The ‘Knowledge-Capital Model' of the multinational firm predicts relationships between affiliate sales and country characteristics. Comment on the empirical tests of the model.
Over the recent decades, the mushroom growth of Multinational Enterprises (MNEs) along with the increasing growth of foreign direct investment (FDI) has been a remarkable feature of globalisation and international trade. The rate of growth of FDI flows in the world economy surpassed the growth rates of world income and international trade (UNCTAD, 2006). Following this astonishing growth of FDI, many economists have attempted to encompass the direction of FDI flows into theories and carry out empirical investigations on the factors which determine the market access and factor-cost motives of FDI. The earlier theories and literature on Multinational Enterprises focused on two aspects; horizontal integration and vertical integration. When MNEs duplicate their production process in many different countries to avoid trade costs/transportation costs, they are said to be horizontally integrated. MNEs locate similar or parallel activities in different countries. (Markusen, 1984; Markusen and Venables, 1998, 2000). When MNEs geographically fragment their production process to take advantage of factor-price differences, they are said to be vertically integrated. Under vertical integration MNEs choose to facilitate their activities in different countries depending on the stage of production (Helpman, 1984). Theoretical and empirical work has tended to slant towards the theory of horizontal firms as they tend to be more abundant in the world (Carr et al, 2001). The recent development of the Knowledge-Capital model (Carr et al, 2001) includes both horizontal and vertical FDIs under a general equilibrium framework. The model permits for the existence of both vertical and horizontal integrations of MNEs by assuming trade costs and different factor intensities across production stages. PUT SOMETHING EXTRA IN THE INTRODUCTION.
The model combines the horizontal motivations of FDI with the vertical motivations; the desire to place production close to the consumers and thereby avoiding any trade costs combined with “...the desire to carry out unskilled labour intensive production activities in locations with relatively abundant unskilled labour,” (Bloningen, Davies, Head, 2002).
In formulation of the Knowledge-Capital model, there are two countries; home (h) and foreign (f) to consider. Two types of goods are taken into account; good Y and good X. The quantity of a homogeneous good produced by competitive firms using the technology of constant returns to scale is represented by good Y while good X is the representation of a good produced by imperfectly competitive firms using the technology of increasing returns to scale and is “…subject to Cournot Competition with free entry and exit,” (Carr el al, 2001). There are two internationally mobile homogeneous factors of production skilled Labour (S) and un-skilled labour (L). Good Y is unskilled labour intensive and good X is generally skilled labour intensive. Furthermore it is assumed that head-quarter services can be separated from the production.
The knowledge capital model is constructed on three principal assumptions. First, the knowledge based and knowledge generating activities are geographically separated from production within the firm. These knowledge based services can be supplied to facilitate production at low costs. Second, the knowledge based activities supplied to the production facilities are skilled labour intensive in comparison to production. This feature of skilled labour intensity may lead to the vertical motivation of FDI, which separates production and establishes it in accordance with factor prices and market size. In direct relation to market size, firms may want to locate in some countries to take advantage of plant-level scale economies. Third, there is a joint input nature of knowledge based services to a certain extent; multiple production facilities can be supplied these services at low marginal costs simultaneously. This assumption creates an incentive through firm level scale economies for horizontal investments, under which the same goods and services are produced in multiple locations (Carr et al, 2001).
There are three more assumptions pertaining the size and composition of fixed costs that are important to make predictions (Carr et al, 2001). First, due to the joint input nature of knowledge based services, there exists multi-plant economies of scale (determining factors for type h firms). As mentioned in a previous assumption that head-quarters can transfer knowledge capital for low costs; the total fixed costs in good X of head-quarters and two production facilities “…is less than double the total fixed costs of a single plant firm,” (Carr et al, 2001). Secondly, there will be more prevalence of skilled labour intensive services in head-quarters (HQ) than production and plant-level production compared to the rest of the economy. The following ranks the skilled-labour intensity of activities:
HQ > HQ + Plant [Integrated X] > Plant > [Y]
The final assumption states that national markets are segmented for goods and unskilled labour is used in transport costs.
Within this model there are six types of firms to consider that combine the characteristics of horizontal and vertical multinationals. The types of firms considered by the knowledge-capital model are as follows:
Type Mh (Mf): These firms establish and sustain plants in both countries and the head-quarters is located in country h (country f). These firms are horizontal multinationals.
Type Nh (Nf): This type of firm constitute national plants and have a plant in and head quarters in country h (country f). Export of goods from these firms to country f (country h) may or may not occur.
Type Vh (Vf): This type of firm is a multinational with vertical motives. They have a single plant in country f (country h) with head quarters located in country h (country f). Export of goods from the plant in country f (country h) to country h (country f) may or may not occur.
From this construction the model predicts the dominance and favourable position of the different type of firms based on different country characteristics:
There will be dominance to national firms from h in the home country h if,
i. country h is large and there is abundance of labour
ii. there is similarity in size and relative endowments across both countries
iii. the transport costs are low
iv. there are high restrictions to invest in country f.
