Kenneth D. West’s research while affiliated with University of Wisconsin–Madison and other places

This paper compares the cyclical and secular behavior of Japanese and US inventories at the aggregate and sectoral level, 1967–1987. While, as is well known, US inventories are sharply procyclical, Japanese inventories are only mildly procyclical. In neither country do inventory and sales move together in the long run, in the sense that the two series do not seem to be cointegrated. In Japan, but not in the US, there is a secular decline in the inventory-sales ratio.

A simple real linear-quadratic inventory model is used to determine how cost and demand shocks interacted to cause fluctuations in aggregate inventories and GNP in the United States, 1947–1986. Cost shocks appear to be the predominant source of fluctuations in inventories and are largely, though not exclusively, responsible for the fact that GNP is more variable than final sales. Cost and demand shocks are of roughly equal importance for GNP. These estimates, however, are imprecise. With different, but plausible, values for a certain target inventory-sales ratio, cost shocks are less important than demand shocks for GNP fluctuations.

Casual examination of annual postwar data on inventories and aggregate output for seven developed countries -- Canada, France, West Germany, Italy, Japan, United Kingdom, United States -- suggests that in these countries the primary function of aggregate inventories is not to smooth aggregate output in the face of aggregate demand shocks. Japan is a possible exception to this generalization.

Hansen and Sargent (1981b) discussed instrumental variables procedures for estimating linear rational expectations models, under the assumption that all variables have zero unconditional means. This note points out two implications if variables instead have nonzero unconditional means: first, there are additional restrictions beyond those noted by Hansen and Sargent (1981b), and second, imposing these restrictions results in more efficient estimates of the parameters that are the focus of Hansen and Sargent (1981b). Explicit formulas are given for the restrictions generated by some commonly assumed deterministic terms.

Empirical examination of some aggregate manufacturing data suggests that order backlogs may help explain two puzzling facts: (1) the variability of production appears to be greater than that of demand, and (2) inventories appear to be drawn down when demand is low, built up when demand is high.

It is shown that GNP will have an autoregressive root very close to unity in a variant of Taylor's (1980a,b) overlapping wage contracts model, for stylized versions of simple money supply rules and plausible values for the model's parameters. In this variant, monetary policy is the only reason for serial correlation in GNP. It is premature, therefore, to conclude, as some authors, have, that the presence of such a root in U.S. GNP is inconsistent with either a stationary natural rate or with nominal shocks playing a major role in the business cycle.

This paper uses a variance bounds test to see whether consumption is too sensitive to news about income to be consistent with a standard permanent income model, under the maintained hypothesis that income has a unit root. It is found that, if anything, consumption is less sensitive than the model would predict. This implication is robust to the representative consumer having private information about his future income that the econometrician does not have, to wealth shocks, and to transitory consumption. This suggests the importance in future research on the model of allowing for factors that tend to make consumption smooth.

A standard efficient markets model states that a stock price equals the expected present discounted valu e of its dividends, with a constant discount rate. This is shown to i mply that the variance of the innovation in the stock price is smalle r than that of a stock-price forecast made from a subset of the marke t's information set. The implication follows even if prices and divid ends require differencing to induce stationarity. The relation betwee n the variances appears not to hold for some annual U.S. stock-market data. The rejection of the model is both quantitatively and statisti cally significant. Copyright 1988 by The Econometric Society.

Under fairly general conditions, ordinary least squares and linear instrumental variables estimators are asymptotically normal when a regression equation has nonstationary right hand side variables. Standard formulas may be used to calculate a consistent estimate of the asymptotic variance-covariance matrix of the estimated parameter vector, even if the disturbances are conditionally heteroskedastic and autocorrelated. So inference may proceed in the usual way. The key requirements are that the nonstationary variables share a common unit root and that the unconditional mean of their first differences is nonzero. Copyright 1988 by The Econometric Society.