اعتماد بیش از حد و مدیریت پرتفولیو واگذار شده

We study the impact of overconfidence on investment decisions by financial institutions. These institutions are characterized by the delegation of investment decisions to portfolio managers and the design of contracts that aim at aligning managers’ incentives with those of the institution. We show that when rational and overconfident agents acquire information of the same precision, overconfident agents trade lower quantities than rational agents. However, overconfidence also generates incentives to overinvest in information acquisition. In such cases, overconfident agents trade larger quantities and take more risk than rational agents. The direct consequence of these results is that, as far as delegated portfolio management is concerned, overconfidence generates high trading volumes only through over-acquisition of information. Based on psychological evidence that overconfidence is generated by a self-attribution bias, our results are consistent with recent empirical evidence about mutual fund managers’ portfolio-rebalancing patterns and changes in mutual funds’ advisory contracts.

It is widely acknowledged that some financial investment decisions are difficult to reconcile with fully rational behavior. Accordingly, the analysis of financial markets in the presence of irrational agents has received increasing attention over the last fifteen years. One form of irrational behavior studied extensively is overconfidence (see Kyle and Wang, 1997, Benos, 1998, Odean, 1998, Wang, 1998, Wang, 2001, Daniel et al., 1998, Daniel et al., 2001 and Caballé and Sákovics, 2003). A common feature of these models is that market participants are individual investors maximizing their expected utility of wealth given their beliefs. Furthermore, the common result is that overconfident investors hold riskier position/trade larger quantities than rational investors. However, a large fraction of company stocks are held by financial institutions. As highlighted by Lewellen (2008), institutions held 68% of the US equity market at the end of 2007. Furthermore, (i) their investment strategies differ from those of individual investors (see Cohen, 1999, Grinblatt and Keloharju, 2000, Cohen et al., 2002 and Ekholm and Pasternack, 2008); and (ii) recent empirical and experimental studies provide evidence of overconfidence on the side of professional investors (see, for example, Gort, 2007, Pütz and Ruenzi, 2008 and Choi and Lou, 2008).1 As a consequence, in order to analyze the impact of overconfidence on financial asset prices, one needs first to study how overconfidence influences the investment strategies of institutional investors. A fundamental characteristic of financial institutions is the delegation of investment decisions to professional money managers. As a consequence of this delegation, agency conflicts may arise. Therefore, the institution will design a compensation contract aimed at mitigating conflicts of interest. Faced with overconfident managers, the institution must also deal with the biases overconfidence may generate.2 Hence, the institution will design a compensation contract that deals with both agency problems and biases associated with overconfidence. In the end, the investment strategy of the manager will depend on his compensation contract. Therefore, in order to analyze how overconfidence influences institutional investments, one must endogenize compensation contracts and study investment strategies resulting from these endogenous contracts. To address this issue, we analyze a delegated portfolio management problem in which a risk-averse financial institution (the principal) hires a risk-averse money manager (the agent) who may be of two types: rational or overconfident. By exerting effort, the agent acquires private information about the value of a risky asset. If the agent is rational, he updates his beliefs in a Bayesian fashion. However, if overconfident, the agent over-estimates the precision of his private signal. Based on his updated beliefs, the agent then makes an investment decision. In this situation, the moral hazard problem faced by the principal has two aspects, overconfidence having an impact on each of them. First, the principal must provide the agent with incentives to exert effort and acquire information. Second, if the principal and the agent have different levels of risk aversion (as we will assume), the principal must calibrate the risk-taking incentives the contract provides to the agent. In order to disentangle the impact of overconfidence on each aspect of the agency problems, we consider two different cases regarding the acquisition of information. First, we assume that effort is a binary choice. It implies that if rational and overconfident agents acquire information, it is of the same precision. As a consequence, the effect of overconfidence on the acquisition of information is neutralized, and only the effect of overconfidence on the portfolio allocation is at work. In this environment, we show that overconfident agents trade lower quantities than rational agents do. A direct implication of this result is that those obtained in the case of private investors (i.e., excessive trading) may not extend to the case of delegated portfolio management. When considering institutional investment, overconfidence may not generate excessive return volatility if the