تغییرات در اثرات سیاست های پولی بر توزیع پویایی های قیمت غیر تراکمی

Based on a time-varying factor-augmented vector autoregression, we demonstrate that the propagation mechanism of monetary policy disturbances differs across disaggregate components of personal consumption expenditures. While many disaggregate prices rise temporarily in response to a monetary tightening in the early part of the sample, there is no evidence of a price puzzle at the aggregate level. The share of disaggregate prices that exhibit the price puzzle diminishes from the early 1980s onwards. There also is evidence of a substantial decline in the dispersion of disaggregate price responses over time. This gradual decrease in cross-sectional heterogeneity of disaggregate price responses is associated with a dampening effect on aggregate real economic activity and a stronger effect on the aggregate price level. We illustrate by means of a multi-sector sticky-price model augmented by a cost channel how key structural parameters would have had to change to match this evolution of sectoral price dynamics.

In formulating policy decisions, central banks not only rely on information about the aggregate economy but also carefully monitor sectoral conditions by conducting business surveys that provide important information about the price-setting process of firms (e.g., Blinder, 1991). For monetary authorities it is crucial to know how their policy actions affect the pricing decisions of firms in various sectors because this determines the effectiveness of monetary policy in stabilizing the economy as a whole. Altissimo et al. (2009), Bils and Klenow (2004), and Clark (2006) show that important differences in inflation dynamics arise at the aggregate and disaggregate level. Lastrapes (2006) and Balke and Wynne (2007) demonstrate that money supply shocks have long-run effects on the distribution of relative commodity prices implying an important degree of monetary non-neutrality. On the other hand, Boivin et al. (2009) make the case that discrepancies between aggregate and sectoral measures of inflation derive from the fact that the bulk of fluctuations in individual prices is due to sector-specific factors and that monetary policy shocks are of minor importance but induce sluggishness in price adjustment. In a recent contribution, De Graeve and Walentin (2011) show that if one takes special features of sectoral pricing behavior into account, the variance and persistence of disaggregate inflation are driven by both aggregate and idiosyncratic disturbances. These examples illustrate the importance of understanding the effects of monetary policy on consumer prices not only at the aggregate level but also at the disaggregate level. The disaggregate effects of monetary policy actions have to be viewed in the context of the Great Moderation. For example, the volatility of inflation and of output have declined considerably since the mid-1980s creating a more stable macroeconomic environment. In addition, the level and persistence of aggregate inflation have reached historical lows.1 However, little is known about the evolution of the underlying components of aggregate price and output measures. Over the last decades, macroeconomic developments such as increased monetary policy credibility, enhanced competition due to globalization, financial innovations, and technological advances might have contributed to alter the price-setting behavior across sectors which ultimately changes the way monetary policy is transmitted to the macroeconomy. The main contribution of this paper is to analyze how disaggregate price dynamics have evolved over time in response to monetary policy shocks in order to inform central bankers about changes in the relative price effects of their policy actions which cannot be inferred from aggregate price measures. Because exploring the monetary transmission mechanism at the disaggregate level requires an empirical framework that lends itself to including information from a large number of macroeconomic indicators and sectoral variables representing various dimensions of the economy, we employ an extended version of the factor-augmented VAR (FAVAR) model introduced in Bernanke et al. (2005). Given the possibility that disaggregate dynamics have changed over time, we extend the FAVAR model to allow for time variation in the coefficients and for stochastic volatility in the variances of the shocks. Previously, changes in the monetary transmission mechanism have been studied for the aggregate economy by incorporating time variation into small-scale VAR models that typically consist of three variables—a short-term interest rate, a measure of real economic activity and inflation (see, e.g., Primiceri, 2005). One problem with this approach is that the amount of information incorporated in these models is relatively limited which has two potential consequences. First, missing variables could lead to biases in the reduced-form VAR coefficients and in the estimates of persistence and volatility. Second, the omission of some variables could hinder the correct identification of structural shocks. One possible manifestation of these problems are impulse response functions that are at odds with economic theory. For instance, Bernanke et al. (2005) raise the issue that if the information set used by the econometrician is smaller than that employed by the monetary authority, then structural shocks and responses to them may be mismeasured because the empirical model excludes some variables that the central bank reacts to. Similarly, Castelnuovo and Surico (2010) and Benati and Surico (2009) building on Lubik and Schorfheide (2004) argue that during periods of indeterminacy, the dynamics of the economy are characterized by a latent variable. Therefore, reduced-form and structural estimates of the VAR model may be biased when estimation is carried out over these periods. Thus, with our estimation framework, we are not only able to provide novel insights about the time-varying responses of disaggregate prices and consumption quantities, but also to re-examine the mixed evidence on possible changes in the responses of the main macroeconomic variables after a monetary policy shock reported in earlier studies addressing some of the shortcomings discussed above. Our empirical results show that the propagation mechanism of monetary disturbances is highly heterogeneous across components of personal consumption expenditures suggesting that monetary policy actions exert an important influence on relative prices in the US economy. We show that a considerable fraction of sectoral prices increases temporarily after a contractionary monetary policy shock, and that the share of disaggregate prices that respond positively gradually declined from the early 1980s onwards. On the other hand, there is no evidence of a price puzzle for any of the aggregate price level measures throughout the sample which suggests that the extra information used in the estimation is helpful in correctly identifying monetary policy shocks. In addition to considerable variation in the median responses of prices and quantities at the disaggregate level, we also find that the cross-sectional distribution of responses has undergone substantial changes over time. In particular, there has been a marked decline in the variance of the distribution of sectoral prices, while the median response has become more negative over time. The variance of the distribution of quantity responses has also declined, but the median response has become smaller in magnitude. This paper is also related to empirical work by Nakajima et al. (2010) who, using alternative datasets and empirical methods building on Balke and Wynne (2007), similarly documented cross-sectional heterogeneity in the disaggregate price responses and the existence of a price puzzle at the disaggregate level after a monetary policy tightening. Our analysis differs from Nakajima et al. (2010) in several dimensions. First, we not only model the aggregate and disaggregate dynamics jointly and do so within a model framework that includes more information relative to medium-scale VAR models of the type used by Nakajima et al. (2010), but we also allow for time variation in the parameters of that model, whereas Nakajima et al. (2010) impose a time-invariant structure. Second, while Nakajima et al. (2010) are interested in contrasting the responses of disaggregate prices to technology and monetary policy shocks, our objective is to gain a better understanding of the monetary transmission mechanism at both the aggregate and disaggregate level. In addition to quantifying the extent of time variation in disaggregate price dynamics, we also study the evolution of the corresponding quantity dynamics and changes in their mutual interaction. Third, we use a version of the theoretical DSGE model proposed by Nakajima et al. (2010) that features heterogeneity in price stickiness across sectors and a cost channel of monetary policy to explore how key model parameters would have had to evolve over time to fit the time-varying dynamics of the estimated disaggregate price and quantity responses, thereby complementing the analysis in Nakajima et al. (2010). We show that the gradual attenuation of the disaggregate price puzzles can in part be explained by decreases in the sensitivity to the cost channel of monetary policy and increases in the degree of wage flexibility. The rest of the paper is organized as follows. Section 2 introduces the empirical methodology adopted in this study and describes our large dataset. Section 3 discusses the time-varying dynamics of selected macroeconomic aggregates and of disaggregate prices and quantities in response to monetary policy shocks, and conducts a sensitivity analysis of the baseline model. Section 4 relates the empirical results to a multi-sector sticky-price model and explores how changes in structural model parameters can account for changes in disaggregate price and quantity dynamics over time. Section 5 offers some concluding remarks.

