The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
ESSAYS ON ISSUES THE FEDERAL RESERVE BANK OF CHICAGO AUGUST 1994 NUMBER 84 Chicago Fed Letter The dollar and the federal funds rate In early February the Federal Open Market Committee announced a tightening of monetary policy; as of June, this has been reflected in a 125basis-point increase in the federal funds rate from 3.0% to 4.25%. In light of this action, recent movements in the dollar against the German mark and Japanese yen seem perplex ing. From February 4 to June 30, the dollar depreciated by 9.9% against the mark and 10.2% against the yen. The conventional wisdom would pre dict that the dollar should strengthen following a tightening of U.S. mone tary policy. Yet the dollar depreciated despite central bank interventions in the foreign exchange markets to support it. To take another example, on Wednesday, May 11, the Bundes bank cut its discount and Lombard rates by 50 basis points each. The Wall StreetJournal characterized this as “an unexpectedly deep” cut.1 The day before, the dollar had traded at 1.6735 marks; by Wednesday’s close in New York, the dollar was trading lower—at 1.6695 marks. This Chicago Fed Letter investigates the empirical relationship between the federal funds rate and the yen/dollar and m ark/dollar exchange rates in the 1980s and 1990s. The evidence from this period shows that sustained and large increases in the federal funds rate led to an appreciation of the dollar, but it often took two years for these effects to take noticeable hold. The effects of monetary policy actions are difficult to detect initially because exchange rate fluctuations over a one-year period are largely composed of unexplained shocks. Nevertheless, movements in the fed eral funds rate since the most recent business cycle peak (mid-1990) ex plain a surprisingly large fraction of the long-term movements in the yen/ dollar exchange rate and somewhat less in the m ark/dollar rate. Forecasting exchange rate movements Accurately forecasting movements in the dollar over short horizons is a difficult exercise. In an influential study, economists Richard Meese and Kenneth Rogoff concluded that so phisticated models of exchange rate determ ination make poor forecasts.2 These economists suggested a thought experim ent that, in the context of today’s foreign exchange markets, could be described as follows: On February 4, Chairman Greenspan announced that “the Federal Open Market Committee decided to in crease slightly the degree of pressure on reserve positions.” This action sent the federal funds rate to 3.25% from its previous average of 3.0%. Immedi ately following the announcem ent, the dollar was trading at 1.752 marks and 109.0 yen. Since the historical pattern of the federal funds rate is to continue rising for some time following an ini tial upward change in direction, what values of the dollar should a sophisti cated analyst forecast for February 4, 1995? Or 1996? Of the models they considered, Meese and Rogoff con cluded that the best forecasting model would simply predict no change, that is, 1.752 marks and 109.0 yen for 1995 and 1996. Only at horizons of about two or three years can the additional information about the behavior of the economy add forecasting power to these models. One limitation of standard forecasting models is that they usually identify changes in monetary policy with changes in monetary aggregates such as Ml or M2. A recent study by Eichenbaum and Evans, however, indicates a stronger empirical rela tionship between the federal funds rate and the dollar.3 When expan sionary (contractionary) changes in monetary policy are measured by unanticipated reductions (increases) in the federal funds rate, the data indicate that the dollar will depreciate (appreciate)—but there is a substan tial delay between the policy actions and the maximal effect on the dollar. Assessing the average relationship between the federal funds rate and the exchange rate For both Germany and Japan, I used a statistical model to analyze three financial variables: the federal funds rate, a short-term interest rate in the foreign country, and the exchange rate. The exchange rate was mea sured in marks per dollar and yen per dollar. The statistical model was a vector autoregression estimated using weekly data. The German mod el was estimated for the period March 1979 through June 1994. Because of subtle statistical issues involving trends in the yen/dollar exchange rate, the Japanese model was estimat ed over the period July 1987 though June 1994. (Estimating the Japanese model over the 1979 to 1994 period would have attributed an even greater explanatory role to the federal funds rate in affecting the yen/dollar ex change rate.) The analysis accounts for changes in U.S. and foreign m one tary policies through exogenous changes in the federal funds rate and the foreign short-term interest rate. This Fed Letter reports only on the effects that changes in the fed eral funds rate have on the value of the dollar.4 1. Effects of FF and YEN