The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
s tn tn n u tK / u u u H tK i a«a ECONOMIC PERSPECTIVES A review from the Federal Reserve Bank of Chicago T h e g e o g ra p h y o f v a lu e ad d ed 2 5 th C o n fe re n c e on B a n k S tru c tu re a n d C o m p e titio n : C o n tro llin g ris k in fin a n c ia l services P u b lic in v e s tm e n t an d p ro d u c tiv ity g ro w th in th e G ro u p o f S even ■ ■ ■ ■" ■ : ... Contents T h e g e o g ra p h y o f v a lu e a d d e d ................................................................ 2 Philip R. Israilevich and W illiam A . Testa The decline of manufacturing in the Northeast and the Midwest has been exaggerated by the peculiarities of data-reporting on value added 2 5 th C o n fe re n c e on B a n k S tru c tu re a n d C o m p e titio n : C o n tro llin g ris k in fin a n c ia l s e r v ic e s ................................................... 13 M ary J. W illiam son Risk management is the key theme as the conference marks its quarter-century P u b lic in v e s tm e n t an d p ro d u c tiv ity g ro w th in th e G ro u p o f S e v e n ................................................................ 17 D avid A . A schauer The investment of public money has positive direct and indirect effects on private sector output and productivity ECONOMIC PER SPEC TIV ES SEPTEMBER/OCTOBER 1989 Volume X llljssu e 5 Karl A. Scheld, Senior Vice President and Director of Research ECONOM IC PER SPEC TIV ES is published by the Research Department o f the Federal Reserve Bank o f Chicago. The views expressed are the authors’ and do not necessarily reflect the views of the management o f the Federal Reserve Bank. Single-copy subscriptions are available free of charge. Please send requests for single- and multiple-copy subscriptions, back issues, and address changes to Public Information Center, Federal Reserve Bank o f Chicago, P.O. Box 834, Chicago, Illinois, 60690-0834, or telephone (312) 322-5111. Articles may be reprinted provided source is credited and the Public Information Center is provided with a copy of the published material. Editorial direction Edward G. Nash, editor, David R. Allardice, regional studies, Herbert Baev, financial structure and regulation, Steven Strongin, monetary policy, Anne Weaver, administration Production Nancy Ahlstrom, typesetting coordinator, Rita Molloy, Yvonne Peeples, typesetters, Kathleen Solotroff, graphics coordinator, Roger Thryselius, Thomas O ’Connell, Lynn Busby, graphics, Chris Cacci, design consultant, Kathryn Moran, assistant editor ISSN 0164-0682 T h e g e o g ra p h y o f v a lu e a d d e d Marketing, R&D, even accounting and legal departments, add value to a manufacturer's product, but that value is attributed solely to the production site, a practice that distorts our understanding of regional manufacturing activity Philip R. Israilevich and W illia m A . Testa Amoco Corporation, a diver correctly measured value added in understand sified manufacturer of chemi ing regional economic behavior. cal and petroleum products, T ak in g s to c k o f m a n u fa c tu rin g refines crude petroleum into It may come as a surprise to some, but we gasoline and other products do not measure manufacturing output by the at such locations as Texas City, Texas, and final sales value of goods such as automobiles, Whiting, Indiana.1 However, many of the tractors, or refined petroleum. Rather, we support services which contribute to the value count only the value that is added by manufac of these refined products are performed at turing companies to raw materials, such as Amoco’s corporate headquarters in Chicago, crude petroleum for gasoline, and intermediate Illinois, and at its research center in Naper components, such as steel and rubber for autos, ville, Illinois. in producing these final manufactured prod The sprawling geography of these activi ucts. Companies engaged in the processes of ties presents a considerable problem in track assembling and transforming these intermedi ing the location of manufacturing across U.S. ate products into finished goods are designated states and regions. In the case of Amoco, how as manufacturers. Their contribution of labor much manufacturing activity should be attrib and capital and entrepreneurship to the na uted to its Chicago area headquarters and tion’s GNP accordingly becomes the nation’s R&D center and how much to its refineries in “value added in manufacturing” or manufac Texas and Indiana? turing output. The U.S. Census Bureau currently attrib Formally, value added is the value of utes all of a company’s manufacturing output products shipped by manufacturers less the to the locations of the production plants, i.e., value of intermediate goods (which is embed the refineries in the Amoco example. While ded in the value of the final manufacturing there may be no one correct method of appor product): tioning output to states and regions, the Census method is clearly inadequate. Consequently, 1) Value Added = Value of Shipments much of what we think we know concerning Materials and Intermediate Goods. the changing geography of manufacturing Value added is, then, a residual, represent across the U.S. may need to be re-examined. ing the incremental value contributed to the In this article, the regional biases inherent product by the manufacturing company (see in the Census measure of manufacturing out Figure 1). Quite correctly, the value of raw put, which is called value added (VA), are Philip R. Israilevich and W illiam A. Testa are econo explained and illustrated. Two potential meth mists at the Federal Reserve Bank of Chicago. ods of correcting the problem are evaluated. The assistance of Tirza Haviv is gratefully We conclude by discussing the importance of acknowledged. F 2 ECONOMIC PERSPECTIVES counted (quite correctly) in the national sum mation of value added.2 The national totals of Measuring value added as a residual value added are not at issue. However, auxil iary activities are wrongly apportioned to states and regions on the basis of operating es For example, tablishment site while neglecting the location value added in manufacturing of the auxiliary establishments. The problem gasoline is, therefore, one of geography and not of sum mation to national industry totals.3 The total VA of each manufacturing company is allo cated to states and regions solely on the basis of where the company’s operating or produc Cost of m aterials & Value of product purchased inputs tion establishments are located. shipped (i.e., gasoline) (i.e., petroleum) However, the geography of the overall company can be quite different from the oper materials and intermediate products is attrib ating establishments where VA is reported. A manufacturing product’s design and engineer uted to the industrial sectors in which they originate, such as mining, construction, serv ing may originate at the company’s R&D cen ter and not at the operating establishment ices, or agriculture. location.4 Similarly, the product’s advertising The current Census method inappropri and image may be fashioned at an out-of-state ately apportions a large part of manufacturing sales office or corporate headquarters of the value added to states and regions. This inap manufacturing company. All these activities, propriately apportioned part is the activity of “auxiliary” establishments of manufacturing which provide services to the operating establishments, do legitimately contribute to a firms, i.e., corporate headquarters, research product’s value. We believe that this contribu and development labs, data processing cen tion to manufacturing output should be ters, and warehouses (see Figure 2). The ac counted at the site of the auxiliary activity. In tivities of auxiliary establishments are practice, no VA at all is reported and FIGURE 2 recorded by auxil Value added to a company’s product takes iary establishments. place throughout the nation... FIGURE 1 T h e a u x ilia ry eco n o m y CORPORATE HEADQUARTERS RESEARCH & DEVELOPMENT .but geographically all value added is attributed to its operating establishments FEDERAL RESERVE BANK OF CHICAGO It is apparent from the payrolls of auxiliary establish ments that the share of VA originating at auxiliary estab lishments is signifi cant. Auxiliary payroll amounted to almost 11 per cent of the nation’s total manufacturing payroll in 1986 (see Figure 3 and Table 1). In individual regions, auxiliary payroll ranged from negligible amounts in several states and 3 the 2-digit SIC (Standard Industrial Classifica tion) code level, the wide-ranging importance of auxiliary payroll is revealed. For example, some industries that fall under the “chemicals industry” banner report over one-fourth of total payroll at auxiliary establishments; some industries in “petroleum and coal products” report over one-third of payroll outside of operating establishments. A u x ilia rie s and regions Standard Metropolitan Statistical Areas (SMSAs) to as high as 49 percent for the State of Delaware and 54 percent in the Stamford, Connecticut, SMS A in 1982. Among the various types of auxiliary ac tivities, administrative and managerial activi ties were most prominent in 1982, followed by general office and clerical, and third by re search, development, and testing (see Figure 4). For individual industries, the evidence on the significance of auxiliary activities is also striking (see Figure 5). Disaggregating total manufacturing into its 19 major components at TABLE 1 Auxiliary establishments for manufacturing firms—1982 Number Share Total m anufacturing A dm inistrative and managerial Office and clerical Research, development, and testing W arehousing 9,676 100.0 7,792 6,157 80.5 63.6 1,967 2,087 20.3 21.6 Electronic data processing 2,357 24.4 Other activities 4,353 44.9 NOTE: Detailed establishm ent data exceed totals and sum to m ore than 100 percent because som e estab lishm ents participate in m ore than one activity. SO U RCE: U.S. Departm ent of Com m erce, Bureau of the Census, 1982 Census o f Manufacturing Subject Series, Vol. 1, p. 1-100. 