We develop a model of the individual term structure of inflation expectations across forecasting horizons. Using the Survey of Professional Forecasters, we decompose disagreement about inflation expectations into individuals' long-term beliefs, private information, and public information. We find that in normal times, long-horizon disagreement is predominantly driven by individuals' long-term beliefs, while short-horizon disagreement stems from private information. During economic downturns, heterogeneous reactions to public information become a key driver of disagreement at all horizons. When forecasters disagree about public information, monetary policy exhibits a delayed response and a price puzzle emerges, underscoring the importance of anchoring inflation expectations.

A coding error in Farmer et al. (2023) (FST) meant that future information leaked into the classification scheme used to identify pockets with elevated return predictability. While the underlying return forecasts are unaffected by the error, their usage to identify predictability pockets ex ante becomes noisier under the corrected classification scheme. A simple modification that smooths the underlying forecasts prior to pocket classification retains all the main conclusions of FST including stronger out-of-sample return predictability inside pockets, economic gains from exploiting this return predictability, and the greater alignment of return predictability pockets with a sticky expectations model than with existing asset pricing models.

Forecasts of professional forecasters are anomalous: they are biased, forecast errors are autocorrelated, and predictable by forecast revisions.Sticky or noisy information models seem like unlikely explanations for these anomalies: professional forecasters pay attention constantly and have precise knowledge of the data in question.We propose that these anomalies arise because professional forecasters don't know the model that generates the data.We show that Bayesian agents learning about hard-to-learn features of the data generating process (low frequency behavior) can generate all the prominent aggregate anomalies emphasized in the literature.We show this for two applications: professional forecasts of nominal interest rates for the sample period 1980-2019 and CBO forecasts of GDP growth for the sample period 1976-2019.Our learning model for interest rates also provides an explanation for deviations from the expectations hypothesis of the term structure that does not rely on time-variation in risk premia.

Using stock market responses to drug development announcements, we measure the $\textit{value of pharmaceutical drug innovations}$ by estimating the market values of drugs. We estimate the average value of successful drugs to be \$1.62 billion and the average value at the discovery stage to be \$64.3 million. We also allow for heterogeneous expectations across several major diseases, such as cancer and diabetes, and estimate their values separately. We then apply our estimates to (i) determine the average costs of drug development at the discovery stage and the three phases of clinical trials to be \$58.5, \$0.6, \$30, and \$41 million, respectively, and (ii) investigate $\textit{drug buyouts}$, among others, as policies to support drug development.

We construct a perpetual youth DSGE model with aggregate uncertainty in which there are dynamically complete markets and agents have Epstein-Zin preferences.We prove that, when endowments have a realistic hump-shaped age-profile, our model has three steady-state equilibria.One of these equilibria is dynamically inefficient and displays real price indeterminacy.We estimate the parameters of our model and we find that a fourth-order approximation around the indeterminate steady-state provides the best fit to U.S. data.Our work interprets the large and persistent generational inequality that has been observed in western economies over the past century as the result of uninsurable income shocks to birth cohorts.

Existing methods for estimating nonlinear dynamic models are either highly computationally costly or rely on local approximations which often fail adequately to capture the nonlinear features of interest. I develop a new method, the discretization filter, for approximating the likelihood of nonlinear, non‐Gaussian state space models. I establish that the associated maximum likelihood estimator is strongly consistent, asymptotically normal, and asymptotically efficient. Through simulations, I show that the discretization filter is orders of magnitude faster than alternative nonlinear techniques for the same level of approximation error in low‐dimensional settings and I provide practical guidelines for applied researchers. It is my hope that the method's simplicity will make the quantitative study of nonlinear models easier for and more accessible to applied researchers. I apply my approach to estimate a New Keynesian model with a zero lower bound on the nominal interest rate. After accounting for the zero lower bound, I find that the slope of the Phillips Curve is 0.076, which is less than 1/3 of typical estimates from linearized models. This suggests a strong decoupling of inflation from the output gap and larger real effects of unanticipated changes in interest rates in post Great Recession.

ABSTRACTFor many benchmark predictor variables, short‐horizon return predictability in the U.S. stock market is local in time as short periods with significant predictability (“pockets”) are interspersed with long periods with no return predictability. We document this result empirically using a flexible time‐varying parameter model that estimates predictive coefficients as a nonparametric function of time and explore possible explanations of this finding, including time‐varying risk premia for which we find limited support. Conversely, pockets of return predictability are consistent with a sticky expectations model in which investors slowly update their beliefs about a persistent component in the cash flow process.

Approximating stochastic processes by finite-state Markov chains is useful for reducing computational complexity when solving dynamic economic models. We provide a new method for accurately discretizing general Markov processes by matching low order moments of the conditional distributions using maximum entropy. In contrast to existing methods, our approach is not limited to linear Gaussian autoregressive processes. We apply our method to numerically solve asset pricing models with various underlying stochastic processes for the fundamentals, including a rare disasters model. Our method outperforms the solution accuracy of existing methods by orders of magnitude, while drastically simplifying the solution algorithm. The performance of our method is robust to parameters such as the number of grid points and the persistence of the process.

I develop a new method for approximating and estimating nonlinear, non-Gaussian state space models. I show that any such model can be well approximated by a discrete-state Markov process and estimated using techniques developed in Hamilton (1989). Through Monte Carlo simulations, I demonstrate that my method outperforms popular existing methods both in terms of accuracy and computational efficiency for empirically relevant sample sizes. This finding has important implications for the estimation of nonlinear dynamic economic models, many of which are computationally intractable using exisiting methods. I illustrate an empirical application of my method by re-estimating the Wu and Xia shadow rate term structure model. When I calculate the shadow rate using the parameter estimates from my proposed filter, I find that it was as much as 2.2 percentange points lower in July 2012 than previous estimates imply.