shocks Effect of FF shock on federal funds rate Effect of YEN shock on federal funds rate basis points basis points Effect of FF shock on yen/dollar Effect of YEN shock on yen/dollar percentage app reciation o f do lla r percentage app reciation o f do lla r 1.5 r 1.5 r 1.0 - ------------- 1 --------------- 1_________ i_________i_________ i_________i_ 1 26 52 78 w e eks 104 130 156 _q 5 _________i__________i_________ i_________i_________ I_________L. 1 26 52 78 104 w eeks 130 156 Note: Impulse response functions estim ated from three-variable vector autoregression of federal funds rate, three-m onth Japanese bill rate, and yen/dollar exchange rate. Data cover July 1987 through June 1994. How do unusual movements in the federal funds rate affect the time path of the dollar? I will refer to such un usual movements as FF shocks. These shocks can be interpreted in the fol lowing way. Much as bond market traders forecast a value for interest rates each week, the statistical model can generate forecasts of the federal funds rate each week on the basis of its past values as well as past values of the foreign interest rate and the ex change rate. Then the new federal funds rate occurs this week. If the forecast is either lower or higher than the actual value, then the forecast is in error and the difference is called an FF shock. When the federal funds rate is unexpectedly high (low), as on February 4, that is a positive (nega tive) FF shock; this can also be re ferred to as a contractionary (expan sionary) policy shock. The first column of figure 1 plots the average effect of an FF shock on the federal funds rate (FF) and on the yen/dollar exchange rate (YEN, mea sured as yen per dollar). An average sized FF shock causes the federal funds rate to increase approximately 13 basis points during the first six months. So a 25-basis-point shock is a two-standard-deviation shock over the 1987-94 sample period. These shocks are larger over the 1979-94 period. This effect persists for about 18 months and then dampens. The ef fect of this approximately 13-basispoint FF shock on YEN is initially small, and even perversely negative for about six months. By the end of the first year, however, the dollar has appreciated by about 0.5%, and this appreciation persists through the end of the third year. These estimates imply that a series of FF shocks total ling 125 basis points could increase the value of the dollar by 5% over a two- to three-year period. Of course, many other things cause exchange rates to fluctuate besides U.S. and foreign monetary policy actions. The statistical model for Japan refers to these as unanticipated changes in the yen/dollar exchange rate (YEN shock) that cannot be ac counted for by fluctuations in U.S. or Japanese short-term interest rates. The model has difficulty capturing the effects of irregular phenom ena such as fears of a trade war, the top pling of the Berlin Wall, the coup attem pt against Gorbachev, or presi dential elections, to name only a few. As a result of this difficulty, the model lumps all such nonm onetary effects into what I refer to as the YEN shock for Japan, or the MARK shock for Germany. The second column of figure 1 dis plays the effect of the YEN shock on FF and YEN. Unanticipated shocks to the yen/dollar exchange rate are large: An average-sized YEN shock increases the value of the dollar by almost 1.5%. The response patterns of the exchange rate arising from the YEN and FF shocks indicate that over shorter horizons, most of the variation in the yen/dollar exchange rate will be due to factors unrelated to the federal funds rate; it turns out that they are mainly due to YEN shocks. Over longer horizons, more of the variation in the yen/dollar exchange rate will be due to monetary policy factors such as FF shocks. These results seem consistent with the forecasting results reported by Meese and Rogoff; that is, information about the stance of monetary policy in the U.S. is not very helpful for forecasting future movements in the yen/dollar exchange rate at horizons under two years. The results for Germany are comparable.5 The yen/dollar and mark/dollar experience, 1987-94 How would the dollar have behaved over the 1987-94 period if the only factors influencing it had been U.S. monetary policy actions? Row 1 of figure 2 suggests an answer for Japan. In the upper left panel, the highly variable blue line is the actual path of the exchange rate from July 1987 through June 1994. The black line is the counterfactual path that the vec- 2. Exchange rate movements explained by FF shock Japanese yen Eleven-week centered moving average yen per d ollar basis points this sense, the dollar’s depreciation against the yen and the mark since February 4 does not seem shocking. Over longer periods, however, these unexplained movements in exchange rates seem to average out. At longer horizons, then, the economic funda mentals of monetary policy seem to provide reasonable predictions for the direction and level of the dollar. —Charles L. Evans 1"What’s news: Business and finance,” Wall Street Journal, May 12, 1994, p. 1 German mark Eleven-week centered moving average 2Richard A. Meese and Kenneth Rogoff, “Empirical exchange rate models of the seventies: Do they fit out of sample?” Journal of International Economics, Vol. 14, No. 1/2, 1983, pp. 3-24. 