• No se han encontrado resultados

CONCLUSIONES Y RECOMENDACIONES

Wang and Blomstrom (1992) developed a model and used it to investigate whether the technological gap was a major factor that determines FDI spillovers. The study investigated whether international technology was transferred from MNCs by means of interaction with a domestic firm. The study used US majority owned foreign affiliates in 33 host countries. The model used for the study began by assuming that technology affected demand. Consumer preferences was represented by a utility function of the form

 

YU

GiYi

U ………2.4

Where, Y is an industry output index, Yi is firm’s output and Gi reflects the attractiveness of firm’s products. Gi Increases in relation to the firm’s technology levelKi. The author assumed a logarithmic utility function, with Gi(Yi)taking the formKia, where a is a positive constant. The U(Y) was expressed as:

) ( ) (Y U KdaYd KafYd U   ) ) ( ( (Kda Yd Kaf Kda Yf U   ……… 2.5

Then the model was monotonically transformed to take the form of ),

( )

(Y LnKd LnYd kaYf

U    ………. 2.6

Where k is the technology gap, defined as the ratio of the foreign firm’s technology level to that of the local firm, subscripts d and f referred to domestic and foreign firm, respectively. The study found out that learning effort of host country was an important factor that determined the rate at which MNCs transfer technology to domestic firms.

Kokko (1994) evaluated the impact of technological gap between domestic firms and foreign firms. The study represented a pioneering contribution in this area. In order to accomplish this task, the study used a detailed industrial data from Mexican manufacturing industry. The study considered three variables; the level of technological complexity, the average capital intensity of MNCs and the

technological gap. The results suggested that an increase in technological gap complexity and capital intensity makes the occurrence of FDI spillovers less likely, but that an influence of the technological gap was neutral. However the study concluded that wide technological gap, together with large foreign market shares generate a less favorable situation for the emergence of spillovers since in this case MNCs may operate without connections with domestic firms. However the study did not take care of time effects and endogeneity problems and this could have affected the robustness of the results.

Girma and Wakelin (2001) conducted a study to find out how interaction between size of the domestic firms and absorption capacity determined benefit of spillovers to domestic firms. The study used established level data taken from the UK census of production. To control selectivity and endogeneity problem, the study used semi parametric approach in the analysis. The conclusion was that large and highly skilled domestic firms do not benefit from foreign presence because they are probably the nearest to foreign multinationals in terms of technology and market share, and may already operate at the technology frontier. However, the group of firms that gain most from foreign presence consisted of small ones with a high proportion of skilled labor.

Kinoshita (2001) conducted a study to find out whether research and Development is a determinant of spillovers to domestic firms in Czech Republic. The study used R&D as a proxy for absorption capacity, as it was considered that

this increases the capacity of domestic firms to imitate new technologies. With statistical information for the Czech Republic, the study confirmed that domestic firms only benefited from foreign firms when they performed R&D actively, that is, when they developed the ability to imitate new technologies. Thus, R&D activity and FDI appeared to be complimentary in the productivity of domestic firms.

2.5 Overview of Literature

From the above literature on FDI spillovers, it can be observed that the pioneering studies that were done by Globerman (1979), and Blomstrom and Persson (1986) used cross sectional data. The results of most of these studies found a positive impact of FDI spillovers to local firms. However, the use of cross sectional data did not allow for time and industry effects. The evidence of positive spillovers from foreign firms may be due to the possibility that MNCs invest in highly productive industries. If, for example, productivity in a given industrial sector is higher than the others, the MNCs will be attracted to that sector. A basic cross- sectional analysis will show that there is a positive and statistically significant correlation between MNCs and productivity of the locally owned firms, which would be interpreted as indicative of spillovers, which may not necessarily be true. Hence, due to failure to control for time effects which might be correlated with but not caused by foreign presence, results obtained may be spurious or biased.

However, the use of panel data which is a mixture of cross-sectional and time series data has enabled researchers to overcome some of those challenges from cross sectional data. According to Green (2006), the fundamental advantage of a panel data set over cross section is that it allows for great flexibility in modeling differences in behavior across the individuals. In addition, panel data gives a larger number of data, increasing degrees of freedom and reducing the problem of multicollinearity among explanatory variables (Green, 2006). Most of the studies undertaken with panel data reveal negative or positive insignificant effects. These studies include, Aitken and Harrison (1999), Pavel (2007) and Jurat (2007). There are few studies that used panel data that produced positive results, examples are Blomstrom and Persson (1986) and Gachino (2007), but depending on certain factors like the absorption capacity and extent of technological gap.

From the literature on the effect of FDI on domestic enterprises, it can be seen that the results have been mixed, some producing positive and others negative results. The studies that do exist restrict themselves to a very small number of countries. In particular, most studies have examined horizontal spillovers and few on vertical spillovers. In addition very few studies have been done in developing countries.

