The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
________ Review_________ Vol. 68, No. 3 M a rch 1986 5 The R esp onse o f Stock Prices to Changes in Weekly M oney and the D iscount Hate 15 The E ffects o f Inflation on C o m m ercial Banks T h e Review is p u b lish ed 10 tim es p e r y ea r by th e R esea rch a n d Public In form ation D ep a rtm en t o f th e F ed era l R eserve Bank o f St. Louis. Single-copy su bscrip tio ns a re available to th e p u blic f r e e o f ch a rg e. Mail re q u ests f o r su bscrip tio ns, back issues, o r a d d re s s c h a n g e s to: R esea rch a n d Public Inform ation D ep a rtm en t, F ed era l R eserve Bank o f St. Louis, P.O. Bop 442, St. Louis, M issouri 63166. T h e views e x p re s s e d a re th o se o f th e individual a uth ors a n d d o not n ecessa rily reflec t official positions o f th e F ed era l R eserve Bank o f St. Louis o r th e F ed era l R eserve System . A rticles h e r e in may b e re p rin te d p ro v id ed th e s o u rc e is cred ite d . Please p rov id e th e Bank's R esea rch a n d Public Inform ation D ep a rtm en t with a copy o f re p rin t e d m aterial. F e d e ra l R e se rv e Bank o f St. L ou is Review M arch 1986 In This Issue . . . The efficient m arkets hypothesis suggests th at financial asset p rices reflect the inform ation available to m arket particip an ts. Consequently, an ticip ated ch an ges in any factors that influence asset prices, like the m oney stock, should already be a cco u n te d for in the quoted interest rates and stock p rices. Only u n an ticip ated ch an ges should lead to ch an ges in interest rates an d stock p rices. In the first p ap er in this Review, “The R esponse of Stock Prices to C hanges in Weekly M oney and the D iscount Rate,” R. W. Hafer investigates several a sp ects of the efficient m arkets h ypothesis: Do stock p rices react only to u n an ticip ated ch an ges in the m on ey stock? Are the m agnitudes of th ese effects different across different m on etary policy p ro ced u res? Do different m easu res of stock p rices react differently to a given u n e x p e cte d ch an ge in M l? Are the different stock p rice m easures equally responsive to a given ch an ge in the d iscou n t rate? Finally, do reaction s to d iscou n t rate ch an ges vaiy a cro ss m on etary policy p ro ced u res? Hafer’s results, b ased on daily d ata from 1977 through 1979, generally do not reject the efficient m arkets hypothesis. He does find, how ever, that the respon se to an u n exp ected in crease in M l is different from an u n e x p e cte d d ecrease in M l for the b ro ad er stock price m easures. Hafer’s evidence also ind icates th at the resp o n se to discou n t rate ch an ges varies a cro ss stock p rice m easu res and m on etary policy p roced u res. * * * While m ost analysts agree that the recen t sh arp red u ction in the rate of inflation has affected financial institutions, they disagree about the direction of the effect. Som e people point to the increase in bank failures as evidence that the sudden drop in inflation h as put a great deal of strain on the w hole credit stru ctu re. O thers think that banks have gained from re d u ce d inflation an d falling interest rates. In the se co n d article in this Review, “The Effects of Inflation on C om m ercial Banks,’’ G. J. Santoni exam ines this issue by analyzing th e relation ship betw een inflation an d th e stock p rices of publicly trad ed banks. The a u th o r’s results indicate that the real sh are prices of banks fall w hen the actu al rate of inflation is g reater than the an ticip ated rate of inflation. Fu rth erm ore, upw ard revisions in the anticip ated rate of inflation are associated w ith declines in the real share p rices of com m ercial banks. Santoni’s evidence suggests that the recent decline in the rate of inflation has had a favorable im p act on banks and that any u n exp ected resurgen ce of inflation w ould be harmful to bank stock. 3 The Response of Stock P rices to Changes in Weekly Money and the Discount Rate II. W. Hafer r J ONSIPF.RABI.F, research hits been devoted lo an alyzing the effects of weekly ch an ges in the m oney stock (Ml) on interest rates and exch an ge rates.1 In general, the results of this research are con sisten t with the efficient m arkets hypothesis, w hich holds that only u n exp ected ch an ges in the m oney stock should significantly affect interest rates and exch an ge rates.- Few of these studies, however, have investigated the reaction of stock prices to the weekly m oney a n n o u n cem en t:1The pu rpose of this article is to p ro vide som e evidence on this effect. R. I/V. Hafer is a research officer at the Federal Reserve Bank of St. Louis. The author would like to thank Stu Allen, Gikas Hardouvelis, Scott Hein, Jan Loeys, Doug Pearce, Jim Schmidt and Jim VanderHoff for their suggestions and comments on earlier drafts of this paper. Thomas A. Pollmann provided research assistance. ’The surveys by Cornell (1983) and Sheehan (1985) contain numer ous references to the literature on this subject. Alternative evidence is presented in Hein (1985), and Falk and Orazem (1985). 3Pearce and Roley (1983, 1985) find that stock prices react only to unanticipated changes in money and, for the most part, show no statistical relationship to either unanticipated or expected move ments of other economic news. Hardouvelis (1985) reports that stock prices react to unanticipated movements in four monetary measures (M1, net free reserves, the discount rate and the discount rate surcharge), as well as three non-monetary measures (trade deficit, unemployment rate and personal income). Although few studies have examined the weekly money/stock-price relationship, numerous studies have studied the longer-term reaction of stock prices to movements in money. Among others, see Sprinkel (1964), Rozeff (1974), Sorensen (1982) and Davidson and Froyen (1982). Various studies also have examined the behavior of stock prices to announcements of different types of information. For example, Schwert (1981) examines the reaction of stock prices to the announcement of inflation data; Fama, et. al. (1969) study the effects of stock splits; Lloyd-Davies and Canes (1978) investigate the effects of stock analysts’ published recommendations; and Niederhoffer (1971) analyses the reaction of stock prices to “world events." This p ap er extend s previous research on the re a c tion of stock prices to m o n etaiy "n e w s” in several wavs. First, it covers a broader tim e period, from Sep tem ber 1977 through D ecem ber 1984, than most p re vious studies. This allows one to test w h eth er the ch an ges in m o n etaiy policy operating p ro ced u res in O ctober 1979 an d O ctober 1982 influenced the re sp on se of stock prices to ch an ges in the m oney stock and the discou n t rate.4 Second, unlike previous studies, this study uses both broad and industiy-specific m easures of stock prices to determ ine if general m arket effects also o c c u r uniformilv across in specific ind u stiy groups.'’ As n oted by King (1966): “. . . it is intuitively appealing to think of incoming information as falling into various classes according to the scope of its effect on the market. There is some news of a monetaiy nature, for example, which is bound to have a market-wide impact on security price. The magnitude of impact need not, however, he the same for all stocks." (p. 140) Although n u m erous studies have attem p ted to d e te r mine optim al groupings of individual stocks based on their relative m ovem ents over time, little has been done to investigate the relative respon se of different stock groups to the sam e piece of eco n o m ic in formation. 4A discussion of the October 1982 change in policy procedures can be found in Wallich (1984) and Gilbert (1985). 5Most previous research focuses solely on the broad market effects. For example, Pearce and Roley (1983) use the Dow Jones Indus trial Average while Pearce and Roley (1985) use the Standard and Poor’s 500 index. In an approach similar to that used in this study, Hardouvelis examines the effect of new information on several different stock price measures. 5 FEDERAL RESERVE BANK OF ST. LOUIS MARCH 1986 Finally, unlike m ost previous work, w hich p re sum ed that the reaction of stock prices wits sym m etric with respect to unanticipated increases o r d ecreases in m oney, this p ap er tests for the sep arate effect of positive and negative u n exp ected ch an ges in M l. This perm its one to test for market efficiency in a som ew hat different m an n er than simply testing for the signifi ca n c e of e x p e cte d an d u n e x p e cte d ch an g e s in money.'' the effect of increases in th e rate at w hich th o se flows are discounted." A large literature ad d resses these conditions, show ing th at they generally are not fulfilled.10 Based on th ese studies, w hich ind icate that stock p rices react negatively to inflation, the exp ected inflation hypothesis suggests that stock p rices should fall following the an n o u n cem en t of an u n exp ected increase in the m oney stock. THEORETICAL MODELS OF STOCK PRICE RESPONSE TO MONETARY NEWS stock as a signal that alters market p ercep tion s of future m o n etaiy policy. Presum ing that the ch an g e is perceived as p erm an en t o r that the Federal Reserve is slow to respon d to u n exp ected deviations in the m oney stock aw av from its target level, interest rates will rise as the public ex p e cts the Fed to offset the u n exp ected in crease in the m oney stock. C on se quently, an u n e x p e cte d increase in the m oney stock implies g reater future tightening of credit availability, w hich results in higher interest rates. B ecau se the higher interest rates red u ce the p resent value of d is co u n ted cash flows, stock p rices are hypothesized to decline. The efficient m arkets hypothesis suggests that, w hen exp ected changes in the m oney stock o ccu r, they have no significant effect on stock p rices b ecau se they already have been in corporated into security p rices. Only u n exp ected ch an ges in the m oney stock affect stock prices accord in g to this h ypothesis.7 Weekly M oney Chanties Several hyp otheses have been suggested to explain w hy stock p rices react to u n exp ected ch an ges in the m oney stock." The e x p e c t e d inflation h y p o th esis sug gests that an u n exp ected increase in the m oney stock increases m arket participants' exp ectatio n s of in flation, leading to higher interest rates via the so-called F isher effect. If the increase in interest rates low ers the present value of corp oration s' d iscou n ted cash flows, stock prices will fall w henever investors observe an u n exp ected increase in the m oney stock. This hypoth esized respon se does not o ccu r, how ever, u n d er c e r tain restrictive conditions. Given perfect m arkets and no taxes, for exam ple, ch an ges in exp ected inflation w ould have no effect on stock prices, b ecau se e x p e cted in creases in nom inal cash flows w ould offset 6Comell, Pearce and Roley (1985) and Hardouvelis provide no evidence on this issue as it relates to stock price effects. Pearce and Roley (1983) present evidence on the response of stock prices when money deviates from announced long-run target ranges. Their tests are based on separating unanticipated money into positive surprises above target, negative surprises below target and all others. In general, their results indicate that the different surprise measures are not statistically different in their effect. There is, however, some evidence of a different effect of positive surprises across the different policy regimes. 