KOBAYASHI Keiichiro: This time, I would like to focus on "Inflation Persistence and Flexible Prices" (*International Economic Review* 46(1): 245-61) by Robert Dittmar, William Gavin, and Finn Kydland (2005).

In analyzing the effect of monetary policy such as quantitative easing, a certain degree of price rigidity is generally assumed, whereby the primary focus of discussions is typically on the extent that monetary policy can resolve inefficiencies stemming from sticky prices. This is how New Keynesian theories, the current mainstream of monetary policy, are constructed. However, the actual frequency at which prices change at a firm level, as demonstrated by empirical findings, is far greater than the frequency consistent with changes in macroeconomic data (Bils and Klenow 2004). In one attempt to solve this problem, Lawrence Christiano created a model that produces results consistent with macroeconomic data, even when price changes occur as frequently as shown in the firm-level empirical studies, by assuming firm-specific capital in a New Keynesian model.

However, some advocate models that reject outright the very thinking underlying New Keynesian models; that price rigidities are the primary concern. The paper I introduce this time is one such argument.

In short, the paper says that the assumption of price rigidities is unnecessary in explaining the persistent change in the inflation rate (the basis of the Phillips curve) observed in actual data. Instead, Dittmar et al. assume an environment in which productivity shocks are persistent when prices are flexible. When a Taylor Rule-type monetary policy is implemented in this environment, the persistence of production shocks is converted into inflation persistence via induced policy responses. They argue that through this mechanism the Phillips curve relationship, in which the inflation rate moves in tandem with output, arises (as a result of monetary policy). Kydland, one of the authors of this paper, is a leading Real Business Cycle theorist who won the 2004 Nobel Prize in Economics. The paper definitely showed his true colors.

Econo-kun: Could you explain the model you referred to as rejecting outright the thinking underlying New Keynesian theories?

KOBAYASHI Keiichiro: The basic structure of the model is that of Real Business Cycle models. They incorporate money as a time constraint for consumers; an active factor that shortens consumers' shopping time. The greater the (real) money stock held by consumers, the less time they need for shopping, thus the more time they can spend working and participating in leisure activities. The level of money demand is a result of the optimization of this saving in shopping time.

The paper presents a number of simulation results. First, when monetary policy decisions are random, with money supply fluctuating randomly around a certain level, the inflation rate exhibits no significant persistence and little correlation (only a modest negative correlation) is observable between the inflation rate and outputs.

Second, Dittmar et al. examined a case in which a central bank implements a Taylor Rule-type monetary policy. In this model, when the coefficient of the inflation rate (ν_{π}) under the Taylor Rule equals or exceeds 1, the price level cannot be determined. This contradicts the findings in Clarida, Gali, and Gertler (2000) in which the price level is indeterminable when ν_{π}< 1. (How Dittmar et al. interpret this difference remains ambiguous but they provide the following explanation: their model assumes a monetary transmission mechanism different that of Clarida et al. Dittmar et al. assumes that monetary policy affects real economy via changes in the expected inflation rate, whereas Clarida et al. assumes that the effect of monetary policy is transmitted via changes in the expected real rate of interest. These differences have led to contradictory results concerning the determinability of price levels.)

In the model which assumes the Taylor Rule-type monetary policy, the persistence of productivity shocks (autocorrelation coefficient ρ= 0.95) are shown to convert into inflation persistence with strong correlation observed between the inflation rate and output. Qualitatively, this result is hardly affected when the coefficient of the inflation rate is changed. (The result of this Taylor Rule model is affected by changes to the coefficient of output ν_{y}; when the value of ν_{y} falls below a certain level, the correlation between the inflation rate and output becomes negative.)

Based on these findings, Dittmar et al. infer that the Phillips curve observed in actual data (the positive correlation between the inflation rate and output) is an "artifact" induced by monetary policy.

Econo-kun: I see. That is quite a provocative idea. Are there any drawbacks to this argument?

KOBAYASHI Keiichiro: The problem with this paper is that very strong persistence of productivity shocks (ρ＝0.95) is assumed and the price level is indeterminable when ν_{π}>1.

As a way to solve this problem, consider a model that incorporates money into the Kiyotaki-Moore (1997) credit cycles model with collateral constraints, and see what happens when a Taylor Rule-type monetary policy is implemented. In their model, changes in the price of land (collateral asset) are transmitted via collateral constraints to affect production activities, and thus, even when production shocks do not persist, both the asset price and output change in a sticky manner (or follow a damped oscillatory motion, to be more precise). Given such persistent changes in the output, it is inferred that a Taylor Rule-type policy generates inflation persistence (through a mechanism similar to that in Dittmar et al.). In this case, strong correlation between the inflation rate and output, such as the Phillips curve, would probably be observed even when the value of ρ is small. Also, if we apply a Taylor Rule-type policy to the Kiyotaki-Moore model, the price level may not become indeterminate even when the coefficient of the inflation rate ν_{π} is above 1. Such results, if obtained, would be consistent with preceding studies and complementary to the hypothesis by Kydland et al. that inflation persistence and the Phillips curve can be explained without sticky prices.

Econo-kun: If the Kydland et al. hypothesis turns out to be correct, what policy implications does this have?

KOBAYASHI Keiichiro: In the world defined in the model of Dittmar et al., monetary policy per se has little or no impact on the real economy. Monetary policy only serves to transmit changes in real economy to the inflation rate. Therefore, it is meaningless to ask what percentage of inflation is optimal in a given state of the economy. That is, there is no need to discuss inflation targeting. According to Kydland et al., the goal of monetary policy should be zero inflation regardless of whether the economy is good or bad because price volatilities impair economic activities (probably through minor destabilizing effects not addressed in their paper). They are strongly against the conventional idea that monetary policy can alleviate the magnitude of macroeconomic business cycles.

Now, think what will happen if we replace their model with Kiyotaki-Moore and incorporate nominal money and a Taylor Rule-type policy. What policy implications would this new model provide? I have yet to do a rigorous analysis but I presume that under this model, monetary policy would impact the real economy through changes in asset prices. And I would say that even in this case the inflation (or deflation) rate of general prices would have almost no direct impact on the real economy. We might very well derive the implication that prices of assets such as land and stock should be the primary target of monetary policy. If these ideas can be properly presented in the model, we can probably say that it is closer to reality (as compared to the New Keynesian models in which only price rigidities are important or the Real Business Cycle model in which monetary policy is meaningless).