T-Test and Econometric Results

3.1 Data and T-test Results

The study uses data from the Asian Development Bank/World Bank (ADB/WB) investment climate survey of urban and rural enterprises in Sri Lanka conducted in 2004 (see ADB/WB, 2005). The ADB/WB survey selected firms on a largely random basis using a stratified simple random sample design. The cross-section data on 205 clothing firms are for the 2003–2004 period. The sample includes 47 foreign-owned and 158 domestic enterprises, which cover a range of market orientation, size classes, and locations in Sri Lanka.

Table 1 [ PDF 79.5KB | 1 pages ] shows the results of T-tests comparing the means of some characteristics of the foreign and domestic clothing firms (including export shares, capacity utilization rates, firm size, age of the firm, replacement cost of capital, imported equipment ratios, CEO education and experience, and share of foreign employees).

The following conclusions may be drawn:

• Foreign firms record better performance than domestic firms. There is a significant difference between the exports-to-sales ratios of the two groups. Sales to export markets account for more than 80 percent of total sales of foreign firms while they only make up about 40 percent for domestic firms.
• Foreign firms are larger in terms of the number of workers employed. On average, foreign firms have 1,060 permanent employees while domestic firms have 259 permanent employees.
• Foreign firms invest more in modern equipment than domestic firms. The means of the replacement value of capital and the shares of new equipment in total equipment are significantly higher in foreign firms than in domestic ones. Almost half of the imported equipment of foreign firms is new while the proportion is only a quarter for domestic firms.
• Foreign firms have better human capital than domestic firms. This is indicated by significant differences in the means of the educational attainment of the CEO, and the years of export experience of the CEO. For instance, the means for the level of educational attainment of the CEO are 5.02 and 3.53, respectively, for foreign and domestic firms (where 5 indicates a completed university-level undergraduate degree and 3 only vocational qualifications).⁴

3.2 Factors Affecting Firm-level Export Performance

A firm-level export function was estimated for Sri Lankan clothing firms using a Tobit model.⁵ The dependent variable is the export-to-sales ratio (EXSH). The full linear model is as follows:

EXSH = f (RVE, WAGE, SKW, CEOED, CEOEXP, FE, SIZE, TI, LOC)

The hypotheses and independent variables are as follows.

Capital is represented by the replacement value of capital per employee (RVE). Within a given activity, a higher level of physical capital in the form of modern equipment is expected to give a firm a competitive advantage. Thus, RVE is expected to be positively associated with export performance.

Human capital is captured by four variables: the skill adjusted wage rate (WAGE) ,⁶ the share of skilled workers in employment (SKW), the level of education of the chief executive officer (CEOED),⁷ and the years of experience of the chief executive officer (CEOEXP). Given the same set of skills, a lower wage in relation to productivity per worker is associated with greater firm-level competitive advantage and exporting. Furthermore, within a given activity, a higher level of human capital is likely to give a firm a competitive export advantage and is expected to have a positive effect on export performance. In this regard, different levels of human capital—the share of skilled workers as well as the chief executive’s educational attainment and experience of exporting activity—are all likely to be important.

Foreign ownership, the share of foreign equity (FE), is expected to have a positive influence on export performance. Access to the marketing connections and know-how of their parent companies as well as accumulated learning experience of producing for export make foreign affiliates better placed to tap international markets than domestic firms (see Dunning, 1993 for a discussion of the ownership advantages of multinationals). Furthermore, foreign firms tend to be larger than domestic firms and therefore better placed to reap economies of scale in production, R&D, and marketing. A large firm will be better able to exploit such scale economies and enjoy greater efficiency in production, enabling it to export more.

Firm size (measured by a dummy variable⁸) is expected to have a positive sign because exporting allows large firms, especially in small economies, to exploit economies of scale in production by relieving the disadvantage of the small home market.

Technological capabilities are measured by a firm-level technology index (TI). We expect TI to be positively associated with export performance because the process of acquiring technological capabilities in enterprises is not just a simple function of years of experience.

Rather, it requires conscious investments in creating skills and information. Such investments would include search, training, and engineering activities. The TI used here is a simple production capability-based variant of indices developed by Wignaraja for Sri Lanka (1998) and Mauritius (2002). The TI was constructed by ranking a clothing firm’s competence across a series of technical functions and the results were normalized to give a value between 0 and 1.⁹ Formal R&D activities are excluded from the TI but are included in the analysis as a determinant of TI (see Section 3.3).

The effect of an urban location is captured by a dummy variable (LOC), which takes the value 1 for firms located in and around Colombo and 0 otherwise. Favorably located firms are likely to have lower transport costs to the country’s main seaport and benefit from externalities (e.g., ready access to suppliers of raw materials and sub-contractors; marketing and other business services; and government services) compared with more distant firms. Thus, LOC is expected to be positively associated with export performance.

