Direction of Outward FDI of Indian Manufacturing Firms: Influence of Technology and Firm Productivity

Abstract

This chapter deals with India’s outward foreign investment (OFDI) flows, investigating econometrically whether the direction of OFDI of Indian manufacturing firms is related to their technical competence and level of productivity. The theoretical model of Aw and Lee (J Int Econ 76:403–415, 2008) is taken as the basic framework for the analysis. Data for about 2400 Indian manufacturing firms are used for the analysis. The data relate to the year 2007–08 or thereabout. The econometric results indicate that a firm with a relatively high level of productivity is more likely to invest abroad than a firm with relatively low productivity. However, the type of relationship between firm productivity and direction of FDI that is expected on the basis of the Aw-Lee model and their empirical findings for Taiwanese electronics firms is not found in the analysis of data on Indian manufacturing firms. The econometric results do not show that the Indian firms that invest in industrialized countries have significantly higher productivity than the firms that invest in developing countries, which is a prediction of the Aw-Lee model. The results for the technology related variables, on the other hand, do provide some support to the Aw-Lee model. There are indications from the econometric results that a relatively greater engagement with technology acquisition activities among Indian firms is associated with investment in industrialized countries. One interpretation of this empirical finding is that the technical competence of a firm is an important factor determining whether it will invest in an industrialized country.

Appendix

4.1.1 Multinomial Logit Model

Consider a situation where a firm decides among J + 1 choices. The choices available to the firm are 0, 1,..J. The base choice is zero. The choice made by the firm (which is denoted by Y) depends on a set of explanatory variables, denoted by x. Under the multinomial logit model, the probability of making choice j is specified as:

$$ {\text{Prob}}(Y_{i} = j) = \frac{{e^{{\beta_{j}^{'} x_{i} }} }}{{\sum\nolimits_{k = 0}^{J} {e^{{\beta_{k}^{'} x_{i} }} } }},\quad j = 0,\;1,..J $$

(4.1)

Note that there is one parameter vector associated with each choice. There is indeterminacy in the model above because if a vector q is added to each of the β vectors, then an identical set of probabilities emerge. To solve this problem, the convenient normalization done is to take the parameters for choice 0 as zero, i.e. β₀ = 0. With this normalization, the probabilities of the choices are obtained as:

$$ {\text{Prob}}(Y_{i} = j) = \frac{{e^{{\beta_{j}^{'} x_{i} }} }}{{1 + \sum\nolimits_{k = 1}^{J} {e^{{\beta_{k}^{'} x_{i} }} } }},\quad j = 1,..J\quad ) $$

(4.2)

$$ {\text{Prob}}(Y_{i} = 0) = \frac{1}{{1 + \sum\nolimits_{k = 1}^{J} {e^{{\beta_{k}^{'} x_{i} }} } }} $$

(4.3)

For each firm i, the J log-odds ratios vis-à-vis the base choice is given by:

$$ \ln \left[ {\frac{{P_{ij} }}{{P_{i0} }}} \right] = \beta_{j}^{'} x_{i} $$

(4.4)

The log-odds ratio between choices j and k can be obtained as:

$$ \ln \left[ {\frac{{P_{ij} }}{{P_{ik} }}} \right] = x_{i}^{'} [\beta_{j} - \beta_{k} ] $$

(4.5)

The model described in Eq. (4.1) above is estimated by the maximum likelihood method given firm-wise observations on x and the choices actually made by them. Further details of the multinomial logit model and its limitations are available in standard econometrics textbooks.