 0.8 (1,325) 3.6 + (1,40) 0.9 (1,76) 5.1* (1,76) joint test 1.1 (2,325) 1.9 (2,40) 0.8 (2,76) 2.5 + (2,76) NOTE: SE is the regression standard error. F„, F„, Fr are F-tests for the joint significance of the sets of parameters involved. Numerator degrees of freedom are: F„ 3, F„ 2, Fy 2. Denominator degrees of freedom are: total period, 325; 1980-81, 40; 82-84, 76; 85-87, 76. R2 is the multiple correlation coefficient. F is the F-test for model significance. R2 and F are taken from the second stage estimates and apply to each equation individually. AR(6) is the test statistic of a Lagrange Multiplier test for residual autocorrelation up to lag six. AR(6) has a chi-square distribution with six degrees of freedom under the null of no autocor relation. + , * and ** denote significance at the 10 percent, 5 percent and 1 percent levels. and m = n = 1 lag for the intervention vari ables in the transitory component model, and N = 6 and M = 4 lags for dollar and EMS inter ventions in the base equation. These specifica tions were chosen because the inclusion of more lags did not improve the model’s ex planatory power, while further restrictions reduced it. own lags are not significant, presumably reflects the Bundesbank’s attempt to build up a larger portfolio of loans to banks with repurchase agreements during this period. Both sets of in tervention variables appear highly significant. In all subsamples and the total sample, the sums of the short-run sterilization coefficients /Jl(j and y 1;j have the expected negative signs. Short-Run Sterilization Based on equation 9, the hypothesis of full short-run sterilization is given by: Table 3 summarizes the result for the tran sitory component equation. The upper panel in dicates that the regressions are highly signifi cant for all samples and explain monthly changes in the transitory component fairly well. The AR(6) statistic shows no signs of residual autocorrelation. The significant negative sum of the AR coefficients agrees with the reversibility o f the short-run operations involved. The only exception in the subsample 1982-84 where the (11) i pul = -1, X y j = -1. J= 1 J= 1 The lower panel of table 3 reports the results of the tests for short-run sterilization. W e test the hypothesis o f full short-run sterilization separately for each type of intervention, and jointly. The hypothesis is not rejected for the total sample. Only for interventions in the EMS is full short-run sterilization rejected in the two SEPTEMBER/OCTOBER 1989 64 Table 4 Results for Long-Run Sterilization (dependent variable Sample 1/79-12/88 1/80-12/81 1/82-12/84 1/85-12/87 own lags F„ 2.3 + 5.4** 0.4 1.3 -0.32 -0.86 0.23 -0.30 F, 1.6 1.6 0.6 4.0* AB) Summary of Estimates R2 AIRS AIRE SE(%) F Fr ^ 2) V* 0.22 2.0 + 0.24 1.08 13.3** 0.64 0.74 1.9 0.66 1.55 5.3** 0.89 -0.20 1.3 0.30 1.41 7.8** 0.84 7.1** 0.83 0.59 7.7** 0.30 0.72 AR(6 4.3 6.8 5.0 6.1 II O M II o N> F-Tests for Long-Run Sterilization (degrees of freedom in parentheses) joint test 1/79-12/88 5.6* 4.8* 3.9* (2,325) (1,325) (1,325) 1/80-12/81 5.3* 6.1* 5.6** (1,40) (1,40) (2,40) 1/82-12/84 0.4 2.3 1.3 (1,76) (1,76) (2,76) 1/85-12/87 9.4** 2.9 + 6.7** (1,76) (1,76) (2,76) NOTE: SE is the regression standard error. F„, F„, F, are F-tests for the joint significance of the sets of parameters involved. Numerator degrees of freedom are: F„ 3, Fp 6, Fr 4. Denominator degrees of freedom are: total period, 325; 1980-81, 40; 82-84, 76; 85-87, 76. R2 is the multiple correlation coefficient. F is the F-test for model significance. R2 and F are taken from the second stage estimates and apply to each equation individually. AR(6) is the test statistic of a Lagrange Multiplier test for residual autocorrelation up to lag six. AR(6) has a chi-square distribution with six degrees of freedom under the null of no autocor relation. + , * and ** denote significance at the 10 percent, 5 percent and 1 percent levels. subsamples with equal relative stances of the mark, 1980-81 and 1985-87. The hypothesis of full short-run sterilization of dollar market in terventions is never rejected. Long-Run Sterilization Table 4 presents a similar summary of the results for the monetary base. Again, all regres sions are found highly significant and without residual