appropriate compensation contract is used. The second case we consider is such that effort is a continuous variable. We show that overconfident agents acquire more precise information and trade larger quantities than rational agents do. The riskiness of investment strategies is then influenced in two ways. First, it is decreased through a lower risk level per unit of investment, and second it is increased through larger risky-asset holdings. We show that the latter effect always dominates the former implying that overconfidence always increases the level of risk undertaken. One could find this result very intuitive since more precise information implies higher expected returns and lower risk per unit traded, hence providing room to trade larger quantities. However, this is not the full picture. The reason is that the return taken into account by the principal when designing a compensation contract is the portfolio return net of compensation expenses, and the cost of providing incentive to exert effort is different for overconfident and rational agents. We show that the principal provides more incentives to exert effort to overconfident agents. As a consequence, for a given portfolio-return volatility, the net-return volatility is smaller if the agent is overconfident than if the agent is rational. It then follows that the principal is willing to provide more trading incentives to an overconfident agent. Hence, comparing the results obtained in the binary and the continuous effort cases, we show that in the case of delegation, overconfidence generates higher trading volume only through over-acquisition of information. Based on psychological evidence that overconfidence is generated by a self-attribution bias, i.e., agents credit themselves for past good performance but blame others for failure, (see, for example, Gervais and Odean, 2001), our results are consistent with several pieces of empirical evidence. First, our results are consistent with those of Pütz and Ruenzi, 2008 and Choi and Lou, 2008 who show that fund managers trade more after a good performance. Second, in our model institutions acquire more information than individual investors. As a consequence, they realize better performances than individual investors, implying that institutions are the market participants who suffer from the stronger confidence reinforcement. Using this result in the market model of Daniel et al. (1998, Section III), we deduce that institutions are the market participants who follow momentum strategies. This is consistent with the findings of Grinblatt and Keloharju (2000) who provide evidence that the most sophisticated agents, foreign funds in their sample and professional money managers in our model, are those following momentum strategies. Third, our results suggest that, as a consequence of overconfidence reinforcement, the incentive component of the compensation contract increases. Therefore, we expect an increase in mutual fund management fees following good performances and no variation in case of bad performances. This is consistent with the findings of Warner and Wu (2005) on the changes in mutual fund advisory contracts. The organization of the paper is as follows. Section 2 reviews the related literature. Section 3 presents the model. Sections 4 and 5 study optimal contracts and investment strategies in the binary and continuous effort cases, respectively. Finally, Section 6 discusses the results and their consistency with empirical evidence, while Section 7 presents our conclusion. All the proofs are contained in the Appendix.

The impact of overconfidence on investment strategies and asset prices has received a lot of attention in the last decade. However, until recently, attention has concentrated on individual investors’ behavior, while one of the main categories of financial market participants are institutions, which are characterized by the delegation of investment decisions to professional money managers. This delegation of tasks requires the design of a compensation contract which influences managers’ investment decisions. In this article, we have shown that results on overconfidence obtained in the case of private investors (i.e., excessive trading volume and risk-taking incentives) may not hold in the case of institutional trading. The reason is that institutions can propose contracts that mitigate risk-taking incentives generated by overconfidence. More precisely, we show that in equilibria in which rational and overconfident agents acquire information of the same precision, overconfident agents trade lower quantities than rational agents. However, if we consider the precision of the information to be acquired as a continuous variable, we have shown that overconfident agents are proposed compensation contracts with a larger incentive component. As a consequence, acquire more precise information, trade larger quantities, and take more risk than rational agents do. Based on psychological evidence that overconfidence is generated by a self-attribution bias, our results are then consistent with recent empirical evidence about mutual fund managers’ investment strategies, namely an increase in portfolio rebalancing after good performances and the use of momentum strategies. Our results are also consistent with evidence that good past performances have a positive impact on the likelihood of increasing management fees.