7As Sheehan (1985) notes, the unexpected change in the money stock provides new information about money stock developments that already have occurred. That is because the money stock is announced with a lag. Thus, the announcement causes market participants to revise their forecasts for future policy actions apart from previously held expectations only if the announced money stock change is different from its expected change. This discussion is based on Cornell and Sheehan. An alternative hypothesis, the p o licy an ticip ation s h y p othesis, views an u n exp ected ch an ge in the m oney Finally, the m o n ey d em a n d o r r e a l activity h y p o th e sis asserts that m oney an n o u n cem en ts provide infor m ation about future m oney d em an d . Suppose that market participants interpret an u n exp ected increase in the m oney stock as a signal that there h as been a perm anent increase in m on ey d em an d . If m oney d e m and d ep en ds on exp ected future outp u t, then the u n exp ected increase in m oney d em an d ind icates that future outp u t will be higher than previously ex p ected ." Consequently, the in crease in e x p ected cash 9This discussion abstracts from the distinction of net monetary credi tors and net monetary debtors. For a discussion of the effects of changes in inflation expectations on each group, see Kessel and Alchian (1962). For a more recent study of the effects of inflation on bank stock prices, used to represent a group of net monetary creditors, see Santoni (1986). 10For example, Feldstein (1980) discusses the effect of taxes; Schwert (1981), Fama and Schwert (1977) and Nelson (1976) examine the inflation-stock price relationship for the United States while Branch (1974) and Cohn and Lessard (1981) provide evi dence from other countries. In general, these studies indicate that unexpected increases in inflation lower stock prices. Results re ported in Kool and Hafer (1986), however, suggest that this result does not hold for earlier time periods. "Fama (1981) argues that expected inflation in previous studies serves as a proxy for expected real activity. Consequently, regress ing stock prices on expected inflation without accounting for ex pected real activity will yield incorrect estimates. Following this line of reasoning, several researchers have used available survey data to study the relationship between expected stock price changes, expected inflation and expected real activity. See, for example, Gultekin (1983), Pearce (1984), Hasbrouck (1984) and Coate and VanderHoff (1985). FEDERAL RESERVE BANK OF ST. LOUIS flows p ro d u ces an increase in stock prices.'- This hy pothesis thus predicts that stock prices sh ould in crease in response to an u n exp ected increase in the m oney stock. Discount Rate Changes Discount rate chan ges m ay lie thought of as an indication of ch an ges in exp ected future m on etaiy policy; d iscou nt rate changes, in o th er w ords, affect financial and stock m arkets primarily through their effect on interest rates and p ercep tion s of future e co nom ic activity.'1' In general, increases in the discou n t rate red u ce stock prices b ecause they presage a tight ening of m o n etaiy policy.14 The move tow ard tighter policy is exp ected to in crease interest rates and re d u ce real eco n om ic activity and, consequently, future co rp o rate cash Hows. Stock prices decline b ecau se the red u ced future cash flows are d iscou n ted at higher interest rates. Some argue that the im p act of a d iscou n t rate change d ep en ds on the Federal Reseive’s cu rren t o p erating p ro ced u re.1' If the Fed is targeting interest rates, ch an ges in the d iscount rate m ay provide no inform ation about future policy that is not already 12lt should be noted that attendant increases in the ex ante real rate are presumably more than offset by expected increases in future real economic activity. 13Batten and Thornton (1985) and Smirlock and Yawitz (1985), for example, each attempt to determine "technical” from “non technical” changes in the discount rate. Batten and Thornton dichot omize discount rate changes into technical or policy-related, based on Federal Reserve statements. Their procedure assumes that the change is entirely technical or policy-related. Smirlock and Yawitz attempt to define technical and non-technical discount rate changes by regressing rate changes on lagged values of the spread between the federal funds rate and the discount rate and lagged values of changes in discount window borrowing. Pre dicted values from this equation constitute the technical (or antici pated) change, while the regression’s error term constitutes the non-technical (or unanticipated) change in the discount rate. Several factors militate against this procedure. First, it does not capture effects not incorporated in the explanatory variables. Sec ond, it may alter the timing of the actual change. Last, it assigns each discount rate change, which generally is 25, 50 or 100 basis points, some estimated value that often does not equal the actual value. In other words, there is always some non-technical change. Because of the problems surrounding these classifications of discount rate changes, we take the changes to be unanticipated. '4There are instances, however, when moves to raise the discount rate have been received by the market as good news, precipitating increases in stock prices. This is discussed in the shaded box on the next page. '5For example, see Roley and Troll (1984) or Smirlock and Yawitz for a discussion of this point. MARCH 1986 in corp orated in interest rates. If the Fed is using a re se rv e g ro w th o p e ra tin g p ro c e d u re , h ow ev er, ch an ges in the discount rate influence interest rates. During the period covered by this study, three differ ent operating p ro ced u res w ere used: interest rate tar geting (pre-O ctober 1979); nonborrow ed reseives tar geting (October 1979 to O ctober 19821; and a borrow ed reseives proced u re, w hich tends to sm ooth m ove m en ts in the funds rate m ore than nonborrow ed re serves targeting (post-O ctober 1982). The em pirical tests below assess the different effects of discou n t rate changes u n d er different policy p roced u res. TESTS OF THE BASIC MODEL The basic equation tested is: It) ASP, = ct„ + P, UM, + 0 , EM, + p., DR, + p, IJKS, + e„ w hen: ASP = the fii-st-difference of the logarithm of the daily stock price index, UM = the u n exp ected dollar ch an ge in M l, EM = the exp ected ch an ge in M l, DR = the ch an ge in the d iscount rate, and DRS = the ch an ge in th e d iscou n t rate su rch arge."1 The efficient markets hypothesis suggests that the estim ated coefficient on exp ected m oney ch an g es If},) should be zero. If d iscou n t rate ch an ges influence stock prices as hypothesized above, p.. and P4 will be negative. Finally, the exp ected sign of (3, differs d e pending up on the hypothesis being tested. The policy anticipations and the exp ected inflation hypotheses suggest that it will be negative; the m on ey dem and hypothesis suggests that it will be positive. Besides investigating the validity of several h yp o th e ses regarding the effects of m oney stock an d discou n t rate ch an ges on stock prices, estim ates of equation 1 can be used to test several o th e r h yp oth eses as well: Are the effects of the exp lan atoiy variables statistically equal a cro ss different m o n etaiy policy regim es? Are the effects sim ilar acro ss stock price ind exes? Are the effects of m oney stock ch an ges on stock prices sym m etrical, as generally is assu m ed ? 16The discount rate surcharge was used by the Federal Reserve during the period from March to May 1980 and again during the period from November 1980 to November 1981. The 1980 imposi tion of the surcharge was part of the credit restraint program enacted by the Carter administration and it was set at 3 percent. During each period when the surcharge was used, it applied to discount window borrowings by banks with deposits of $500 million or more that borrow frequently. Because the surcharge did not apply to all borrowing, it is included as a separate variable in the regression equations presented below. 7 FEDERAL RESERVE BANK OF ST. LOUIS MARCH 1986 Timing D iscount Rate Changes Studies using d iscou n t rate ch an ges often take the an n ou n cem en t date as the day it becom es “public." How one tim es the publication of the new information, however, m ay alter the em pirical results. F o r exam ple, th ere are o ccasion s on w hich discount rate ch an ges are an n ou n ced d uring the trading day and, as su ch , the actu al an n ou n cem en t day will differ from the day it is rep o rted by the financial press. B ecause of this, w e en ter the d is coun t rate change on the trading day that the change becom es effective. To illustrate how sensitive the results are to ch an ges in the timing, the SP500 equation w as reestim ated for the pre-O ctober 1979 period, defining the an n ou n cem en t date as that day w hen the dis coun t rate ch ange ap p eared in the the Wall S treet Journal. An exam ination of the d ata revealed that, on Friday, N ovember 1 ,1 9 7 8 , the d iscou n t rate w as raised 100 basis points during the trading day. Ac co u n ts in the Wall S treet Jo u rn a l on M onday attrib ute the stock price rally on Friday to the an n o u n ce To assess these questions, equation 1 w as estim ated using daily stock price data from Septem ber 23, 1977, through D ecem b er 31, 1984. T hree zero-on e dum m y variables w ere u sed to differentiate the periods asso ci ated w ith alternative m o n etaiy policy regim es. Thus, D1 = 1 from S eptem ber 23, 1977, through O ctob er 5, 1979, zero elsew here; D2 = 1 from O ctober 5, 1979, through O ctober 15, 1982, zero elsew here; and D3 = 1 after O ctober 15, 1982, zero elsew here. Interaction term s are form ed by multiplying ea ch exp lan atoiy variable by these dum m y variables.17 F or exam ple, D1UM rep resents the effect of UM in the first sub p e riod, D21IM the effect during the seco n d and so on. Table 1 p resen ts the results of estim ating equation 1 using