How (Not) to Identify Demand Elasticities in Dynamic Asset Markets
Jules H. van Binsbergen, Benjamin David, and Christian C. Opp
How much do investors change their portfolio holdings when asset prices change? This question is central to understanding how financial markets work and how central bank policies such as quantitative easing (QE) affect prices. If the price of a bond or stock exogenously rises by 1%, do investors meaningfully reduce how much they want to hold, or do they barely respond? Economists summarize this behavior using a single number called the demand elasticity: how sensitive investor holdings are to price changes, holding the asset’s underlying characteristics fixed.
That number matters because it effectively determines how much buying or selling is required to move prices. If demand is highly elastic, investors quickly sell into price increases, so large purchases by an institution like a central bank are needed to push prices up. In contrast, if demand is inelastic, even small trades can have big price effects.
A widely cited recent literature has concluded that institutional investors are almost perfectly inelastic, a result that has greatly puzzled researchers. Using data on portfolio holdings and price changes, these studies find that even large price movements are accompanied by very small shifts in investor positions. If correct, this would imply that asset markets are extraordinarily easy to move and that financial institutions are held back in their responses by large frictions.
Ourpaper shows that this conclusion rests on a methodology, borrowed from the industrial organization (IO) literature, that fails to account for how financial assets differ from ordinary goods. Unlike most consumer products, a stock or bond can be easily re-traded by its owner, which means its future resale value is a central part of what investors are buying. Raising today’s price increases the cost of the asset, but raising tomorrow’s expected price simultaneously increases the benefit of owning it, since the investor can sell at that higher price. Yet the methods used in the existing literature effectively treat these assets as if resale were not an option, and as if the only relevant change affecting today’s demand were the increase in today’s price.
What is more, our paper shows that the price movements used to measure demand in this literature do more than shift today’s price. They tend to persist or even predict further price increases. Yet when a price rise also signals higher future returns, rational investors will want to buy more, even though the asset has become more expensive. Observing little selling (or even buying) in such episodes does not indicate inelastic or friction-ridden demand; it simply reflects that the asset has also become more attractive due to its higher resale value. In contrast, the existing literature has interpreted this behavior as evidence of substantial behavioral and institutional frictions.
Our paper further characterizes how important intervention dynamics are for the effectiveness of central bank policies like QE. The longer a central bank credibly commits to supporting prices, the less it needs to buy today. For instance, extending the expected duration of a 1% price support from one week to one year reduces the required initial purchases by almost 98 percent.
In sum, our paper’s broader message is that analyzing demand in asset markets requires methods that properly account for investors’ dynamic trading opportunities. What has been interpreted as inelasticity is, to a large extent, a consequence of an incorrect static view of investor demand. When price changes also alter future resale values, portfolio behavior reflects shifting investment opportunities rather than a lack of responsiveness to prices.