The multinationals with horizontal motives from country h will be dominant in h if,
i. there is similarity in relative size and endowment across both countries
ii. the transport costs in country h are very high
There will be dominance to vertical multinationals from country h in country h if,
i. the size of country h is small and there will be an abundance of skilled labour in the country in comparison to country f
ii. the transport costs in country h are low.
Due to the lack of data on the number of types of firms, it is only possible for the knowledge-capital model to generate predictions on the relationship between affiliate production and country characteristics (Carr et al, 2001). We assume that production is directly related to sales. These predictions are made by running simulations on the model. The simulation results are presented by three Edgeworth box diagrams. The Y-axis represents the total world endowment of skilled labour and X-axis represents that of unskilled labour. The south-west corner of the box diagram represents the origin of country h while that of country f is represented at the north-east corner. Output in both countries affiliate plants represents affiliate production.
1, shows a classic saddle and is simulated with high trade costs in both countries. When relative endowments are identical (SW-NE diagonal), then there will be largest affiliate sales if countries have similar income. This is because the production is dominated by horizontal firms. Half of the production of good X would be affiliate sales if only horizontal multinationals exist.
1 also shows that if one country is small as well as skilled labour abundant, affiliate sales are highest. The reason is that vertical multinationals are producing goods in unskilled labour abundant country. If only vertical multinationals existed, all the production of good X would be by affiliates. Affiliate sales are lowest when there are similarities in relative endowments but countries differ in size, and the skilled-labour abundant country is large. In both cases national firms with head-quarters located in the large country dominate the production of good X.
2 shows only the production plants in country f which are owned by country h. An inverted U-shaped curve is seen along the SW-NE diagonal. The affiliate production is highest when country h is small and skilled-labour abundant.
3 shows that high trade costs favour those firms with horizontal motives. Affiliate production tends to be highest when countries have similar relative endowments and high transportation costs. Vertical multinationals are harmed by low trade costs and affiliate sales fall with differences in relative endowments and high trade costs.
Many studies have empirically tested the predictions of affiliate sales and have found mixed evidence in support or against the Knowlege-Capital model (Braconier et al, 2005). The study by Carr et al (2001) estimated the model which consisted of skill differences defined as the ratio of skilled labour to total labour force in a home country, h minus that in a foreign country f. The model also consisted of an interaction term between skill differences and market size differences. Carr et al. (2001) used bilateral panel data on US inbound and outbound affiliate sales in 1986-1994. 36 other countries are also used with one year's data. Real sales, GDP Sum, GDP difference, GDP difference squared, Skill difference, Skill difference squared, Investment Cost Host, Trade Cost Host, Trade Cost Parent are the variables used for estimating the model in CMM (2001), where real sales the dependent variable is the volume of affiliate sales in millions, GDP sum is the sum of GDP in the parent and host country, GDP difference is the difference in the GDP of host and parent country, the skill difference is the difference between the ratio of skilled labour to total labour force in the parent country and that in the host country, investment cost host is an index from zero to 100 of impediments in investment in the host country, Trade cost host/parent is an index from zero to 100 of the restraints of protectionism in host/parent country and distance is the distance in kilometres of each country's capital city from Washington.
The empirical results of the CMM (2001) proved the following important aspects of the Knowledge-Capital Model:
1. An increase in the trade costs of the host countries would raise production by affiliates of parent country firms in all cases.
2. When incomes converge between the US and the j'th country affiliate sales would increase in both directions.
3. The skill labour abundance would attract the production affiliates of the US to the host country.
4. Affiliate production has an income elasticity of more than one. A bilateral increase in income for both parent and host country would increase affiliate production by more than proportionally.
The empirical results strongly support the Know-ledge capital theory of FDI. It is suggested from the “...strong statistical fit of the model that bilateral variables explain much of the variable in affiliate sales,” (Carr et al, 2001).
The study by CMM (2001) study gives empirical support for the effects of all variables related to FDI except for Trade Cost and Skilled Difference Squared. That is, the study strongly supports the theory of horizontal multinational. The Vertical nature is not strongly supported by the study. However a series of papers (Markusen and Maskus, 2002; Carr et al, 2003) undertake studies particularly on the predictive power of the Knowledge-Capital model on vertical motivations of multinationals. In order to solve this puzzle of vertical integration, these papers selected different types of samples (different countries and different time periods) for the study and made slight modifications to the CMM (2001) model. But in spite of all these attempts, the issue of vertical integration still remains as an unsolved puzzle. All these empirical attempts are robustly in support of horizontal integration of multinationals while the vertical aspect is weakly supported by these studies.
Hence, modifications on the existing models and alternate ways of estimation have come to fill the gaps of already existing studies. Bloningen, Davies and Head, 2002 points out the misspecification of the terms measuring differences in skilled-labour abundance, which are “...key variables that identify vertical multinational motivations,” (Bloningen, Davies and Head 2002). After correcting the model, the model seems to be suitable to support the horizontal motivations of the multinationals strongly.