The Dynamic Distribution in the Fixed Cost Model: An Analytical Solution
New!! (short paper)
I derive an analytical solution to the Kolmogorov forward equation for a fixed cost model. This is a challenging PDE, because the dynamic distribution depends on the flow of resetting agents, which is endogenously determined by the distribution itself. I show there is a shortcut that allows the flow function to be derived without first finding the entire distribution of agents. This shortcut is also valuable because many aggregate variables can be written in terms of the flow function alone. As an example, I solve the canonical menu cost model. In it, the analytical solution uncovers effects that are inconsistent with existing approximation methods. Specifically, the effects of shocks are both size and history dependent. These nonlinearities are substantial; if a monetary shock is sufficiently large, it can even reverse the sign of the effect on output.
Optimal Policy Without Rational Expectations: A Sufficient Statistic Solution
New!!
How should policymakers respond to mistakes made by agents without rational expectations? I demonstrate in a general setting that the optimal policy is determined by a sufficient statistic: agents' belief distortions. This result is both simple and only semi-structural: in order to calculate policy from the belief distortion, the policymaker does not need to know the whole macroeconomic model. They only need to know how beliefs and policies distort decisions. Crucially, they do not even need to know how expectations are formed; they only need to measure them. Next, I study several examples. In a behavioral RBC model, the optimal policy is to tax capital when agents are overly optimistic about future returns. In a behavioral New Keynesian model, the optimal policy is to raise interest rates when agents misperceive the economy to be running hot.
The Rise of AI Pricing: Trends, Driving Forces, and Implications for Firm Performance
(with Min Fang, Zheng Liu, and Yajie Wang)
New!!
We document key stylized facts about the time-series trends and cross-sectional distributions of AI pricing and study its implications for firm performance, both on average and conditional on monetary policy shocks. We use the universe of online job posting data from Lightcast to measure the adoption of AI pricing. We infer that a firm is adopting AI pricing if it posts a job opening that requires AI-related skills and contains the keyword "pricing". At the aggregate level, the share of AI-pricing jobs in all pricing jobs has increased by more than tenfold since 2010. The increase in AI-pricing jobs has been broad-based, spreading to more industries than other types of AI jobs. At the firm level, larger and more productive firms are more likely to adopt AI pricing. Moreover, firms that adopted AI pricing experienced faster growth in sales, employment, assets, and markups, and their stock returns are also more sensitive to high-frequency monetary policy surprises than non-adopters. We show that these empirical observations can be rationalized by a simple model where a monopolist firm with incomplete information about the demand function invests in AI pricing to acquire information.
Incomplete Information and Investment Inaction
(with Cheng Chen, Min Fang, Takahiro Hattori, and Eugenio Rojas)
New!!
How do investment friction and information frictions interact? We study this question in a stylized continuous time model of heterogeneous firms facing incomplete information and irreversible investment. We analytically characterize how the information friction distorts firms' decision rules and stationary distribution. The two frictions interact in rich and substantial ways. At the firm level, noisier information shrinks a firm's inaction region and reduces the elasticity of investment to productivity. In the aggregate, it increases steady-state capital, increases capital misallocation, and attenuates the effect of productivity shocks on investment. Finally, we test and confirm these predictions using Japanese administrative data that match firms' forecasts to their balance sheets, incomes, and expenditures.
Macroeconomic Models with Incomplete Information and Endogenous Signals
Revise and Resubmit - Journal of Economic Theory
This paper characterizes a general class of macroeconomic models with incomplete information, which feature endogenous signal processes. These models may exhibit multiple or no equilibria, and standard algorithms can fail to converge. I introduce an Information Feedback Regularity condition to impose discipline on these models. If the condition holds, the model has desirable properties: a computable equilibrium must exist, and if stable, it is the globally unique one. I also develop an algorithm to solve the general model and provide computational resources.
Identifying News Shocks from Forecasts
(with Philip Barrett)
Submitted
We propose a method to identify the anticipated components of macroeconomic shocks in a structural VAR. We include empirical forecasts about each time series in the VAR. This introduces enough linear restrictions to identify each structural shock and to further decompose each one into "news" and "surprise" shocks. We estimate a VAR on US time series using forecast data from the SPF, CBO, Federal Reserve, and asset prices. Unanticipated fiscal stimulus and monetary policy shocks have typical effects that match existing evidence. In our news-surprise decomposition, we find that news drives around one quarter of US business cycle volatility. News explains a larger share of the variance due to fiscal shocks than for monetary policy shocks. Finally, we use the news structure of the shocks to estimate counterfactual policy rules, and compare the ability of fiscal and monetary policy to moderate output and inflation. We find that coordinated fiscal and monetary policy are substantially more effective than either tool is individually.
Equilibrium Determinacy with Behavioral Expectations
Submitted
Behavioral expectations affect determinacy in macroeconomic models. Relaxing rational expectations can make models more or less well behaved, depending on the behavioral assumptions. In some cases, multiplicity is created; in other cases, multiplicity is eliminated. Is it possible to tell exactly when there are multiple solutions? Yes: I derive a Behavioral Blanchard-Kahn sufficient condition that ensures a unique equilibrium exists. If and only if this condition or a Sunspot Admissibility condition hold, then a model's solution must be unique. These conditions depend on the spectrum of the behavioral expectation operator. I describe how to check these conditions for an arbitrary behavioral expectation, and illustrate with a large variety of popular types of expectations, heuristics, and information frictions. As an example, I demonstrate that a large class of behavioral expectations imply a unique solution to the New Keynesian model with an interest rate peg, including all strictly backwards-looking heuristics. Another class of expectations imply that asset prices exhibit non-fundamental volatility in a standard model.
Firestorm: Multiplicity in Models with Full Information
Submitted
FIRE models can feature multiple equilibria when causality is relaxed. I demonstrate how self-fulfilling equilibria arise and explore their implications in asset pricing and business cycle models, arguing that the standard FIRE framework may need to be re-evaluated.