سیاست های پولی درونزا و چرخه کسب و کار

The correlations and volatilities of real variables seem to be stable over time, but the relation between real and nominal variables is unstable. Presumably, one important factor behind this observation is the nature of money supply. In this paper, I look at a business cycle model where the central bank sets money supply to minimize the volatility of inflation and output. I find that small changes in the central bank's preferences can generate large changes in the derived money supply rule and in correlations between real and nominal variables. Although wages are assumed to be sticky, changes in the money supply rule do not generate any major changes in the behavior of real variables.

It is widely acknowledged that money, inflation, and output are positively correlated over the business cycle. The behavior of real variables seems to be stable, but there is clear evidence that the relations between real and nominal variables change over time. In a large sample of countries, Backus and Kehoe (1992) find real variables to behave similarly in different subperiods while the behavior of money, inflation, and the price level is changing. Gavin and Kydland (1996) document these facts for U.S. post-war data. Presumably, variations in the monetary policy is one important explanation to these observations. Even if money does not have any major real effects, changes in money supply certainly have a large impact on nominal variables. If the central bank takes real variables such as output and unemployment into consideration when deciding on money supply, nominal and real variables will be correlated just because of the central bank's reactions to changes in these variables, and if the money supply rule changes, so will correlations between real and nominal variables. In the present paper, the central bank does indeed take the real economy into consideration when deciding on monetary policy. More precisely, I solve for the money supply rule that minimizes the central bank's loss function over inflation and output variability in a dynamic stochastic general equilibrium model. There are shocks both to productivity and in the money supply process. Wages have to be set before contemporaneous shocks and central bank decisions are observed. Hence, unanticipated changes in money supply have real effects. I find, as did Gavin and Kydland (1996), that changes in the money supply rule can induce large changes in the business cycle behavior of nominal variables. The present paper adds to Gavin and Kydland's analysis by showing that money supply rules can change substantially when central bank preferences change. I find that the quantitative effects that monetary policy has on real variables are small but significant enough to make the optimal money supply rule change considerably when the central bank's weight put on output stability changes. The paper thus shows that sizeable variations in the central bank reaction function can be a reality. The reason for the instability of the optimal money supply rule is that the central bank faces a trade-off between output and inflation stabilization. When the central bank puts much weight on output stability, its response to a negative productivity shock is as follows. The central bank observes the shock and increases money supply directly. Since nominal wages are assumed to be sticky, this action will decrease real wages and thus stimulate employment. Wage contracts will then be renegotiated, so the central bank cannot exploit the Phillips curve in later periods. Instead, the central bank contracts money supply in successive periods to decrease inflation. This leads to a temporary decrease in the distortionary effects from inflation and stimulates real activity. When, on the other hand, the central bank puts much weight on inflation stability its reactions are different. The central bank does not exploit the Phillips curve at all. Instead it contracts money supply in order to dampen the inflationary tendencies caused by the productivity shock. Compared to the first scenario, the central bank's willingness to use the timing of the inflation tax as an instrument to stabilize output has decreased. The paper has implication both for empirical and theoretical research on the role of money in the business cycle. When trying to estimate, for example, a vector autoregression including both real and nominal shocks in the system, one must be careful in controlling for changes in monetary regimes. Ideally, one should use short time series for periods of stable monetary policy. Moreover, using high-frequency data (as do for example Bernanke and Mihov, 1995) is an advantage since then, arguably, the central bank cannot influence contemporaneous output. The main implication for theoretical modeling is that we should not expect there to be one business cycle behavior of nominal variables, but rather one behavior for each monetary regime. Before going on to the model and its implications, I will shortly comment on earlier literature in this field. Methodologically, my approach is akin to the real business cycle framework. The model I work with is not purely ‘real’, though, since there are money supply shocks and wage rigidity. My attempt to introduce money supply in this framework is not new, but until recently a common critique against real business cycle models was their absence or ignorance of monetary issues. Some articles allowed for money, in particular King and Plosser (1984), but the focus was still on the real economy and productivity shocks. Lately, though, several attempts to incorporate effects of monetary policy in dynamic general equilibrium models have been done, for example Cooley and Hansen 1989 and Cooley and Hansen 1995 and Huh (1993).1Cooley and Hansen (1995) assume that money supply is exogenous and follows an AR(1) process. In reality, however, the central bank reacts to changes in the economic environment when they decide on the monetary policy. This has been captured in the paper by Huh. He postulates a reaction function for the central bank, and this is fitted to actual data. In a recent paper, Gavin and Kydland (1996) first document that the volatility and cross-correlations of real variables have been stable in post-war U.S. data but that the correlations between real and nominal variables have changed over time. They then look at a model with a transactions motive for holding money, and experiment with different money supply rules. As expected, they find that changes in the money supply rule have large effects on the correlations between real and nominal variables, but that the behavior of real variables is unaffected by the experiments. To model monetary policy out of the general equilibrium framework has been typical for research in the real business cycle tradition so far. In this paper, I will assume that the central bank sets monetary policy to minimize a loss function over inflation and output. The central bank is assumed to dislike both inflation in itself and fluctuations in output and inflation. The main difference between my setup and earlier dynamic equilibrium models with money is that money growth was typically fitted to actual data in previous work, whereas I let monetary policy be the equilibrium outcome of the model used. Since the model is only a simplification of the true economy it will be more relevant to relate monetary policy to the model than to data, if we want to learn anything about how monetary policy works and how it (at least theoretically) drives the business cycle. A limitation of the approach is that the central bank's preferences are not derived from the preferences of the agents in the economy. The paper is organized as follows. In Section 2, I present the model used in the paper. Then, in Section 3, I calibrate the model and look at its business cycle properties. In Section 4, I look at how changes in central bank preferences affect the bank's behavior and the business cycle properties of the simulated economy. Section 5 concludes.

The effects of anticipated and unanticipated monetary policy have for a long time been a controversial issue in economics. The observation that correlations between real and nominal variables are significant in magnitude is not enough to conclude a causality from money to output or vice versa. Theoretically, these correlations can, for example, be due to nominal rigidities, i.e. that money causes output. It could also be the case that money demand responds to real activity, i.e. that real variables cause fluctuations in nominal variables. In models trying to explain the money–output correlations, money supply has often been neglected. In this paper, I have worked from the starting point that the money supply process is the most important source of fluctuations in nominal variables. Therefore, money supply will also be an important factor behind the relationship between real and nominal variables if, which seems to be the case, the central bank takes the real economy into account when deciding on money supply. In order to study these issues, this paper has endogenized the central bank's money supply decisions in a dynamic general equilibrium model of macroeconomic fluctuations. The central bank has some power to stabilize inflation and output in the model. To achieve this stabilization it has to react to changes in the real and nominal environment. I find that the money supply process, as expected, is an important determinant of the joint behavior of real and nominal variables. I also find that small changes in the central bank's desire to stabilize output relative to inflation cause large changes in the implied money supply rule and in the behavior of nominal variables, but real variables are mostly unaffected. An interesting next step on this research agenda would be to explicitly model the stochastic nature of central bank preferences. Empirical research has indicated that preferences or objectives actually do change and this paper has shown that such changes can have important effects on the conduct of monetary policy. My presumption is that preference shocks are more important than other monetary policy shocks, and that such shocks will be a necessary ingredient in a succesful theoretical model of the co-fluctuations of money and output.