Recent global financial crisis has raised some critical questions about the current status of the financial system, including management of systemic risk at both macro and micro level, integrity of the financially regulatory regime, and the overall quality and mechanism of financial infrastructures. Among many other salient issues especially for the academia in the analysis of the cause of the crisis, the asymmetry in an incentive compensation system within the financial industry in particular has been singled out as one of the serious issues to be addressed because the asymmetry in incentive compensation only creates incentives for risk-taking while it fails to penalize financial managers when bets turn sour. Raghuram Rajan (2008) refers to this risk-taking incentive as “It's heads I win, tails you lose.” However, the incentive for risk-taking can account for only a necessary condition for a financial crisis, not a sufficient condition. Even risky investments, as long as they are diversified, for example, can hardly make financial markets unstable. In fact, it seems more likely to take the combination of investors' both incentives to take excessive risky investments and herding behavior to ignite a financial crisis. This paper, therefore, theoretically investigates the effect of the asymmetry in incentive compensation for fund managers on their herding behavior in investment, based on the canonical model of Bikhchandani, Hirshleifer and Welch (1992, hereafter abbreviated as BHW). Without a rigorous examination, indeed, it is very difficult to accurately assess the effect as to whether it is positive or negative. For example, if fund managers sufficiently take into account their predecessors' activities in their decisions, they are likely to discount the informational value of those activities and the possibility of information cascade will be, therefore, lower. On the contrary, if the effect of the incentives for individual fund managers to take risky investments is greater than that of the depreciation of the informational value of the predecessors' investment activities, the asymmetry in incentive compensation can precipitate the information cascade and herd behavior in investment. Thus, our theoretical model analyzes the effect of asymmetry in incentive compensation for fund managers on their herding behavior by introducing asymmetric costs related to investment decision. In other words, these costs could be regarded as the opportunity costs in investment decision. In fact, these costs play key roles in our model; if these costs are assumed to be less than half like in our model, fund managers anticipate higher rates of return than that in the benchmark case by making decision of riskier investment than the market portfolio. This means that the opportunity costs in the model are relatively lower than those in the benchmark case. It is, thus, important to note that if these costs are assumed to be just half, as in a benchmark case, they are the opportunity costs when fund managers make any investment decisions that show the same risk with the market portfolio. All other fundamental assumptions are similar to the BHW (1992). We have shown the following theoretical results. First, with asymmetric incentive compensation in which fund managers' gains for success are larger than their losses for failure, the fund managers are more likely to be optimistic compared to the case of symmetric incentive compensation. Second, with asymmetric incentive compensation, wrong up-cascades are more likely to occur and less likely to stop, compared with wrong down-cascades. In our model, a wrong cascade is defined as information cascades occurring in the opposite direction of ex-post investment state. On one hand, a wrong up-cascade is a cascade where ex-post investment state is bad while investment decision has been made by individual fund managers. On the other hand, a wrong down-cascade is a cascade where ex-post investment state is good while investment decision has not been made by individual fund managers. Third, with asymmetric incentive compensation, wrong up-cascades start earlier than the case with symmetric incentive compensation does. These theoretical results imply that fund managers are, in fact, likely to hold too optimistic approach to their investment with asymmetric incentive compensation. Our findings further imply that the asymmetry in the incentive compensation scheme may exacerbate wrong cascades and herd behavior in investment, thus causing the financial system to be more unstable. Therefore, it is necessary to consider this asymmetry carefully in designing in order for the fund managers' incentive compensation scheme to generate desirable effects. Since the recent global financial crisis, regulatory reforms or provisions of best practice of compensation scheme for prudent risk management have been widely discussed at both international and domestic levels. For example, the Financial Stability Board has proposed principles for sound compensation practice and its implementation standards, focusing on governance structure, pay structure and risk alignment, disclosure of compensation scheme, and supervisory issues. The Korean government is also preparing policy measures related to the compensation practice in order to curb risk-taking in financial institutions, which is along the lines with the recommendation of Financial Stability Board. Our findings imply that regulatory authorities should consider more appropriate compensation schemes that reduce this asymmetry such as deferring incentive pay so that it is better linked to long-term performance.