5. Conclusions We examined the evolution of the US monetary policy transmission mechanism over time at both the aggregate and disaggregate level using an extended version of the factor-augmented VAR model proposed by Bernanke et al. (2005) that incorporates time variation in the parameters. Our empirical framework includes information from almost 600 macroeconomic and financial indicators as well as disaggregate price and quantity variables representing various sectors of the economy. By allowing for time variation in the coefficients and stochastic volatility in the variances of the shocks, we are able to characterize the time-varying dynamic responses to a contractionary monetary policy shock for macroeconomic aggregates and for sectoral prices and quantities in personal consumption expenditures. Our analysis enhances the understanding of the behavior of disaggregate prices in response to monetary policy disturbances, but also of the role of price-setting behavior in the propagation mechanism. It also sheds light on how this role has evolved in response to changes in the macroeconomic environment. Consistent with some related studies using alternative datasets and empirical methods, we find no evidence of a price puzzle for any of the aggregate price level measures. However, at the disaggregate level, many sectoral price responses display a temporary price puzzle during the 1970s, which diminishes gradually from the early 1980s onwards. We also find a shift in the relationship between the responses of prices relative to consumption volumes which tends to suggest that over time nominal adjustment of the economy following a monetary policy tightening increasingly takes place by prices rather than by quantities. This finding at the disaggregate level coincides with a decline in the real effects of monetary policy surprises on economy-wide real activity measures and a greater sensitivity of aggregate price responses. Apart from considerable heterogeneity in the median responses of prices at the disaggregate level, we also observe important changes in the cross-sectional distribution of price responses over the sample period. While the price dispersion diminishes over time, it widens considerably over the forecast horizon and is very persistent in the long run. Our analysis adds to a growing literature that tries to bring evidence on the heterogeneity of sectoral prices to bear on our understanding of the transmission of structural shocks to the aggregate economy by providing new empirical insights on the relevance of time variation for aggregate and disaggregate dynamics after monetary policy shocks. To illustrate the potential sources for these time-varying dynamics, we used a version of the multi-sector DSGE model proposed by Nakajima et al. (2010) that features heterogeneity in price stickiness across sectors and a cost channel of monetary policy to investigate which changes in structural parameters are required to account for changes in the cross-sectional distribution of prices and quantities over time. By means of simulation exercises, we showed that the gradual attenuation of disaggregate price puzzles is consistent with a decrease in the importance of the cost channel and in the degree of nominal wage stickiness. Therefore, positive sectoral price responses following a restrictive monetary policy shock may not be a statistical artifact, but have economic content. They may indicate the existence of additional channels of transmission at the disaggregate level, which are often missing in standard macroeconomic models employed in policy analysis. However, the channels analyzed here are not the only possible mechanisms that are compatible with the empirical findings documented in this paper. An interesting avenue for future research could be to explore alternative channels that might be at work.