4 In studying the corporate organization of the manufacturers, some regional analysts have recognized that diverse activities are undertaken within companies and industries in producing a single product. Moreover, these activities are often located at sites away from each other—even across state borders and regional divisions. Industry studies by economic geographers have documented the spatial separation of ac tivities within single corporate entities. For example, the R&D functions of pharmaceuti cal companies in Great Britain have been studied. One study reports that basic re search—that of a generally applicable na ture—is frequently undertaken at large central ized R&D facilities of large pharmaceutical companies. At the same time, specific and applied R&D is overwhelmingly conducted at the production plant site (Howells 1984). Studies of manufacturing establishments have also reflected the cumulative importance of such establishment specialization to regions. Jusenius and Ledebur (1976) were among the first to document specialization in the U.S. South by branch production plants of U.S. manufacturing companies. More recently, Malecki (1985) has examined regional spe cialization in corporate headquarters versus branch plants across U.S. regions for four high-tech industries: computers, semiconduc tors, medical instruments, and computer soft ware. But despite this wide recognition of regional specialization in diverse manufactur ing activities, data covering VA in manufac turing has continued to be allocated to U.S. regions according to the location of production activity alone. The observed geographic distribution of auxiliary activity varies quite widely across states and across metropolitan areas.5 More over, a cursory view of the distribution of ECONOMIC PERSPECTIVES auxiliary payroll suggests a systematic bias across the U.S. (see Figure 6). States in the New England and Middle Atlantic regions are home to very large numbers of auxiliary estab lishments. Similarly, individual Northern states including Illinois, New Jersey, Michi gan, Ohio, and Pennsylvania display manufac turing sectors which are highly intensive in auxiliaries. Meanwhile, states in the South and especially those of the East South Central Region have a dearth of auxiliary locations, tending instead to specialize in operating es tablishments. Accordingly, we would expect that, in measuring manufacturing output, the North and Midwest actually have greater lev els than currently reported while manufactur ing activity in the South is overstated. FIGURE 4 Activities at auxiliary establishments (percent of employees) Administrative 39.3% Data processing Warehousing A fo rm a l te s t FIGURE 5 Auxiliary payroll share by industry percent 0 10 20 30 Petroleum and coal products Chemicals and allied products Tobacco products Instruments and related products Electric and electronic equipment Food and kindred products Leather and leather products Total manufacturing Stone, clay, and glass products Machinery, except electrical Misc. mfg. industries Textile mill products T ransportation equipment Rubber and misc. plastic products Apparel and other textile products Paper and allied products Printing and publishing Lumber and wood products Primary metal industries Fabricated metal products FEDERAL RESERVE SO U R C E: U.S. Department of Commerce, Bureau of the Census. BANK OF CHICAGO 40 It is reasonable to expect that the Census VA is under estimated in states which specialize in auxiliary estab lishments and overestimated in states with high concen trations of operating estab lishments. However, the problem may be insignificant if the proportion or split of activity between auxiliaries and operating units is largely the same in each state and SMSA. If such is the case, the difference between the Census and true VA will be insignificant; i.e., operating establishment activity serves as a good allocator of total manufacturing output of companies to SMSAs and states. To test whether the Census method has a strong bias in overlooking the site locations of auxiliary estab lishments, a formal hypothe sis can be constructed. The current Census method of estimating VA as the resid ual between value of ship ments and materials at oper ating establishments is equivalent to assuming that either: 5 FIGURE 6 1. the auxiliaries make no contribution to VA; or 2. the auxiliaries locate in close proportion to operating establishments with re spect to their effect on VA. The first assumption can be rejected since we have seen that the auxiliaries’ payroll com prises a sizable part of total VA (see Figure 3). The second assumption can be tested if we assume that region-to-region variations in VA of both types of units, operating and auxiliary establishments, can be approximated by the variations in their respective payrolls. Based on assumption 2, we then can formulate the following null hypothesis: HO: the Census-determined VA and true VA are the same. If true, this hypothesis implies that the elasticities of VA with respect to auxiliary unit and operating unit payrolls are the same. A dollar of either auxiliary payroll or operating payroll will contribute equally to a region’s manufacturing VA. The null hypothesis can then be formally tested using the following ordinary least squares (OLS) regression equation: 6 2)7 V = c + b aA + b oO where: V = VA in logarithmic form. A = payroll for auxiliaries in logarithmic form. O = payroll for operating units in logarithmic form. Equation 2 was estimated for both SMSAs and states. There were 172 SMSAs and 46 states which disclosed auxiliary payroll. The estimated results are: SMSAs: c = 1.149 b = 0.031 b =0.941 (12..3) a (2.4) ° (49.3) adj. R2= 0.97 n=172 States: c = 1.197 b = 0.006 b =0.961 (9.5) a (0.3) ° (39.3) adj. R2 = 0.99 n=46 NOTE: Numbers in parentheses are t-statistics. For SMSAs, coefficients for auxiliary and operating units payrolls are both significant and strongly different (bu is 30 times smaller than b ). This means that estimated elasticities of VA (bu and bo ) with respect to payroll in auxiliaries and operating units are very dif ferent. This leads to the rejection of the H hy- ECONOMIC PERSPECTIVES pothesis.6 For states the rejection of the H hypothesis is even more obvious, since bo is positive and significant while ba is insignifi cantly different from zero. Therefore the hy pothesis that bo is infinitely larger than ba can not be rejected. To test the H hypothesis, we had to as sume that the payrolls of operating and auxil iary establishments parallel their respective VA for each state and metro area. However, if this assumption is relaxed, it is still evident that the H would be rejected. It is inconceiv able that differences in the payroll/value-added ratio could offset the large differences between the elasticities of auxiliary unit and operating unit payrolls that were uncovered in the re gression estimation. S e cu lar and c yc lic al bias There are reasons to believe that manufac turing value added, as currently measured, dis torts our view of both long-term regional manufacturing growth and also of the nature of manufacturing activity over the course of the business cycle. Over the long term, the pay roll of employees at auxiliaries has been grow ing steadily for the past 25 years, now ac counting for almost 11 percent of the total industry payroll in comparison to 6 percent around 1960 (see Figure 3). To the extent that growth in auxiliary activity is skewed toward particular regions, long-run growth in manu facturing across regions will be biased there. For example, in a region experiencing greater growth in auxiliary activities than in other manufacturing activities, output growth re ported by the Census is likely to be biased downwards over time. As a case in point, the Great Lakes Region, i.e., Minnesota, Wiscon sin, Illinois, Michigan, Indiana, and Ohio, has maintained its national share of payroll at manufacturing auxiliary establishments from 1963 to 1986 even while its share of national share of total payroll and output declined. Distortion of output changes over the course of the business cycle can also be dem onstrated. Analysts have long puzzled over the severity of the business cycle in manufac turing regions (Borts 1960; Bolton 1978). In general, they have found that, due to the sensi tivity of durable goods sales during business downturns, manufacturing regions undergo wide fluctuations in economic activity over the course of the business cycle. FEDERAL RESERVE BANK OF CHICAGO In measuring the volatility of any region with the Census VA, cyclical volatility will be overstated. VA is based on fluctuations in ac tivity at operating establishments over time. But operating or production activities will likely be more cyclical than the manufacturing sector overall, thereby overstating cyclical swings. This further implies that a greater intensity of auxiliary activities in a region will magnify the cyclical bias. One hypothesized reason for heightened volatility of operating establishments in com parison to auxiliary establishments concerns the differing firm behavior affecting semi skilled versus highly-skilled workers over the course of the business cycle. With downturns in sales, production workers are more likely to be laid off in comparison to more highly skilled or white collar workers at auxiliary facilities (Williamson, et. al. 1975). If em ployees at auxiliary establishments acquire “firm-specific” skills to a greater extent than production workers at operating establish ments, it will be advantageous for the firm to retain auxiliary workers even when their pres ence is not required by current production levels. If skills are firm-specific and not trans ferable by the employee to other firms, the firm must partly pay for training. Accord ingly, firms will be reluctant to lay off these workers during downturns for fear that they will need to train new workers once economic conditions begin to improve. For the problem at hand, this means that manufacturing activity appears to be more volatile than it actually is because manufactur ing shipments gyrate with the business cycle. However, the presence of auxiliary workers (who tend to be retained during downturns) suggests that actual manufacturing activity (including R&D, strategic planning, etc.) con tinues even while production activities are curtailed. From a geographical perspective, this cyclical reporting bias would tend to be greater at locations of higher auxiliary concen trations where a higher percentage of auxiliary activity fails to be recorded. Evidence to the effect that auxiliary activ ity undergoes milder cyclical swings than overall manufacturing activities can be seen by regressing the share of the nation’s employ ment at auxiliary establishments on the busi ness cycle and other variables: 7 3)' AUX = c + b tT + b gG + b yY where : AUX = current year share of auxiliary employment in total manufactur ing employment. T= annual