th ese interaction term s and the various stock price in d exes.IS ’'Because the discount rate surcharge variable enters only during the second subsample, no interaction term is necessary. Also, the choice of October 15 for the 1982 policy change is arbitrary, since published accounts of the procedural change do not provide an exact date. 18Note that equation 1 is estimated without day-of-the-week variables. Previous analysis by Pearce and Roley (1985) using the same data finds that the presence or absence of these variables did not influence their results. Given this evidence and the fact that we are using the same data, we also omit day-of-the-week variables. Other evidence on the existence of day-of-the-week effects, much of it 8 m ent, suggesting that the increase in the d iscou n t rate reaffirmed the m arket’s p ercep tion that the Fed w as resolved to rein in m oney grow th an d to red u ce the possibility of future inflation. If w e ch an ge only this one an n ou n cem en t d ate from the day it b e cam e effective (Friday) to the day it ap p eared in the Wall S treet Jo u rn a l (Monday), the estim ated rela tionship b ecom es ASP500, = 0.04 - 0.098 UM, - 0.872 DR, (1.14) (2.18) (2.05) R2 = 0.015 SE = 0.713 DW = 1.68 Two points should be m ade. First, the use of the effective day is theoretically preferable to dating the an n ou n cem en t by its a p p earan ce in the financial press. Second, the em pirical effects of discou n t rate ch an ges on stock prices ap p ear to be quite sensitive to small timing changes. In this exam ple, changing one observation reverses the sign and significance of the discou n t rate variable. As the efficient m arkets hyp oth esis predicts, the exp ected ch an ge in M l (EM) d oes not significantly affect stock price ch an ges. The results in table 1 indi cate that none of the 15 estim ated coefficients on exp ected m oney is statistically significant at th e 5 p ercen t level of significance. The test results in table 2 also b ear out the efficient m arkets h ypothesis, as the rep orted F -statistics can n o t reject the hyp oth esis that the coefficients on the exp ected ch an ge in m oney togeth er are insignificantly different from zero. Thus, the hypothesis that the estim ated coefficients on e x p ected m oney are zero a cro ss the different m on eta ry re g im es is not rejected by the data. As p red icted by the e x p ected inflation an d policy anticipations hypotheses, but co u n te r to th e m on ey conflicting, are reported in French (1980) and Gibbons and Hess (1981). Following Pearce and Roley (1985) and Hardouvelis, we also included several other measures of economic “news” as explana tory variables in equation 1. Those results indicated that stock prices, irrespective of the index, responded to monetary announce ments more reliably than the other measures, such as unexpected inflation, economic activity or unemployment. Because the results of these tests do not extend the analysis already provided by Pearce and Roley, we do not report them here. MARCH 1986 FEDERAL RESERVE BANK OF ST. LOUIS Table 1 Estimates of Equation 1 Index Variable SP500 SP400 SPTRAN SPUTIL SPFIN Constant 0.025 (1.20) 0.025 (1.14) 0.027 (0.98) 0.012 (0.86) 0.018 (0.80) D1UM -0.100 (1.81) -0.104 (181) -0.164 (2.29) -0.037 (1.02) -0.060 (1.03) D2UM -0.124 (3.84) -0.120 (3.58) -0.067 (1.60) -0.129 (6.05) -0.149 (4.37) D3UM -0.112 (2.46) -0.114 (2.41) -0.091 (1.54) -0.119 (3.96) -0.125 (2.60) D1EM 0.072 (1.14) 0.077 (1.18) 0.105 (1.29) 0.036 (0.87) 0.033 (0.50) D2EM 0.072 (1.37) 0.073 (1 34) 0.101 (1.49) 0.048 (1.40) 0.031 (0.56) D3EM 0.034 (0.74) 0.037 (0.77) -0.021 (0.34) 0.022 (0.72) 0.094 (1.92) D1DR 0.916 (1.77) 1.038 (1.91) 1.189 (1.76) 0.259 (0.76) 0.944 (1.72) D2DR -0.597 (2.35) -0.599 (2.27) -0.332 (1.01) -0.282 (1.69) -0.448 (1.67) D3DR 1.208 (1.54) 1.315 (1.61) 1.412 (1.38) -0.269 (0.52) 0.336 (0.41) -0.432 (2.66) -0.421 (2.48) -0.501 (2.36) -0.543 (5.08) - 0.658 (3.83) DRS R2 0.020 0.018 0.009 0.039 0.021 SE 0.877 0.913 1.165 0.596 0.952 DW 1.79 1.82 2.00 1.98 2.01 P — — 0.19 (8.31) 0.26 (11.36) 0.23 (9.88) NOTE: Absolute value of t-statistics are reported in parentheses. R2Is the coefficient of determination adjusted for degrees of freedom; SE is the regression standard error; DW is the Durbin-Watson test statistic; and p is the estimate of the first-order serial correlation coefficient. The dependent variables are measured as first-differences of the logarithms of Standard and Poor’s 500 (SP500), 400 stock index (SP400), the transportation index (SPTRAN), the utility index (SPUTIL) and the financial index (SPFIN). The right-hand-side measures are unexpected changes (UM) and expected changes in M1 (EM), based on the Money Market Services, Inc. survey. DR and DRS represent the percentage change in the Federal Reserve's discount rate and surcharge rate, respectively. The terms D1, D2 and D3 represent (0,1) dummy variables where D1 = 1 from September 23, 1977, through October 5, 1979,0 elsewhere; D2 = 1 from October 5, 1979, through October 15, 1982, 0 elsewhere; and D3 = 1 from October 15, 1982, to December 31,1984, 0 elsewhere. d em and hypothesis, u n an ticipated ch an ges in Ml (UM) generally have a statistically significant, negative im pact on stock p rices. F o r instance, an unanticipated $1 billion increase in M l red u ced the grow th rate of the SP500 and the SP400 by about 10, 12 and 11 basis points acro ss the three periods tested . The results in table 2 provide supporting evidence that un antici pated m oney stock ch an ges do affect stock p rices. The results on line 3 reject th e claim that u n an ticip ated ch an ges in M l have no effect; the results on line 4, w hich test the equality of the estim ated coefficients across the different policy periods, indicate that one can n ot reject coefficient stability at the 5 p ercen t level. These results show that only u n exp ected ch an ges in 9 FEDERAL RESERVE BANK OF ST. LOUIS MARCH 1986 Table 2 Hypothesis Test Results Index/F-statistics Hypothesis SP500 SP400 SPTRAN SPUTIL SPFIN D1EM = D2EM = D3EM = 0 1.23 (0.30) 1.24 (0.29) 1.33 (0.26) 1.08 (0.36) 1.41 (0.24) D1EM = D2EM = D3EM 0.19 (0.83) 0.17 (0.84) 1.20 (0.30) 0.16 (0.85) 0.46 (0.63) D1UM = D2UM = D3UM = 0 8.04 (0.00) 7.31 (0.00) 3.40 (0.02) 17.81 (0.00) 8.96 (0.00) D1UM = D2UM = D3UM 0.08 (0.92) 0.03 (0.97) 0.68 (0.51) 2.48 (0.08) 0.88 (0.42) D1DR = D2DR = D3DR = 0 3.68 (0.01) 3.81 (0.01) 2.00 (0.11) 1.24 (0.29) 1.97 (0.11) D1DR = D2DR = D3DR 5.15 (0.01) 5.47 (0.00) 2.97 (0.05) 1.03 (0.36) 2.78 (0.06) NOTE: Marginal significance levels are reported in parentheses. Variable definitions are found in table 1. m oney reliably influence the behavior of stock prices an d that there ap pears to be no statistically significant change in this resp on se across the different m on etary policy regim es.19 T h e g en e ra l h y p o th e sis ab o u t d is c o u n t ra te ch an ges on stock p rices does not fare so well as the hypothesis about the effects of un an ticip ated ch an ges in M l. D iscount rate ch an ges generally had a positive but not statistically significant (5 p ercen t level) effect on stock p rices before O ctober 1979 an d after O ctob er 1982. This result does not su p p ort the view th at dis co u n t rate in creases should negatively influence stock prices. It does, however, sup p ort the notion that, d u r ing periods in w hich m on etary policy em p h asizes the behavior of the federal funds rate, the discou nt rate m ay not im part relevant policy inform ation not al ready co n tained in, for exam ple, the federal funds rate.20 The results for the O ctober 1979 to O ctober 1982 period indicate that ch an ges in the d iscount rate result in stock price m ovem ents generally con sisten t w ith the hypothesis d escribed above. Changes in the d iscou n t rate have a significant (one-tailed) negative 19This evidence is in sharp contrast to the results from studies examin ing the interest-rate/money relationship over this period. “ It should be noted that the discount rate changes during the preOctober 1979 period are positive and significant at the 10 percent level for all of the indexes except SPUTIL. For a possible explana tion of this result, see the shaded insert on page 8. Digitized for10 FRASER effect on all indexes during this period, e xcep t for the SPTRAN index. The size of the estim ated coefficients, however, is low er for the m ore narrow ly defined in dexes than it is for th e broad SP500 an d SP400 m ea sures. Thus, a 100 basis-point in crease in the d iscount rate during this period led to a 60 basis-point decline in the grow th rate of SP500 and SP400, but only a 28 basis-point drop in the grow th rate of the SPUTIL index. W hy are the effects of d iscou n t rate ch an ges so different across the different policy regim es? Prior to O ctober 1979, m ovem ents in the federal funds rate directly conveyed inform ation about ch an ges in pol icy objectives, thus making the inform ational co n ten t of d iscou n t rate ch an ges red u n d an t.21 A sim ilar argu m ent can be m ade about the p ost-O ctob er 1982 p e riod, since the sw itch from a non b orrow ed to a b or row ed reserves targeting p ro ced u re is sim ilar to a policy that sm ooths m ovem ents in th e federal funds rate.22 The finding that the estim ated coefficients on 21The evidence in the shaded insert on page 8 and accounts of discount rate changes in the Wall Street Journal do not support the gross generality of this view. “ If a borrowings target (referred to as the borrowings assumption) is used and the primary determinant of discount window borrowing is the federal-funds-rate/discount-rate spread, increases in the funds rate, ceteris paribus, necessitate an increase in reserves since borrowings will otherwise increase. Thus, reserves are injected, the funds rate falls and borrowings return to their desired level. This policy scenario suggests that movements of the federal funds rate after October 1982 again directly reflect policy objectives. For a more complete discussion, see Gilbert. FEDERAL RESERVE BANK OF ST. LOUIS d iscount rate ch an ges are insignificantly different from zero during the two different policy regim es suggests that the m arket's p ercep tion of ch an ges in the discoun t rate m ay not be any different after O cto ber 1982 than it w as before O ctober 1979. The test results in table 2 indicate that the discount rate change is an im portant variable in explaining the behavior of the broad indexes, but is less so for the m ore specialized groups. In fact, the rep orted Fstatistic for the SPUTIL index indicates that w e can n ot reject the hypothesis that ch an ges in the d iscou n t rate together have no significant effect. The test results also reveal that the effect of discou nt rate chan ges is not equal acro ss regim es at reasonable levels of signifi ca n ce (except for SPUTIL). Finally, the estim ated coefficient on the discount rate su rcharge (DRS) is highly significant and negative for each of the stock p rice indexes tested. The m agni tude of the effect on the broad stock price m easures is sim ilar to that found by Pearce an d Roley (1985); in addition, all of the stock p rice m easures are affected. In fact, unlike the results for the discount rate, w hich tend to have a sm aller effect on the n arrow er indexes, a ch an ge in the su rcharge rate actually had a larger im pact on the narrow indexes. To sum m arize, the hypothesis that only u n an tici pated changes in m on ey negatively influence the m ovem ent of stock p rices can n o t be rejected . This finding, w hich holds for m ost of the stock price in dexes used and tim e periods tested, su p p orts the efficient m arkets hypothesis, rejects the m oney d e m an d hypothesis and corrob orates earlier results b ased solely on the use of broad stock p rice indexes. It also show s that the effect of d iscou n t rate changes varies am ong the p articu lar indexes and over the p eri ods tested. Thus, although policy regim e ch an ges do not ap p ear to influence the m arket ’s reaction to u n an ticipated changes in m oney, the evidence suggests that the inform ation conveyed through d iscou n t rate changes varies across policy regim es. Symmetry Hypothesis Tests Analysts generally assu m e that positive and nega tive u n anticipated ch an ges in m oney have sym m etri cal influences on stock p rices.