Skip to main content

Asset Prices and Macro Policy when Agents Learn

Final Report Summary - APMPAL (Asset Prices and Macro Policy when Agents Learn)

To better understand asset prices and related policy issues we provide an alternative to the standard modelling approach of consumers and investors’ expectations.

It is well understood that the price of a given asset must depend on the fundamental variables that influence the asset payoffs. For example, a stock price should depend on the current and expected dividend. A standard premise is that stock prices reflect an objective measure of the discounted value of current and future dividends. This gives rise to the so- called efficient market hypothesis. There is little room for policy interventions in financial markets under this hypothesis, as these interventions would only disrupt prices and markets.

It is generally thought that efficient prices are an outcome of rational behavior. Related to this is the concept of rational expectations, where investors are assumed to know what is the correct price for each level of dividend. Under this view, since financial market participants in the real world are indeed very sophisticated, stock prices are likely to be efficient and policy makers should abstain from intervening in the market.

But stock prices in practice are extremely volatile. It has been shown that this variability is very difficult to explain under the efficient market hypothesis. We have shown in various ways that the observed stock price volatility can be explained under an alternative hypothesis, namely, that investors can not perfectly understand how dividends are related to prices, but they have a very good model of how prices are formed, and their behavior is fully optimal given their knowledge. We call this concept “internal rationality”. This allows us to formalize an intuitive concept: if investors expect a stock price to increase this drives the current stock price up, so that further increases in actual and expected stock prices will follow, much in the way that stock price bubbles and busts seem to occur in the real world. A similar story explains housing bubbles and busts.

We find ample empirical evidence that supports our hypothesis. The volatility of stock prices becomes very easy to explain even with extremely simple models: stock prices display large swings just because agents expect these fluctuations to occur. The observed dynamics of survey expectations about stock returns is inconsistent with efficient markets but it agrees with our hypothesis.

In this setup stock prices are not efficient even if investors are very sophisticated, they always take correct decisions given their information, and they have a very good understanding of the way markets work. This gives a different angle on asset price formation and it leaves much more room for policy intervention. Interest rate changes should take stock price fluctuations into account, since low interest rates can fuel stock or housing bubbles. To the extent that stock price fluctuations are related with investment fluctuations, intervening in the stock market can improve resource allocations. We find, however, that a financial transaction tax will have undesirable effects, as it will also lower stock prices even further in periods of stock busts and, therefore, it will increase total volatility. Policy markes should monitor return expectations, as this detects asset price movements that are unjustified by fundamentals.

Our project has also studied issues of government portfolio choice, redistribution of wealth through taxation, inflation, exiting the euro, risk-sharing, bank opacity and how these issues are affected by learning and partial information by the government.