time trend 1958 to 1986. G= year-over-year percentage growth in U.S. gross domestic product in constant dollars (1982=100). Y= a binary variable; one for census year, zero otherwise. FIGURE 7 Change in auxiliary payroll and GDP growth rate percent change RHO = autoregressive parameter.7 Results of the maximum likelihood esti mation procedure are: c =-2.48 b = 0.0013 b =-0.42 b =-0.0006 (-17.7) (18.1) 8 (-2.3) V (-0.06) RHO = 0.34 (1.72) n = 29 adj. R2 = 0.97 D-W =1.83. We included the binary variable Y for two reasons. During census years, questionnaires are addressed to each establishment while, during non-census years, Annual Survey of Manufactures (ASM) questionnaires are ad dressed to company headquarters. Second, during census years the entire population is observed, while in an ASM year observations are sampled. For these reasons one could argue that these two types of observations would have different results. The regression does not confirm this argu ment. The regression does confirm that there is a significant positive linear relationship between the share of auxiliary employment and time which means that the demand for auxiliary services increases in the long run for total manufacturing. In addition, a significant negative sign for the variable G, a proxy for the business cycle (i.e.,the short run effect), lends support to the hypothesis that business downturns tend to raise the share of manufacturing employment at auxiliaries (see also Figure 7). Apparently, the employees of operating establishments are more likely to be laid off than the employees of the auxiliaries. Thus, in both the long run and the short run, the Census VA may exert a strong re 8 N O TE: Year-to-year difference in share of auxiliary jobs is expressed as a multiple of 1,000. G D P growth rate is expressed as a percent. gional bias relative to the true but unknown manufacturing output. C o rrec tin g th e p ro b lem Since the strong statistical difference be tween the Census and true VA is evident and important, the next question is whether the true VA can be estimated with greater accu racy. Two approaches can be identified. We argue that one of them, already being used, is deficient while the other holds great promise. The Bureau of Economic Analysis, U.S. Dept, of Commerce, attempts to rectify the misapportionment of VA by manipulating ag gregate regional data with national ratios (BEA 1985). However, their methodology to do so can only be correct under some highly stringent assumptions. As their first step, BEA multiplies each state’s VA (for a given industry) by a national factor which nets out the VA contribution made by auxiliary establishments. This adjust ment can only be correct if the proportion of total VA contributed by auxiliaries is identical for each state. In a second stage of estimation, the BEA method re-allocates the nation’s VA of auxilia ries to states, adding it back into the estimated VA of operating establishments. For each in dustry, the method assumes that each state or region has the same relation between auxiliary VA and auxiliary payroll as the nation. Then the re-allocation of VA to states and regions is ECONOMIC PERSPECTIVES performed according to the reported payrolls of auxiliaries of each industry in the state. The key assumption of this second stage, that VA can be spatially allocated in propor tion to payroll for broad industry categories, is not necessarily erroneous. But it is an assump tion that remains untested. Only an analysis using the micro data can validate or reject the second BEA assumption. The deficiencies of using aggregate data strongly suggest the use of Census data at the establishment level to re-compile VA for states and regions. One obvious but unworkable method would be to sum the factor payments at each establishment—both operating and auxiliary establishments alike. (VA is equiva lent to the sum of factor payments including wages, rental, capital costs, and profits.) Un fortunately, this approach must be discarded because several individual data items on factor payments are not gathered by the Census. However, using existing data from the Census, the analysis can be conducted at the company level. The Census collects payroll and other data on each establishment. The Enterprise Statistics Division subsequently combines these data to portray company struc ture. Each company can be viewed as a unit of observation composed of both operating and auxiliary facilities. The true VA for the over all company (and each product) is known from existing data (using the residual method). The remaining problem is to apportion each com pany’s VA according to the contribution of each of its establishments. For the companies with an intricate and integrated structure, the flow of services from auxiliary to operating units may be difficult to determine. This problem is compounded be cause many operating establishments are de fined by a single industry code, yet produce products outside that industry as well. There fore, an auxiliary service provided to an oper ating unit will have to be subdivided into as many components as there are products pro duced by the operating unit. No data series of such detail exists to determine service corre spondence between operating and auxiliary units. However, by combining companies into an industry sample set, one can estimate the relationship between auxiliary and operating units in creating VA using econometric tech niques. Finally, stepping back once again and viewing each establishment separately, data FEDERAL RESERVE BANK OF CHICAGO can be recombined to arrive at better estimates of VA for SMSAs and states according to the locations of their auxiliary and their operating establishments. Im p lic a tio n s fo r reg io n al research A correct accounting of manufacturing output will significantly affect the outcome of current regional research on the existing distri bution of manufacturing in the U.S.; on the importance of manufacturing to regional eco nomic bases; on the movement of manufactur ing activity across regions over time; on re gional productivity differences; and finally, on the determination of the linkages between aux iliary services and operating units located in different regions. To illustrate, a heated debate focuses on whether the nation’s manufacturing sector has been diminishing in recent years. The ques tion has been raised, in particular, for the na tion’s manufacturing intensive regions—espe cially the Midwest (Hill and Negrey 1987; Schnorbus and Giese 1987). As evidenced by declining shares of employment and income, the Midwest has lost a significant share of the nation’s manufacturing activity. However, revised VA may indicate that the losses have been overstated. If, as several studies have suggested, the older industrial belt has retained auxiliary activities even while production operations have decentralized (Jusenius and Ledebur 1976; Giese and Testa 1988), the method by which VA is currently reported would have failed to notice it. Generally speaking, regions which have witnessed a relative decline (or rise) in the share of manufacturing vis a vis other industry sectors probably are understating (or overstat ing) the extent that manufacturing fortunes influence the regional economy. The revised VA may also contribute to a better understanding of the growth process among regions. Some analysts believe that the spread of manufacturing from the North east-Midwest manufacturing belt to outlying U.S. regions has taken place within a “product cycle” process (Norton and Rees 1979).8 His torically, the Northeast-Midwest served as the nation’s innovative center, creating new tech nologically-advanced industries. Over time, in order to economize on costs, these industries have decentralized their routine production op erations to the peripheral regions of the South 9 and West. Initially, growth in peripheral re gions was composed of branch plant open ings—usually production plants of companies headquartered in the Northeast and Midwest Regions. A recent acceleration in manufactur ing growth in peripheral regions may reflect a reversal in regional roles; the Southwest and West finally having reached a critical mass of technology and infrastructure so as to spawn their own high-growth industries. The division between VA attributable to auxiliaries versus operating establishments for each region could be used to test for the changing specializations of regions over time. A more precise measure of output may also change conclusions of papers devoted to measuring regional productivity (Hulten and Schwab 1984; Beeson 1987). While several different measures of productivity have been examined, they all focus on a region’s manu facturing output in relation to inputs such as labor and capital. To the extent that the ob served output trends are not reliable, conclu sions regarding regional performance and competitiveness will not be reliable. Our data, for instance, suggests that productivity in a number of Northeast and Midwestern states is understated, i.e., the numerator, VA, is under estimated, in these studies. One of the more intractable problems in modeling regional economies has been the observation of the economic linkages and trade flows between regions in services. The inter regional flow of goods can be observed from Census of Transportation data while the flow of services cannot. The corporate linkages be tween operating establishments and auxiliaries of manufacturing companies would fill in part of this puzzle. Accordingly, interregional input-output models, which attempt to exam ine the economic linkages across regions, could be specified more fully. Estimated rela tionships can be expressed in the form of ex ports flowing from regions with auxiliary services to regions with operating units. This information can be incorporated into the multiregional input-output model, which would allow analysts to estimate the effect of the change in the output of the operating units for one region on the auxiliary employment for another region. In a broader context, observing whether these operating-auxiliary linkages are increas ing in distance over time would reflect on the question of whether, because of enhanced transportation and communication ability, the service sector can be thought of as an “export base” for regions. Over time, are regions with specialized service sectors serving customers that are farther and farther apart? FOOTNOTES 1Am oco’s activities are also large in energy exploration and development. These activities constitute value added in the mining, services, and other sectors. 