Abstract
This paper attempts to examine the role of technological efforts and firm size in determining the export behaviour of firms belonging to the basic chemical industry in India. The basic chemical industry is an important industry that provides intermediate chemicals to firms operating in diverse industries in both India and abroad. In this study technological efforts have been considered in terms of in-house R&D, import of embodied technology, and import of disembodied technology. Three different econometric models, namely, the Tobit, the two-part (Probit + Truncation) and the sample selection (Heckman), have been used for estimation and the results have been compared. The authors find evidence in support of the view that export behaviour of the firm can be modelled in a more appropriate manner using a two-part or a sample selection model rather than the popular Tobit model. The results of the econometric exercise confirm that technological efforts, firm size and other firm-specific characteristics are important in explaining the export behaviour of the firms.
The authors gratefully acknowledge the comments and suggestions given by Prof. K. L. Krishna and the anonymous referees on the earlier drafts of the paper. The errors that remain are their own.
Notes
1 Organic chemicals (also called hydrocarbons) are formed with the building blocks of carbon and hydrogen molecular structures and are petroleum-based, whereas inorganic chemicals are essentially salts of metals and non-metals derived from minerals and ores or occur naturally.
2 The websites for NSE and BSE are http://nseindia.com/ and http://www.bseindia.com/, respectively. One can get more information on the securities markets in India from ISMR (Citation2007).
3 We have access only to the STATA (Version 10) statistical package. Hence, we can use the Probit, Tobit, Truncation and sample selection (Heckman) models but not the Lee & Maddala (Citation1985) model. At the same time, one of our aims is to demonstrate how the two-part model or the sample selection model is more appropriate to study the export behaviour of firms than the Tobit model (which is widely used in India). STATA (Version 10) is sufficient for this purpose.
5 The data on manufacturers of fertilizers and nitrogen compounds (NIC-98 code 2412 at the four-digit level) have been excluded, as most of the firms under this subcategory are government owned. The initial unbalanced sample consisted of 1842 observations. A sample of 847 observations (which is around 46% of the initial unbalanced sample) remained after dropping the observations that were incomplete on some of the firm characteristics.
6 In STATA 10 a robust option is available for the Probit, Truncation and Tobit models but not for the Heckman two-step sample selection model. However, for the Heckman two-step sample selection model robust standard errors can be obtained using a bootstrapping procedure.
7 See Heckman (Citation1979), Maddala (Citation1983) and Greene (Citation2002) for more details on the inverse Mills ratio (λ) and its significance in the estimation of the sample selection model using the Heckman two-step procedure.