autocorrelation. When tested separate ly, the two sets of intervention variables enter these equations significantly only in the 1985-87 subsample. However, if we test the significance of the dollar and the EMS interventions jointly, 23The joint tests yield the following F-ratios (degrees of freedom in parentheses): 1.5 (10,325) for the total sample, 2.1* (10,40) for 1980-81, 1.3 (10,76) for 1982-84, and 4.8** (10,76) for 1985-87. http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis w e find that the interventions together had significant consequences for German monetary base growth in 1980-81 and 1985-87.23 The sum of coefficients p2ii and y2;J have the anticipated, positive signs in the total sample and these two sub samples. The lower panel o f this table reports the results of testing for full long-run sterilization: N (12) I J= 1 M p2>l = 0, X y j = 0. J= 1 W e reject full long-run sterilization for both types of interventions in the total sample and the two subsamples when the mark’s move- 65 Estimates With Switching Parameters Table 4 indicates that the Bundesbank’s long-run sterilization of foreign exchange interventions varied substantially between the subperiods considered here. But the estimates in these tables may suffer from small sample problems, as each subperiod is fairly short. To reduce this problem, I re-estimate the system using the total sam ple, 1979 to 1988, and modify the base equation to allow for regime shifts over time. For this purpose, I define a dummy variable with the value one for 1980, 1981, and 1985 to 1987, and zero elsewhere, and multiply it by the lagged intervention vari ables AlRS^j and AlREt j. The resulting variables are added as additional regressors to the base equations. The coefficients on these new regressors estimate the differ ence between the impact of interventions on monetary base growth in 1979, 1982-84, and 1988 vs. the remaining years.1 The results of this step are reported in the table below. The first line of this table con firms the significance of the new regressors and, hence, of the change in regression para meters between the subperiods. Foreign ex change market interventions had a significant long-run impact on base money growth in the years of 1980 to 1981 and 1985 to 1987. The second jia rt o f the table reports the estimates I/J2. and Zy2. and the results of the tests for long-run sterilization. Long-run sterilization prevailed in 1979, 1982-84 and 1988. However, long-run sterilization is strongly rejected for the remaining years, when the mark had similar relative positions in the EMS and the dollar market. Thus, the table corroborates the results o f the earlier tables. periods. A similar modification for the transitory compo nent seems unnecessary, as we find full short-run sterilization uniformly for all subperiods. 1Treating the two periods 1980-81 and 1985-87 equally is justified by the result in table 4 that the sums and Xy2j in the base equation are similar for these Estimates of Base Equation with Shifting Parameters Dependent Variable: AB; Sample 1979 to 1988 F 0.4 (6) F, 0.3 (4) F„ 0.4 (10) F,d 2.2* (6) K 3.2* (4) K 2.5** (10) SE(°/o) 0.94 F 10.8** (24) AR(6) 6.0 R2 0.73 Tests for Long-Run Sterilization -0 .0 7 0.4 -0.05 0.5 0.61 10.7** 0.49 5.4* NOTE: F„, F, and F „ are F-tests for joint significance of the intervention variables, x = AIRS, z = AIRE'. Numbers in parentheses are numerator degrees of freedom. Denominator degrees of freedom are 315 (94 for F). “ d” denotes tests and parameter estimates for the subperiods 1980-81 and 1985-87 for which the dummies are one. Numbers below the sum-of-parameter estimates are F-values for the null of full long-run sterilization, with one numerator degree of freedom. See previous tables for further symbols. SEPTEMBER/OCTOBER 1989 66 ments against the dollar and the EMS were in the same direction. Thus, the impact of foreign exchange interventions on monetary base growth was significant in times when the two exchange rate constraints required intervention in the same direction.24 The empirical results for short-run and longrun sterilization are thus very different. In con trast to short-run sterilization, long-run steriliza tion, which is the more relevant issue for mone tary policy, did