For a cluster-robust t -statistic under cluster heterogeneity we establish that the cluster-robust t -statistic has a gaussian asymptotic null distribution and develop the effective number of clusters, which scales down the actual number of clusters, as a guide to the behavior of the test statistic. The implications for hypothesis testing in applied work are that the number of clusters, rather than the number of observations, should be reported as the sample size, and the effective number of clusters should be reported to guide inference. If the effective number of clusters is large, testing based on critical values from a normal distribution is appropriate.

We study the effect of income uncertainty on consumption in a model that includes precautionary saving. In contrast to previous studies, we focus on time-series variation in income uncertainty. Our time-series measure of income uncertainty is constructed from a panel of forecasts. We find evidence of precautionary saving in that increases in income uncertainty are related to increases in aggregate rates of saving. We also find evidence that anticipated income growth rates have less explanatory power for consumption growth rates after conditioning on income uncertainty. The evidence indicates the presence of forward-looking consumers who gradually adjust precautionary savings in response to changing income uncertainty.

We develop a method for measuring the foresight agents have. We first dichotomize an agent's information at current date t into knowledge up to date t 1 f and expectations after t 1 f . We then form a residual-based test statistic that allows us to compare prediction errors for econometric models based on different values of f . We illustrate the method, examining investment around tax reforms to measure the foresight firms have about tax policy. In this illustration, current investment appears to reflect currently available information but little foresight other than foresight of enacted policy changes.