In Kenya, Gachino (2007) was limited only to spillovers in the manufacturing sector. However, in the past two decades, the sector composition of FDI has

shifted sharply away from extractive industries and manufacturing and towards services (UNCTAD, 2004). In 1990, some 47 percent of outward FDI stock was in service industries. By 2003, this figure increased to 67 percent (UNCTAD, 2004). The shift to service is being driven by the general move in many developed economies away from manufacturing and towards service industries. Therefore, we cannot ignore the investment in this sector that is growing at such a high rate. This shows that it is important to do a study that cuts across all the sectors in Kenya before drawing a conclusion on how spillovers have affected domestic firms.

On empirical literature on labour diversity, many governments in recent times have introduced affirmative action policies in addition to the ban on discrimination. This is in order to promote equality and this affects firms hiring decision. Conversely some countries hesitate to introduce any affirmative policies arguing that affirmative policies could be counterproductive for both the discriminated and the business groups. This has necessitated studies on labour diversity and its effect on productivity. Earlier studies have looked at labour diversity in only one dimension, but in recent studies the multi dimensionality of labour diversity is captured (see for instance studies by Hamilton, Nickerson and Owan (2004) and Parrota, Pozzolli and Mariola (2010).)

In Kenya the studies that have been conducted like Alesina and Ferrara (2002) have only analyzed one dimension of labour diversity which is ethnicity.

Therefore this study has used a multi dimensional aspect of labour diversity by looking at three aspects i.e. skills, ethnicity diversity and gender diversity and the eventual different implications related to each of these dimensions in terms of spillovers and productivity. This will be useful in guiding the policy makers on issues to do with hiring in public service or firms.

CHAPTER THREE RESEARCH METHODOLOGY 3.1 Introduction

This chapter presents the research design, theoretical framework on production and spillovers theory and empirical model adopted as per each objective of the study. The data collection, method of analysis and the variables used are also explained.

3.2 Research Design

The purpose of this study was to investigate the effects of FDI spillovers on domestic firm’s productivity. There is no comprehensive study of firm productivity that has been based in Kenya hence the total number of foreign and domestic firms is unknown. In addition, not all firms are registered with Kenya Investment Authority (KIA) and hence the population of the firms is unknown. However, the study used a list of registered firms from KIA combined with another list from Kenya Institute for Public Policy Research and Analysis (KIPPRA). The total number of firms was 1140 as represented in Table 3.1. To get a representative sample, the study used a formula developed by Cohran (1975). 2 2 ) 1 ( e p p z n  ……… 3.1

Where n is the desired sample size, z is standard normal deviate at the required confidence level which was 1.96 at 95% level of significance, p was the proportion in the target population that was estimated to have the characteristic being estimated. According to Fisher et al (1943), if there is no estimate available of the proportion in the target population assumed to have the characteristic of interest, 50 percent should be used. Hence in this study p was assumed to be 0.5 and e, is the level of precision and for this study, it was taken to be 5 percent. Applying this formula, the desired sample size was;

385 05 . 0 5 . 0 * 5 . 0 * 96 . 1 2 2   n ……… 3.2

Therefore the target sample size was 385 firms. Sampling was done from the total number of firms in each city i.e. Nairobi, Nakuru, Mombasa and Kisumu. The four cities were chosen because, from the list obtained, they hosted most of the foreign and domestic firms. The Table 3.1 shows the total number of firms listed, the targeted sample and the achieved target of firms in the four cities.

Table 3.1: Number of Firms as per list from KIA and KIPRRA.

Source: Constructed From the List of Firms at KIA and KIPPRA

Location Total Number of Firms. Targeted sample Percentage Achieved Target Nairobi 812 193 50 100 Mombasa 148 97 25 46 Kisumu 120 57 15 30 Nakuru 60 38 10 28 Total 1140 385 100 204

As evident in Table 3.1, since Nairobi had the highest number of registered firms, 50 percent of the targeted firms were from Nairobi; Mombasa took 25 percent as it was the second largest host of firms, Kisumu 15 percent and Nakuru 10 percent. From the list of the firms in the four cities, stratified sampling was done whereby each city acted as a stratum. Then from each stratum, simple random sampling was done according to the required number of firms in each stratum. The selected firm had to have more than 20 employees since the study targeted small, medium and large firms (Ngugi and Musengele, 2008). However from the targeted firms, only 204 firms had complete information required for the three years, therefore firms with incomplete questionnaire were discarded. Since a panel data of three years was used, equivalent to 612 observations, the achieved sample was a good representative of the total number of firms.

The firms targeted were from three different sectors i.e. manufacturing, agricultural and service sectors. The basis of classification was their functions (see operational definition of terms pg. xii). Panel data was then used in the analysis. This was an improvement onearlier studies that had used traditional techniques of cross sectional data. Panel data is an inclusion of both cross sectional and time series data; hence it is believed to capture the dynamics of change in its analysis. Whereas micro dynamic and macro dynamic effects typically cannot be estimated using cross-sectional data, a single time series data set usually cannot provide

precise estimates of dynamic coefficients either, but the use of panel data helps in getting precise estimates of dynamic coefficients.

Documento similar