not generally hold throughout the 1980s. The shaded insert on the opposite page illustrates how this important conclusion is missed if the analysis fails to recognize the Bundesbank’s distinction between short-run and long-run sterilization. D o EMS Interventions and Dollar Interventions Have the Same Effects? Did the Bundesbank's sterilization procedures treat interventions in the EMS significantly dif ferent from interventions in the dollar market? To answer this question, we test the hypotheses: N <1 3 > 1 Pui = J= 1 1 Y ltj and J 1 02,J = J= 1 = 1 M 1 Y 2,V J= 1 meaning that the extent of short- and long-run sterilization is the same for both types of in terventions. The F-statistics pertaining to these tests for the total sample and the three sub samples are all well below the 10 percent significance levels.25 Thus, we do not reject the hypothesis that, apart from differences in tim ing, sterilization is the same on both markets. Even though the extent of sterilization was the same in the dollar market and the EMS, dollar market and EMS intervention had dif ferent implications for monetary growth in Ger many during the 1980s. This result is due to the “ Finding a significant impact of interventions on monetary base growth does not necessarily imply that interventions became an obstacle to monetary targeting. For example, if the Bundesbank wanted to hold a constant share of its total assets in international reserves, the latter would have to grow in line with the monetary base [e.g. Scholl (1983), p. 120]. In this case, such reversed causality between base growth and growth of international reserves could lead to similar regression results as those above. How ever, we know from figure 1 that the net international reserves positions of the Bundesbank were subject to large swings, particularly during the years when long-run sterilization did not occur, which makes this interpretation implausible. http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis marked differences in the time profiles of these interventions. EMS interventions have occurred on a large scale mostly around realignments of the central parities. As the realignments involv ed revaluations of the mark, the Bundesbank generally experienced large inflows o f interna tional reserves before realignments, due to in terventions supporting weak currencies. But the new central parities were usually chosen such that the mark would be temporarily in a rela tively weak position in the EMS after the re alignment. The realignments triggered outflows of reserves, which offset the initial expan sionary effect.26 This general pattern is clearly demonstrated in figure 1. Similar, self-reverting tendencies did not, in general, occur for interventions in the dollar market. The long-lasting changes in dollar reserves shown in figure 1 suggest that in terventions in the dollar market, especially in the two critical periods of 1980-81 and 1985-87, had more permanent effects. These observations indicate that the monetary control implications of EMS and dollar market interventions have been very different: non-sterilized EMS in terventions cause temporary deviations of monetary base growth from what is warranted by the monetary target. In contrast, dollar market interventions cause more permanent deviations and, particularly in the two critical periods, have contributed significantly to the over- and under-shooting of the monetary target.27 IM PLIC A T IO N S FO R M O N E T AR Y PO LIC Y The fact that the Bundesbank fully sterilizes foreign exchange interventions in the short run but does not generally do so in the long run can be interpreted in two ways. Incomplete long-run sterilization may reflect the lack of suf- 25The following F-values are computed for this test (degrees of freedom in parentheses). Equal short-run sterilization: 0.2 (1,325), 2.8 (1,40), 0.1 (1,76), 2.2 (1,76) for 1979-88, 1980-81, 1982-84, 1985-87, respectively. Equal long-run sterilization: 0.0 (1,325), 0.0 (1,40), 0.1 (1,76), and 1.1 (1,76). For the base equation with shifting parameters, the F-value for equal long-run sterilization is 0.2 (1,315) for the interactive dummy terms. 26See Bundesbank, Monthly Report, November 1988, p. 34. 