Discussion on the Paper ”Technology shocks and monetary policy : assessing the Feds performance” Irene S´anchez, Milim Kim , Guillem Roig March 4, 2010

Macroeconomics II Professor: Franck Portier

1

1

Introduction

In this report we analyze the work of Gali et al(2002)[3]. We proceed as follows: in section two we present the main advantages of using a sticky-price model compared with the traditional model which leads to the Traditional Phillips Curve. In section three we expose what are the main drawbacks of using the sticky-price model in terms of the dynamics and we present a stickyinformation model following Mankiw (2002)[1]. In the sticky information model, the dynamics are more in accordance to central bankers general wisdom and what we find empirically. In section four we briefly discuss the simple Taylor rule that the authors expose in their work. We conclude our work by exposing the main problems of model simulation by using the Structural Var procedure because of the little economic theory that the this procedure entails and we expose if technology shocks are the only ones explaining the permanent effect on labour productivity.

2

Traditional Phillips Curve Vs. New Keynesian Phillips Curve

The New Keynessian Model introduces a new view of what the literature calls the traditional Phillips Curve: The New Keynesian Phillips Curve. The following equation represents the traditional Phillips Curve. πt = πt−1 + δ y¯t + εt

(1)

This equation has two main characteristics: First, inflation is backward looking, past inflation matters to determine current inflation. Second, the correlation between the actual inflation and the past output is positive ρ(πt , yt−1 ) > 0 meaning that output leads inflation. However, the New Keynesian Phillips Curve represented in the following equation has the opposite characteristics: πt = βEt (πt+1 ) + κxt

(2)

First, inflation is forward looking, so what matters to determine current inflation is the expectation today of future inflation. Second, current inflation is positively correlated with future output instead of past output ρ(πt , yt+1 ) > 0. This later result means that inflation leads output and not the other way around as the traditional Phillips Curve postulates. The intuition of the forward looking component behind the New Keynesian Phillips Curve relays in the fact that in a framework with sticky prices, inflation appears because at each period, there are some firms that are resetting their prices and, as shown in the following equation this decision depends on current and anticipated marginal costs, which are in general unrelated to past inflation. Pt∗ = µ + (1 − βθ)

∞ X

(βθ)k (mcnt+k )

(3)

k=0

However, there are some critics against the robustness of the New Keynesian Phillips Curve. Despite the fact that the NKPC has a strong theoretical micro-foundation, the traditional Phillips Curve fits better with the data. Some authors have tried to fit the data better and they developed what is called an Hybrid PC. πt = γb πt−1 + γf Et (πt+1 ) + δ(yt − y¯t )

2

(4)

The data appears to be more consistent with a backward looking specification of the Phillip Curve rather than with a forward looking. Some revisions of this claim has been made by many authors see for instance, Gali (2001) [8]. He concludes that the Hybrid model is wrong in specifying which is the variable that has to be used to define the output gap of the NKPC. They have used mistakenly the detrended GDP as a proxy of the output gap. The previous is not a good measure because it is a smooth variable and theoretically speaking, the natural level of output is not defined as a smooth function of time, but as a function that fluctuates with any shock and it is quite volatile. To solve this incompatibility problem some authors have proposed to use the real marginal cost as a proxy of the output gap. Then, when they estimate the NKPC using the right measure consistent with the theory behind the NKPC, they found that indeed the NKPC seems to fit with the data much better than the traditional PC.

3

Drawbacks of the Sticky-Price model and advantages of the Sticky-Information model

The sticky price information model in which firms follow time-contingent price adjustment rules has been proved to be far form central bankers general wisdom and its dynamic properties are contradictory to many empirical evidence. A sticky price-information model generates the New Keynesian Phillips Curve that we presented in the previous section (Eq:2) As we have mentioned before, in this model inflation is forward-looking and depends on the current expectation about future inflation1 . Therefore, announced and credible disinflation will cause booms rather than recessions. Empirically, this has not been the case in the last decade in countries like Japan where a deflation has been followed by a tremendous recession in the economy. General wisdom says that when agents expect deflations in the economy they postpone their purchasing decisions which creates a negative demand shift leading to a recession. Another critique is that output can be increased forever if agents expect deflation. The model predicts that there exist a path of money creation that enriches an economy in real terms forever. McCallum and Lucas (1973)[10] have criticized that the nominal part of an economy cannot have a permanent effect on the real side and therefore, the prediction mentioned above seems implausible. It is also important to mention that the price-sticky model has low inflation inertia, there is low persistency of inflation. Although the price level is sticky, inflation rate changes quickly. As a result, the model fails to explain the fact that shocks to monetary policy have a delayed and gradual effect on inflation. It is also the case that reducing the level of inflation has no cost in the economy since agents can coordinate to have low inflation costless and regardless of past inflation. In order to have a more accurate dynamic prediction, economist have developed a micro-founded model based on sticky-information rather that on sticky-prices. In this model, firms can adjust prices at every period but they gather information about the state of the economy slowly over time. The resolution of the model gives the following ”Sticky-Information” Phillips Curve.

