In full-information estimates, long-run risks explain at most a quarter of p/d variance, and habit explains even less
Critical Finance Review (forthcoming), with Andrew Chen and Rebecca Wasyk
Many consumption-based models succeed in matching long lists of asset price moments. We propose an alternative, full-information Bayesian evaluation that decomposes the price-dividend ratio (p/d) into contributions from long-run risks, habit, and a residual. We find that long-run risks account for less than 25% of the variance of p/d and that habit’s contribution is negligible. This result is robust to the prior, including priors that assume long-run risks in consumption and highly persistent habit. However, the residual mostly tracks decades-long movements in p/d. At business cycle frequency, long-run risks explain about 70% of the movements of p/d while habit’s contribution stays negligible.
The Role of Learning for Asset Prices and Business Cycles
Journal of Monetary Economics (forthcoming)
I examine the implications of learning-based asset pricing in a model in which firms face credit constraints that depend partly on their market value. Agents learn about stock prices, but have conditionally model-consistent expectations otherwise. The model jointly matches key asset price and business cycle statistics, while the combination of financial frictions and learning produces powerful feedback between asset prices and real activity, adding substantial amplification. The model reproduces many patterns of forecast error predictability in survey data that are inconsistent with rational expectations. A reaction of the monetary policy rule to asset price growth increases welfare under learning.
- published version
- FEDS working paper version
- Dynare add-on for solving learning models with conditionally model-consistent expectations
Unemployment Insurance and International Risk Sharing
European Economic Review (2019), with Stéphane Moyen and Nikolai Stähler
We discuss how cross-country unemployment insurance can be used to improve international risk sharing. We use a two-country business cycle model with incomplete financial markets and frictional labor markets where the unemployment insurance scheme operates across both countries. Cross-country insurance through the unemployment insurance system can be achieved without affecting unemployment outcomes. The Ramsey-optimal policy however prescribes a more countercyclical replacement rate when international risk sharing concerns enter the unemployment insurance trade-off. We calibrate our model to Eurozone data and find that optimal stabilizing transfers through the unemployment insurance system are sizable and mainly stabilize consumption in the periphery countries, while optimal replacement rates are countercylical overall. We also find that debt-financed national policies are a poor substitute for fiscal transfers.
Asset Price Beliefs and Optimal Monetary Policy
Revise and Resubmit, AEJ: Macroeconomics, with Colin Caines
We characterize optimal monetary policy when agents have extrapolative beliefs about asset prices that induce inefficient fluctuations in asset prices, aggregate demand and investment. We find that the optimal monetary policy raises interest rates when expected capital gains or the level of current asset prices is high, but does not eliminate deviations of asset prices from their fundamental value. When the asset is in elastic supply, optimal policy also leans against the wind, tolerating low inflation and output when asset prices are too high. Optimal policy can be reasonably approximated by simple interest rate rules that respond to capital gains. Our results are robust to a wide range of belief specifications.
A Factor Structure of Disagreement
with Edward Herbst
We document new facts about how forecaster disagreement comoves across a wide range of variables. Instead of compressing disagreement into statistics of dispersion, we use individual response data in a panel dynamic factor model to summarize the comovement of disagreement in the Survey of Professional Forecasters. We offer an interpretation of the extracted factors through a heterogeneous information model. Up until the Great Moderation, the factors describe negative comovement of individual output and inflation expectations, which can be interpreted as disagreement about the supply side of the economy. In recent years and particularly during the Great Recession, the comovement has turned positive, which can be interpreted as disagreement about the demand side of the economy. Disagreement about monetary policy plays a minor role. We also document that nowcast and medium-term forecast disagreement are systematically related.
Learning and Misperception: Implications for Price-Level Targeting
with Martin Bodenstein and James Hebden
Given the mixed success with forward guidance policies in the aftermath of the financial crisis, monetary policy strategies that target the price level have been advocated as a more effective way to provide economic stimulus in a deep recession when conventional monetary policy is limited by the zero lower bound on nominal interest rates. Yet, the effectiveness of both forward guidance and price-level targeting strategies depends on a central bank's ability to steer agents' expectations about the future path of the policy rate. We develop a flexible method of learning about the central bank's reaction function from observed interest rates that takes into account the limited informational content at the zero lower bound. When agents learn, switching from an inflation targeting to a price-level targeting strategy at the onset of a recession may not yield the desired stabilization benefits but make matters worse. Nevertheless, agents can eventually learn the systematic monetary policy response of a price-level targeting strategy, an advantage not shared by ad-hoc forward guidance strategies including temporary price-level targeting.
Impulse-based Computation of Optimal Policy Problems
I propose a new computational procedure to solve for monetary policy counterfactuals in linear models with an occasionally binding zero-lower bound (ZLB) constraint. The procedure is fast and neither requires knowledge of the structural or reduced-form equations of the model, its state variables, nor its shock processes. All that is required is a "baseline" projection of the variables entering the policy problem and impulse responses of these variables to news shocks to the monetary policy instrument, computed under an arbitrary instrument rule. A practical insight that emerges is that central banks need not be concerned about misspecification of the structural models they use for monetary policy analysis, as long as the effects of monetary policy actions on the variables of interest to policymakers are correctly specified.
The Effect of Asset Price Learning in RBC and Labour Search Models
It is plausible that subjective investor beliefs play a role in determining asset prices, but do they also affect the business cycle? I add learning about stock prices to the canonical real business cycle and the labour search and matching models. In so doing, I develop a new method to model small departures from rational expectations, which I call conditionally model-consistent expectations. Adding learning to the real business cycle model improves some asset price properties but leads to counterfactual comovement between consumption, output and stock prices. The search model with learning however has realistic business cycle and asset price properties and a sizeable amount of amplification. In particular, adding learning substantially reduces the need to rely on wage rigidity to explain the observed magnitude of unemployment fluctuations.