Liquidity Matters: Money Non-Redundancy in the Euro Area Business Cycle Antonio Conti January 21, 2010 Abstract While New Keynesian models label money redundant in shaping business cycle, monetary aggregates are found relevant in predicting output, prices and interest rates or correctly identifying monetary policy shocks. This paper provides new evidence of money non redundancy in Euro area business cycle, contrary to most of recent literature. We study the dynamic effects of liquidity shocks under different traditional identifying methods, assessing their relevance compared to other sources of fluctuations. To shed some lights on money s role we propose to adopt an agnostic approach on the effects of liquidity shocks, thus letting data answer. Results show that shocks to monetary aggregates raise output, prices and policy rate, having a permanent effect on the latter. Hence, (i) money can not be omitted in modeling economic structure; (ii) liquidity is a powerful transmission in the short run, even more effective than interest rate channel. Our findings point to a reexamination of theoretical models and seem to support the emphasis given by European Central Bank (ECB) to the monetary pillar. JEL classification E52; E58; C32. Keywords: monetary aggregates; liquidity shocks; structural VAR; sign restrictions. This paper was written while the author was visiting student at ECARES, Université Libre de Bruxelles, whose kindly hospitality is gratefully acknowledged. PhD student at Sapienza University of Rome, Department of Economics. Piazzale Aldo Moro, 5, 00185, Roma, Italy. E-mail address: ant.conti@gmail.com; antonio.conti@uniroma1.it 1
Extended abstract This paper analyzes money s role in the Euro Area (EMU) business cycle using Vector Autoregressions (VAR). Whether money plays a structural role or it is just - at its maximum - an indicator variable is an open issue in current macroeconomics, the answer having clearly very different implications for economic analysis and implementing policy. 1 The key point highlighted by literature (Nelson, 2002) is the presence of an explicit money term in the aggregate demand equation and the impact of money growth on inflation, i.e. a monetary motive in the Phillips Curve. While many papers have been produced in order to assess this issue, conventional wisdom excludes money from small macroeconomic models giving the edge to the New Keynesian (NEK) model (Rotemberg and Woodford, 1997; Clarida, Galì and Gertler, 1999): the triplet composed by output gap, inflation and interest rate is completely determined and fully characterizes steady state and dynamics in such models under an interest rate targeting rule, i.e. a Taylor rule (1993). Money is just redundant, and so money demand equation. 2 In NEK models the absence of money in the structural log-linearized form derives from the separability assumption in households preferences on consumption and real balances: however, Woodford shows that even removing this not plausible assumption, real balances play such a quantitatively small role in affecting output gap, inflation and interest rate to be considered negligible. One can thus take the separability assumption as a reasonable approximation. Identical results are obtained by Ireland for US economy, Andres, Lopez-Salido and Valles (2006, hereafter ALSV) for the Euro Area and by Fujiwara (2007) for Japan. 3. These papers conclude that there is no room for money directly playing a structural role in the business cycle: monetary policy only acts through 1 For more details on this debate see Nelson (2003). 2 See Woodford (2003) for an analytical explanation of why money does not affect output, inflation and interest rate dynamics. 3 All these papers construct a DSGE model in which a microfounded money demand is derived allowing for non separability between consumption and real balances in the households utility function, and then perform a state-space representation with estimation computed by Kalman Filter Maximum Likelihood: Favara and Giordani (2009) show some serious pitfalls of this method in capturing the real balances effect. 2
