ABSTRACT
Following Maddala (1987), the study employs random-effect panel data regression model with Tobit specification for analysing the determinants of TE. Use of Tobit model is necessitated by the fact that the TE estimates lies in the limited range of zero and one. The use of panel data improves the efficiency of econometric estimates on account of larger number of observation compared to the individual data set of cross-section or time series. Besides improving the efficiency, the application of panel data model in this study shall enable us to control for time invariant firm-specific heterogeneity in TE arising from the unobserved firm-specific characteristics. The empirical form of the Tobit model is symbolically represented by the following equation:
TE jt = b0 + b1 FCDjt + b2 FSjt + b3 IMDTjt + b4 CAPIjt + b5 IMIGjt + b6 XDjt + b7 RDIjt + b8 AMIjt + b9 SZjt + b10 AGEjt + b11 NWIjt + b12 IMCjt + b13 YD02+ · · · + b18 YD07+ uj + vjt,
j = 1, . . . . , 178 and t = 1, . . . , 7; TEjt = TE∗jt if TE∗jt . 0, TEjt = 0 if TE∗jt ≤ 0
(6)
where the term uj are unobserved stochastic heterogeneity varying across groups but not over time and distributed as ujI˜N (0, s2u). The error term vjt vary across groups and over times and vitI˜N (0, s2v ). The term uj are assumed to be uncorrelated with vjt and the explanatory variables in Equation (6). The model represented by Equation (6) is estimated by ML estimation technique.
6. Results and discussions
