No abstract

No abstract

IN THEIR 1982 STUDY, MCFARLAND, Pettit, and Sung [16] (MPS) found thatconsecutive daily foreign exchange rates conform to neither a normal nor a nonnormal stable process and that day-of-the-week data could be adequately described by a series of different stable distributions.' The authors have also documented a "weekday effect" in the foreign exchange market; Mondays and Wednesdays offered higher average price changes than did Thursdays and Fridays. In a more recent study, Jaffe and Westerfield [9] documented negative Monday returns for the currencies of the U.K., Canada, and Australia.In estimating the parameters of the distribution, MPS have utilized the Fama and Roll [5, 6] method, which assumes a symmetric distribution. Jaffe and Westerfield [9] report only estimates of the sample moments. The sample moments reported in MPS's Table II (part of which is reproduced in Table I here) reveal that most currencies, especially the minor currencies (the Australian dollar, the Spanish peseta, and the Swedish krona), have significant skewness. The results indicate that the asymmetric stable Paretian distribution may be more appropriate for the overall daily returns. This hypothesis is consistent with Calderson-Ressel and Ben-Horim [1] and So [24], both of whom observe that daily spot rates of the Australian dollar, the Japanese yen, the British pound, and the Spanish peseta tend to be asymmetrically distributed. The asymmetric distribution may also be more appropriate for the price changes on individual days of the week if the skewed overall distributions are caused by large price changes that occur on certain days of the week. Some studies reveal that skewness may have important impacts on estimating the location estimate. For example, Rozelle and Fielitz [18] find that the Fama-Roll estimate of the location is sensitive to skewness. Leitch and Paulson [11]also indicate that, when skewness is nonzero, the "fatness" of the tails in the distribution skew either to the right or the left; hence, the location parameter is either to the right or left of the median.2 Moreover, McCulloch [14] and Simkow-* Department of Finance, Southern Illinois University at Edwardsville. The author thanks Neil Myer, Radcliffe Edmonds, and two anonymous reviewers for comments and suggestions. He is also grateful to Richard Nyerges and Neil Myer for editorial assistance. Cary Chao has provided excellent research assistance. This project is supported by the Graduate School, Southern Illinois University at Edwardsville. The author is solely responsible for any errors in this study. l See Mandelbrot [12], Fama [4], and McCulloch [13] for discussion of the stable distribution. 2 Elton, Gruber, and Kleindorfer [3] have demonstrated that the use of log price relatives also implies maximal left skewness.

No abstract

We study the efficacy of forbearance using a real options approach. Our model endogenizes moral hazard embedded in credit risk undertaken by the bank. The bank's interest rate risk is modeled as duration mismatch. Other modeling improvements over previous studies include such features as stochastic interest rates and deposits, continuous interest payments on an ongoing deposit portfolio, and a stochastic forbearance period. We find that the bank does have an incentive to engage in undue risk taking. Even in the presence of moral hazard, however, forbearance can still be a desirable course of action in reducing the FDIC's expected liability. In addition, the capital ratio plays an extremely important role in determining the fair insurance premium. Finally, using the mismatch of asset and deposit durations as the correct measurement of interest rate risk, our model reveals that an optimal asset variance may exist for a particular bank, contrary to what the contingent claims framework would predict. Therefore, we resolve the puzzle that banks in practice do not increase asset risk to take full advantage of the limited liability. Copyright The Journal of Risk and Insurance, 2004.