Session 1: Asset Pricing: Models, Solution Methods, and Applications

This paper develops a theory of the elasticity of preference for early resolution of uncer-tainty (PER) that parallels the Arrow-Pratt measure of risk aversion in expected utility theory. We demonstrate that the local welfare gain of early resolution of uncertainty is equal to the product of the elasticity of PER and the conditional variance of continuation utility. We illustrate how asset market data can be used to estimate the elasticity of PER and how this measure can be used to compute the welfare gain for various experiments of early resolution of uncertainty.

This paper examines how important news can effectively move stock prices without trading. Using macroeconomic announcements released outside regular Chinese stock market hours, we demonstrate that calendar time, independent of trading, plays a critical role in information transmission and price discovery at market reopening. We further show that this effect is driven by retail investors’ overnight learning over time, as evidenced by increased post-announcement news flow and active information acquisition on social media platforms. Our findings indicate that overnight learning helps level the information playing field among heterogeneous investors, leading to more efficient price discovery and reduced information asymmetry at market opening. Retail investors benefit particularly from this process, as demonstrated by increased trading activity, higher trading profit, and reduced return reversals during the trading day compared to overnight returns.

Subjective expectations and asset prices both revolve around distorted probabilities. Subjective expectations are expectations under biased probabilities, and asset prices are expectations under risk-neutral probabilities. Given this link, asset pricing tech-niques designed to estimate a Stochastic Discount Factor (SDF) can be used to estimate a Subjective Belief Factor (SBF) a distortion that characterizes many subjective expectations, even for non-nancial variables. Further, the Subjective Belief Factor can be used to characterize asset prices, by separating the roles of beliefs and prefer-ences/risk. Using the Survey of Professional Forecasters and Blue Chip, we nd that dierences between subjective expectations and statistical expectations for 24 macroe- conomic variables can be summarized (average R2 of 50%) by a single SBF related to real GDP growth and the T-bill rate. This SBF accurately replicates dierences across the 24 variables in the serial correlation of forecast errors and under/overreaction. Ap- plying this SBF to cross-sectional stock returns, we nd it accounts for the majority of excess returns for the Fama-French factors and explains about two thirds of the variation in returns across 176 anomalies, while the remaining third is attributed to preferences/risk. Our results support models like diagnostic expectations and robust control in which agents' beliefs across dierent variables are characterized by a single probability distortion.

We document a new stylized fact about inelastic demand in financial markets: larger un-informed demand shocks have smaller per-unit price impact (i.e. smaller price multipliers). That is, we find total price impact is concave in the size of demand shocks. This finding reveals an important dimension of endogenous variation in price multipliers. Since many existing theories imply convex or linear price impact, our finding helps discipline potential theories of inelastic demand. Based on these insights, we propose a nonlinear asset de-mand system with endogenous price elasticities of demand. The estimated demand system demonstrates this concavity is quantitatively important: extrapolating local price multiplier estimates may overstate the impact of large quantity shifts on prices.

This paper provides a guide for using causal inference with asset prices and quantities. Our framework revolves around an elementary assumption about portfolio demand: ho-mogeneous substitution conditional on observables. Under this assumption, standard cross-sectional instrumental variables or difference-in-difference regressions identify the relative demand elasticity between assets with the same observables, the difference be- tween own-price and cross-price elasticity. In contrast, identifying aggregate elasticities and substitution along specific characteristics requires joint estimation using multiple sources of exogenous time-seriesvariation. The same principles apply to the estimation of multipliers measuring the price impact of supply or demand shocks. Our assumption maps to familiar restrictions on covariance matrices in classical asset pricing models, encompass demand models such as logit, and accommodate rich substitution patterns even outside of these models. We discuss how to design experiments satisfying this condition and offer diagnostics to validate it.

We develop a unified framework to study the term structure of risk premia of nontrad-able factors. Our method delivers level and time variation of risk premia, uncovers their propagation mechanism, and is robust to misspecification and weak identification. Most macroeconomic factors are weakly identified at the quarterly frequency, but have increasing (unconditional) term structures with large risk premia at business cycle horizons. More-over, macro risk premia are strongly time-varying and countercyclical. Our framework also recovers the term structure of forward equity yields. We show that it is strongly countercyclical and closely matches the observed values implied by dividend strip data.

This paper examines the term structure of interest rates and the key drivers of the risk premium, identifying the main factors that influence the premium in-vestors require for bearing economic risks within a micro-founded New Keynesian model featuring Calvo pricing. We address two central questions: What are the key factors shaping the term premium and the inflation risk premium, and how can fun-damental economic shocks be classified as ’good’ (reducing the premium) or ’bad’ (increasing it)? We present a novel global solution that improves our understanding of how investors price these risks in Treasury markets. By solving the representative investor’s fundamental pricing equation and applying the Feynman-Kac formula, we compute moments of key macroeconomic variables for direct comparison with em-pirical data. Additionally, we connect our findings to affine term structure models and provide identifying restrictions for economic shocks in empirical work.

