Research
Research
Working Papers
Measuring Information Frictions with Surveys of Expectations: How Revised Data Bias Estimates
Abstract: I study how data signals of the past affect the prevailing method of estimating information frictions. Existing literature measures the relationship between average forecast errors and forecast updates to estimate the degree of information frictions, or slow response of forecasts to news. I introduce a revised data signal into the noisy information model and find this theoretic relationship now suffers from omitted variable bias and cannot be estimated by OLS. The bias is not due to noise in the signal but due to the dynamic property of the revised data which introduces additional past dependency into forecasts and breaks the Markovian flow of information. I use the model to propose a new specification which corrects the bias. When I estimate the corrected specification on data from the Survey of Professional Forecasters, I find the degree of information frictions is 33% different from existing estimates on average. I apply my results toward two applications to demonstrate the economic significance of this bias. Under my corrected estimates I find incomplete information theories more closely match the data, while the efficacy of forward guidance as a monetary policy tool is significantly reduced.
Now or Later: Optimal Data Provision with Data Revisions and Strategic Incentives
Abstract: When statistical offices (S.O.) are budget constrained they face a trade-off: provide imprecise yet timely data about the present, or wait to provide more precise yet delayed data about the past. I study this decision in the context of a beauty-contest economy where end-users of data want to produce forecasts that are both accurate and similar to one another. The S.O.'s welfare-maximizing choice of current versus revised signal precisions depends on the accuracy of private information and strength of strategic incentives. The S.O. prioritizes the revised signal when individuals have poor private information, yet as it becomes more precise the S.O. increases the budget devoted to the current signal until eventually it does not provide a revised signal. When private information and strategic incentives are sufficiently high, this reverses: the S.O. increases resources towards the revised signal in order to reduce the crowding-out of private information caused by the current signal. A larger budget constraint makes the S.O. reduce the precision of the revised signal in favor of the current signal, as the S.O. effectively treats the former as an inferior good.