πt = λ

∞ X

(1 − λ)j Et−1−j (πt + α∆xt ) + κxt

(5)

j=0

In this equation, inflation depends on output, expectations of inflation, and expectations of output growth, which is represented by the term ∆xt . Notice that, expectations are still important 1

Inflation is forward looking due to price stickiness. This is because when firms expect a higher future inflation they adjust they prices upwards today in order to keep their relative prices tomorrow

3

but what matters are past expectations of current economic conditions. The key difference with respect to the simple backward-looking model is that agents form expectations rationally and credibility plays a role to reduce the cost of disinflation. The dynamics that this model predicts is aligned with conventional wisdom. First, it predicts that disinflations create recessions in the economy. Second, there is inertia in inflation. Mankiw et al. (2002)[1] showed that monetary shocks have their maximum effect on inflation with a substantial delay. Finally, the sticky information model is able to explain the so called ”acceleration phenomenon” in which vigorous economic activity is correlated with rising inflation.

4

On the coefficients of the Simple Taylor Rule

In the paper the authors present the following equation: rt = rr + φπ πt + φx xt

(6)

Notice that in the previous equation it appears the output gap xt . However, this variable depends on the natural level of output which is not observable in practice. Moreover, a Monetary Policy simple rule has the following characteristics: • Rule that CB’s can achieve in practice. • They depend on observable variables. • No necessity to know the real value of the parameters. • Approximation to the optimal rule. So, in the paper they still include the natural output gap, instead of replace it for an observable variable. However, this is a minor problem since they choose an appropriate combination of the parameters. This means that in a basic New Keynesian model, a Simple Taylor Rule that respond aggressively only to inflation movements can be a good approximation for the optimal policy rule. Policy makers usually define the optimal policy rule as the one that in equilibrium leads to the lower welfare losses. However, it is known that it is not desirable to stabilize inflation aggressively. rt = rr + φπ πt + φx x∗t

(7)

where x∗t = yt − y and φπ = 1.5, φx = 0

5

Estimation with structural Var

The question that we discuss in this section is if structural VAR is a useful method for analyzing the dynamic effect of technology shock. In his paper, Gali estimated the dynamic effect of technology shocks by exploiting three assumptions : 1. Technology shocks are a linear combination of a finite number of lags of past data. 2. Technology shocks are orthogonal to other structural shocks. 4

3. Technology shocks are the only shocks that have long run effects on labor productivity. But there are some weaknesses: The SVAR approach begins with the idea that it is possible to obtain impulse responses from the data using only a minimal amount of economic theory. These impulse response functions are the responses of the model economic system to innovations in various shocks. The hope is that the SVAR assumption nests most business cycle model, at least approximately, so that the impulse responses obtained from the VAR sets the standard for the theory: any promising model must produce impulse responses similar to those from the VARs. Hansen and Sargent (1991)[7] pointed out that the main problem with the SVAR approach is that it uses a priori too little economic theory. Without more economic theory it seems to be impossible to determine the answer to basic issue like ”for what questions will a short lag length be reasonable? What variables should we include in the VARs?” Even if we believe that Gali’s identification assumption is true, CKM(2004)[6] numerical example shows that Structural VAR is distorted. The main finding of the differenced specification of hours (DSVAR) literature is that a technology shock leads to a fall in labor input. CKM (2004)[6] generates data from an economic model, apply the SVAR procedure to this data and ask whether the impulse responses identified by the SVAR procedures are close to the model impulse responses. The conclusion is that this shock leads to a decline in hours. This conclusion is mistaken because in the economic model an innovation to technology leads to a rise in hours. This shows that a researcher who applied the VAR procedure to an economic model would systematically draw the wrong inference.

6

Technology shocks and Var

In this section we address the following question: Does only technology shocks have a permanent effect on the level of labor productivity? Gali does not attempt to identify sources of fluctuations other than exogenous variations in technology. Identifying restriction is that only technology shocks may have a permanent effect on the level of labor productivity. Gali (1999)[5] showed that VAR’s based evidence • Variance of output and hours explained by technology shocks: 5 and 7 percent respectively. • Correlation of output and hours conditional on technology shocks: −0.08. • Variance of output and hours explained by non-technology shocks: 95 and 93 percent respectively. • Correlation of output and hours conditional on non-technology shocks: 0.96. It means that • Technology shocks have a negligible role in explaining the variance of output or hours at business cycle frequencies. • Dominant driving force: preference/demand shock. 5

Moreover, why are we just think this kind of force has only short term effect? We conclude that the assumption of VAR analysis saying that only technology shock affect on the labor productivity level contradicts to the result which is derived from VAR method. And we suggest that the assumption has to be changed appropriately.

6

References [1] N. Gregory Mankiw, R. Reis Sticky information versus sticky prices: a proposal to replace the New Keynesian Phillips Curve. The Quarterly Journal of Economics, (2002). [2] M. Eichenbaum, D.M. Fisher Testing the Calvo model with Sticky Prices. NBER Federal Reserve of Chicago, (2003). [3] J. Gali et al. Technology shocks and monetary policy: assessing the Feds performance. , (2002). [4] J. Gali On the role of technology shock as a source of business cycles: some new evidence. (2004). [5] J. Gali, P. Rabanal Technology Shocks and Aggregate Fluctuations: How Well Does the RBC Model Fit Postwar U.S. Data? (1999). [6] Chari, Kehoe-McGrattan A Critique of Structural VARs Using Real Business Cycle Theory (2004). [7] Hansen and Sargent Two difficulties in interpreting vector autoregression (1991) [8] J. Gali New perspectives on Monetary Policy: inflation and business cycle CREI and Universitat Pompeu Fabra. (2001). [9] J. Gali Gali slides: Universitat Pompeu Fabra Academic year 2009. [10] R. Lucas Some international evidence on output-inflation trade-off AER (1973) 326-334.

7