Expectations of returns and cash flows are linked to firms’ labor search decisions. Using a dataset that covers the near-universe of online job vacancy postings, we show that vacancy rates negatively predict returns and positively predict cash flows in the crosssection of firms and industries. The predictive power of vacancy postings is stronger for firms facing less fa- vorable labor market conditions. By incorporating the supply and demand information of the aggregate labor market, we construct a new measure of employment value that strongly pre- dicts aggregate stock and bond market returns, even in the presence of other known predictors. A general equilibrium, production-based asset pricing model with heterogeneous firms that are subject to firm-specific labor market conditions accounts for our empirical findings.

I present a consumption-based asset pricing model in which the representative agent se-lectively recalls past fundamentals that resemble current fundamentals and updates be-liefs as if the recalled observations are all that occurred. This similarity-weighted selec-tive memory jointly explains important facts about belief formation, survey data, and realized asset prices. Subjective expectations overreact and are procyclical, the subjec-tive volatility is countercyclical, and the subjective risk premium has a low volatility. In contrast, realized returns are predictably countercyclical, highly volatile, and unre-lated to variation of objective risk measures. My results suggest that human memory can simultaneously account for individual-level data and aggregate asset pricing facts.

I analyze asset tax reforms in an overlapping generations economy with prefer-ence heterogeneity and find that asset taxes have a limited impact on consumption and asset heterogeneity, even when the tax revenues are redistributed to the less wealthy. Higher asset taxes raise both the risk-free rate and the equity premium, largely offsetting their redistributive benefits. The equity premium increases not only because taxing risky assets reduces the wealthy’s willingness to bear risk but also because these assets are reallocated to less wealthy individuals with a lower risk tolerance. Calibrating the model to the US economy, I show that a 10 percent tax on risky assets increases the equity premium from 5.2 percent to 6.2 percent, reduces con-sumption Gini only from 0.33 to 0.31, and asset Gini from 0.55 to 0.54. The reduction in inequality is significantly larger if asset price changes are not taken into account.

Heterogeneous agent models in continuous time (HACT) have become a workhorse in macroeconomics, but incorporating aggregate risk remains a ma-jor computational challenge. Existing methods often rely on local approxima-tions or restrict models to quasi-stationary settings, limiting their ability to capture nonlinear macroeconomic dynamics. This paper introduces a novel approach that globally solves HACT models with aggregate risk by leverag- ing the master equation. Our approach transforms this non-standard partial differential equation (PDE) into a high-dimensional yet standard PDE using Polynomial Chaos expansions, enabling the application of advanced but off- the-shelf deep learning techniques from the applied mathematics literature. Specifically, we adapt the Deep BSDE method to solve this equation efficiently. To demonstrate its applicability, we solve a two-asset HACT model with aggre-gate risk featuring adjustment costs and collateral constraints. By overcoming a key computational barrier, our method enables the study of fully nonlinear heterogeneous agent models, opening new avenues for macroeconomic research.

We examine how ambiguity aversion shapes real estate market responses to long-run sea-level rise (SLR) risk. We link theory and empirics to quantify ambiguity aversion, and explore how this parameter impacts real estate markets and raises homeowners’ willingness to invest in climate change adaptation. Using a novel dataset of projected inundation times for over two million coastal homes, we show that housing prices reflect both expected SLR risk and uncertainty across climate scenarios. We estimate an ambiguity aversion parameter, and find that this shifts probability weights toward worst-case scenarios, substantially increasing the weight on the most extreme SLR projections. Our paper provides the first market-based estimates for ambiguity aversion parameters in the field. We further use our model and estimates to provide new estimates for very long-run discount rates.

We uncover financial professionals’ mental models---the reasoning they use to explain their quantitative forecasts. We organize our analysis around a framework of top-down and bottom-up attention, where analysts endogenously choose both a valuation method and how to allocate attention across variables. Using the near-universe of 2.1 million equity analyst reports, we collect the valuation methods analysts adopt to compute their price targets. To measure attention, we then prompt large language models (LLMs) on a subset of over 300,000 reports to extract 11.8 million lines of reasoning---each combining a topic, valuation channel, time horizon, and sentiment. To validate the reliability of our output, we introduce a multi-step LLM prompting strategy and new diagnostic tools. We document five main findings. (1) Analysts exhibit sparse and rigid mental representations, which they adjust following large forecast errors. (2) The choice of valuation methods and topic focus is closely linked. (3) Attention allocation across variables plays a bigger role than valuation methods in explaining both changes in valuations over time and disagreement across analysts. (4) Biases in analysts' forecasts are driven by over-reaction to firm-specific features and under-reaction to macro-related ones. (5) These biases translate into asset prices: topics analysts overreact (underreact) to predict lower (higher) realized returns.