-1 To test this hypothe23Little research into the symmetry of the effects is available. Although Pearce and Roley (1983) and Roley and Troll (1983) test for the effects on interest rates when money changes are above or below stated policy targets, this does not directly address the hypothesis. Also, Pearce and Roley (1983) present similar tests for stock prices. MARCH 1986 sis, w e again use zero-on e dum m y variables to generate the ap p rop riate interaction term s that differ entiate positive an d negative obseivations of UM. Table 3 p resen ts the results of this test. Positive values of UM are d en oted by UM ( + ) ; negative values by UM ( —).M Negative u n exp ected ch an ges in m on ey have no statistically significant effect on stock p rices using the SP500, SP400 and the SPrRAN indexes. In each case, the rep o rted t-values are quite small, as are the estim ated coefficients. In tests of the equality of the coefficients on the positive an d negative values of UM, w e find that, for th ese three stock p rice m easures, the t-statistics are large enough to reject equality at the 5 p ercen t level. It ap p ears that only positive values of UM have significant effects on ch an g es in these stock prices; the grow th rates of these indexes fell by 16 basis points for a $1 billion surprise in M l. This result suggests that m arket efficiency is violated.-'1 The SPUTIL and SPFIN results indicate that both positive an d negative values of UM have similar, statis tically significant effects on the stock p rice ch an ges. In these instances, the calculated t-statistics to test coef ficient equality are well below any accep tab le level of significance. The sym m etrical resp o n se of utility and financial stocks to an u n an ticip ated increase o r d e crease in the m oney stock indicates that th ese stocks are relatively m o re sensitive to interest rate an d p rice level m ovem ents than o th e r stocks. SUMMARY The results of this study generally su p p o rt the ef ficient m arkets hypothesis. Based on evidence from several different stock p rice indexes, u n an ticip ated ch an ges in m on ey have a statistically significant effect on stock p rices. E xp ected ch an ges in m oney never display a statistically significant effect. The estim ated effect of u n an ticip ated ch an ges in m oney did not differ across alternative m on etary policy regim es. One 24Values of zero are included in the UM ( + ) data. Discount rate variables are omitted from the SPUTIL equation, because the evi dence in table 2 indicates that they are not significant (Jointly) at any reasonable level. It should be noted that reestimation of the equa tions in table 3 using a seemingly unrelated regression procedure does not alter the conclusions reached in this section. 25Gikas Hardouvelis, in private correspondence, suggests the fol lowing scenario. Consider the median forecaster facing the money announcement with equal probability that the announced M1 figure will be above or below the forecast. Given the results in table 3, the strategy is to sell before the announcement, since a positive sur prise in money will lower stock prices while a negative surprise has no statistical effect. If such response persists, market efficiency is violated. 11 FEDERAL RESERVE BANK OF ST. LOUIS MARCH 1986 Table 3 Results for Symmetry Test Index Variable SP500 SP400 SPTRAN SPUTIL SPFIN Constant 0.048 (2.18) 0.049 (2.16) 0.067 (1.90) 0.024 (1.20) 0.040 (132) UM( + ) -0.162 (5.07) -0.164 (4.93) -0.166 (4.01) -0.120 (5.75) -0.153 (4.56) UM(-) -0.046 (1.24) -0.041 (1.08) 0.006 (0.13) -0.086 (3.65) -0.072 (1.89) D1DR 0.937 (1.81) 1.059 (196) 1.195 (1.77) 0.955 (1.74) D2DR - 0.604 (2.39) -0.607 (2.30) -0.340 (1.03) -0.446 (166) D3DR 1.232 (1.57) 1.341 (1.64) 1.432 (1.40) 0.394 (0.47) -0.403 (2.49) -0.391 (2.32) —0.467 (2.21) 2.48* 2.52* 2.75* 1.07 1.58 DRS t -0.551 (5.22) -0.637 (3.71) R2 0.022 0.021 0.015 0.039 0.025 SE 0.876 0.912 1.164 0.597 0.952 DW 1.79 1.82 2.00 1.98 2.01 P — — 0.19 (8.38) 0.27 (11.73) 0.23 (9.94) NOTE: The reported t-statistic is based on testing the hypothesis that UM (+) = UM (-). An asterisk denotes significance at 5 percent level. All other terms are defined in table 1. result that does not su p port the efficient markets h ypothesis is the finding that the effects of u n an tici pated m on ey ch an ges are asym m etric: only positive values of u n an ticip ated ch an ges in m oney ap p e a r to have a significant im p act on the SP500, SP400 and SPTRAN m easures. The etfects of discoun t rate ch an ges on stock prices vaiy with ch an ges in m on etaiy policy p ro ced u res; their influence also lessened as the stock price index b ecam e narrow er, fn general, d iscou n t rate ch an ges have significant negative effects on stock prices only from O ctober f 979 to O ctober 1982, a period ch a ra c te r ized by a m o n etaiy policy that focu sed on controlling n onborrow ed reserves. Before an d after that period, discoun t rate ch an ges convey little additional infor m ation about policy. 12 REFERENCES Batten, Dallas S., and Daniel L. Thornton. “The Discount Rate, Interest Rates and Foreign Exchange Rates: An Analysis with Daily Data,” this Review (February 1985), pp. 22-30. Branch, Ben. “Common Stock Performance and Inflation: An Inter national Comparison,” Journal of Business (January 1974), pp. 48-52. Coate, D., and James VanderHoff. “Stock Returns, Inflation and Real Output,” unpublished manuscript, Rutgers University (Janu ary 1985), forthcoming Economic Inquiry. Cohn, Richard A., and Donald R. Lessard. “The Effect of Inflation on Stock Prices: International Evidence,” Journal of Finance (May 1981), pp. 277-89. Cornell, Bradford. “The Money Supply Announcements Puzzle: Review and Interpretation,” American Economic Review (Septem ber 1983), pp. 644-57. Davidson, Lawrence S., and Richard T. Froyen. “Monetary Policy MARCH 1986 FEDERAL RESERVE BANK OF ST. LOUIS and Stock Returns: Are Stock Markets Efficient?” this Review (March 1982), pp. 3-12. cation of Second-Hand Information," Journal of Business (January 1978), pp. 43-56. Falk, Barry, and Peter F. Orazem. “The Money Supply Announce ments Puzzle: Comment,” American Economic Review (June 1985), pp. 562-64. Niederhoffer, Victor. “The Analysis of World Events and Stock Prices, ”Journal of Business (April 1971), pp. 193-219. Fama, Eugene F., Lawrence Fisher, Michael C. Jensen and Richard Roll. “The Adjustment of Stock Prices to New Information,” Inter national Economic Review (February 1969), pp. 1-21. Fama, Eugene F., and G. William Schwert. “Asset Returns and Inflation,” Journal of Financial Economics (November 1977), pp. 115-46. Fama, Eugene F. “Stock Returns, Real Activity, Inflation, and Money,” American Economic Review (September 1981), pp. 54565. Feldstein, Martin. “Inflation and the Stock Market,” American Eco nomic Review (December 1980), pp. 839-47. Nelson, Charles R. “Inflation and Rates of Return on Common Stocks," Journal of Finance (May 1976), pp. 471-82. Pearce, Douglas K. “An Empirical Analysis of Expected Stock Price Movements," Journal of Money, Credit and Banking (August 1984), pp. 317-27. Pearce, Douglas K., and V. Vance Roley. “The Reaction of Stock Prices to Unanticipated Changes in Money: A Note," Journal of Finance (September 1983), pp. 1323-33. _______ . “Stock Prices and Economic News," Journal of Busi ness (January 1985), pp. 49-67. French, Kenneth R. “Stock Returns and the Weekend Effect,” Journal of Financial Economics (March 1980), pp. 55-69. Roley, V. Vance, and Rick Troll. “The Impact of New Economic Information on the Volatility of Short-Term Interest Rates," Federal Resen/e Bank of Kansas City Economic Review (February 1983), pp. 3-15. Gibbons, Michael R., and Patrick Hess. “Day of the Week Effects and Asset Returns,” Journal of Business (October 1981), pp. 57996. _______ _ “The Impact of Discount Rate Changes on Market Interest Rates,” Federal Reserve Bank of Kansas City Economic Review (January 1984), pp. 27-39. Gilbert, R. Alton. “Operating Procedures for Conducting Monetary Policy," this Review (February 1985), pp. 13-21. Rozeff, M. S. “Money and Stock Prices: Market Efficiency and the Lag in Effect of Monetary Policy,” Journal of Financial Economics (September 1974), pp. 245-302. Gultekin, N. Bulent. “Stock Market Returns and Inflation Fore casts,” Journal of Finance (June 1983), pp. 663-73. Hardouvelis, Gikas A. “Macroeconomic Information and Stock Prices,” unpublished manuscript, Columbia University (Septem ber 1985). Santoni, G. J. “The Effects of Inflation on Commercial Bank Stock Prices,” this Review (forthcoming 1986). Schwert, G. William. "The Adjustment of Stock Prices to Informa tion About Inflation,” Journal of Finance (March 1981), pp. 15-29. Hasbrouck, Joel. “Stock Returns, Inflation, and Economic Activity: The Survey Evidence,” Journal of Finance (December 1984), pp. 1293-310. Sheehan, Richard G. “Weekly Money Announcements: New Infor mation and Its Effects,” this Review (August/September 1985), pp. 25-34. Hein, Scott E. “The Response of Short-Term Interest Rates to Weekly Money Announcements: Comment," Journal of Money, Credit and Banking (May 1985), pp. 264-71. Smirlock, Michael, and Jess Yawitz. “Asset Returns, Discount Rate Changes, and Market Efficiency,” Journal of Finance (September 1985), pp. 1141-58. Kessel, Reuben A., and Armen A. Alchian. “Effects of Inflation,” Journal of Political Economy (December 1962), pp. 521-37. Sprinkel, Beryl