Introduction Directory UMM :Data Elmu:jurnal:J-a:Journal of Economics and Business:Vol51.Issue2.Mar1999:

On the Relevance of Distinctions Between Anticipated, Unanticipated Expansionary, and Unanticipated Contractionary Monetary Policy Joonsuk Chu and Ronald A. Ratti It has been found that distinctions among positive innovations, negative innovations, and anticipated monetary policy change are relevant for explaining movement in real output. An asymmetry in the effects of anticipated expansionary and anticipated contractionary monetary policy on output also was found to be statistically significant, and the null hypothesis of no asymmetry in stimulativecontractionary policy was rejected. There is evidence that unanticipated stimulative, unanticipated contractionary, anticipated stimu- lative, and anticipated contractionary monetary policy each have statistically significant effects on output. Recognition of these asymmetries was found to make the finding of non-neutrality more likely. © 1999 Elsevier Science Inc. Keywords: Expansionary; Contractionary; Monetary policy JEL classification: E52, E58

I. Introduction

The policy ineffectiveness proposition advanced by Lucas 1973 and Sargent and Wallace 1975 conjectures that anticipated nominal changes have no effect on real output. The hypothesis that differentiation between anticipated and unanticipated change in nominal variables is important for explaining movement in real output was tested by Barro 1977, who found support for the neutrality hypothesis. In contrast, Mishkin 1982 found that when lag length was increased in the nominal variables, the neutrality hypothesis was Youngsan University of International Affairs, Yangsan, Korea JC; Department of Economics, University of Missouri-Columbia, Columbia, Missouri RAR. Address correspondence to: Dr. R. A. Ratti, University of Missouri-Columbia, Department of Economics, 118 Professional Building, Columbia, MO 65211. Journal of Economics and Business 1999; 51:109 –131 0148-6195 99 –see front matter © 1999 Elsevier Science Inc., New York, New York PII S0148-61959900029-0 rejected. In a further extension, Frydman and Rappoport 1987 presented evidence casting doubt on the relevance of the distinction itself between anticipated and unantic- ipated nominal change in explaining real output. Work by De Long and Summers 1988 and Cover 1992 has drawn attention to a different asymmetry and concluded that negative-money shocks have a greater effect on output than positive-money shocks. 1 Macklem et al. 1996 found evidence for Canada that asymmetric effects on output exist in anticipated policy. 2 This paper examines the relevance of distinctions among anticipated monetary policy change, unanticipated positive-money change, and unanticipated negative-money change in explaining movement in real output. The role of a possible asymmetry in policy actions, e.g., stimulative versus contractionary policy, is also considered. 3 Growth in M1 will be one of the measures of monetary policy. The spread between the commercial paper rate and the Treasury bill rate will be another. Work by Friedman and Kuttner 1992 indicates that spread contains highly significant information about movement in real output. Kashyap et al. 1993 viewed spread as a proxy for the stance of monetary policy. A tight monetary policy causes firms to compete for funds, leading to an increase in the issuance of commercial paper. Hence, the commercial paper rate rises more than the Treasury bill rate when money is tight. 4 The change in the federal funds rate will also be used as a measure of monetary policy. This will allow the issue of possible asymmetry in the effects of anticipated policy to be tested. 5 Bernanke and Blinder 1992 have argued that innovations in the federal funds 1 Cover 1992, p. 1261 went so far as to state that “positive-money shocks have no effect on output, whereas negative-money shocks cause output to decline.” Cover noted that this outcome is consistent with either a rigidly vertical aggregate supply curve and sticky prices in the face of unexpected changes in demand, or with a situation in which wages are sticky downwards but flexible upwards and a vertical aggregate supply curve at the point of full employment. A zero effect of positive money shocks is, of course, not predicted by either the new classical model with rational expectations, in which prices are flexible, or the nonclassical rational expectations model with contracts. Recent work by Lucas 1990, Christiano 1991, and Fuerst 1992 within a cash-in-advance real business-cycle framework, provides some motivation for assuming an asymmetry in the effect of positive and negative money shocks, but not a basis for concluding that positive shocks are unimportant. 2 A number of papers have addressed ways in which there may be asymmetric effects of policy on the real economy. This includes work by Bernanke and Gertler 1989, Caplin and Leahy 1991, and Ball and Mankiw 1994. Empirical evidence of asymmetric effects of monetary policy has been reported by Morgan 1993, Huh 1994, Thoma 1994, Ammer and Brunner 1995, and Garcia and Schaller 1995. Macklem et al. 1996 provide an excellent survey of work on asymmetric effects of monetary policy. 3 In this paper, monetary policy innovations are formed by using a method used by Mishkin 1982, Cover 1992, and Macklem et al. 1996. This involves a forecast equation for the monetary policy indicator. Residuals from this equation are then used as innovations to be used in an output equation. This method of construction contrasts with that used to identify policy shocks in a large literature using structural vector autoregression VAR models [see, for example Sims 1980, 1992; Bernanke 1986; Bernanke and Blinder 1992]. Cochrane 1995 argued that the VAR technique implicitly assumes that only unanticipated shocks matter. Bernanke and Mihov 1995 and Cochrane 1995 maintained that it is important to allow for the possible effect of anticipated money on output. In addition, Macklem et al. 1996 stressed that when asymmetric effects are being considered, scarcity of degrees of freedom becomes even more of a problem with the VAR approach. 4 Friedman and Kuttner 1992 held that the gap between the Commercial Paper rate and the Treasury bill rate better captures information about occurrences in financial markets relevant for the determination of output than movements in an interest rate or fluctuations in money. In another paper, Friedman and Kuttner 1993 point out that monetary policy is only one of several factors that can account for movement in spread and its relationship to output. Kashyap et al. 1993 have shown that monetary policy influences a firm’s mix of external financing and that this implies that a loan supply channel of monetary policy exists. 5 We are grateful to two referees for suggesting that we consider asymmetry in anticipated policy between stimulative and contractionary components, and also in policy action overall, as between stimulative and contractionary operations. Exploration of these issues with the other measures of monetary policy is not possible anticipated growth in M1 is nearly always positive, and anticipated spread is always positive. 110 J. Chu and R. A. Ratti rate are a good measure of changes in monetary policy and are informative about future movements in real activity. 6 It was found in our study that distinctions between positive innovations, negative innovations, and anticipated monetary policy change are relevant for explaining move- ment in real output. There is evidence that unanticipated expansionary monetary policy is just as likely to have a statistically significant effect on output as is unanticipated contractionary monetary policy. 7 These results appear to be robust across different measures of monetary policy, different specifications of the monetary policy and output equations, and over different sample periods. For monetary policy measured by change in the federal funds rate, an asymmetry in the effects of anticipated expansionary and anticipated contractionary monetary policy on output was found. The null hypothesis of no asymmetry in stimulativecontractionary policy was rejected. Anticipated expansionary monetary policy and anticipated contrac- tionary policy were each found to have statistically significant effects on output. A major finding of the study is that allowing asymmetries in anticipated and unanticipated monetary policy between stimulative and contractionary components makes the finding of neutrality of money less likely. In Section II, the model and the hypotheses to be considered are presented. Empirical results for growth in money and for spread are presented in Sections III and IV, respectively. The issue of asymmetry in stimulativecontractionary policy is taken up in Section V, when the change in the federal funds rate is used as the measure of monetary policy. Section VI is the conclusion.

II. The Model and Hypotheses