Motivated by the systematic differences in macroeconomic expectations across US households and the complexity of rational-expectations equilibria when the state of the economy is highly dimensional, we study optimal information choice in neoclassical economies with heterogeneous agents and incomplete markets. Incentives to acquire information about the current state of the economy differ strongly by wealth level and employment status. Because some households lose very little from uninformed choices, standard rational expectations equilibria are typically not robust to small costs of information. When we include optimal dynamic information choice in a standard quantitative economy with unemployment risk and incomplete markets (Krusell and Smith, 1998), we find expectational patterns similar to those observed in US micro data. Both wealth inequality and business cycle volatility are substantially larger than with full information. Moreover, policies have an additional transmission channel through their effect on information choice. For example, a wealth tax decreases information acquisition and increases both aggregate volatility and inequality.

This paper develops a tractable theory of cautious expectations. We impose the constraint that agents have to estimate the optimal weight on information in an otherwise standard class of linear dynamic economies. Within this framework, we show that expectations optimally feature dampened responses to new and prior information. Our theory has several similarities to models of limited attention. However, our theory is crucially consistent with the broad-based predictability of forecast errors and overreactive expectations that have otherwise called into question attention-based models. We illustrate the consequences of our framework in a standard consumption-savings problem, which shows that cautious expectations can help account for empirical evidence on the marginal propensity to consume and amplify precautionary savings.

Modern solutions of macroeconomic models with wealth inequality and aggregate shocks often rely on the assumption of limited but common information among house- holds. We show that this assumption is inconsistent with rational information choice. To do so, we embed information choice into a workhorse heterogeneous-agent economy. First, we demonstrate that the benefits of acquiring more precise information about the current state of the economy depend crucially on household wealth. Second, because of such heterogeneous benefits to information acquisition and the strategic substitutabil- ity of savings choices, equilibria in which households acquire the same information do not exist for plausible costs of acquiring information. Our results further imply that a representative-agent equilibrium may not exist even in the absence of exogenous sources of wealth heterogeneity.

Professional forecasts are often used to gauge household and firm expectations. Recently, the average of such forecasts have been argued to support rational expectation formation with noisy private information. We document that individual forecasts of US GDP and inflation in the Survey of Professional Forecasters overrespond to both private and public information, contradicting, prima facie, the assumption of noisy rational expectation formation. We generalize two alternative models of forecaster behavior that focus on strategic diversification and behavioral overconfidence, respectively, to dynamic environments with noisy private information. We find that both models predict overresponses, but only the overconfidence model is simultaneously consistent with a substantial overreaction to public information.

We document simultaneous over- and under-responses to new information by households, firms, and professional forecasters in survey data. Such behavior is inconsistent with existing theories based on either behavioral bias or rational inattention. We develop a structural model of information choice in which people base expectations on observables that can reconcile the seemingly contradictory facts. We show that optimally-chosen, asymmetric attention to different observables can explain the co- existence of over- and under-responses. We then embed our model of information choice into a microfounded macroeconomic model, which generates expectations consistent with the survey data. We demonstrate that our model creates over-optimistic consumption beliefs in booms and predictability in consumption changes.

The past two decades have seen a considerable increase in the amount of public information provided by policy makers. This paper proposes a novel rationale against such disclosure. Unlike other providers of public information, a policy maker can condition a policy instrument, such as a tax or an interest rate, on his information. This option allows the policy maker to control the influence of his information on market outcomes without the added obfuscation associated with partial disclosure. As a result, the exclusive use of a policy instrument is preferable in simple models in which externalities render full disclosure suboptimal. I show how this rationale extends from an abstract game to a micro-founded macroeconomic model in which firms learn from market prices.

A common view states that central bank releases can decrease a central bank’s own information about the economy and can be harmful if about inefficient disturbances, such as cost-push shocks. This paper shows how neither is the case in a micro-founded macroeconomic model in which households and firms learn from central bank releases and the central bank from the distribution of firm prices. Central bank releases make private sector and central bank expectations closer to common knowledge. This helps transmit dispersed information between the private sector and the central bank. As a result, the release of additional information can decrease the central bank’s own uncertainty and be beneficial, irrespective of the efficacy of macroeconomic fluctuations. A calibrated example suggests that the associated welfare benefits can be substantial.