27This accords well with Dudler’s (1988) assessment of the EMS and dollar market constraints. The result for the EMS also accords well with the findings about German dominance in the EMS in Fratianni and von Hagen (1990). 67 A Pitfall: Testing for Sterilization Using the Total Domestic Component The distinction between short- and long-run sterilization, derived from the Bundesbank’s operating principles, is crucial to assess the impact of its exchange rate policies on mone tary control in Germany. Had we looked only at the transitory component, we would have concluded that the Bank fully sterilized all in terventions; this conclusion, which has been reached by most studies of the Bundesbank, would then suggest that exchange rate con straints have not interfered with Germany’s monetary targeting. The tests for long-run sterilization, however, show that this conclu sion is erroneous. To further demonstrate this point, I replace the base equation in the regressions cited in the text by an equation for the total domestic component, AD = AB - AIR. Following the ficiently effective monetary control instruments, i.e., an institutional deficiency. Alternatively, in complete long-run sterilization may reflect that, under certain circumstances, the Bundesbank is willing to give in on its monetary target and lend more weight to exchange rate considera tions in its decisions. The Bank itself has argued repeatedly that the latter is true.28 Specifically, violations of the monetary target ranges have been justified e* post as necessary to reduce ex change rate pressures. From this perspective, the results tell us something about how the relative weights o f exchange rate targets and the monetary target in the Bank’s decisions vary over time: The monetary target dominates as long as the mark performs differently in the EMS than against the dollar, but the monetary target becomes subordinate if the mark is either strong or weak against both currency groups.29 The estimates in table 4 indicate that, under these circumstances, an intervention of, say, DM 1 billion, will raise the monetary base per 28See Scholl (1983), p. 124, Dudler (1988), p. 74, Schlesinger (1988), p. 11, Bundesbank, Monthly Report, May 1988, p. 20, and Rieke (1984), p. 53. procedure used in the previous studies, I allow for a contemporaneous effect and two lags of the intervention variable. This yields the following estimates o f Z/?2j, the total im pact of dollar market interventions: -1.06 for the total sample, -0.86 for 1980-81, -1.64 for 1982-84, and -1.25 for 1985-87. For the EMS interventions, the total impact o f interven tions is estimated as Zy2j = -1.02 for the total sample, -1.11 for 1980-81, -0.99 for 1982-84, and -1.18 for 1985-87. In no case do I reject the hypotheses of complete sterilization. These results, similar to those for the tran sitory component, indicate that short-run fluctuations in the total domestic component and its short-run co-movements with the in tervention variable are dominated by changes in the transitory component. manently by DM 500 to 600 million over the next two quarters. The Effectiveness o f “Sterilized” Intervention The empirical finding of a significant dif ference between short-run and long-run sterilization of foreign exchange interventions also sheds some light on how interventions af fect exchange rates. Asset market theories of exchange rate determination hold that sterilized interventions have no exchange rate effect if market participants regard assets in the coun tries concerned as perfect substitutes. The reason is that such interventions do not effec tively change the composition of private in vestors’ portfolios, and, hence, cannot change relative asset prices. Leaving the two money supplies unaffected, such interventions merely exchange perfect substitutes. On this basis, the assertion of earlier studies that the Bundesbank perfectly sterilizes interventions in the dollar 29Neumann (1984) concludes that the Bank raises the weight on the monetary target in response to increasing exchange rate uncertainty. SEPTEMBER/OCTOBER 1989 68 market would imply that such interventions are ineffective, since the degree of capital mobility between Germany and the U.S. is high and it seems plausible to assume that short-term, in terest bearing assets in the