2Another problem, which we will not address in this essay, concerns the fact that this Census Bureau definition of VA also includes the value of services purchased by the manu facturing company from either outside service companies or other manufacturers. Also, the Census does not subtract the materials costs of auxiliary establishments. Both of these practices lend an upward bias to the Census concept o f VA.3 3Others have taken up the possible biases in the national measures o f manufacturing output (Mishel 1988). Mishel argues that manufacturing growth has been overstated at the national level by the BEA. This results from a failure to properly deflate the value of intermediate components in manufacturing over time. Foreign-source components are routinely deflated by a domestic price deflator— a proce dure that Mishel believes has understated the foreign con 10 tent of domestically manufactured goods and concurrently overstated the value of domestic manufacturing activity. 4With existing data collection procedures, distinguishing auxiliaries from similar activities that take place at operat ing establishments is somewhat capricious. Often, by the choice of the survey respondents, auxiliary activities that take place at the same site as the operating establishment can be combined and reported as one. In this paper, we single out auxiliary establishments because they are most likely to be located at different sites from operating estab lishments; the nature o f the problem is most easily commu nicated by making the auxiliary versus non-auxiliary dis tinction. However, a skewed distribution of support activi ties versus operating establishments of multi-plant manu facturing companies across the U.S. would result in the same problem. Support services are often located at the same site as production activities. 5Here are the summary statistics for states and SMSAs in 1982: ECONOMIC PERSPECTIVES Auxiliary payroll / total payroll Low Mean Std. deviation High n States 0.083 0.080 0.498 0 46 SMSAs 0.099 0.088 0.534 0.005 172 ‘For formal testing of the equality between ba and b t coefficients, we proceed as follows. Equation 2 can be rewritten as: V = c + ba(A + O) + yoO = c + baA + (ba + yo )0 . It is obvious that the equality between two coefficients cannot be rejected if yo is insignificantly different from zero. [See Pindyck and Rubinfeld (1981)]. In both SMSAs and states yg had t-statistics o f 30 and 23 respectively, which strongly rejects the hypothesis o f equality between two coefficients in both cases. 7OLS estimation resulted in a D-W statistic of 1.33, falling within the uncertain region. After first-order correction for serial correlation, the D-W statistic was 1.83. 8Some analysts have long maintained that regional econo mies can be understood by focusing on “export base”, the key industries for which the region produces and trades with the rest of the nation or world. Typically, the export base has comprised manufacturing, mining, and agriculture although many service sectors are now also receiving such recognition. For seminal discussions see Andrews (1953), North (1955), and Tiebout (1956). REFERENCES Andrews, Richard B., “Mechanics of the Ur ban Economic Base,” Land Economics, Vol. 29, 1953, pp. 161-167. Beeson, Patricia, “Total Factor Productivity Growth and Agglomeration Economies in Manufacturing, 1959-73,” Journal of Regional Science, Vol. 27, 1987, pp. 183-199. Bolton, Roger, “Review of Literature on Re gional Econometric Models and Regional Business Cycles,” working paper, Williams College, Williamstown, Mass., 1978. Borts, George H., “Regional Cycles of Manu facturing Employment in the United States, 1941-1953,” Journal of the American Statisti cal Association, Vol. 55, 1960, pp. 151-211. Garnick, Daniel H., “The Regional Statistics System” in Modeling the Multiregional Eco nomic System, F. Gerard Adams and Norman J. Glickman, eds., Lexington Books, Lexing ton, Mass., 1979, pp. 25-55. Giese, Alenka S., and William A. Testa, “Can Industrial R&D Survive the Decline of Production Activity: A Case Study of the Chi cago Area,” Economic Development Quar terly, Vol. 2, 1988, pp. 326-338. Hill, Richard Child and Cynthia Negrey, “Deindustrialization in the Great Lakes,” Ur ban Affairs Quarterly, Vol. 22, 1987, pp. 580-597. FEDERAL RESERVE BANK OF CHICAGO Howells, J.R.L., “The Location of Research and Development: Some Observations and Evidence from Britain,” Regional Studies, Vol. 18, 1984, pp. 13-29. Hulten, Charles R., and Robert M. Schwab, “Regional Productivity Growth in U.S. Manu facturing: 1951-78,” American Economic Re view, Vol. 74, 1984, pp. 152-161. Jusenius, Carol L., and Larry C. Ledebur, A Myth in the Making: The Southern Economic Challenge and Northern Economic Decline, Washington D.C., U.S. Dept, of Commerce, Economic Development Administration, No vember 1976. Malecki, Edward J., “Industrial Location and Corporate Organization in High Technology Industries,” Economic Geography, Vol. 61, 1985, pp. 345-69. Mishel, Lawrence R., Manufacturing Num bers, Economic Policy Institute, Washington, D.C.,1988. North, Douglas C., “Location Theory and Regional Economic Growth,” Journal of Po litical Economy, Vol. 63, 1955, pp. 243-58. Norton, R.D., and J. Rees, “The Product Cycle and the Spatial Decentralization of American Manufacturing,” Regional Studies, Vol. 13, 1979, pp. 141-151. 11 Pindyck, R.S., and D.L. Rubinfeld, Econom etric Models and Economic Forecasts, McGraw-Hill, New York, 1981. Schnorbus, Robert H., and Alenka S. Giese, “Is the Seventh District’s Economy Deindus trializing?,” Federal Reserve Bank of Chicago, Economic Perspectives, Vol. 11, No. 6, November/December 1987, pp. 3-9. Tiebout, Charles M., “Exports and Regional Economic Growth,” Journal of Political Econ omy, Vol. 64, 1956, pp. 160-64. U.S. Department of Commerce, Bureau of the Census, Annual Survey of Manufactures, 1959-62, 1964-66, 1968-71, 1973-76, 1978-81, 1983-86, U.S. Government Printing Office, Washington D.C. 12 U.S. Department of Commerce, Bureau of the Census, Census of Manufactures, 1958, 1963, 1967, 1972, 1977, 1982, U.S., Govern ment Printing Office, Washington, D.C. U.S. Department of Commerce, Bureau of Economic Analysis, Experimental Estimates of Gross State Product by Industry, Bureau of Economic Analysis Staff Paper 42, BEASP85-042, U.S. Dept, of Commerce, May 1985. Williamson, O.E., M.L. Wachter, and J. Harris, “Understanding the Employment Relation: The Analysis of Idiosyncratic Exchange,” The Bell Journal of Economics, Vol. 6, 1975, pp. 250-278. ECONOMIC PERSPECTIVES 2 5 th C o n fe ren c e on B an k S tru c tu re a n d C o m p e titio n : C o n t r o llin g ris k in fin a n c ia l s e rv ic e s M ary J. W illiam son Risk management has always been a major challenge for the financial services industry. Today, however, the increas ing number of failures of distressed depository institutions seems to indicate that managing risk has become more difficult. At the 25th annual Conference on Bank Structure and Competition, sponsored by the Federal Reserve Bank of Chicago, several industry leaders discussed their recommenda tions for controlling risk in today’s environ ment. These participants shared several points of emphasis and presented some personal concerns about regulation, supervisory inter vention, and deposit insurance. D iffe re n t p ersp ectives The panelists were in practical agreement about the fundamental issues affecting the industry, and all agreed that regulation has been used excessively to control risk. Each, however, had a different perspective on risk and, therefore, advocated different approaches for managing it. “Banking by definition is the management of risk,” began Federal Reserve Board gover nor John LaWare. This ex-banker-tumedregulator said that he resents the underlying assumption inherent in the regulatory structure that bankers do not know as much as legisla tors or regulators about how to run a bank. This false assumption has fostered excessive regulation and has created an anti-competitive atmosphere, said LaWare. He added, “it is increasingly creating a disadvantage for the FEDERAL RESERVE BANK OF CHICAGO American banking system in world markets, to say nothing about domestic markets.” Accord ing to LaWare, “supervision, rather than regu lation, ought to be the focus” for controlling bank risk. Continental Bank Corporation chairman Thomas Theobald agreed with LaWare that regulation has gone too far. Taking a broad perspective on the future of the financial serv ices industry, Theobald said that the business of banking will likely undergo “colossal re structuring,” but it is not appropriate for “cen tral planners,” i.e., legislators and regulators, to decide “the finer points” of the restructur ing. “I don’t think . . . a sincerely motivated, highly intelligent, nationally interested bunch of people in Washington . . . are going to be able to design the proper response to these changes.” Rather, according to Theobald, those decisions belong with the market partici pants—the consumers and the producers of fi nancial services. Early in the Conference, Carter H. Golembe, chairman and managing director of The Secura Group, asked, “Why is the market so distrusted as an efficient regulator of bank ing?” He conjectured that the reasons are that first, history has painted American banking during the first century and a half as “a chaotic black hole that was cured only by the estab lishment of the Federal Reserve System . . . and federal deposit insurance;” and second, “the market can be a brutal regulator.” Mary J. W illiam son is deputy librarian at the Federal Reserve Bank of Chicago 13 Federal Home Loan Bank Board member Lawrence J. White said that “depositories are special.” According to White, their liabilities are special, and that is why they are insured and why controlling the risk of depository institutions is so important. But, like the other panelists, White did not advocate regulation as a primary tool to control risk. Rather, White preferred risk-based capital requirements and risk-based deposit insurance premiums as well as better and earlier supervisory intervention. R eg u lation and re-re g u latio n Regulation is one approach to controlling risk, and according to the panelists, it is the approach most often used—and overused—in the banking and thrift industries. Said White, we regulate “with a vengeance.” Many regu lations, originally designed to protect