Empirical evidence on time-varying risk attitudes

My thesis focuses on the risk-taking behavior of financial agents, aiming particularlyat better understanding how risk attitudes can change over time. It alsoexplores the implications that these changes have on financial markets, and on theeconomy as a whole.The first paper, which is a joint work with Kim Oosterlinck and Andrey Ukhov,studies how risk aversion of financial markets’ participants is affected by the SecondWorld War. The literature links extreme events to changes in risk aversion but failsto find a consensus on the direction of this change. Moreover, due to data limitationsand difficulties in estimation of risk aversion, the speed of the change in risk aversionhas seldom been analyzed. This paper develops an original methodology to overcomethe latter limitation. To estimate changes in attitude toward risk, we rely on thedaily market prices of lottery bonds issued by Belgium. We provide evidence on thedynamic of risk attitude before, during and after the Second World War. We findsubstantial variations between 1938 and 1946. Risk aversion increased at the outbreakof the war, decreased dramatically during the occupation to increase again afterthe war. To our knowledge, this finding of reversal in risk attitude is unique in theliterature. We discuss several potential explanations to this pattern, namely changesin economic perspectives, mood, prospect theory, and background risk. While theymight all have played a role, we argue that habituation to background risk mostconsistently explains the observed behavior over the whole period. Living continuouslyexposed to war-related risks has gradually changed the risk-taking behavior ofinvestors.In the second paper, I derive a measure of risk aversion from asset prices andanalyze what are its main drivers. Given the complexity of eliciting risk aversionfrom asset prices, few papers provide empirical evidence on the dynamics of riskaversion in a long-term perspective. This paper tries to fill the gap. First, I providea measure of risk aversion that is original, both because of the length of its sampleperiod (1958- 1991) and the methodology used. I study the relationship betweenthis new measure of risk aversion and several key economic variables in a structuralvector autoregression. Results show that risk aversion varies over the period. Aworsening of economic conditions, a decrease in stock prices or a tighter monetarypolicy lead to an increase in risk aversion. On the other hand, an increase in riskaversion is linked to a larger corporate bond credit spread and has an adverse effecton stock prices.The third paper explores the impact of asset price bubbles on the riskiness offinancial institutions. I investigate the effect of a real estate boom on the financialstability of commercial banks in the United States using exogenous variations intheir exposure to real estate prices. I find that the direction of the effect dependson bank characteristics. Although higher real estate prices have a positive impacton bank stability on average, small banks and banks that operate in competitivebanking markets experience a negative effect. I reconcile these findings by providingevidence that higher real estate prices benefit commercial banks by raising the valueof collateral pledged by borrowers but at the cost of an increase in local bankingcompetition. This increase in competition affects banks that have a low marketpower more severely, which explains why small banks and banks facing a high degreeof competition display relatively lower stability during a real estate boom.