Abstract: This paper derives fundamental limits to identifying asset demand from
observational data. We establish a trilemma: it is impossible to maintain that
(i) prices satisfy no arbitrage, (ii) investors value assets for their payoffs, and
(iii) asset demand curves are invariant to exogenous asset supply shocks. That
is, one cannot use supply shocks to move along an asset demand curve with-
out shifting it. The only exception is the knife-edge case in which the asset
menu consists of Arrow securities. In realistic settings, demand elasticities
thus necessarily reflect theoretical assumptions rather than the data alone.
Abstract: We document that investors lack conviction as to what the stock market price should be, absent seeing it. This is at odds with most asset pricing models, which assume that investors are uncertain about the future, but understand what the current market price should be and whether the actual price is different from this. Through experiments, we find that online participants, MBA students, and asset-management professionals who are provided with information about fundamentals are unable to identify extreme deviations from market prices. When directly asked, both asset-management professionals and LLM reasoning models do not think it is possible to use information to reliably estimate the level of the market within 10\% without seeing the prevailing price. We document that professional return forecasts, published trading strategies, and investment advice are largely agnostic about the market price level. This suggests that return expectations are often formed, and many investment decisions are made, without conviction as to the prevailing market price level. We call this channel price agnostic demand. This represents a significant limit to arbitrage which can help explain a number of puzzling empirical patterns in asset prices such as why prices exhibit excess volatility and why markets appear inelastic.
Discussant: Andrea Tamoni, University of Notre Dame
Abstract: We model a uniform-price auction for safe assets in which dealers trade in both primary and secondary markets, whereas long-term investors hold to maturity. Risk-averse dealers demand a risk premium for uncertainty about post-auction gains. Higher demand risk and heterogeneity in holding costs lower dealers’ demand elasticity relative to long-term investors. Dealers’ inelastic demand at the auction predicts higher post-auction returns, but auction cycles are longer when long-term investors are also inelastic. Unique data on Swiss Treasury auctions with bidder identities validate our model’s predictions on how investor composition shapes safe asset valuations and post-auction price dynamics.
Discussant: Jason Allen, University of Wisconsin-Madison