W. 1964). King, Benjamin F. “Market and Industry Factors in Stock Price Behavior,” Journal of Business (January 1966), pp. 139-90. Sorensen, Eric H. “Rational Expectations and the Impact of Money upon Stock Prices,” Journal of Financial and Quantitative Analysis (December 1982), pp. 649-62. Kool, Clemens J.M., and R. W. Hafer. “Inflation and Stock Prices: A Long Term View,” Federal Reserve Bank of St. Louis Working Paper No. 86-001 (January 1986). Lloyd-Davies, P., and Michael Canes. “Stock Prices and the Publi Money and Stock Prices (Richard D. Irwin, Inc., Wallich, Henry C. “Recent Techniques of Monetary Policy," Fed eral Reserve Bank of Kansas City Economic Review (May 1984), pp. 21-30. APPENDIX Data Definitions Money The exp ected change in the m on ey supply (EM) is the m edian forecast obtained from M oney Market Services, Inc. (MMS). Since 1977 this firm has co n d u cted a weekly tele phone survey of 50 to 60 governm ent securities dealers to obtain their forecast of the change in M l. Before February 8,1980, the survey w as co n d u cted tw ice each week, initially on Tuesday, with a follow-up call on Thursday, allowing resp o n d en ts to alter their original guess. From February 1980 through February 1984, however, the survey w as co n d u cte d only on Tuesday, b ecau se of the Federal Reserve’s shift in announcing the weekly M l figures from T hursday to Friday after noon. Beginning February 1984, w hich co rresp on d s to the change from lagged to co n tem p o ran eo u s reserve accou n tin g and w ith the an n ou n cem en t day again being ch an ged from Friday to T h u rsd ay afternoon, 13 FEDERAL RESERVE BANK OF ST. LOUIS MMS o nce again used two surveys: the initial poll on the Friday im m ediately following the T hursday m oney a n n ou n cem ent an d again on the following Tuesday. F o r this study, w e use the forecasts fr om the Tuesday survey. The data used here are those from P earce and Roley (1985) as u p d ated by Doug P earce. We w ould like to thank him for making th ese as well as the actu al M l data available. Actual changes in weekly M l, w hich ap p ear in the Federal Reserve’s H.6 statistical release, are m easured as the first an n ou n ced value m inus the first revised estim ate of the previous w eek’s level. Due to the changing definition of M l during o u r sam ple, the following p ro ced u re w as followed to obtain a series consistent w ith that being forecast by the survey re sp ondents: Until February 1980, w e use the old defini tion of M l. From February 1980 through N ovember 1981, w e u se the actual M1B m easure (not the M1B value that w as “shift-adjusted’’ for the introd uction of NOW accou n ts). Finally, from November 1981 through the en d of o u r sam ple, w e u se the cu rren t definition of M l. Given the actu al an d exp ected series for m oney, u n an ticip ated changes in M l (UM) are m easured as actu al less exp ected . Digitized for 14FRASER MARCH 1986 Stock Prices The stock price indexes used in this study are daily close values of the broad Standard and P oo r’s (SP) 500 and 400 indexes, as well as the industry-specific in dexes for transportation (SPTRAN), utilities (SPUTIL) an d financial institutions (SPFIN). In e a ch instan ce, the stock p rice ch an ge is m easured as th e difference of the logarithm s. Discount Rate Changes in the Federal Reserve’s d iscou n t rate and the su rch arge are m easu red in p ercen tage poin ts; that is, a 100 basis-point ch an ge in eith er rate is m easu red as 1.0. Our m easu rem en t of th e d iscou n t rate change, unlike that in som e studies, follows the Federal Re serve’s official dating p roced u re, changing w h en one of th e 12 Federal Reserve Banks h as the approval of the Federal Reserve Board to ch an ge its rate. The d ata used h ere is based on the day th e new rate is in effect, not w hen the new rate is an n o u n ced in the financial press. The Effects of Inflation on Com m ercial Banks G. J. Santoni p ■ EOPLE disagree about the effect of the recen t decline in inflation on U.S. financial institutions. Som e claim that the "sudden drop in inflation . . . put the co u n try ’s w hole credit stru ctu re u n d er great strains that are becom ing increasingly ap p aren t.” 1 One of the m ore im portant indicators of the “strain ,” accord in g to this argum ent, is th e increase in bank failures.2 O thers argue that financial institutions have been the “beneficiaries of disinflation and falling inter est rates,” pointing out that com m ercial bank earnings increased as the inflation rate fell.3 This article d iscu sses the effect of inflation on co m m ercial banks by analyzing the relationship betw een inflation and the m arket value of bank capital. INFLATION: A BR IEF EXPLANATION Inflation is an increase in the general price level, and is typically exp ressed as an annual p ercen tage rate of change. F o r exam ple, the GNP deflator (one index of the general price level) rose from 1.00 in 1982 to 1.038 in 1983, then in creased to 1.081 in 1984. In flation averaged 3.8 p ercen t during 19 8 2 -8 3 and about 4.1 p ercen t during 1 983-84. The average annual rate over the tw o-year interval w as about 3.9 p ercen t.4 G. J. Santoni is a senior economist at the Federal Reserve Bank of St. Louis. Thomas A. Poiimann provided research assistance. Inflation d ep reciates the value of m oney. An in flation rate of 4.0 p ercen t m ean s that the dollar falls in value at an annual rate of 4.0 p ercen t in term s of the goods it will buy. BANKS AND NOMINAL FINANCIAL INSTRUMENTS Inflation is im p ortan t for banks b ecau se they typi cally deal in nom inal financial instrum ents, that is, instrum ents den om in ated in fixed dollar am ounts. F o r exam ple, w hen a bank m akes a loan, it a cce p ts nom inal financial instru m en ts (notes, m ortgages, com m ercial p ap er an d o th e r financial securities) as evidence of the debtor's obligation to th e bank. W hen a bank borrow s, it issues nom inal financial instrum ents to creditors (deposit liabilities, a cce p ta n ce s an d d e bentures) as evidence of its obligation. An Important Characteristic While nom inal financial in stru m en ts differ from one an oth er in m any r esp ects, they share one im portant ch aracteristic: their paym ents are fixed in nom inal value, that is, in term s of dollars. Nominal instru m en ts make up the bulk of bank assets and liabilities. F u r therm ore, banks are typically net cred itors in nom inal instrum ents b ecau se their nom inal assets exceed their nom inal liabilities (see appendix 1 for a theoreti cal explanation).’ 'See Shaky Credit Structure (1985). 2lbid. 3See Corporate Earnings Uneven (1985). “There are various methods of computing average annual rates of change. The method employed in this paper assumes continuous compounding. The rates are calculated by dividing the difference between the natural logarithms of the price level at the two points in time by the number of intervening years and multiplying by 100. 5See Alchian and Kessel (1977a) and Kessel (1956). Of course, banks have real assets and liabilities as well (land, buildings, office equipment, equities, etc.). These, however, make up a very small portion of bank portfolios and are irrelevant in assessing the effect of inflation on banks. 15 FEDERAL RESERVE BANK OF ST. LOUIS MARCH 1986 Some Balance Sheet Data Table 1 uses d ata from the con solid ated balance sh eet of dom estically ch artered com m ercial banks to illustrate their net position in nom inal instruments.® Nominal assets are calculated by subtracting the val u es of bank prem ises, o th er real estate ow ned, the equities of o th er firms ow ned, and investm ent in sub sidiaries from total assets. T h ese item s are su b tracted b ecau se the m arket p rices of real assets vaiy directly with the price level. Nominal liabilities are the sum of total deposits, subordinated n otes an d debentures, federal funds p u rch ased, interest-bearing d em an d n otes, m ortgage indebtedness, all o th er liabilities for b orrow ed m oney an d the value of preferred stock. Preferred stock, w hich is sim ilar to a bond, is included as a nom inal liability b ecau se it is an obligation of the bank to pay a fixed stream of dollars. The table 1 d ata indicate that, in the aggregate, the nom inal assets of these banks exceed th eir nom inal liabilities. The excess of nom inal assets over nom inal liabilities am o u n ts to 66.3 p ercen t of bank capital.7 Table 1 Net Nominal Assets of All Commercial Banks (in billions of dollars) Nominal assets = total assets - real assets Total assets Less real assets: Equity ownership in other firms $10.7 Bank premises 39.7 Investment in subsidiaries 1.9 Real estate other than premises 6.9 Nominal assets 59.2 $2,621.8 Nominal liabilities1 = total liabilities + preferred stock Total liabilities Preferred stock Nominal liabilities Net nominal assets = nominal assets Nominal assets Nominal liabilities $2,681.0 $2,504.5 1.0 $2,505.5 nominal liabilities $2,621.8 2,505.5 Net nominal assets Net nominal assets as a percent of equity $ 116.3 66.3% ANTICIPATED INFLATION AND BANKS An increase in an ticip ated inflation raises the n om i nal interest rate. This in creases the n u m b er of dollars that cred itors o r debtors w ho are tran sactin g in nom i nal financial instru m en ts exp ect to receive o r pay w hen loans m atu re (see sh ad ed insert), ff th ese e x p e c tations are realized, all nom inal values will be higher at m aturity. Table 2 show s this effect on the balance sh eet of a hypothetical bank. The exam ple assu m es that all of the bank's borrow ing and lending co n tra cts w ere negotiated with the exp ectatio n that th e rate of inflation over the next two y ears w ould be 5.0 percen t. The bank's loan co n tra cts have a tw o-year life. Its borrow ing co n tracts m atu re at the en d of ea ch y e a r an d are renegotiated at the existing interest rate. Reserve requirem ents against all d eposits are 10.0 p ercen t. F o r simplicity, the real interest rate is assu m ed to be zero so the nom inal interest rate on bank loans is 5.0 percent." The interest rate on bank deposits is 4.5 6This calculation should be considered as illustrative only, since book values rather than market values are used. 7As above, capital excludes outstanding preferred stock. 8The assumption about the real rate has no qualitative effect on the results. See appendix 1. 16 NOTE: Data are as of third quarter 1985. 