John W. Keating

The long-run relationship between inflation and output in postwar economies

Abstract We investigate the relationship between inflation and real output in a large sample of postwar economies. Our methodology is to use a structural vector autoregression to estimate the response of the level of real output to permanent inflation shocks separately for each country. We find that a permanent shock to inflation is not associated with a permanent movement in the level of real output for most countries in our sample. The main exceptions are certain low inflation countries, in which permanent inflation shocks permanently increase the level of output. We also find that permanent inflation shocks do not permanently influence real output growth rates in our sample.

Identifying VAR models under rational expectations

Abstract This paper shows that the exclusion restrictions used to identify structural vector autoregressions (SVARs) generally yield inconsistent parameter estimates under rational expectations. I develop an alternative method of identifying rational-expectations models within the SVAR framework that circumvents the need of imposing stringent identifying restrictions on the dynamic processes for technologies, preferences, and other mechanisms (e.g., adjustment costs). These strong restrictions may explain why rational-expectations models are frequently rejected by the data. Two empirical examples demonstrate the feasibility and potential advantages of this alternative modeling strategy.

Structural information in recursive VAR orderings

Abstract This paper investigates structural models that will permit a Cholesky decomposition of the covariance matrix of VAR residuals to identify some structural impulse response functions. Cholesky decompositions are found to be useful identification tools for the set of partially recursive structural models. A partially recursive structure is defined as any block recursive system where the equations in one block can be recursively ordered and where the structural shocks are uncorrelated. Using this class of models, we derive necessary and sufficient conditions for the moving average representation from a Cholesky decomposition to identify structure. The paper concludes by discussing implications of these results for empirical research.

Macroeconomic modeling with asymmetric vector autoregressions

VARs typically employ the same number of lags for each variable. Consequently, they often estimate many insignificant coefficients. The “asymmetric VAR” (AVAR), defined as a VAR for which each variable may have a unique number of lags, is shown to be the reduced form for a general linear structure. A particular economic structure is developed to compare and contrast parameter estimates from AVAR and VAR specifications. Qualitatively similar results are obtained from each model. Parameters in the AVARs frequently have smaller standard errors than the VAR, suggesting AVARs may obtain more efficient estimates. Important questions about macroeconomic structure are addressed with these models.

Permanent and Transitory Shocks in Real Output: Estimates from Nineteenth Century and Postwar Economies

This paper reexamines Olivier J. Blanchard and Danny Quahs (1989) aggregate supply/demand model interpretation of output shocks using data from ten countries. Although postwar data support their interpretation, the model is not supported by nineteenth=century data. In the postwar period, permanent output shocks cause the price level to move in the opposite direction (like supply shocks) and temporary output shocks cause the price level to move in the same direction (like demand shocks). But with pre-World War I data, permanent output shocks typically cause price level movements in the same direction. The results are systematic and cannot be explained by data problems.

The dynamic effects of aggregate demand and supply disturbances in the G7 countries

This paper uses post-World War II and pre-World War I data on output and the unemployment rate from the G7 countries to estimate Blanchard and Quahs (1989) model. Their model is identified by assuming that permanent movements in output obtain from aggregate supply shocks and that aggregate demand shocks have temporary effects on output. We find that these demand shocks typically explain more output variance in the post-World War II period than in the pre-World War I period. This result is consistent with the view that price adjustment was slower in the latter half of the twentieth century. A second finding is that supply shocks cause the unemployment rate to rise for some countries and fall for others. This result implies that at least two different kinds of supply shocks are at work. However, if a country is subject to various kinds of supply shocks which have qualitatively different effects on the unemployment rate, Blanchard and Quah have shown that their model is unable to identify structural effects. The paper concludes by discussing implications of this result for macroeconometric research.

Structural inference with long-run recursive empirical models

This paper investigates conditions under which a long-run recursive model can be used to identify a structure. Economists frequently employ this type of empirical model. I define the class of long-run partially recursive structures. If an economic system is a member of this class, then certain long-run recursive empirical models will obtain some of the structural impulse response functions. This sufficient condition for a structure is first shown in a vector autoregression. A well-known example from the literature is used to illustrate this particular class of structures and to present some useful applications of the result. Then the result is shown in models of cointegrated time series. Necessary conditions for a long-run recursive model to identify structure are addressed in the conclusion.

