Data uncertainty: Empirical evidence, general-equilibrium implications, and hedging strategies

In the first chapter we document the empirical properties of revisions to major macroeconomic variables in the U.S., over the period 1966–2000. We find that these revisions do not have a zero mean, which indicates that the initial announcements by statistical agencies are biased. We also find that the revisions are quite large compared to the original variables. They are predictable using the information set at the time of the initial announcement, which means that the initial announcements of statistical agencies are not rational forecasts. We also provide some evidence that professional forecasters ignore this predictability. Our findings suggest that data revisions in the U.S. do not satisfy simple desirable statistical properties.;In the second chapter, using the empirical results from the previous chapter as the motivation, we study the effects of data revisions in a general equilibrium framework. We find that the presence of data revisions, or data uncertainty, creates a precautionary motive and causes significant changes in the decisions of agents. We also find that the model with revisions captures some aspects of the business cycle dynamics of the US data better than the benchmark model with no revisions. Using our model we measure the cost of having data revisions to be about ;In the third chapter we take the first step in analyzing ways of hedging the risk of data revisions. We show that we can construct portfolios of assets, which are maximally correlated with revisions to macroeconomic variables. The average correlation of the returns of these hedging portfolios and the underlying revision risk are in the order of 0.50, which is very promising in terms of hedging performance.