'All bank liabilities are nominal. SOURCE: Board of Governors of the Federal Reserve System p ercen t. The sp read betw een the bank’s borrow ing an d lending rates is n e ce ssa iy co m p en sation for the requirem ent to hold n on -in terest-earn in g reserves (see ap p en dix 1). All assets an d liabilities are valued at m arket p rices so assets m inus liabilities, o r capital, rep resen ts the m arket value of the bank, $200, in this exam ple. The general level of p rices is 1.0 in panel A. Panel B show s the balan ce sh eet of the bank at the en d of the first y e a r assum ing th at the an ticip ated inflation is realized an d that nothing else h as o ccu rre d to ch an ge th e a cco u n t b alances. Acci-ued interest on deposit liabilities is $45, while $50 interest h as a ccru e d on bank loans. A portion of the bank’s interest e a rn ings ($4.50) m u st be ad d ed to reseives to co v er the increase in deposits. The nom inal value of bank ca p i tal has in creased to $210.00, but its real value is u n ch an ged ($200.00). A sim ilar result o ccu rs in panel C, w hich show s the balance sheet at the end of the secon d y ear. The an ticip a te d inflation h as no real effect o n th e bank’s capital and, therefore, on the w ealth of its stockholders. MARCH 1986 FEDERAL RESERVE BANK OF ST. LOUIS Forecastin g Inflation Anticipated and Unanticipated Inflation Many eco n om ic tran saction s require a co m m it m ent to exch an ge m oney at som e future time. Credit transaction s are a good exam ple of this. Since inflation red u ces the future value of m oney, it pays people (both potential b orrow ers and lenders) to tiy to forecast inflation over the relevant time period. This forecast is called anticipated inflation.1 As the nam e suggests, anticipated inflation is forward-looking. It is the rate of ch an ge in the general price level that people think will o c c u r during som e specific future tim e period. Of course, the a ccu ra cy of inflation forecasts d e pends on future events an d circu m stan ces that are unknow n w hen the forecast is m ade. C on se quently, these forecasts generally will be “w ron g.”Any difference betw een actu al (or realized) inflation and anticip ated inflation is called unanticipated inflation. U nanticipated inflation is known only with hindsight. B ecause it is known only after the fact, it plays no role in people's decisions, ft is im portant, however, in assessing w h eth er the d eci sions p rodu ced profits o r losses. Anticipated Inflation and Interest Rates The nom inal interest rates quoted in financial m arkets are form ed in th e p ro cess of con tractin g betw een borrow ers an d lenders. They indicate the nu m b er of dollars the b orrow er m ust pay to the cred itor in the future in exch an ge for a given n u m ber of present dollars. If borrow ers and lenders exp ect the value of the dollar to d ep reciate in term s of the goods it will buy over th e life of the loan (i.e., if they anticipate inflation), the nom inal interest rate specified in the loan co n tra ct will take a cco u n t of this. The interest rate will be sufficiently high to co m p en sate for the e x p e cte d d ep reciation in the value of the dollar.3 To illustrate, su p p ose the real interest rate is 3 p ercen t and the an ticip ated rate of inflation over the com ing y e a r is 5 p ercen t. People think that it wall take $1.05 one y e a r from now to p u rch ase the goods that $1.00 will buy today. A loan of $1,000 for one y e a r will require a paym ent of $1,081.50 ( = $1,000.00 x 1.03 x 1.05) at m aturity. This implies a nom inal interest rate of 8.15 p ercen t.4 The an tici pated real value of this am ou nt at m aturity is $1,030 ( = $1,081.50/1.05), w hich is the sum of the principal ($1,0001 and the real retu rn ($30). 3See Fisher (1965), pp. 1-100, who relates the nominal interest rate, i, to the ex ante real interest rate, r, and the anticipated rate of inflation, n*, as follows: i 'See Alchian and Allen (1977), p. 490. They note that “though odd names are given to inflation (creeping, galloping, runaway, and hyper-), a critical distinction is between unanticipated and antici pated inflation.” 2Economic theory suggests that, although “wrong,” the forecasts will not consistently over- or underpredict; that is, forecasts will be unbiased. See Fisher (1954), pp. 36-37; Fisher (1907), p. 213; and Fama (1970) for discussions of business foresight and effi cient markets. UNANTICIPATED INFLATION AND BANKS If the realized rate of inflation exceed s the an tici pated rate, the price level h as risen u n exp ectedly. T he u n exp ected increase in the price level cau ses a p ro portional red u ction in the exch an g e value of both nom inal financial assets and liabilities in term s of real goods. B ecause banks are typically net cred itors in nom inal instrum ents, bank ow ners lose w ealth w hen = r+ it* + (r)(iT*). Fisher’s hypothesis regarding the formation of the nominal inter est rate is an approximation to the "true” relationship; it ignores risk premiums and the effect of taxes on interest income and assumes that the anticipated rate of inflation is held with certainty. See Darby (1975) and Kochin (1981). The ex ante real rate is the premium in terms of real goods that creditors expect to receive (and borrowers expect to pay) expressed as a percentage of the principal of the loan. See Fisher (1954) and Santoni and Stone (1981) for discussions of the real rate of interest. 48.15 = [($1,081.50/$1,000.00)-1]100. there is unan ticip ated inflation (that is, w hen bank capital declines) * Table 3 p resen ts a n um erical exam ple of this effect. The assu m p tions in table 3 are the sam e as those in table 2 excep t that people are surp rised by a 10 p er cen t increase in the p rice level in the first y ear. The exam ple assu m es that the surprise is interp reted as a 9See Keynes (1923), pp. 18-19, and Alchian and Kessel (1977b). 17 FEDERAL RESERVE BANK OF ST. LOUIS MARCH 1986 T a b le 2 The Effect of an Anticipated Inflation of 5.0 Percent Panel A: The current balance sheet. The price level is 1.0. Assets: Reserves Loans and securities Premises $ 100.00 1,000.00 100.00 Liabilities: Deposits Capital: $1,000.00 200.00 $1,200.00 $1,200.00 Panel B: The balance sheet one year hence. The price level is 1.05. Assets: Reserves Loans and securities Premises $ 104.50 1,045.50 105.00 Liabilities: Deposits Capital: $1,045.00 210.00 $1,255.00 $1,255.00 Real value of capital = $210/1.05 = $200 Panel C: The balance sheet two years hence. The price evel is 1.1025. Assets: Reserves Loans and securities Premises $ 109.20 1,093.07 110.25 Liabilities: Deposits Capital: $1,092.02 220.50 $1,312.52 $1,312.52 Real value of capital = $220.5/1.1025 = $200 one-tim e-only deviation in the p rice level so that the inflation forecast for the secon d y e a r rem ains at 5.0 percent. Panel B show s the effect of the un an ticip ated in flation. The bank’s nom inal assets an d liabilities are unaffected by the inflationaiv surprise. The increase in these dollar values is fixed by co n tract. In co n trast, the nom inal value of the bank’s real assets, th e p rem ises, increases by the realized rate of inflation, 10.0 p ercen t. The nom inal value of capital increases from $200 to $215. The value of the bank in term s of the real goods for w hich it can be exchanged, however, d e clines to $195.45. The u n anticip ated inflation cau ses the real value of the bank to fall by $4.55."' 10An unanticipated decrease in the price level produces symmetrical results in that the real wealth of bank owners is increased. 18 Panel C show s the bank’s b alan ce sheet at the en d of the seco n d y ear. The real value of the bank rem ains at $195.45, indicating that th e one-tim e inflationary su r prise p ro d u ces a p erm an en t red u ction in the real value of the bank even though th e rate of inflation in subsequent y ears retu rn s to 5.0 p ercen t. THE INTEREST RATE AND INFLATION A bank’s nom inal financial assets an d liabilities typi cally m atu re at different dates. At any given m om ent, the m aturity dates of a bank’s assets generally extend beyond those of its liabilities." In o th e r w ords, an "More precisely, the duration of the bank’s receipt stream exceeds the duration of its payment stream. For discussions of duration, see Samuelson (1945), p. 19; Bierwag, Kaufman and Toevs (1983); Maisel and Jacobson (1978); and Santoni (1984). MARCH 1986 FEDERAL RESERVE BANK OF ST. LOUIS T a b le 3 The Effect of an Unanticipated Inflation Panel A: The current balance sheet. The price level is 1.0. Assets: Reserves Loans and securities Premises Liabilities: Deposits Capital: $ 100.00 1,000.00 100.00 $1,000.00 200.00 $1,200.00 $1,200.00 Nominal assets - nominal liabilities = ($100 + $1,000) - $1,000 = $100 Real value of capital = $200/1.0 = $200 Panel B: The balance sheet one year hence. The price level is 1.10. Assets: Reserves Loans and securities Premises $ 104.50 1,045.50 110.00 Liabilities: Deposits Capital: $1,045.00 215.00 $1,260.00 $1,260.00 Real value of capital = $215/1.10 = $195.45 Panel C: The balance sheet two years hence. The price level is 1.155. Assets: Reserves Loans and securities Premises $ 109.20 1,093.07 115.50 Liabilities: Deposits Capital: $1,092.02 225.75 $1,317.77 $1,317.77 Real value of capital = $225.75/1.155 = $195.45 interest rate change affects the paym ent stream obli gated by the bank’s liabilities before it affects the bank's receip t stream . Consequently, an increase in the interest rate red u ces the exp ected net stream of dollar receip ts as the bank’s cred ito rs renegotiate for the higher interest rate, while the interest rate earn ed by the bank on its existing loans is locked up. Of cou rse, the loans eventually m atu re and are renegoti ated at the higher nom inal rate, but the bank’s capital is red u ced nonetheless. An Illustration Table 4 illustrates the effect of a ch an ge in the nom inal interest rate on bank capital. The exam ple assu m es the interest rate in creases b ecau se antici pated inflation in creases. The qualitative effect illus trated by the exam ple, however, results from the change in the interest rate, regardless of w hat p ro d u ced the change. In this exam ple, an ticip ated inflation at the tim e the bank initially co n tra cts with its cred itors an d debtor's is 5.0 p ercen t. The bank’s co n tra cts with its creditors m atu re in one year, while its loans m atu re at the end of the secon d y e a r and can n o t be renegotiated before maturity. As in the previous exam ples, the bank’s loans and 19 FEDERAL RESERVE BANK OF ST. LOUIS MARCH 1986 T a b le 4 The Effect of a Change in Anticipated Inflation Panel A: The current balance sheet. The price level is 1.0. Assets: Reserves Loans and securities Premises $ 100.00 1,000.00 100.00 Liabilities: Deposits Capital: $1,000.00 200.00 $1,200.00 $1,200.00 Panel B: The balance sheet one year hence. The price level is 1.05. Assets: Reserves Loans and securities Premises $ 104.50 1,045.50 105.00 Liabilities: Deposits Capital: $1,045.00 210.00 $1,255.00 $1,255.00 Real value of capital = $210.00/1.05 = $200.00 Panel C: The balance sheet two years hence. The price level is 1.155. Assets: Reserves Loans and securities Premises $ 113.90 1,090.65 115.50 Liabilities: Deposits Capital: $1,139.05 181.00 $1,320.05 $1,320.05 