the interest rate channel. Furthermore, Svensson (2003) clearly states it is apparent that there is not much of a role for money in the transmission mechanism, whereas its potential most useful function is the one of an useful indicator which contains information about the state of the economy. 4 Lippi and Neri (2007), in an estimated structural DSGE model of the Euro Area, show that no useful information role emerges for monetary aggregates. However, many empirical papers raise doubts or counterintuitive arguments on this result, either stating a role for monetary aggregates in identifying monetary policy shocks either emphasizing their forecasting properties for output, prices and interest rates. Leeper and Roush (2003) recommend to "put money back in monetary policy models", as a consequence of the fact that the way of modelling money significantly affects output and inflation, while Poilly (2007) stresses the importance of the way money is introduced in the identification scheme, finding it non neutral in estimating DSGE models. As for the second strand of literature, Hafer, Haslag and Jones (2007) show that M1 and M2 can not be omitted for US economy in the (backwardlooking) IS curve such as in Rudebusch and Svensson (2002), even once taken into account output-gap own lags and real interest rate ones, so that money helps for predicting economic activity and prices. In a complementary work Hafer and Jones (2008) present international evidence of the predictive power of money growth on GDP gap; thus, they suggest that money, independently of real interest rate, generally exerts a significant impact on the output-gap. Andres, Lopez-Salido and Nelson (2009) show that money demand has a forward-looking nature and that it is a key element in anticipating future variations of natural interest rate. Finally, Favara and Giordani (2009) find that money directly influences the triplet output, prices and interest rate, therefore recommending to reconsider money s role in macroeconomic models: specifically, they argue that shocks to broad monetary aggregates have substantial and persistent effects on output, prices and interest rates, which is completely at odds with the theoretical prediction of flat impulse response 4 Gerlach and Svensson (2003) find that real money gap has predictive power for EMU inflation, but no more than output gap, and, however, they downgrade the money growth indicator proposed by European Central Bank (ECB) in its two pillars strategy. 3
functions implied by NEK model. This paper analyzes money s role in the Euro Area, trying to shed some new light on the debate between the two opposite paradigms, the one stating money redundancy and the other finding money relevant in describing business cycle. If possible, assessing money s role is even more crucial for EMU economy because of the ECB two pillars monetary policy strategy. Given the aim of the paper, it s quite natural using a multivariate environment, in which a monetary aggregate augments the standard three-equations model. Furthermore, policy reasons related to the 2008 financial crisis make the question relevant: the massive liquidity injection performed by ECB and FED; the proper analysis of monetary transmission channels in a zero lower bound interest rate environment; possible consequences and implications for the exit strategy and the design of a new monetary strategy. We first show that, contrary to the conventional wisdom of money redundancy, liquidity shocks account for a relevant change in dynamic response of Euro Area output, inflation and interest rate in a traditional framework validating analogous results obtained by Favara and Giordani (2009) on US data. A one standard deviation liquidity shock is responsible for raising output of almost 0.3 points: the effect seems to be quite persistent, vanishing only after 11 quarters. Prices go up after 4 quarters to a peak of 0.8 points, while the policy rate is raised to a maximum of 50 basis points after almost 9 quarters, before starting slowly decline. Interestingly, this last response to a liquidity shock seems suggesting the presence of a monetary pillar in the ECB reaction function. Furthermore, variance decompositions show that money explains up to a 20% of output volatility, while accounting for 35% of prices and short-term rate. All these findings seem consistent with the long-run linkage between money and prices. However, our main contribution stems from extending Favara and Giordani (2009) results from a three-ways perspective. First, we allow for simultaneity between monetary aggregates and interest rate, which is denied in the Choleski identification with money ordered last. Second, we move to a full identification context, in order to compare the importance of liquidity shocks to other sources of business cycle, thus recovering also a supply shock, a demand shock, a monetary policy shock. Second, we achieve identification 4
