research
Working papers
Capital Gains Taxation and Asset Price Volatility (replaces “Capital Gains Taxation, Learning and Bubbles”)
Reject and Resubmit at the American Economic Review. Resubmitted!
Abstract. Do capital gains tax cuts destabilize or stabilize asset prices? In an asset pricing model with heterogeneous agents and realization-based taxation, a tax cut has two opposing effects. It dampens volatility by reducing realization-based trading frictions, but also amplifies it by strengthening the pass-through from expectations to prices, fueling self-fulfilling fluctuations. Estimated on U.S. stock-market data, the model implies that the sequence of tax cuts since the 1970s triggered a net increase in volatility of about +35%, driven primarily by stronger belief-to-price pass-through. Policy experiments suggest a tax on unrealized gains robustly reduces volatility, whereas a financial transaction tax has mixed effects.
Abstract. We study the macroeconomic implications of a debt-financed sovereign wealth fund (SWF) that invests in the domestic stock market. In a stochastic general equilibrium model with distortinary taxation, we show that the SWF can improve fiscal sustainability despite initially increasing public debt. The SWF generates efficiency gains by reducing distortionary taxation, which stimulates economic growth, but at the cost of amplifying macroeconomic volatility. Paradoxically, this increased volatility benefits the government by deteriorating the hedging properties of stocks, which raises the equity premium that the SWF captures, while simultaneously triggering a flight-to-safety effect that lowers government borrowing costs. These combined effects—higher economic growth, increased stock returns, and reduced interest expenses—enhance fiscal sustainability. The optimal size of the SWF results from balancing efficiency gains against increased risk, with maximum size being optimal under constant marginal efficiency gains of lower taxes. Certain versions of Central Banks’ asset purchase programs can be interpreted as a de facto SWF.
Learning to Be Rich: How Expectations amplify Wealth Inequality (with Adrian Ifrim and Janko Heineken) (replaces “Heterogeneous Expectations and Wealth Inequality”)
[New Draft here!]
Abstract. We document systematic heterogeneity in stock-market beliefs across the wealth distribution: low-wealth households are persistently too pessimistic, while high-wealth households’ expectations are approximately unbiased. We develop a heterogeneous-agent model in which belief dispersion arises endogenously through learning from experience, generating a feedback loop: good past returns foster optimism, induce higher equity shares, and lead to higher future returns that further reinforce optimistic beliefs. The calibrated model matches key features of the joint distribution of expectations, portfolio returns, and wealth. Heterogeneous beliefs increase the top 1% wealth share by 50% relative to homogeneous expectations. Methodologically, we show that Internal Rationality—where households learn directly about the law of motion for prices rather than forecasting entire distributions—makes heterogeneous-agent models with aggregate risk both more realistic and tractable.
Abstract. Simultaneously accounting for key features of bond markets —high excess return volatility, flat yield volatility term structures, and Expectations Hypothesis violations— remain challenging for equilibrium models. Additionally, we document that survey expectations systematically overestimate how current yield slopes predict future excess returns, rejecting the standard assumption of full information rational expectations. We show that these observations arise naturally from a parsimonious model where investors are Internally Rational—they optimize given their understanding of bond pricing but must learn the true pricing function. When learning about how slopes affect future prices, their beliefs become self-reinforcing: high expected slopes drive up long bond prices, validating initial beliefs and creating persistent price inertia. This single mechanism explains both the stylized facts and survey evidence. Methodologically, we extend Internal Rationality to multiple asset, which requires to specify the joint perceived distribution of all asset prices.