Table 2 [ PDF 88.8KB | 1 pages ] shows the estimated Tobit models. Estimated equation (1) presents the general model discussed above and equation (2) the reduced form with only the significant variables. Following testing for multicollinearity and heteroscedasticity, the results of equation (2) are considered.¹⁰ The pseudo R2 in equation (2) is acceptable for a cross-section model. Of the nine independent variables, six are significant (mostly at the 1% level) and have the expected sign.

Strikingly, FE is significant and positive, which indicates that foreign firms are more successful exporters than domestic firms. The explanation lies in a combination of access to marketing connections and know-how of their parent companies, accumulated learning experience of producing for export, and economies of scale linked to firm size. The correct sign and significance of SIZE underlines the links between firm size, ownership, and exporting.

TI is significant (at the 10% level) and positive, emphasizing that conscious investments in skills and information to use imported technologies efficiently contributes to export performance. More generally, this finding suggests that domestic technological activity and foreign ownership are complements rather than substitutes in developing export capabilities at firm level.

Two of variables for human capital (SKW and WAGE) are significant and have the correct signs. Within a given activity, a higher level of human capital in the production and a lower wage rate in relation to productivity give a competitive export advantage. The chief executive’s educational attainment (CEOED) and experience (CEOEXP) show no significance.

RVE shows no significance, which may be due to difficulties in measuring the replacement value of capital.

LOC is significant and positive. A location near Colombo provides an export advantage due to lower transport costs to the seaport and the benefits of numerous locational externalities.

3.3 Factors Affecting the Firm-level Technology Index

A firm-level technology function was estimated for the Sri Lankan clothing sample using an OLS model. The dependent variable was the technology index (TI). The full linear model is:

TI = f (SIZE, AGE, AGESQRD, PROF, CEOED, TRNG, R&D)

The hypotheses and independent variables are as follows.

Firm size, represented by total employment (SIZE), is expected to have a positive sign. The returns from capability acquisition are higher where a firm has a larger volume of sales to spread the fixed costs of capability acquisition and larger firms can have more specialized manpower and equipment. As foreign firms tend to be larger than local ones, firm size may also capture the influence of foreign ownership and their ownership advantages.

The a priori relationship between age of firm (AGE) and TI remains ambiguous. Theoretically a positive relationship with TI may be expected because years of accumulated experience can crudely capture “learning by doing” amongst other things. A negative sign is likely, however, for foreign firms that use superior imported technology and enjoy access to international markets but began operations recently. To test for the presence of a non-linear impact on TI, the square of age of the firm (AGESQRD) was used as well.

Human capital, captured by the share of university-level educated employees (PROF) and the level of education of the chief executive officer (CEOED), is expected to have a positive sign. Better educated chief executives and a larger base of university-educated workers can have a significant influence on technological capabilities through more effective search, engineering, and research activities.

Training (measured by expenditure on employee training as a percentage of sales, TRNG), is expected to have a positive sign. Explicit employee training is crucial during enterprise start-up for creating the requisite capabilities to use new production technologies. As technologies evolve, a continuous process of re-training is needed to supply the technical and managerial skills needed by new process and product innovations.

Research and development (represented by R&D expenditure as a percentage of sales) is expected to be positively related to TI. Efforts to acquire production capability in clothing firms can be supplemented by more formal in-house technological effort by technical manpower directed at new product designs, new fabrics (e.g., synthetics or rubber/cotton mixes), process adaptation, and trouble shooting.

Table 3 [ PDF 71.4KB | 1 pages ] shows the estimated OLS model results. Equation (1) is the general model and (2) and (3) are the reduced form models. Following testing for multicollinearity and heteroskedasticity, we consider the results of equation (2). ¹¹ The R2 in equation (2) is reasonable for a cross-section model. Five of the seven independent variables are significant mostly at the 1% level.

Firm size, university-educated employees, and R&D have a positive and significant relationship with TI. The correct sign on the firm size variable suggests that the different explanations for the firm size effect are valid. It is also likely that firm size may reflect foreign ownership. The positive sign on the university-educated workers variable suggests that higher-level skills are related to building technological capabilities. The positive sign on the R&D variable indicates that formal R&D efforts complement efforts at enhancing production capability.

Age is shown to have a non-linear impact on TI. A plausible explanation for this finding is the fact that foreign-owned firms (with superior technology and market access) began operations recently. Furthermore, with learning by doing, local firms gradually acquire technological capabilities. There may also be a minimum age that has to be reached before a domestic enterprise accumulates the requisite level of technological capabilities for export markets.

Meanwhile, the CEO’s education level and training expenditure show no significance. Further work is needed to find explanations for these results.

Download this Discussion Paper [ PDF 248KB| 18 pages ].

Post a Comment

We welcome your feedback on this publication. Post a comment. ADBI is not obliged to acknowledge or publish comments and may abridge or edit them before web posting.

Comment(s)

There are [0] comment(s) for this entry. Post a comment.