the safety and soundness of the financial system, now are considered by some to be outmoded, anticompetitive, and too stringent. Furthermore, Theobald pointed out that regulations do not always work as planned. He noted that the thrift industry has “just man aged to lose $100 to $200 billion in a beauti fully regulated business.” He added that this loss is greater than the cost of all the land acquisitions throughout the history of the American republic. LaWare said that regulations can create inefficiencies and used the interstate banking formula as an example. He asked why banks operating in a multi-state environment should be burdened with the operating restrictions of each state in which they operate. LaWare contemplated the possibility of interstate bank holding companies operating under one set of federal rules. This, he said, could stimulate managerial and operating efficiencies rather than replicate the whole regulatory structure in each state. While all panelists agreed that regulation is not the best way to control risk, LaWare expressed serious concern that the thrift crisis, bank failures, and scandals in the investment banking industry “have created a counterbal ance to what was beginning to be a very healthy tendency on the part of Congress to deregulate the financial industry. . . .What we do not need now is a re-regulation binge.” Paul Horvitz, professor of banking and finance at the University of Houston, observed at the Conference that both the regulated and regula 14 tors have learned from their mistakes and that, given the proper incentive, these human errors will not be repeated. Nevertheless, Horvitz emphasized that the regulatory system does need some reforming, although not extensive restructuring. S upervision and in te rv e n tio n Rather than regulation, said LaWare, “we need intelligent supervision doing an in-depth job of monitoring what is going on in all these institutions and the authority to move quickly and peremptorily when something goes wrong.” Supervisory attention should concen trate on institutions that threaten the insurance system. LaWare emphasized aggressive moni toring and authority to intervene quickly to change the course of action. Fellow Federal Reserve Board governor, Manuel Johnson, earlier had said “to prevent problem banks from becoming threats to the safety net and the financial system, it is necessary to give exam iners stronger tools.” Rather than legislate against risky behav ior, which would constitute credit rationing and asset allocation, LaWare recommended improvement in the supervision of banks. For example, LaWare suggested that examiners of financial institutions that are involved in highly leveraged finances need to determine that the proper credit policies are in place and that limits on the proportion of the portfolio that can be dedicated to this kind of lending have been established. As Joseph A. Manganello, Jr., an executive vice president at Bankers Trust Company, said, “Don’t make the same bet in your whole portfolio.” In addition, directors should be informed and approve what is going on so that there is some feeling that there is control over the risk. This method is more effective than legislation, which is inflexible and hard to manage, con cluded LaWare. In fo rm a tio n system s White agreed that there is a need to strengthen the ability of regulators to intervene before an institution becomes insolvent. Insur ance losses would decrease if supervisory authorities could force recapitalization before insolvency and subsequent loss to the deposit insurance corporations occurred. Accurate information, however, is crucial to early intervention. Current information ECONOMIC PERSPECTIVES systems make it difficult to detect risk expo sure. In fact, financial reporting is based on accounting methods that do not necessarily provide an adequate assessment of present conditions or the value of assets. White, a strong advocate of market value accounting, said that relying on generally accepted ac counting principles (GAAP) for banks may indicate financial soundness when market value measures would indicate otherwise. For example, book value measures of capital can be a very misleading measure of an institu tion’s ability to absorb losses. George Benston, professor of finance, accounting, and economics at Emory Univer sity, said that “the accounting system was not and is not designed to present economic values that regulators, economists, and investors might use. . . . It’s to control the use of re sources, particularly cash.” Yet, a crucial piece of information for controlling risk and learning about risk is market information. According to Benston, market value account ing is generally difficult to do, “but not for banks” because of the nature of banks’ assets and liabilities. “There really is no substitute for market value accounting,” said White. Although initially “it won’t be perfect,” it would be “a whole lot better than GAAP ac counting.” GAAP is inadequate and will be come increasingly divorced from economic reality, said White. Insurers and regulators need to have a better idea, even if approxi mate, of the market value of the assets and liabilities of financial institutions. James Annable, chief economist at First National Bank of Chicago, said, however, that information between the regulator and the regulated is so unbalanced that “a cost-effec tive regulatory process may not be possible to design.” Therefore, deregulation may be the best alternative. R isk-based c a p ita l and insurance p rem iu m s In the sense that capital is akin to an insur ance deductible, risk-based capital require ments and deposit insurance premiums go hand-in-hand. As White pointed out, “every auto insurance company in the land will charge a lower premium . . . if you take out a larger deductible. And the same principle ought to apply to deposit insurance premi ums.” These two means of controlling risk FEDERAL RESERVE BANK OF CHICAGO were discussed by the panelists and strongly advocated by White. “Capital is going to be the focus of man aging risk in the financial industries,” pointed out LaWare. Capital adequacy has played a central role in controlling the risk of individual institutions because capital protects the deposit insurance funds by reducing any incentives to take risks. The definitions of capital and acceptable capital requirements are frequently modified and studied by the regulators, and the need to reform and substantially tighten capital re quirements has been acknowledged throughout the industry. Recent risk-based capital guide lines, which incorporate off-balance-sheet items into the capital requirements, are cer tainly a step in the right direction. Theobald observed, however, that the financial services industry is overcapitalized, while some individual institutions are under capitalized. The banking industry has never earned more than 10 percent on equity capital, while the rest of American industry is earning 15 to 18 percent. “This is an unsustainable situation,” said Theobald. “Now I understand that the regulators want to see more capital, but I think what they really want to see is more capital per enterprise. . . .You can’t say you want more capital in the industry when it’s already earning a nonmarket clearing return.” While more capital would lead to a lower premium under a typical insurance scheme, deposit insurance is not typical in that all insti tutions are charged a flat rate. Therefore, the current system overprotects some depositors, while it encourages other institutions to take on higher risks. White commented that he finds it “absurd that the [deposit] insurers do not and cannot charge premiums that are also risk-based.” White also said that practicing co-insur ance, i.e., cutting back on coverage, is fine if bank runs are not a problem. He said, how ever, that he believes in 100-percent deposit coverage and employing other tools to control risk. Theobald disagreed: “What started off as a life vest is now a luxury yacht. We need to limit the deposit insurance . . . I submit that there is no logic that will get you away from the fact that if we don’t limit deposit insurance we’re going to forever be fighting futile cen tral planning of the financial business.” 15 C o m p etitive n es s Theoretically, restrictions on financial activity prevent financial institutions from taking excessive risk. In practice, however, these restrictions increase risk when they pre vent institutions from adapting to the changing needs of their customers. One type of restric tion is the “firewall,” which legally and opera tionally separates banking activities of a hold ing company from nonbanking activities. “Firewalls that are too high can indeed create risks and inefficiencies, rather than minimize them,” said Dennis Weatherstone, president of J.P. Morgan & Company, during the Conference. Referring to investment and commercial banking, he said “the business we do today weaves the two together so tightly that we really have to rip the fabric to separate the threads.” Nevertheless, firewalls require that an investment banking subsidiary and a commercial banking affiliate maintain “sepa rate capital, different people, and duplicate support functions.” Manuel Johnson conceded that “firewalls will lead to some sacrifice of synergies,” but he said that firewalls are neces sary to protect the safety net. LaWare addressed the issue of expanded powers in light of one aspect of the safety net, 16 deposit insurance. He said that he supports the idea of a financial services holding company. If insured banks are isolated from nonbank affiliates, LaWare noted, there should be no limit to other businesses those affiliates could get into. In particular, LaWare said, as many others have, that such financial activities as insurance, real estate, and securities are appro priate for financial services holding compa nies. But LaWare added, “an industrial corpo ration cannot own a bank and a bank cannot own an industrial corporation.” This separation of commerce and banking needs to be reexamined. There may be better and cheaper access to capital markets by com bining the two. The outcome of the current debate over controlling risk will significantly affect the strength of financial organizations in the years to come. Fundamental reform is needed for insuring deposits and regulating financial institutions. The ongoing appraisal of all risks facing the management of bank funds regardless of size and status is an impor tant priority. The