by means of sign restrictions (Uhlig, 2005; Peersman, 2005), which enables us in adopting an agnostic point of view on money s role. Identifying sign restrictions are presented in table 1. y π s m Supply + unrestricted + Demand + + unrestricted + M on.p olicy + + unrestricted Liquidity unrestricted unrestricted unrestricted + Table 1: Identifying sign restrictions, agnostic approach. An expansionary supply shock is defined as a shock having a positive effect on output (and money demand) and a negative impact on prices; an expansionary demand shock is defined as a shock raising output and prices; an expansionary monetary policy shock is a demand shock having the further requisite of a negative effect on the policy rate; finally, consistently with the agnostic point of view, an expansionary liquidity shock is supposed only have a positive effect on liquidity, while leaving unrestricted all other responses. Finally, as for robustness checking, we naturally derive the identifying restrictions from the theoretical model in ALSV (2006), allowing not only for a Taylor rule augmented by the monetary aggregate, but for a two pillars Phillips curve too (Gerlach, 2004), so to be as consistent as possible with the theoretical framework which ECB had in mind when two pillars strategy was designed. A number of results is illustrated. First, money can not be omitted in modeling economic structure because it significantly and persistently affects output, prices and interest rates, as showed by impulse response functions. Second, liquidity is a powerful channel of monetary transmission in the short run, even more effective than policy rate in affecting output. Third, liquidity shocks explain a small share of short-term inflation volatility, while their role becomes more relevant in the long-run. Fourth, consistently with the former point, policy rate setting is affected by liquidity shocks at a longer 5
horizon. Last, our findings call for a reexamination of NEK models and seem to support the emphasis given by ECB to the monetary pillar: but this is left to further research. References [1] Andres, J., D. Lopez-Salido, J. Valles (2006). Money in an estimated business cycle model of the Euro Area. The Economic Journal 116, pp. 455-477. [2] Andres, J., D. Lopez-Salido, E. Nelson (2009). Money and the natural rate of interest: structural estimates for US and Euro area. Journal of Economic Dynamics and Control. [3] Clarida, R., J. Gali and Mark Gertler (1999). The Science of monetary policy: a New Keynesian perspective. Journal of Economic Literature 37, pp. 1661-1707. [4] Favara, Giovanni and Paolo Giordani (2009). Reconsidering the role of money for output, prices and interest rates. Journal of Monetary Economics doi:10.1016/j.jmoneco.2009.01.002. [5] Fujiwara, I. (2007). Is there a direct effect of money? Money s role in an estimated monetary business cycle of the Japanese economy. Japan and the World Economy 19, pp. 329-337. [6] Gerlach, Stefan (2004). The two pillars of European Central Bank. Economic Policy 19, pp. 393-439. [7] Gerlach, Stefan and Lars E.O. Svensson (2003). Money and inflation in the euro area: A case for monetary indicators? Journal of Monetary Economics 50, pp. 1649-1672. [8] Hafer, R.W., J. Haslag, G. Jones (2007). On money and output: is money redundant? Journal of Monetary Economics 54, pp. 945-954. 6
[9] Hafer, R.W. and Garett Jones (2008). Dynamic IS curves with and without money: An International Comparison. Journal of International Money and Finance 27, 4, pp. 609-616. [10] Ireland, Peter (2004). Money s role in the business cycle. Journal of Money, Credit and Banking 36, 6, pp. 969-983. [11] Leeper, Eric M. and Jennifer Roush (2003). Putting "M" back in monetary models. Journal of Money, Credit and Banking 35, No 6, pp. 1217-1259. [12] Lippi, Francesco and Stefano Neri (2007). Information variables for monetary policy in an estimated structural model of the euro area. Journal of Monetary Economics 54, pp. 1256-1270. [13] Nelson, E. (2002). Direct effect of base money on aggregate demand: theory and evidence. Journal of Monetary Economics 49, 4, pp. 687-708. [14] Nelson, E. (2002). The future of monetary aggregates in monetary policy analysis. Journal of Monetary Economics 50, 5, pp. 1029-1059. [15] Poilly, C. (2007). Does Money Matter for the Identification of Monetary Policy Shhocks: A DSGE perspective. NER Banque de France n 184. [16] Rudebusch, G. and L. E. O. Svensson (2002). Eurosystem Monetary Targeting 46. European Economic Review 46, pp. 417-442. [17] Svensson, L. E. O. (2003). Comment on: The future of monetary aggregates in monetary policy analysis. Journal of Monetary Economics 50, 5, pp. 1061-1070. [18] Uhlig, H.(2005). What are the effects of monetary policy on output? Results from an agnostic identification procedure. Journal of Monetary Economics 52, 381-419. [19] Woodford, M.(2003). Interest and Prices: Foundations of a Theory of Monetary policy, Princeton University Press. 7