THE TIME-VARYING EFFECTS OF PERMANENT AND TRANSITORY SHOCKS TO REAL OUTPUT

Annual changes in volatility of U.S. real output growth and inflation are documented in data from 1870 to 2009 using a time varying parameter VAR model. Both volatilities rise quickly with World War I and its aftermath, stay relatively high until the end of World War II and drop rapidly until the mid to late-1960s. This Postwar Moderation represents the largest decline in volatilities in our sample, much greater than the Great Moderation that began in the 1980s. Fluctuations in output growth volatility are primarily associated with permanent shocks to output while fluctuations in inflation volatility are primarily accounted for by temporary shocks to output. Conditioning on temporary shocks, inflation and output growth are positively correlated. This finding and the ensuing impulse responses are consistent with an aggregate demand interpretation for the temporary shocks. Our model suggests aggregate demand played a key role in the changes in inflation volatility. Conversely, the two variables are negatively correlated when conditioning on permanent shocks, suggesting that these disturbances are associated primarily with aggregate supply. Our results suggest that aggregate supply played an important role in output volatility fluctuations. Most of the impulse responses support an aggregate supply interpretation of permanent shocks. However, for the pre-World War I period, we find that at longer horizons a permanent increase in output is generally associated with an increase in the price level that is frequently statistically significant. This evidence suggests aggregate demand may have had a long-run positive effect on output during the pre-World War I period.

A Model of Monetary Policy Shocks for Financial Crises and Normal Conditions

In late 2008, deteriorating economic conditions led the Federal Reserve to lower the federal funds rate to near zero and inject massive liquidity into the financial system through novel facilities. The combination of conventional and unconventional measures complicates the challenging task of characterizing the effects of monetary policy. We develop a novel method of identifying these effects that maintains the classic assumptions that a central bank reacts to output and the price level contemporaneously and may only affect these variables with a lag. A New-Keynesian DSGE model augmented with a representative financial structure motivates our empirical specification. The equilibrium model provides theoretical support for our choice of different series to replace variables that were popular in models of monetary policy but became problematic in the aftermath of the 2008 financial crisis. One of our most important innovations is to utilize the Divisia M4 index of money as the policy indicator variable. The model is bolstered by its ability to produce plausible responses to a monetary policy shock in samples that include or exclude the recent crisis period.

Forecast Design in Monetary Capital Stock Measurement

We design a procedure for measuring the United States capital stock of money implied by the Divisia monetary aggregate service flow, in a manner consistent with the present-value model of economic capital stock. We permit non-martingale expectations and time varying discount rates. Based on Barnett’s (1991) definition of the economic stock of money, we compute the U.S. economic stock of money by discounting to present value the flow of expected expenditure on the services of monetary assets, where expenditure on monetary services is evaluated at the user costs of the monetary components. As a theoretically consistent measure of money stock, our economic stock of money nests Rotemberg, Driscoll, and Poterba’s (1995) currency equivalent index as a special case, under the assumption of martingale expectations. To compute the economic stock of money without imposing martingale expectations, we define a procedure for producing the necessary forecasts based on an asymmetric vector autoregressive model and a Bayesian vector autoregressive model. In application of this proposed procedure, Barnett, Chae, and Keating (2005) find the resulting capital-stock growth-rate index to be surprisingly robust to the modeling of expectations. Similarly the primary conclusions of this supporting paper regard robustness. We believe that further experiments with other forecasting models would further confirm our robustness conclusion. Different forecasting models can produce substantial differences in forecasts into the distant future. But since the distant future is heavily discounted in our stock formula, and since alternative forecasting formulas rarely produce dramatic differences in short term forecasts, we believe that our robustness result obviates prior concerns about the dependency of theoretical monetary capital stock computations upon forecasts of future expected flows. Even the simple martingale forecast, which has no unknown parameters and is easily computed with current period data, produces a discounted stock measure that is adequate for most purposes. Determining an easily measured extended index that can remove the small bias that we identify under the martingale forecast remains a subject for our future research. At the time that Milton Friedman (1969) was at the University of Chicago, the “Chicago School” view on the monetary transmission mechanism was based upon the wealth effect, called the “real balance effect” or “Pigou (1943) effect,” of open market operations. Our research identifies very large errors in the wealth effects computed from the conventional simple sum monetary aggregates and makes substantial progress in the direction of accurate measurement of monetary-policy wealth effects.