An Optimization-Based Econometric Framework for the Evaluation of Monetary Policy
Summary (3 min read)
2. The Effects of Monetary Policy Shocks under the Current Policy Regime
- The complete estimated system also allows us to compute the response of output, inflation, and interest rates to a monetary policy shock.
- These impulse responses are plotted in the three panels of Figure 1 .
- Responses are plotted for a one-standard-deviation shock that raises the funds rate unexpectedly by about 0.8%.
- Nonetheless, they give us an idea of the features that their structural model should possess in order to be consistent with the data.
- In particular, a monetary tightening should temporarily lower both output and inflation; and these effects should occur only with a lag of a couple of quarters-so that the effects on output and inflation largely occur after short-term interest rates have returned to their normal level.
8. Though these cannot be read off from the regression reported in
- These results agree qualitatively with those that emerge from several recent VAR exercises, including Christiano, Eichenbaum, and Evans (1994) and, more relevantly (because he considers a three-variable VAR similar to ours), Cochrane (1994) .
- In one of the many exercises reported in his paper, Cochrane (1994) estimates a VAR over the period 1959 to 1992 that includes quarterly observations of the federal funds rate as well as of the logarithms of output and of the price level.9.
- When he includes a trend and computes the monetary policy shock by supposing that the systematic component of policy lets the federal funds rate at t react to output and the price level at t, he gets very similar impulse responses.
3. A Simple Model of Output and Inflation Determination
- The authors assumption thus amounts to a limiting case of a time-to-build model, in which the bulk of the expenditure connected with a project initiated at the beginning of period t -1 occurs during period t.
- Note that the authors could give an information-lag interpretation to this restriction upon household purchases: households choose their overall level of purchases in period t with knowledge of period-t -1 goods-market conditions, but before learning about period-t -1 money-market conditions, or about period-t conditions in either goods markets or money markets.
- This quantity appears as the Lagrange multiplier associated with constraint (3.4).
- It is measured in units of period-t utility flow per dollar.
- As is standard in models of monopolistic competition, the authors assume that an individual supplier regards itself as unable to affect the evolution of the variables Yt and Pt, and so chooses its own price, taking the evolution of those variables as given.
In our computation of the equilibrium responses to shocks in subsequent sections, we make use of a log-linear approximation to the equilibrium conditions of our model, expanding in terms of percentage deviations of various stationary state variables
- The source of the real effects of monetary policy in their model is an assumption of decision lags in price setting.
- Following Calvo (1983) , the authors assume that prices are changed at exogenous random intervals.
- Of those who get to choose a new price, a fraction y start charging the new price during that period, whereas the remaining fraction 1y must wait until the next period to charge the new price, because (owing to a different organization of the markets for these goods) they must post their prices a quarter in advance.
- These assumed delays explain why no prices respond in the quarter of the monetary disturbance (as assumed in their identification of the policy shocks), and why the largest response of inflation to a monetary shock takes place only two quarters after the shock.
- (The price chosen by all suppliers that choose on the basis of the same information set will be the same.).
4. Estimation of Model Parameters
- It is also worth remembering that the predictions of their model on this score (as on others) are themselves uncertain.
- In particular, they depend on the estimated coefficients of the monetary policy rule, which can hardly be estimated with great precision.
- The authors method, which has estimated the monetary policy rule without any reference to the implications of the coefficients of this rule for the nature of the theoretical impulse responses of output and inflation to a monetary shock, makes it particularly unlikely that the theoretical impulse responses will match the unrestricted VAR estimates.
- A joint estimation strategy (in which the coefficients of the policy rule and the structural parameters are jointly chosen so as to match theoretical with estimated impulse responses) might well improve the fit of the impulse response of inflation.
5. Identification of Shock Processes
- The authors construct time series for the three stochastic disturbances Et, Ct, and YA.
- The authors further show how the VAR can be used to infer the stochastic process that generates these variables.
- Finally, the authors show how to construct the response of their three endogenous variables to the shock processes for any given monetary policy rule.
- This allows us to construct counterfactual histories that, according to their model, would have taken place if the monetary authority had followed a different rule.
Note first that (2.3) can be premultiplied by T-' to yield
- The authors denote by Zt the vector whose elements are the model's theoretical predictions concerning the elements of Zt, the vector of historical time series.
- Use of the monetary policy rule implied by the VAR ensures that the authors can perfectly reconstruct the behavior of interest rates as long as they are also able to match the behavior of output and inflation.
- It is important to stress that this ability to reconstruct the historical series does not imply that their structural model is correct.
- These modified Z_t-and et are then substituted into (5.5) to construct the historical sequence of real disturbances.
6. Simulation of Alternative Monetary Policy Rules
- Even sharper contrasts between policy rules are possible if the authors vary the coefficients of the "Taylor rule.".
- The increased response to deviations of output from trend is predicted to reduce the variance of output fluctuations to about a tenth of its value under the historical policy regime.
- This stabilization of output, however, is accompanied by increased volatility of inflation and short-term nominal interest rates.
- (A counterfactual historical simulation assuming this policy rule is shown in Figure 4 .).
- This raises the obvious question of which policy rule results in more desirable patterns of fluctuations.
7. The Welfare Loss from Price-Level Instability
- Table 2 also provides a nice contrast between this constrained-optimal policy and the "Taylor rule" whose coefficients are 06 = 10 and Oy = 0.
- These two policies induce about the same variance of inflation and output while also having similar losses from variability L. However, the simple "Taylor rule" achieves this by having interest rates react aggressively to inflation, and this leads interest rates to be very volatile.
- The authors constrained-optimal rule, by contrast, allows interest rates to be less variable by tailoring the dynamic response to shocks more appropriately.
Did you find this useful? Give us your feedback
Citations
5,370Â citations
4,540Â citations
4,250Â citations
3,990Â citations
3,914Â citations
References
8,580Â citations
"An Optimization-Based Econometric F..." refers background or methods in this paper
...It is also able to match the negative correlation of output with lagged nominal interest rates that King and Watson (1996) find cannot be explained by an optimizing model with Calvo-style staggered price setting similar to our own. Fuhrer (1997b) also draws attention to this correlation and suggests that a "backwards looking" IS curve is needed to explain it....
[...]
...supply equation obtained from models such as those of Rotemberg (1982) and Calvo (1983), called by Roberts (1995) "the New Keynesian Phillips Curve....
[...]
...Following Calvo (1983), we assume that prices are changed at exogenous random intervals....
[...]
8,414Â citations
6,161Â citations
5,379Â citations
3,027Â citations