panelists agreed that the fi nancial industry must adapt information, regu lation, and supervisory controls to avoid unrea sonable and excessive risk. ECONOMIC PERSPECTIVES P u b lic in v e s tm e n t and p r o d u c tiv ity g ro w th in th e G ro u p o f S e v e n A general shift in government spending priorities— from capital investment to consumption— has negatively affected productivity in the G-7 industrial countries D avid A . A schauer Public policies to promote economic growth and interna tional competitiveness have traditionally been focused on savings and private invest ment in plant and equipment. And with good reason. In the words of Martin Feldstein, “an increase in the saving rate is the key to a higher rate of economic growth and a faster rise in the nation’s standard of living. . . . [T]he evidence is overwhelming that countries with high rates of saving and investment are the ones in which productivity, income and the standard of living rise most rapidly.”1 Such a focus leads to specific policy ini tiatives to boost the national savings rate as well as to stimulate private capital accumula tion. Among these initiatives are consump tion-based tax systems, individual retirement accounts, preferential tax treatment of long term capital gains, accelerated depreciation of physical capital assets, and investment tax credits. While economists quibble about the quantitative importance of these savings and investment incentives, they are in near unani mous agreement on their qualitative signifi cance for economic growth. However, there is another potential “sup ply-side” avenue by which public policy may be able to exert significant influence on the process of sustained economic expansion. What the above policies have in common is that they work through the tax system to affect either the supply of loan funds—savings—or the demand for those funds—private invest B FEDERAL RESERVE BANK OF CHICAGO ment in capital goods. Instead, we might look to the opposite side of the government’s budget, at the composition of public expendi ture and the possible effects various budget policies may have on private sector productiv ity and economic growth. In this paper, I distinguish between the public consumption and public investment and argue that this distinction is as important for economic growth calculations as the analogous calculation on the private side of the economy. Public nonmilitary investment—which I take as a proxy for a public infrastructure of roads, highways, mass transit, airports, port facilities, and the like—is argued to have positive direct and indirect effects on private sector output and productivity growth. The direct effect on private sector output growth arises from the availability of public capital to support private sector production; roads, highways, and airports allow the distri bution of goods and services throughout na tional and international markets. The indirect effect evolves from the complementarity be tween private and public capital in privatesector productive activity; an increase in the stock of public capital raises the return to private capital which, in turn, serves to spur the rate of expansion of the private-sector capital stock.2 Supporting these arguments, I offer empirical evidence of a positive effect of public investment on private investment and private output growth. David A. Aschauer is a senior econom ist at the Federal Reserve Bank of Chicago. 17 T rend s in p u b lic e xp e n d itu re In all the Group of Seven (G-7) industrial ized countries, the growth in gross domestic product (GDP) per employed person—labor productivity growth—has fallen over the last twenty years. Productivity growth for these countries taken together averaged 4.0 percent per year during 1960-68, 3.2 percent during 1968-73, 1.4 percent during 1973-79, and 1.5 percent during 1979-86. In each of the G-7 countries, productivity growth during the 1970s and 1980s was some 50 percent less than that attained during the 1960s. At the same time, there was wide dispersion in aver age productivity growth across these countries. For instance, between 1960 and 1986, Japan achieved a productivity growth rate of 5.5 percent per year, West Germany one of 3.2 percent per year, and the United States one of only 1.2 percent per year. Figure 1 depicts trends in public net (of depreciation) investment during the years 1967 to 1985 for the major industrialized econo mies.3 Three broad features stand out. First, in five of the seven countries, the ratio of pub lic investment spending to gross domestic product trended downward; in the United States (from 1.7 percent of GDP in 1967 to 0.3 percent by 1985), in West Germany (from 3.1 percent to 1.5 percent), in France (from 3.5 percent to 1.6 percent), in the United Kingdom (from 3.9 percent to 0.7 percent), and in Can ada (from 3.1 percent to 1.0 percent). In Ja pan, public investment as a share of GDP rose from 3.8 percent in 1967 to 4.1 percent in 1985, peaking at 5.8 percent in 1979. In Italy, public investment climbed from 2.8 percent in 1971 to 3.3 percent in 1983 and then declined slightly to 3.1 percent in 1985. Second, there exists fairly wide differ ences in some of the public investment ratios across countries. While public investment absorbed some 5.1 percent of gross output in Japan over this time period, the United States devoted a much smaller output share to up grading its public capital stock, less than 1.0 percent. In between are to be found the Euro pean countries of France, Italy, the United Kingdom, and West Germany along with Can ada. Finally, there seems to be no pursuit of countercyclical public works policies; for example, in the United States the public in vestment ratio was 0.7 percent in 1973 and 1974, 0.6 percent in 1975 and 0.4 percent in 1976 while it was 0.3 percent in 1980, falling to 0.1 percent in 1981 and 1982. On the other hand, no downward shift in government consumption spending—inclusive FIGURE 1 Public investment as a share of gross domestic investment: 1967-1985 percent of GDP 18 ECONOMIC PERSPECTIVES of military spending—is apparent in the data for these countries. As can be seen in Figure 2, the ratios of public consumption to gross domestic product rose in all countries, with the exception of the United States, and in most cases by 2 or 3 percentage points. In the United States, no clear trend is readily discern ible, although public consumption was close to one percentage point lower in 1985 than it had been in 1967. These statistics paint an interesting picture of government spending priorities in the G-7 countries over the roughly twenty-year period from 1967 to 1985. Generally speaking, while public investment slid downward, public con sumption climbed. What, if any, effect might this alteration in government budget shares have had on output and productivity growth across these countries? I argue that public capital—particularly infrastructure capital investments such as roads, highways, dams, water and sewer systems, mass transit, airport facilities, and the like—is a vital input to the private production process. If this is the case, then the general shift in budget priorities away from capital accumulation toward consump tion may offer a partial explanation for the productivity decline experienced by the G-7 industrial economies. M e th o d o lo g y I assume a neoclassical production tech nology whereby private sector output is ob tained by application of labor services to pri vate and public capital stocks. As shown in the box, this framework leads to the following regression equation = b0 + b, * Dn, + b2 * ir, , + b3 * gir, , + b4 *D cu t where: Dp, Dp, = labor productivity growth; Dn, = em ployment growth; ir,, = ratio of private net investment to gross domestic product (lagged one year); gir, , = ratio of public nonmilitary net investment (also lagged); and Dcu, = rate of change in capacity utilization. According to standard restrictions on the production func tion, we expect b} to be estimated negatively. Simply stated, the application of more laborers to given quantities of private and public capi tal stocks lowers the productivity of labor. On the other hand, given the number of workers, raising the amounts of private or public capital should, on average, make each worker more productive, so we also expect b2 and b1to be estimated positively. As labor productivity growth is highly procyclical—rising in booms and falling in recessions—it is likely we will FIGURE 2 Public consumption as a share of gross domestic product: 1967-1985 percent of GDP FEDERAL RESERVE BANK OF CHICAGO 19 Estimating productivity growth In algebraic form, we have the production technology The relationship between the two variables is given by y, = f(ne k,-p k§i-r cu.) ir = (k/y)*Dk where: where ir = ratio o f (private) investment to gross output. As long as the capital-to-output ratio, kty, is fairly stable the ratio o f investment spending to output, ir, will be a good proxy for growth in the capital stock. The obvious extension o f the public side is left undiscussed. We finally write the equation to be estimated empirically as yt = private sector output during year t; nt = em ployment during the same year; kt ) = the private capital stock at the beginning of year t; kgt = the public nonmilitary capital stock also as o f the start of year t; and cut = the rate o f utilization o f capacity in production. This last variable is entered to cap ture shocks to the production technology as well as to convert capital stocks into flo w s o f capital serv ices. Unfortunately separate estimates o f private and public capital stocks are currently unavailable for the Group o f Seven industrial nations; however, we can finesse this data deficiency by shifting the emphasis from the level o f production to the growth in production. First, by assuming a loga rithmic form for the production technology we may derive the expression D yt = a0+ a,*Dnt + a2*Dk( + a3*Dkgt + a4*Dcut where: Dxt denotes the percentage growth rate o f variable x during period t. In this form, we can employ a proxy for growth in capital stocks, i.e., the ratio of investment, private and public, to gross output. find b4 is positive. We now confront the data with the above equation to see if they perform according to our theoretical expectations. E m p irical results I estimated the equation on data gathered for the Group of Seven countries over the period 1966 to 1985. Detail on these data are given in the Appendix. In general, the data provide