The market reaction to the forgiveness of deferred taxes due to the repeal of the domestic international sales corporation: expectations regarding subsequent export related tax incentives

International Journal of Business Research, Jan, 2008 by Sid Howard Credle, Sharad Maheshwari, Jacob Angima

Security Return Model

The main model used to generate returns in this study conditions the return generating process on the occurrence of an event using Conditioned Ordinary Least Square (COLS). The model is described as follows:

[R.sub.it] = [alpha] [[beta].sub.i][R.sub.mt] 3/[summation]/d = 1 [[psi].sub.idt] [[delta].sub.dt] [[epsilon].sub.it] (1)

Where, [R.sub.ir] is the known return on security i in period t, with t=1,2, ... T. [R.sub.mt] is the CRSP value-weighted return on the market index on day t, and [delta] is a two-day dummy variable corresponding to each calendar event and the day before, which is 0 if no announcement is made and 1 if an announcement is made for the first 30 events, and a 0 for two days before event 31, and 1 for the three day period centered on the earnings announcement date where d=event periods 1,2,3. The notation [[alpha].sub.i] is the intercept for firm i. The [[beta].sub.i] variable represents a stationary market coefficient with cov([R.sub.i][R.sub.m])/var([R.sub.m]), [[psi].sub.idt] is the event coefficient summed over events d, or all events D, and [e.sub.it] = N (0,[sigma]) error term.

4. RESULTS

H(1) is tested using a STUDENT's t test conducted for the providing, non-providing, and full sample. Tests of H(2) compare the means of the providing and non-providing firms using the sum of event coefficients over the stated time period.

Time Period Analysis

Analysis was conducted on the sum of event coefficient during the study time periods. The results of tests of H(1) are noted in Table 3.

The table indicates that all events with the exception of the disclosure of earnings resulted in significant announcement effects. The noted full sample effects were -2.80, 1.90, .892, and -1.83 for the Tax deliberations period, the FASB deliberations period, and the disclosure period, and over all events, respectively. These results suggest that the change from DISC to FSC was costly for all exporters. It also appears that the change was less of a detriment for firms that provided deferred taxes. Non-providing firms experienced negative returns during tax deliberations, calendar, and throughout all event periods. The t values for these periods were -2.63, -2.25, and -2.31, respectively, all significant at one percent. During the calendar events and over the study period, H(1) is rejected. The t test comparison of H(2) was not significant. On average sample DISC firm returns moved in the same direction over the study period, therefore, H(2) is not rejected.

Cross-Sectional Dependency

Since sample firms share common industries, export-related activity and events, it is reasonable to assume that cross-sectional correlation in the disturbance terms may exist using the COLS model. Positive cross-sectional correlation in the data would result in reduced standard errors and a bias towards rejection of the null. To test for the sensitivity of our previous results, coefficients were re-estimated using a Generalized Linear System (GLS), which assumes a common time series/cross-sectional random effects model with a covariance structure.