two countries are very close substitutes. However, the Bundesbank does find sterilized interventions in the dollar market effective, and recent empirical research confirms its claim30. The distinction between perfect short-run and imperfect long-run sterilization becomes crucial in resolving this puzzle. If market participants understand that full sterilization holds only in the short run, while long-run sterilization is—or may be—incomplete, interventions would change market participants’ expectations about the future growth of the money supply, and this change in expectations would lead to ex change rate movements even if the interven tions were fully sterilized initially. That is, sterilized interventions act as a signal about future central bank behavior. This is the essence of Mussa’s (1981) expectations argument o f sterilized interventions.31 As Dominguez (1989) points out, market participants generally seem to be aware of central bank activities in the dollar market, so that they are able to read the intended signal. This lends some further plausibility to the expectations argument. The Bundesbank and the EMS The results provide some insights into the functioning of the EMS during the 1980s. They reject the popular hypothesis that sterilization o f EMS interventions makes German monetary policy independent from policies in the EMS and puts the burden of adjustment to balance o f payments problems on weak currency coun tries.32 The evidence suggests that the long-run independence o f German money supply growth from influences in the EMS, shown in Fratianni and von Hagen (1990), should be rather at tributed to the way realignments have been engineered in most cases. Our results suggest that, with its dual exchange rate constraint, the Bundesbank may be more likely to accept or 30Dominguez (1989). 31Rieke (1984), p. 45, argues that the Bundesbank interprets the effectiveness of sterilized intervention on the basis of a signaling argument. The same argument from the Bank is reported by Funabashi (1988), p. 34. http://fraser.stlouisfed.org/ BANK OF ST. LOUIS Federal Reserve FEDERAL Bank of St. RESERVE Louis even pressure for realignments to revaluate the strong mark in the EMS when the dollar is weak than when it is high, since the mark’s strength in both markets reduces the scope for offsetting interventions. This may explain why realignments have indeed been preceded by periods of dollar weakness in most cases.33 CONCLUSIONS Our results can be summarized as follows: Over the entire period under consideration, I cannot reject the hypothesis that the Bundes bank sterilizes foreign exchange interventions completely in the short run. However, complete intervention sterilization does not generally oc cur in the long run. Foreign exchange market interventions affect German monetary base growth significantly when the mark’s move ments against the dollar and the EMS curren cies are in the same direction, that is, when the mark is relatively strong or relatively weak in both markets at the same time. This was the case in five out of the 10 years from 1979 to 1988. In contrast, exchange rate policies have no significant effect on German monetary con trol when the mark has unequal relative posi tions on the two markets. This suggests that, generally, persistent net intervention affects monetary control. The results indicate that the Bundesbank's dollar policies and participation in coordinated interventions since 1985 have contributed significantly to the excess growth of the Ger man money supply, relative to the Bank’s monetary targets in the second half of the 1980s. The present revival of inflation in Germany—in the spring of 1989, inflation in terms of the cost of living index was at an an nual rate of 2.7 percent, up from zero in 1987—marks one cost of this monetary overexpansion and, consequently, of the policies leading to it. This result suggests that interna tional policy coordination schemes that focus on exchange rates are more costly than they were previously believed to be. 32Giavazzi and Giovannini (1987), p. 253. 33Giavazzi and Giovannini (1986). 