strong support for the idea that public investment is a critical determinant of labor productivity growth. An increase in the level of public nonmilitary investment by one per cent of gross output yields a gain in productive growth of about 0.4 percent per year. The strong positive relationship between public investment and productivity growth is robust to changes in the set of countries included in the data sample and after consideration of the effects of oil shocks in the 1970s. 20 Dpt = b0 + bj*Dn( + b2*ir , + b3*girt , + b4*Dcut where: Dpt = D y -D n t = labor productivity growth and so b, = (a ,- l). Under the standard assumptions o f a positive but diminishing marginal product of labor, we expect to find bt to be negative. We also as sume a complementarity between labor and the services o f private and public capital stocks. Thus, by raising the stocks o f either private or public capital— given labor input— the productivity of labor should be boosted, so we expect b2 and b3 to be positive. Further, it is likely that the capacity utilization rate— proxying for technological shocks as well as converting capital stocks into flows o f capital services— will enter the final expression positively. Table 1 contains the basic set of estimated relationships between the level of public in vestment and productivity growth. The public investment variable is exclusive of military capital expenditures; is expressed relative to the level of gross domestic product; and is lagged one period. I believe this variable to be a good proxy for the percentage growth in the nonmilitary public capital stock during the previous period. The productivity growth variable measures labor productivity growth as the percentage growth rate of gross domestic output per employed person in each of the Group of Seven industrialized economies. Column 1 of Table 1 illustrates the strength of the independent effect of public investment on the growth rate of labor produc tivity. A one-percentage-point increase in the share of GDP devoted to public capital accu mulation is associated with a 0.73-percentage- ECONOMIC PERSPECTIVES TABLE 1 Public investment and productivity growth in the Group of Seven (dependent variable—Dp) 1 2 3 4 5 6 c 0.68 (0.41) -0 .2 1 (0.41) 0.02 (0.66) -0.33 ( - 0.46) -0 .2 1 (0.39) 3.02 (1.63) g ir 0.73 (0.14) 0.44 (0.13) 0.59 (0.18) 0.51 (0.21) 0.41 (0.13) 0.34 (0.14) 0.22 (0.06) 0.13 (0.07) 0.20 (0.08) 0.24 (0.05) 0.12 (0.07) -0 .3 5 (0.08) -0 .2 9 (0.09) -0 .6 4 (0.17) -0 .3 2 (0.08) - 0 .3 5 (0.08) 1.61 (0.15) 1.28 (0.16) 1.67 (0.21) 1.58 (0.14) 1.51 (0.15) ir Dn Dcu d74 -1.83 (0.60) d79 -1 .2 6 (0.60) - 0 .1 3 (0.06) gcr R2 0.17 0.58 0.46 0.48 0.61 0.59 SER 2.21 1.57 1.46 1.47 1.51 1.55 NOB 129 129 91 72 129 129 C o lum n C o lum n C o lum n C o lum n C o lum n C o lum n 1 d isp la ys th e basic re la tio n sh ip betw een p u blic in ve stm e n t and p ro d u c tiv ity g ro w th . 2 is th e basic eq u a tio n in the text. 3 excludes Japan and th e U nited States fro m th e sam ple. 4 excludes Japan, th e U nited States, and Canada fro m the sam ple. 5 a llo w s d u m m y variab les to capture th e effects o f oil shocks. 6 a llo w s a separate effect o f g o v e rn m e n t co n su m p tio n spending. N O T E: F ig u re s in p a re n th e s e s rep re se n t th e sta n d a rd error. point rise in the labor productivity growth rate. The standard error of 0.14 yields a ninety-five percent confidence interval which lies well above zero, namely (0.45, 1.01). The public investment variable alone is capable of ex plaining 17 percent of the variation in produc tivity growth across time and countries. Column 2 expands the list of variables allowed to influence productivity growth to include private investment, growth in total employment, and capacity utilization. As with the public investment variable, private invest ment is expressed relative to GDP and is lagged one year to proxy for previous growth in the private capital stock. The capacity utilization variable is entered in the attempt to convert growth in the stocks of public and private capital (captured by gir and ir, respec FEDERAL RESERVE BANK OF CHICAGO tively) into service flows from these stocks. While the estimated coefficient on public investment is markedly reduced—from 0.73 to 0.44— it still is statistically significant at better than a ninety-nine percent level. The private investment variable enters positively, suggest ing that a one-percentage-point increase in the ratio of private capital accumulation to gross domestic product will raise productivity growth by an amount equal to nearly onequarter of a percentage point. Consistent with the expectation of a diminishing marginal productivity of labor, a one-percentage-point increase in the rate of growth of total employ ment lowers the rate of growth of labor pro ductivity by somewhat more than one-third of a percentage point. Within the organizing context of a Cobb-Douglas production technol 21 ogy, the coefficient on total employment should equal unity minus labor’s share in gross domestic product; the estimated coefficient therefore suggests that labor’s output share was some 65 percent—a reasonable estimate.4 Finally, as expected, the capacity utilization variable bears a positive relationship with productivity growth. Columns 3 and 4 of Table 1 exhibit the robustness of the estimated relationship by limiting the samples to exclude the United States and Japan (Column 3) and to include only the four major European economies (Col umn 4). Excluding the United States and Ja pan—the countries with the lowest and highest public investment ratios during this period— does not erode the relationship between public investment and productivity; indeed, the esti mated coefficient on public investment is increased from 0.44 in the full sample to 0.59 in the limited sample. There is a sizable re duction in the coefficient associated with pri vate investment, however, and the adjusted coefficient of determination is reduced from 58 percent to 46 percent. Focusing on the European countries of France, Italy, the United Kingdom, and West Germany, the relationship between public investment and productivity growth remains significantly positive, although the estimated standard error of the coefficient rises by a non-trivial amount. The period of analysis, 1966 to 1985, includes years in which there were significant “supply-side” disruptions to production in the highly industrialized economies. Most obvi ous are the oil price shocks of late 1973 and 1979. Column 5 allows for the separate ef fects of these oil price shocks by including dummy variables for 1974 (the first year in which the effect of the first major oil price shock would be apparent) and 1979. As ex pected, the dummy variables are significantly negative, indicating that productivity growth fell by more in those years than can be ex plained by the private capital and public in vestment variables and employment growth. The estimated coefficients on these latter vari ables, however, are not altered in an important way from those in Column 2 and the adjusted coefficient of determination rises only a small amount, from 58 percent to 61 percent. Column 6 illustrates that the ratio of gov ernment consumption—measured residually by subtracting public investment from total government spending on goods and services— bears a marginally significant negative rela tionship with productivity growth. A onepercentage-point increase in the share of gross domestic product devoted to government con sumption is estimated to reduce labor produc tivity growth by somewhat more than onetenth of a percentage point. Note that this result, in conjunction with the positive associa tion between productivity growth and public investment, indicates that countries should be able to achieve substantial productivity gains by holding fixed their tax revenues and alter ing the composition of government spending away from public consumption and toward public nonmilitary capital accumulation. Thus, the results of Table 1 are fully com patible with the idea that public investment is a necessary input to the private production process. Without sufficient investment in a public infrastructure of roads, local transporta tion, airports, and port facilities, the task of private-sector production becomes much more exacting in terms of sacrifice of either current consumption or leisure activities. Of course, this is not the only possible explanation for the positive association of public investment and labor productivity. One could argue, for example, that the statistical correlation is the reverse—that public invest ment slumps in periods of low productivity and (presumed) reductions in tax revenues and is stepped up in times of prosperity and more generous growth in revenues. In econo mists’ language, public investment would be considered a “normal” good. This argument, however, has a number of hurdles that it must clear. First, the public (and private) investment variable is lagged one year. Statistically, it is therefore a predetermined variable; this re duces the force of the reverse causation argu ment to some degree. Second, as Column 6 indicates, while there is a positive association between public investment and productivity, there is a negative association between public consumption and productivity. The counterar gument thus must explain why public con sumption, unlike public investment, appears to be an inferior good. Third, the estimated coef ficients in Column 2 are all of the right sign and of a reasonable economic magnitude from 22 ECONOMIC PERSPECTIVES a technological standpoint; it seems unlikely that this is a mere happenstance. Finally, the results in Table 2 provide more concrete evidence against the reverse causation hypothesis. In these equations, the public investment variable has been purged of its direct relationship with the level of eco nomic activity by prior regression on the rate of growth of gross domestic product. The residuals from this estimated equation are then used in place of the “raw” public investment variable in the regressions reported in Table 2. Column 1 shows the simple relationship be tween productivity growth and public invest ment, purged of its income growth component, to be statistically strong and positive. Column 2 allows for the additional effects of private investment, employment, and capacity utiliza tion. As in Table 1, the relationship between public investment and labor productivity growth is attenuated but still of