Real value of capital = $181.00/1.155 = $156.71 deposits are $1,000. The lending rate is 5.0 percent, an d th e borrow ing rate is 4.5 p ercen t. Panel A show s the bank’s initial balance sheet. Panel B show s the balance sheet at the end of th e first y e a r assum ing that the realized rate of inflation during the first y e a r w as 5.0 p ercen t, the sam e as the an ticip ated rate. Panel C show s the b alance sh eet at the en d of the secon d y e a r assum ing that the anticip ated rate of inflation w as revised upw ard to 10.0 p ercen t at the beginning of the secon d year, just before th e bank renegotiated its co n tra cts with d e p o s ito r . The ex a m ple assu m es that the realized rate of inflation during the seco n d y e a r m atch es the an ticip ated rate. The in crease in an ticipated inflation cau se s the nom inal interest rate to rise to 10.0 p ercen t during the secon d year, while the interest rate on bank deposits Digitized for20 FRASER in creases to 9.0 p ercen t. At th e en d of th e secon d year, these deposits will am ou nt to $1,139.05 ( = $1,045.00 x 1.09). T he bank, how ever, is prevented from raising the interest rate on its existing loans ( = $1,000) by the term s of its co n tra ct. T h ese loans co n tin u e to yield 5.0 p ercen t in the se co n d year, a ccn jin g earnings of $50.00 during th e y e a r. Of co u rse, th e bank ca n m ake new loans of $45.50 at the beginning of the secon d y ear. These new loans, w hich result from th e net interest earnings th e bank obtained the first y ear, are m ad e at the h igher interest rate (10.0 percen t) and a ccru e earnings of $4.55 ( = $45.50 X .10) at y ear-en d . Since bank deposits in creased during the secon d year, som e of th e bank’s interest earnings m u st be used to in crease reserves. The in crease in bank d e posits am o u n ts to $94.05 ( = $1,139.05 — $1,045.00), so FEDERAL RESERVE BANK OF ST. LOUIS reseives m ust increase to $113.90 o r by $9.40 ( = $94.05 X .10). As a result, the bank’s loan a cc o u n t aty e a r-e n d is $1,090.65 ( = $1,000.00 + $50.00 + $45.50 + $4.55 $9.40). Bank prem ises in crease in nom inal value by the realized rate of inflation and am ou nt to $115.50 ( = $105.00 x 1.10) a ty e a r-e n d . Note that both the nom i nal an d real value of capital decline. The real value of capital falls by $43.29 to $156.71 ( = $181.00/1.155). As th e table 4 exam ple show s, a ch an ge in the interest rate can have a fairly substantial effect on the bank w hen the m aturities of the bank’s assets and liabilities differ. In this p articu lar exam ple, the real value of capital declined by about 22.0 p ercen t w hen the interest rate doubled. An in crease in an ticip ated inflation affects banks in a w ay that is qualitatively the sam e as u n an ticip ated inflation. This is b ecau se the upw ard revision in an tic ipated inflation that o ccu rs at the end of the first y e a r w as not forecast at the beginning of the y ear. If people had anticipated inflation of 5.0 p ercen t the first y e a r an d 10.0 p ercen t the seco n d y ear, the rate of interest on tw o-year loans w ould not have been 5.0 percen t. Rather, it w ould have been higher to reflect the fact that an ticip ated inflation averages 7.2 p ercen t acro ss the tw o y ears. U nanticipated inflation and ch an ges in anticip ated inflation have sim ilar effects b ecau se both reflect a m isguess about inflation. To recap the main points so far, th e previous d iscu s sion suggests that inflation affects the real capital value of banks through two chann els. First, capital value falls w hen the actu al rate of inflation exceed s the anticip ated rate. This is called u n an ticip ated inflation. Second, capital value falls w hen the an ticip ated rate of inflation is revised upw ard, b ecau se this cau ses nom i nal interest rates to rise u n exp ectedly. The reverse o ccu rs if the actu al rate of inflation falls short of the: anticip ated rate o r if the anticip ated rate of inflation is revised dow nw ard. SOME ESTIMATES These im plications can be exam ined by observing the effect of inflation on various indexes of the stock p rices of publicly trad ed banks. Stock p rices are used as a proxy for the capital value of banks b ecau se they rep resent the m arket’s assessm en t of the p resent value of the future net receip ts banks aie exp ected to generate. The relationship betw een ch an g es in the real stock prices of banks and the o th er variables is assu m ed to take the form show n in equation 1: MARCH 1986 (1) ALnlV/P), = C + aALny, + (iALnli + Sir* + it*), + -yir;' e AL t]tt*, w here V/P C y (i —tt*) = = = = the real p rice of bank stock, a con stan t, real incom e, the nom inal interest rate less an ticip ated inflation, tt" = unan ticip ated inflation, w hich is th e dif ference betw een realized inflation, tt, and anticipated inflation, tt*. tt“^U. Equation 1 exp resses real stock p rices, real incom e, the interest rate residual an d the ch an ge in an tici pated inflation in term s of annualized p ercen tage ch an ges. The u n an ticip ated rate of inflation is the difference betw een two annualized p ercen tage rates of ch an ge: the realized rate of inflation an d th e an tici p ated rate. The estim ates use quarterly data from the first quarter of 1962 through the fourth q u arter of 1984. Anticipated and Unanticipated Inflation M easuring an ticip ated inflation is a problem . Since w e only observe actu al inflation, various analysts have used different m eth od s to estim ate an ticip ated in flation. This study estim ates an ticip ated inflation o ne quar ter ah ead by em ploying a tim e-series forecast of in flation. This m ethod gen erates pred iction s of inflation solely on the basis of its past behavior.13The difference betw een the actual rate of inflation and the rate fore cast by the model is interp reted as the em pirical co u n terp art of u n an ticip ated inflation, t t " . B ecau se the real price of bank stock is e x p ected to be inversely related to unan ticip ated inflation, the coefficient of tt" should be negative. A nticipated inflation, tt*, an d ch an ges in the real p rice of bank stock are theoretically u n related ; co n se quently, the coefficient of this variable sh ou ld be zero. C hanges in anticip ated inflation ch an ge interest rates, 12See Hafer and Hein (1985). 13Roughly, the technique accounts for the past pattern of inflation by estimating a model that provides a description of the process that generated the observed series. Past observations of inflation are then used along with the information contained in the time-series model to forecast inflation one period ahead. For further discussion of time-series models and their properties, see Pindyck and Rubinfeld (1981), pp. 469-573, especially pp. 469-70 and 493-97. For further discussion of the model employed here, see appendix 2. 21 FEDERAL RESERVE BANK OF ST. LOUIS however, and these interest rate ch an ges are exp ected to be inversely related to ch an g es in stock prices. E stim ates of the ch an ges in anticip ated inflation are obtained directly from the inflation fo recasts.14 MARCH 1986 an d ch an ges in the difference betw een the nom inal rate an d the estim ate of an ticip ated inflation, A(i —-it*), ca n be includ ed separately in the regression eq u a tion .1'’ The e x p e cte d sign of ea ch of th ese term s is negative. The Business Cycle Real incom e grow th w as included as an explanatory variable to con trol for the effect of the business cycle on bank earnings. Business exp an sion s increase the real quantity of bank loans, secu rities an d deposits, w hich is thought to have a positive im p act on the exp ected earnings stream . The em pirical co u n terp art of real incom e used in the regressions is gross national p ro d u ct (GNP) divided by the GNP deflator. The e x p e cted sign of the coefficient of this term is positive. The Interest Rate The p rices of bank stocks are exp ected to be related to ch an ges in the interest rate. The interest rate will vary with chan ges in the cx ante real interest rate, ch an ges in incom e tax laws, ch an ges in risk prem ium s an d ch an ges in an ticip ated inflation. Since the interest rate includes all of these factors, ch an g es in it can n o t be readily attributed to the effect of any one of them . The qualitative effect of a ch an ge in the interest rate on stock prices, however, is the sam e regardless of the so u rce. The exp ected sign of the coefficient of interest rate ch an ges is negative. Since this p ap er focu ses on the effect of inflation, the following estim ates attem p t to isolate the effect of a ch an ge in an ticip ated inflation. As m en tioned above, an estim ate of anticip ated inflation, -it*, is p ro d u ced by the tim e-series forecast of inflation. W hen this esti m ate is su b tracted from the nom inal interest rate, the residual is an estim ate of the nom inal interest rate e x c lu d in g a n tic ip a te d in flatio n . C o n s e q u e n tly , ch an ges in the estim ate of an ticip ated inflation, Att*, ’“Changes in the interest rate are expected to be positively related to changes in anticipated inflation. To check this, changes in the Aaa bond rate, AR, and changes in the 3-month Treasury bill rate, ARS, were regressed on the estimate of the change in inflation expecta tions. The results are presented below. AR, = .09 + .13Air*, (1.93) (3.03)* ARS, = .06 + .36 Air*, (.58) (3.71)* DW = 1.65 DW = 1.79 R2 = .10 R2 = .14 Although both coefficients are less than one, they are both positive and significantly different from zero. The estimated coefficient of Air* is larger in the equation for the short-term interest rate, which suggests that the inflation forecast used here is a better estimate of short-run expectations. Digitized for 22 FRASER Controlling f o r Problem Loans In addition to the above variables, th e estim ated equations include a d u m m y variable to con trol for the effect that re ce n t Latin A m erican loan problem s have had on bank stock p rices. D uring th e early p art of 1982, it b ecam e ap p aren t that certain Latin A m erican co u n tries w ould have difficulty h onoring th eir obligations to U.S. banks. In O ctob er an d November 1982, the cen tral bank of Brazil began borrow ing heavily from the E xch an ge Stabilization F u n d of the U.S. Treasu ry; M exico began draw ing heavily on its sw ap arran ge m en t w ith the Federal Reserve System in April of the sam e y ear. News rep o rts on the exten t of the problem co n tin u ed to su rface for about three qu arters. The period du m m ied begins in the first q u arter of 1982 and exten d s throu gh the third q u arter of 1982, w hen it b ecam e evident th at the U.S. governm ent w ould take an active role in resolving th e problem ."1Th e e x p e cte d sign of the du m m y is negative. The Estimates Table 5 p resen ts the regression results. E stim ate 1 exam ines the effect of inflation on an ind ex of th e real sh are p rices of banks located outside New York City. Estim ate 2 d oes th e sam e thing for New York City banks.17 Th e signs of the estim ated coefficients are as e x p ected . The estim ates indicate th at u n an ticip ated in flation an d ch an ges in the an ticip ated rate of inflation are inversely related to ch an g es in the real p rice of bank stock. As exp ected , the estim ated coefficient of an ticip ated inflation is not significantly different from zero in a statistical sense. 