69 REFERENCES Bofinger, Peter. Das Europaische Wahrungssystem und die geldpolitische Koordination in Europa, Kredit und Kapital 21 (1988), p. 317-45. Deutsche Bundesbank. Annual Reports 1979-1987, Frankfurt. ________ Monthly Reports, various issues, Frankfurt. _______ . Geldpolitische Aufgaben und Instrumente, Frankfurt (1985). Camen, Ulrich. FRG Monetary Policy Under External Con straints, 1979-84, Centre for European Policy Studies, Work ing Paper 21, Brussels (1986). Dominguez, Kathryn. Market Responses to Coordinated Cen tral Bank Intervention, Working Paper, Carnegie-Rochester Conference on Public Policy (April 1989). Dudler, Hermann Josef. “ Monetary Policy and Exchange Mar ket Management in Germany,” in Bank for International Settlements, ed., Exchange Market Intervention and Monetary Policy, (Basle, 1988). Fratianni, Michele, and Jurgen von Hagen. “ The EMS—Ten Years After,” in Carnegie-Rochester Conference Series on Public Policy, vol. 32 (Amsterdam: North Holland, 1990). Funabashi, Yoishi. Managing the Dollar: From the Plaza to the Louvre (Washington, D.C.: Institute for International Economics, 1988). Giavazzi, Francesco, and Alberto Giovannini. “ The EMS and the Dollar,” Economic Policy 2 (1986), pp. 455-73. _______ . “ Models of the EMS: Is Europe a Greater Deutsch mark Area?” in Ralph C. Bryant and Richard Portes, eds., Global Macroeconomics, (London: McMillan 1987). Mastropasqua, C., S. Micossi and R. Rinaldi. “ Interventions, Sterilization and Monetary Policy in the European Monetary System Countries, 1979-87,” in F. Giavazzi, S. Micossi and M. Miller, eds., The European Monetary System (Cambridge University Press, 1988), pp. 252-87. Mussa, Michael. The Role of Official Intervention (Group of Thirty, 1981). Neumann, Manfred J.M. “ Intervention in the Mark/Dollar Market: The Authorities’ Reaction Function,” Journal of In ternational Money and Finance 3 (1984), pp. 223-40. ________ “ Implementation of Monetary Policy in Germany,” Working Paper (University of Bonn, 1988). Obstfeld, Maurice. “ Exchange Rates, Inflation, and the Steri lization Problem: Germany, 1975-81,” European Economic Review 21 (1983), pp. 161-89. Rieke, Wolfgang. “ Die Rolle von Interventionen als Bestimmungsfaktor der Wechselkurse beim ‘Floating,’ ” in Werner Ehrlicher and Rudolf Richter, eds., Devisenmarktinterventionen der Zentralbanken, (Berlin: Duncker and Humblot, 1984). Roubini, Nouriel. “ Sterilization Policies, Offsetting Capital Movements and Exchange Rate Intervention Policies in the EMS,” chapter 4 of unpublished Ph.D. dissertation (Harvard University, 1988). Schlesinger, Helmut. “ Recent Experiences with Monetary Policy in the Federal Republic of Germany,” in Karl Brun ner and Manfred J.M. Neumann, eds., Inflation, Unemploy ment and Monetary Control (Berlin: Duncker & Humblot, 1979), pp. 303-20. _______ . “ The Setting of Monetary Objectives in Germany,” in Paul Meek, ed., Central Bank Views on Monetary Targeting (Federal Reserve Bank of New York, 1983). _______ . “ Das Konzept der Deutschen Bundesbank,” in Werner Ehrlicher and Diethard B. Simmert, eds., Wandlungen des geldpolitischen Instrumentariums der Deutschen Bundesbank (Berlin: Duncker & Humblot, 1988). Scholl, Franz. “ Implications of Monetary Targeting for Ex change Rate Policy,” in Paul Meek, ed., Central Bank Views on Monetary Targeting (Federal Reserve Bank of New York, 1983). von Hagen, Jurgen. “Alternative Operating Procedures for Money Stock Control in West Germany - An Empirical Evaluation,” Weltwirtschaftliches Archiv 124 (1988), pp. 89-109. SEPTEMBER/OCTOBER 1989 Federal Reserve Bank of St. Louis Post Office Box 442 St. Louis, Missouri 63166 The Review is published six times p er year b y the Research and Public Information Department o f the Federal R eserve Bank o f St. Louis. Single-copy subscriptions are available to the public f r e e o f charge. Mail requests f o r subscriptions, back issues, or address changes to: Research and Public Information Department, Federal R eserve Bank o f St. Louis, P.O. Box 442, St. Louis, Missouri 63166. The views expressed are those o f the individual authors and do not necessarily reflect official positions o f the Federal R eserve Bank o f St. Louis o r the Federal R eserve System. Articles herein may b e reprinted provided the source is credited. Please provide the Rank’s Research and Public Information Department with a copy o f reprinted material.