quantitative and statistical importance. Column 3 allows for dummy variables for 1974 and 1979 with only a minor change from the results of Col umn 2. In Column 4, private investment is also purged of its direct association with out put growth, with the result a significantly lower estimated relationship between private investment and growth in output per employed person. Finally, Column 5 adds in the ratio of public consumption to GDP. As with the re sults in Table 1, the estimated relationship between productivity growth and the share of government O TABLE 2 consumption in gross output is negative, but now at a consid Cyclically adjusted investment and productivity growth in the Group of Seven erably diminished level of (dependent variable—Dp) statistical significance. Table 3 contains reduced 1 2 3 4 5 form estimates of the relation ship between private invest 2.34 0.62 0.54 c 2.51 2.88 (0.20) (0.44) (0.43) (0.17) (1.62) ment, public investment, and 0.72 0.42 0.38 0.53 0.37 public consumption over the gir (0.11) (0.12) (0.13) (0.11) (0.11) same sample. Column 1 shows ir 0.23 0.25 0.14 0.15 a rise in public investment of 1 (0.05) (0.05) (0.06) (0.01) percent of gross domestic prod Dn -0 .2 9 -0 .2 7 -0.21 -0 .3 0 uct is associated with an in (0.08) (0.08) (0.09) (0.08) crease in total investment (pub 1.54 1.51 1.46 1.48 Dcu lic plus private) of 2.5 percent (0.15) (0.15) (0.16) (0.15) age points, or an increase in d74 -1.65 private investment of 1.5 per (0.60) cent of output. Column 2 cal d79 -1.11 culates that a rise in govern (0.59) ment consumption of one per gcr -0 .0 9 cent of gross output depresses (0.07) national investment by 0.59 of a percentage point. The effect R2 0.21 0.59 0.61 0.53 0.59 of public investment on na 2.14 1.64 SER 1.55 1.49 1.54 tional investment is reduced NOB 121 121 121 121 121 substantially, from 2.5 to 1.4 Column 1 displays the basic relationship between cyclically percentage points. This last adjusted public investm ent and productivity growth. result is due, no doubt, to the Column 2 is the basic equation in the text w ith cyclically strong negative relationship adusted public investment. between public investment and Column 3 allow s dum m y variables to capture the effects of oil shocks. consumption and associated Column 4 is the basic equation w ith cyclically adjusted omitted variable bias in Col private and public investment. umn 1. Columns 3 and 4 repeat Column 5 allow s a separate effect of governm ent consum p the previous regressions but tion spending. with public and total investNOTE: Figures in parentheses represent the standard error. FEDERAL RESERVE BANK OF CHICAGO 23 C onclusion TABLE 3 Public and private investment (dependent variable—ir) 1 2 3 4 c 5.04 (0.46) 17.46 (1.34) -0 .0 6 (0.21) 6.20 (0.98) gir 2.50 (0.16) 1.40 (0.17) 2.27 (0.15) 1.66 (0.16) gcr -0.38 (0.06) -0.59 (0.06) R2 0.65 0.79 0.65 0.74 SER 2.58 1.98 2.28 1.97 NOB 129 129 129 129 Column 1 shows the basic relationship between public and private investment. Column 2 displays a separate effect of goverm ent consum ption. Columns 3 and 4 duplicate Columns 1 and 2, but w ith cyclically adjusted investment. NOTE: Figures in parentheses represent the standard error. ment ratios which are purged of their correla tion with the growth rate of gross domestic product. As can be seen, the positive associa tion of national investment with public invest ment and the negative relationship with public consumption is maintained. There exists a strong, positive correlation between various productivity measures and public nonmilitary capital expenditure. Aschauer (1988) has established this correla tion for annual United States data over the period 1949-1985 and Barro (1989) has at tained similar cross-sectional results for a sample of 72 countries.5 Further, Garcia-Mila and McGuire (1987) have found a statistically significant positive association between gross state product and public capital—high ways and educational structures—for the 48 contiguous states. The contribution of this paper is to expand this list of results and to offer evidence against the “reverse causation” hypothesis that low productivity growth tows in its wake low pub lic capital expenditures. Table 2 contains results which establish a positive correlation between labor productivity growth and public investment even after the latter variable has been purged of its economic growth compo nent by previous regression on the growth rate of gross domestic product. On this basis, I submit that public capital is a vital ingredient in the recipe for economic growth and rising standards of living. FOO TNOTES LSee Martin Feldstein, “A National Savings President,” Wall Street Journal, November 21, 1988, p. A 14. 2See David A. Aschauer, “Government Spending and the ‘Falling Rate o f Profit’,” Federal Reserve Bank of Chicago, Economic Perspectives, May/June 1988 for elaboration and supporting evidence for the United States. 3For Italy, data on public consumption and public invest ment is available only after 1970. sHowever, Barro suggested that this relationship is due to the reverse causation discussed above. He also estimates a public-capital-stock-to-output ratio and, upon regressing the growth in output (per person) on this estimated variable, finds that while the relationship is still positive, it is not statistically significant at conventional levels. By his own admission, however, his public capital stock measures are subject to large errors in measurement. Indeed, for the United States (for which there are direct estimates of public capital) his measure deviates by 50 percent from its actual value. 4In the United States, the ratio o f employee compensation to gross domestic output equalled 58 percent in 1966 and 60 percent in 1985. 24 ECONOMIC PERSPECTIVES REFERENCES Aschauer, David A., “Government Spending and the ‘Falling Rate of Profit’, ” Federal Re serve Bank of Chicago, Economic Perspec tives, Vol. 12, May/June 1988, pp. 11-17. __________, “Is Public Expenditure Productive?,” Journal of Monetary Economics, Vol. 23, March 1989, pp. 177-200. Feldstein, Martin, “A National Savings President,” Wall Street Journal, November 21, 1988, p. A14. Garcia-Mila, Theresa and Therese McGuire, “The Contribution of Publicly Pro vided Inputs to States’ Economies,” unpub lished mss. dated July 10, 1987. Barro, Robert J., “A Cross Country Study of Growth Saving, and Government,” Harvard University, January 1989. DATA APPENDIX Dp = growth in real gross domestic prod uct per person employed (OECD Historical Statistics). gcr = government final consumption ex penditure relative to gross domestic product (OECD National Accounts). Dn = growth in total employment (OECD Historical Statistics). Dcu = rate of change of capacity utiliza tion. Raw data are as follows: for the United States, Canada, France, West Germany, and Italy, rate of capacity utilization; for Japan, judgment on capacity utilization; for the United Kingdom, percent of firms operating at full capacity (OECD Main Economic Indica tors). The raw data have been normalized to account for differences in mean values and volatility across countries. gir = public gross fixed capital accumula tion minus consumption of fixed capital ex pressed relative to gross domestic product (OECD National Accounts). This variable is lagged one year. ir = private gross fixed capital accumula tion minus consumption of fixed capital ex pressed relative to gross domestic product (OECD National Accounts). This variable is lagged one year. FEDERAL RESERVE BANK OF CHICAGO 25 CURRENT RESEARCH As part of the ongoing research at the Federal Reserve Bank of Chicago, there are in-depth studies available on a variety of topics. Recent studies have covered such timely issues as bank deregulation, banking risks, the infrastructure, foreign trade policy, unemployment insurance, and regional development. The STAFF MEMORANDA series were occasional papers prepared by members of the Research Department for comment and review by the aca demic community. Although the series was discontinued in December 1988, a limited number of the studies are still available. A few recent papers included: Risk and Solvency Regulation of Depository Institutions: Past Poli cies and Current Options. George G. Kaufman (SM-88-1); A Note on the Relationship between Bank Holding Company Risks and Nonbank Activity. Elijah Brewer III (SM-88-5); The WORKING PAPER SERIES includes research studies covering three areas—regional economic is sues, macroeconomic issues, and issues in financial regulation. Current research has studied a number of areas, such as: Unemployment Insurance: A State Economic Development Perspective.William A. Testa and Natalie A. Davila (WP-89-2); The Opening of Midwest Manufac turing to Foreign Companies: The Influx of Foreign Direct Investment. Alenka S. Giese (WP-89-3); Is Public Expenditure Productive? David Aschauer (SM-88-7); Determining Manufacturing Output for States and Regions. Philip R. Israilevich and William A. Testa (WP-89-4); Imperfect Information and the Per manent Income Hypothesis. Abhijit V. Banerjee and Kenneth N. Kuttner (SM-88-9); A New Approach to Regional Capi tal Stock Estimation: Measurement and Performance. Alenka S. Giese and Robert H. Schnorbus (WP-89-6); Does Public Capital Crowd Out Private Capital? David Aschauer Why Has Illinois Manufacturing Fallen Behind the Region? William A. Testa (WP-89-7); (SM-88-10); Imports, Trade Policy, and Union Wage Dynamics. Ellen Rissman (SM-88-11). Technical Change, Regulation, and Economies of Scale for Large Com mercial Banks: An Application of a Modified Version of Shepard’s Lemma. Douglas D. Evanoff, Philip R. Israilevich, and Randall C. Merris (WP-89-11); Back of the G-7 Pack: Public Invest ment and Productivity Growth in the Group of Seven. David A. Aschauer (WP-89-13); Are Some Banks Too Large to Fail? Myth and Reality. George G. Kaufman (WP-89-14). 26 ECONOMIC PERSPECTIVES Copies o f the WORKING PAPER SERIES and STAFF MEMORANDA, as well as a complete listing o f all studies and their availability, can be ordered from the Public Information Center, Federal Reserve Bank o f Chicago, P.O. Box 834, Chicago, Illinois, 60690-0834, or telephone (312)322-5111. ECONOMIC PERSPECTIVES BULK RATE P ublic Inform ation C enter Federal Reserve Bank o f Chicago P Q. Box 834 Chicago, Illinois 60690-0834 U.S. POSTAGE PAID CHICAGO, ILLINOIS PERMIT NO. 1942 D » N o t F orw ard AtJdress C o rre c tio n Requested R eturn P o stag e G u a ra n te ed FEDERAL RESERVE BA N K OF CHICAGO