3Martin Eichenbaum and Charles F. Evans, “Some empirical evidence on the effects of monetary policy shocks on exchange rates,” Federal Reserve Bank of Chicago, working paper, No. 92-32. Note: Calculations based on three-variable vector autoregression. tor autoregression predicts the ex change rate would have followed if the YEN and Japanese monetary poli cy shocks had been identical to zero. The yen/dollar exchange rate path predicted by the FF shocks matches many of the long-term movements in the actual exchange rate over this period. From November 1988 to April 1990, the dollar appreciated 31%; the FF shocks statistically ac counted for about 40% of that appre ciation. From April 1990 through June 1994, the dollar depreciated by about 60%; the FF shocks captured roughly half of that fall. The upper right panel displays the smoothed FF shock series, where negative (positive) numbers represent unanticipated policy easing (contraction). The shocks were predominantly positive in 1988, and the dollar appreciated in 1989. The shocks were mostly nega tive in 1989, and the dollar began to depreciate in 1990. In 1992 and 1993, the shocks were more negative than positive, and the dollar depreciated from 1992 through June 1994. In spite of the reasonably close fit, how ever, the FF shocks failed to predict much of the high-frequency variability in the yen/dollar exchange rate. The results for the m ark/dollar (MARK) are substantially weaker but still interesting (see row 2 of figure 2). From 1987 to 1989, the average level of the dollar was about 1.8 marks, the level implied by the FF shock; from mid-1992 to the present, the average level of the dollar has been about 1.6 marks, the level implied by the FF shock. As with Japan, however, the forecasted path from the FF shocks misses the high-frequency variability in MARK by an enormous amount. Summary Over short horizons, exchange rates can follow many paths that do not correspond to the predictions from recent monetary policy actions. In 4Twenty-six lagged observations for each variable were included in the threevariable vector autoregression. The foreign interest rate variable is a threemonth government bill rate. The feder al funds shock is an orthogonalized shock from the vector autoregression. For a fuller discussion of the effects of foreign monetary policy shocks on the dollar and a discussion of some technical issues, see Charles F. Evans, “Interest rate shocks and the dollar,” Economic Perspectives, Vol. 18, No. 5, September/ October 1994, forthcoming. 5These are discussed in Evans, op. cit. Karl A. Scheld, Senior Vice President and Director of Research; David R. Allardice, Vice President and Director of Regional Economic Programs; Janice Weiss, Editor. Chicago Fed Letter is published monthly by the Research Department of the Federal Reserve Bank of Chicago. The views expressed are the authors’ and are not necessarily those of the Federal Reserve Bank of Chicago or the Federal Reserve System. Articles may be reprinted if the source is credited and the Research Department is provided with copies of the reprints. Chicago Fed Letter is available without charge from the Public Information Center, Federal Reserve Bank of Chicago, P.O. Box 834, Chicago, Illinois, 60690, (312) 322-5111. ISSN 0895-0164 Expansion in Midwest m anufacturing activity slowed in recent months, ac cording to the Midwest M anufacturing Index. A significant decline in light vehicle assemblies (on a seasonally adjusted basis) played an im portant role, partly because of special constraints unrelated to dem and. Even so, appliance industry output has also softened in recent m onths after strong gains inl993. While growth has slowed, purchasing m anagers’ surveys suggest that manufac turing activity continued to percolate at a high level. Inventory building and strength in investment helped bolster industrial output in the region. How ever, renewed expansion in consum er spending on durable goods is probably necessary for m anufacturing activity to remain this strong. Sources: The Midwest Manufacturing Index (MMI) is a composite index of 15 industries, based on monthly hours worked and kilowatt hours. IP represents the Federal Reserve Board industrial production index for the U.S. manu facturing sector. Autos and light trucks are measured in annualized physical units, using seasonal adjustments developed by the Board. The purchasing managers’ survey data for the Midwest are weighted averages of the produc tion components from the Chicago, Detroit, and Milwaukee Purchasing Managers’ Associa tion surveys, with assistance from Bishop Associ ates and Comerica. U lttZZ (2Ig) P£80"06909 siouini ‘oSeaipc F88 xog O d J01U33 uopEimojui apqnj OOVOIH3 TO XNW 3A33S3H 3V33Q33 iC T P q p c j c h u m p )