15Actually, the data entry is one plus the difference between the nominal rate and anticipated inflation. This is necessary because the difference is negative in some quarters during 1971, 1975 and 1976. See Brown and Santoni (1981) for a discussion of some problems associated with this method of separating the nominal rate into its various components. ,6On February 2, 1983, the chairman of the Federal Reserve Board addressed the House Committee on Banking, Finance and Urban Affairs regarding the problem and measures to deal with it. See Volcker (1983). 'This was done because Standard and Poor’s reports the data this way. MARCH 1986 FEDERAL RESERVE BANK OF ST. LOUIS Table 5 Estimating the Effect of Inflation on the Price of Bank Shares, Sample Period: I/1962-IV/1985 Estimate 1: ALnBK/P = -4.29 + 1.79ALny - 4.72ALn(1 + i (.47) (2.21)* (3.91)* it*) - 1.48-ir1' + .05ir* -,12ALmr* - 11.75 DUM (3.19)* (.15) (3.01)* (2.64)* RSQ = .34 DW = 1.68 Estimate 2: ALnBKNY/P = 5.94 + 1.10ALny - 5.51 ALn(1 + i-n*) (. 64 ) ( 1.33 ) (4 . 44 )* 1 .34 t t " + ,35n* - ,12ALmr* (. 99 ) (2 . 97 )* (2 .81 )* 6.05 DUM ( 1 .32 ) RSQ = .31 DW = 1.74 where: BK/P BKNY/P y i ir“ ir* DUM = = = = = the Standard and Poor’s index of the real share prices of banks located outside New York City the Standard and Poor’s index of the real share prices of New York City banks Real Gross National Product the corporate Aaa bond rate unanticipated inflation anticipated inflation = 1 during 1/1982-111/1982 and zero otherwise NOTE: Absolute values of t-scores appear in parentheses. * = significantly different from zero at the 5 percent level. The coefficient of the d u m m y variable has the e x p ected sign in both estim ates but is not significant in estim ate 2. In the ca se of estim ate 1, th e coefficient is significant an d its point estim ate is fairly large, sug gesting that the grow th in real stock prices w as about 12 p ercen t lower, on average, during the first three quarters of 1982, o th e r things th e sam e. This m ay be som ew hat m isleading since the con fiden ce interval for this coefficient ranges from —2.9 to —20.7. Implications f o r Banks The average forecast of inflation (tt*) g enerated by the tim e-series m odel during 1984 w as about 4.0 per cen t. This fell to about 3.5 p ercen t during 1985, result ing in a 13.5 p ercen t d rop in an ticip ated inflation (ALmr*). The table 5 estim ates suggest that this raised the real stock p rices of the banks in the sam ple bv about 1.6 p ercen t ( = —13.5 X — .12). In addition, the decline in the actual rate of inflation exceed ed the decline in an ticip ated inflation. As a result, u n an tici p ated inflation averaged about —.85 p ercen t in 1985, raising the real stock prices of banks by an additional 1.2 p ercen t ( = — .85 X 1.4). In sum , the real stock p rices of banks in creased by about 3.0 p ercen t, ceteris paribus, as a co n seq u en ce of the fall in anticipated inflation in 1985 and b ecau se th e actu al rate of in flation in 1985 w as even low er than an ticip ated. CONCLUSION This p ap er exam ines the effect that inflation has on the share p rices of co m m ercial banks. The results indicate that the real sh are p rices of banks are in versely related to both u n an ticip ated inflation — that is, deviations in the realized rate of inflation from its anticip ated rate — and ch an g es in an ticip ated in flation. C ontrary to som e claim s, this evidence indi ca te s that bank shareh old ers have benefited from the recen t decline in the rate of inflation an d that any u n exp ected resurgen ce of inflation w ould be harmful. REFERENCES Alchian, Armen A., and William R. Allen. Exchange and Production: Competition, Coordination and Control, 2nd ed. (Wadsworth Pub lishing Company, Inc., 1977), pp. 490 - 94 . 23 FEDERAL RESERVE BANK OF ST. LOUIS Alchian, Armen A., and Reuben A. Kessel. "Effects of Inflation,” Economic Forces at Work (Liberty Press, 1977a), pp. 363-96. _______ and_______ _ “Redistribution of Wealth Through Infla tion,” Economic Forces at Work (Liberty Press, 1977b), pp. 397412. Bierwag, G. O., George G. Kaufman and Alden Toevs. “Duration: Its Development and Use in Bond Portfolio Management,” Finan cial Analysts Journal (July/August 1983), pp. 15-35. Brown, W. W., and G. J. Santoni. “Unreal Estimates of the Real Rate of Interest,” this Review (January 1981), pp. 18-26. “Corporate Earnings Uneven,” New York Times, November 4,1985. Darby, Michael R. “The Financial and Tax Effects of Monetary Policy on Interest Rates,” Economic Inquiry (June 1975), pp. 26676. Fama, Eugene F. "Efficient Capital Markets: A Review of Theory and Empirical Work,” Journal of Finance, Papers and Proceedings (May 1970), pp. 383-417. Fisher, Irving. pp. 1-100. Appreciation and Interest (Augustus M. Kelley, 1965), MARCH 1986 Kessel, Reuben A. “Inflation-Caused Wealth Redistribution: A Test of a Hypothesis,” American Economic Review (March 1956), pp. 128-41. Keynes, J. M. A Tract on Monetary Reform (Macmillan and Com pany, 1923). Kochin, Levis. 'The Term Structure of Interest Rates and Uncertain Inflation" (University of Washington, 1981; processed). Maisel, Sherman J., and Robert Jacobson. “Interest Rate Changes and Commercial Bank Revenues and Costs,” Journal of Financial and Quantitative Analysis (November 1978), pp. 687-700. Pindyck, Robert S., and Daniel L. Rubinfeld. Econometric Models and Economic Forecasts, 2nd ed. (McGraw-Hill Book Company, 1981), pp. 469-573. Samuelson, Paul A. “The Effect of Interest Rate Increases on the Banking System,” American Economic Review (March 1945), pp. 16-27. Santoni, G. J. “Interest Rate Risk and the Stock Prices of Financial Institutions,” this Review (August/September 1984), pp. 12-20. ________ The Theory of Interest (Kelley and Millman, 1954). _______ , and Courtenay C. Stone. “Navigating Through the In terest Rate Morass: Some Basic Principles,” this Review (March 1981), pp. 11-18. _______ _ “The Shaky Credit Structure.” The Rate of Interest (The Macmillan Company, 1907). Hafer, R. W., and Scott E. Hein. “On the Accuracy of Time-Series, Interest Rate, and Survey Forecasts of Inflation," The Journal of Business (October 1985), pp. 377-98. 24 Washington Post, November 4,1985. Volcker, Paul A. Statement Before the Committee on Banking, Fi nance and Urban Affairs, House of Representatives, February 2, 1983. FEDERAL RESERVE BANK OF ST. LOUIS MARCH 1986 APPENDIX 1 Some Banking Arithm etic In large part, a bank’s exp ected stream of net reve nue is generated by its holdings of nom inal assets and liabilities. These are its loans, L, w hich earn the market interest rate, i„„ and its deposits, D, on w hich interest, i„, is paid. In addition, ow ners have invested capital, I, of w hich a fraction, a , m ust be held as non-interestearning reserves against deposits, and the rem aining fraction, (1 — a), is held in eith er nom inal o r real assets that are exp ected to yield the m arket rate, i,„. The following assu m es that this rem ain der is held entirely in net real assets. The exp ected stream of net revenue, R, is given in equation 1: (II R = i„,L — i„D + i„,(l — all. If the required reserve ratio is p, required reseives are pD = a l. The quantity of loans and d eposits the bank ca n generate are L = □ = al/p. In equilibrium, the exp ected revenue stream of the bank m ust equal the alternative stream of earnings that could be ob tained by investing the capital at the m arket interest rate. This is show n in equation 2: (2) R = i,„L — i„D + i,„(l — all = i,„cd + i,„(l — all. The capital value of the bank, K, is show n in equation 3: (31 K = — = a l + (1 - all = I. i,„ Equation 3 exp resses the capital value of the bank as the p resent value of the stream of exp ected revenue. In equilibrium, K = I. If K wore g reater than I, resou rces w ould be attracted to banking sin ce the capital value of forming a bank w ould e x ceed the opportunity cost. If K w ere less than I, resou rces w ould leave the industiv. The Equilibrium Interest Rate on Bank Deposits Substituting pD for a l in equation 2 and noting that D = L, the equilibrium interest rati; on bank deposits is given in equation 4: 14) i„ = (1 — p!i,„. Banks as Net Creditors in Nominal Assets Net nom inal assets, NNA, are nom inal assets m inus nom inal liabilities. The bank’s nom inal assets are the sum of its loans and reserves, while the bank’s d e posits are its nom inal liabilities. Assum ing equilib rium, these are given in equation 5: (5) NNA = L + pD - D = pD. Under these assum ptions, the bank is a net cred ito r in nom inal assets to the extent of its reserve holdings. APPENDIX 2 A Tim e-Series F o re ca st of Inflation While the initial obseivation for the regressions re ported in the text is first q u arter 1962, the d ata period used to develop the forecast of inflation (as m easured by the GNP deflator) extend s back to first q u arter 1948. A backw ard extension is n ecessaiy to get th e fo recast ing model started. Since the period covered is quite long, a rough ch eck of the data w as m ade to d eterm ine if the p ro cess that generated the tim e series ch an ged materially over the period I/1948-IV/1985. To do so, a m odel was first estim ated for I/1948-IV/1965 an d these results w ere co m p ared with the results obtained from esti m ates for I/1966TV/1985. The GNP deflator ap p ears to be a seco n d -o rd er h om ogeneou s p ro cess that can be m odeled as ARIMA (0, 2, 1). The estim ated m odels for the two periods are rep orted below. C alculated tstatistics ap p ear in p aren th eses, and B is a backward shift operator, i.e., (1-B) X, = X, — X,_,. FEDERAL RESERVE BANK OF ST. LOUIS III948-IV/l 965 AT.nP, = - .107 + (1 - .49B)e, (.58) (4.73) Chi-square (2, 24) = 16.07 HI966-IV/l 985 A-LnP, = .003 + (1 - .48B)e, (.03) (4.70) C hi-square (2, 24) = 26.63 26 MARCH 1986 A m odel w as then estim ated for th e period I/1948-IV/ 1961, and a forecast of inflation for 1/1962 w as m ade. The difference betw een the realized inflation rate for 1/ 1962 and this forecast is in terp reted as th e em pirical co u n terp art of tt". The forecast for the next quarter, 11/1962, is g en er ated by adding the realized inflation rate for 1/1962 to the d ata an d re-estim ating the m odel throu gh 1/1962 an d proceed in g as above. The p ro cess w as rep eated for each q u arter through IV/1985.