Research Database

This paper describes and evaluates P-Star (P*), a new method to forecast inflation trends which was introduced by the Federal Reserve Board of Governors in the summer of 1989. The paper examines how well P* would have done, compared with eight other forecasting methods, had all of these methods been used to forecast inflation in the 1970s and 1980s. P* turns out to be not an exceptionally good or bad way to forecast inflation.

It is argued that economists ought to recognize that modeling in different styles will be appropriate for different purposes or different stages in the development of an area of economics. As an example, the paper displays simulations of a stochastic general equilibrium model which shed light on the interpretation of widely discussed small macroeconomic vector autoregressive models connecting monetary variables to output and prices.

This model extends one originally constructed by Robert Litterman in 1980 and used continuously since then to prepare quarterly forecasts. The current version is 3 variables larger than Litterman’s original model, and it now allows time variation in coefficients, predictable time variation in forecast error variance, and non-normality in disturbances. Despite this elaboration the model in a sense has just 12 parameters free to fit the behavior of 9 variables in 9 equations. The paper reports the model structure and summarizes some aspects of its recent forecasting performance.

This paper discusses at an undergraduate level how forecast rationality can be tested. It explains that forecasters should correctly use any relevant information they knew in making their predictions. It shows that forecast rationality can be tested by determining whether the forecasters' prediction errors are predictable. After addressing what data and methods can be used for testing rationality, the paper presents tests of the price-forecast rationality of individual professional forecasters. Unlike results of previous studies, the test results show that those forecasters' price predictions appear to be rational.

There are two major differences between Japan and the United States in the way saving is calculated in their national accounts. First, depreciation in Japanese national accounts is based on historical costs, which leads to an understatement of true depreciation and hence an overstatement of net saving. Second, government capital formation is not included in U.S. saving. This article adjusts the official Japanese saving numbers by evaluating depreciation at replacement costs and excluding government capital formation from saving. Doing so significantly reduces the apparent gap between the national saving rates of the two countries. Since 1970 Japan's national saving rate has been declining to the stationary U.S. rate. This trend, however, has been reversed in recent years. In contrast, Japan's wealth-to-income ratio (excluding land), after declining in the late 1950s, has been rising toward the U.S. ratio and has reached the U.S. level in 1987.

Stochastic inflation affects the risk characteristics, measured by the equity premium and the correlation of the equity’s return with consumption, in a fundamental way. The riskiness of a dollar-denominated asset depends on two conditional covariances: the covariance of the marginal rate of substitution (MRS) with the equity price and the covariance of the MRS with the rate of appreciation in the purchasing power of money. The second covariance may take either sign which becomes significant when the risk characteristics of the dollar-denominated asset are compared with the risk characteristics of an indexed asset constructed in a real version of the model.

The effects of stochastic inflation on the assets’ risk characteristics are studied in a parameterized version of a cash-in-advance asset-pricing model. The growth rates of the endowment and monetary transfer evolve according to a VAR. The equity price is a geometric distributed lead of log–normally distributed random variables; an algorithm to express the price as an explicit function of the state variables is described.

This paper evaluates Hayashi's conjecture that Japan's postwar saving experience can be accounted for by the neoclassical model of economic growth as that country's efforts to reconstruct its capital stock that was severely damaged in World War II. I call this the reconstruction hypothesis. I take a simplified version of a standard neoclassical growth model that is in widespread use in macroeconomics and simulate its response to capital destruction. The saving rate path implied by the model differs significantly from the path taken by actual Japanese postwar saving data. I discuss several model modifications which would reconcile the reconstruction hypothesis with Japan's postwar saving experience. For the reconstruction hypothesis to be credible requires independent evidence on the empirical plausibility of the model modifications. It is left to future research to determine whether that evidence exists.

The paper employs three different types of identifying restrictions to calculate the impulse responses for the trivariate series composed of the U.S. unemployment level, real GNP and the money stock. The first two are the zero restrictions, arising from the assumption of the delayed information pattern available in forming a money reaction function. The third assumes a particular simplified structural model. The paper shows that the impulse response patterns are generally insensitive to these alternative specifications. Similar exercises are carried out for the bivariate series composed of the U.S. and the unemployment level.

This paper examines several grounds for doubting the value of much of the special attention recently devoted to unit root econometrics. Unit root hypotheses are less well connected to economic theory than is often suggested or assumed; distribution theory for tests of other hypotheses in models containing unit roots are less often affected by the presence of unit roots than has been widely recognized; and the Bayesian inferential theory for dynamic models is largely unaffected by the presence of unit roots. The paper displays an example to show that when Bayesian probability statements and classical marginal significance levels diverge as they do for unit root models, the marginal significance levels are misleading. The paper shows how to carry out Bayesian inference when discrete weight is given to the unit root null hypothesis in a univariate model.

The state vector in the innovation representation is asymptotically the most efficient instrumental variable estimator for the observation matrix C. The paper compares small sample properties of IV estimators for C, the dynamic matrix A and other matrices with the system theoretic estimators described in Aoki (1987) by a small scale Monte Carlo simulations. The IV estimators appear to be about the same as the system theoretic ones as far as their small sample properties are concerned. The covariance matrix of the state vector calculated from the IV point of view are also compared with the solutions of the Riccati equations. The simulation results show that they have quite similar sample means and standard deviations. This method of calculating the state vector covariance matrices may be computationally faster than solving the Riccati equation by the Schur decomposition algorithm.

Previous business cycle models have made the assumption that all the variation in the labor input is either due to changes in hours per worker or changes in number of workers, but not both. In this paper, both vary. We think this a better model for estimating the contribution of Solow technology shocks to aggregate fluctuations. We find that about 70 percent of U.S. postwar cyclical fluctuations are induced by variations in the Solow technology parameter.