THE EFFECTS OF TAXES AND RATES OF RETURN ON FOREIGN DIRECT INVESTMENT IN THE UNITED STATES: SOME ECONOMETRIC COMMENTS
| Autor: | N. R. Vasudeva Murthy |
|---|---|
| Rok vydání: | 1989 |
| Předmět: | |
| Zdroj: | National Tax Journal. 42:205-207 |
| ISSN: | 1944-7477 0028-0283 |
| DOI: | 10.1086/ntj41788789 |
| Popis: | IN (1988) a recent has article presented of this some Journal econometric , Young (1988) has presented some econometric evidence on the effects of domestic taxes and rates of return on foreign investment in the United States for the period 195384, using a modified version of the Hartman model (1984). 1 Young regresses foreign direct investment in the United States (I) on the after-tax rate of return actually realized by the foreign investors in the United States (r(l t)), overall after-tax rate of return on U.S. capital applicable to foreigners (r'(l t)), net-of-tax rate of return received by domestic investors relative to that received by foreigners on the same investment in the U.S., ((1 t')/(l t)) and on U.S. GNP. In some of Young's regression equations, there is the possible presence of autocorrelation between the residuals in the time-series data used. In this note, an attempt is made to adjust for the presence of autocorrelation using an alternative estimator and the resulting findings are contrasted with Young's conclusions. Young's results are presented in Table 1. Standard errors are presented in parentheses. Strictly speaking, he has not tested for the presence of autocorrelation in his equation. If one of the regressors is the lagged value of the dependent variable, then the application of Durbin's h-test or m-test is appropriate to detect autocorrelation [See Durbin (1970).] It is evident from Table 1, that autocorrelation is present in equations II and III. For equation V, using the m-test the null-hypothesis of no autocorrelation cannot be rejected. However, applying the Lagrange multiplier test [see Engle (1984)], we can reject the null-hypothesis of no autocorrelation at the 25 percent level. The usual Cochrane-Orcutt procedure is not employed here to correct for auto*Creighton University, Omaha, Nebraska 68178. correlation because it has been found that in equations with lagged dependent variables, if the Cochrane-Orcutt procedure is used, each of the estimated parameters of the explanatory variables will be inconsistent and under certain circumstances they may even be biased. [See Aschheim and Tavlas (1988).] As a remedy, the maximum likelihood procedure (ML) is recommended. The maximum likelihood estimation (ML) has been used to correct for autocorrelation in equations II, III and for comparison ML estimates of equation V and VI and the results are reported in Table 2. 3 Since the maximum likelihood estimators are more efficient than the ordinary least-squares estimators, most of the ML standard errors are smaller than |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|