Home | Business News | Browse by Publication | I | International Advances in Economic Research

Openness and the efficiency of FDI: a panel stochastic production frontier study.

Article Excerpt
Introduction

In the past 30 years, we have witnessed worldwide trade liberalization, globalization of commerce and integration of a diverse set of economies. Open economies interact with one another in global product and capital markets. A country's trade balance captures the flow of goods and services traded on product markets. On the other hand, Foreign Direct Investment (FDI) and Foreign Portfolio Investment (FPI) represent the flow of physical and financial capital across national boundaries.

As far as the flow of capital is concerned, an issue of great concern to policymakers, international organizations and economists is the potential effect of FDI on long-term economic growth. This subject has been studied extensively at both the theoretical and empirical levels (Aitken and Harrison 1989; Bengoa and Sanchez-Robles 2003; Blomstrom et al. 1992; Blonigen 2005; Borensztein et al. 1995; Chowdhury and Mavrotas 2006; Ciruelos and Wang 2005; Damijan et al. 2003; Frimpong and Oteng-Abayie 2006; Lipsey 2000, 2002; Kohpaiboon 2002). The general consensus appears to be that FDI contributes to economic growth through several channels, the most important of which is perhaps technology transfer.

In an influential book on the consequences of trade barriers, Bhagwati (1978) argued that FDI contributes to growth by enhancing economic efficiency and that this effect is larger in economies that promote outward-oriented trade policies (export promotion) relative to those that pursue inward-oriented strategies (import substitution). A number of studies have tested the "Bhagwati hypothesis" empirically and have found support for it (Balasubramanyam et al. 1996; Kohpaiboon 2002).

In this paper, we use a version of the stochastic production frontier model that allows us to estimate technical inefficiency indices and specify their conditional mean as a function of FDI and its interaction with the degree of openness of the economy so as to test the Bhagwati hypothesis. Using maximum likelihood and an annual panel of 46 countries in different stages of development for the years, 1981-2001, we jointly estimate a translog frontier and the associated mean technical inefficiencies. Our findings suggest that increased FDI increases potential output in both developed and developing countries but the effect is more profound in the former economies. We also find that increased FDI reduces technical inefficiencies the more open the economy, but that this effect holds only for developed economies. Thus, our findings provide qualified support for the Bhagwati hypothesis as they reveal that the efficiency-enhancing effect of FDI depends not only on openness to international trade but also on the degree of development of the host country.

"Econometric Methodology" presents the econometric approach used in this study. "Model, Data, and Results" specifies the empirical model, describes the data, and presents the results. "Summary and Suggestions for Further Research" summarizes this work and draws some conclusions.

Econometric Methodology

The stochastic production frontier (SPF) model can be presented in the context of the following log-linear functional form (Aigner et al. 1977; Meeusen and van den Broeck 1977):

[y.sub.i] = [x.sub.i][beta]' + [[epsilon].sub.i], i = 1, 2,..., N (1)

where [y.sub.i] is the logarithm of output of firm (country or industry) i; [x.sub.i] is a 1 x (k+1) row vector whose first element is 1 and the remaining elements represent the logarithms of the k inputs used by the ith firm; and [beta]' is a (k + 1) x 1 column vector of unknown parameters. The random error term, [[epsilon].sub.i], is the difference of two independent random variables: a classical error term, [v.sub.i], and a non-negative random variable, [u.sub.i], that captures technical inefficiencies in firm i:

[[epsilon].sub.i] = [v.sub.i] - [u.sub.i] (2)

where the error variance is given by:

[[sigma].sub.[epsilon].sup.2] = [[sigma].sub.v.sup.2] + [[sigma].sup.2] (3)

While [v.sub.i] is typically assumed to be iid.n(0, [[sigma].sub.v.sup.2]), the choice of a distribution for [u.sub.i] is arbitrary. In practice, the truncated normal, half normal, gamma and exponential distributions have been used. In this context, the technical efficiency of the ith firm, [TE.sub.i], is the ratio of observed output of firm i divided by its efficient output represented by the estimated production frontier:

[TE.sub.i] = [y.sub.i]/exp([x.sub.i][beta]') = exp(-[u.sub.i]) (4)

The technical efficiency coefficient in Eq. 4, which is bounded between zero and one, is unobservable because [u.sub.i] is unobservable. Battese and Coelli (1988) show that the best estimator of exp(-[u.sub.i]) is its conditional expectation, E[exp(-[u.sub.i])|[[epsilon].sub.i]].

The above model, which implicitly assumes cross-sectional...