Prema-chandra Athukorala

Research School of Pacific and Asian Studies

Australian National University

The structuralist assertion that MNEs involved in export-oriented production in developing countries suppress real wage growth leading to an unequal distribution of gains from the internationalisation of production is examined through a case study of Malaysia. When appropriately controlled for other determinants of inter-industry differences in wage growth, there is no empirical evidence to suggest that MNE presence in negatively related with real wage growth. The observed relatively slow growth of average real wages of workers in foreign-owned export industries can be explained in terms of the ongoing process of wage convergence among industries and the increased exposure of domestic manufacturing to world competition through greater export orientation.

JEL Classification: F23, L60, O53

Employment implications of the involvement of multinational enterprises (MNEs) in domestic manufacturing has long been a source of controversy in Developing countries (DCs). A common ground for attack on MNEs involved in import-substitution production is that the technology introduced by them is highly capital intensive, and therefore tends to reduce the employment potential of industrialisation. It is also alleged that MNEs pay abnormally high wages to local workers leading to further intensification of the overall capital intensity bias in industry and unequitable distribution of gains from industrialisation. By reverse reasoning, MNEs involved in export-oriented industries are expected to yield a favourable labour market outcome as their choice of technology and wage policy tend to be much in line with the comparative advantage in international production of the given host country. This perception is reflected in the new conventional wisdom that export-oriented foreign direct investment (EOFDI) deserve significant encouragement in developing countries.

The critics of MNEs are, however, not sanguine about the desirability of this policy advocacy. They argue that, while global sourcing activities of MNEs may generate new jobs in host countries, there are some untoward effects of MNE participation in export-oriented industries which lead to unequal distribution of gains from such activities between host and home countries. One such alleged adverse effect relates to the perceived tendency of export-oriented MNEs to restrain real wage growth in a given production location compared to their import-substituting (domestic-market oriented) counterparts and indigenous firms. Export-oriented MNE affiliates in developing countries, because of their use of labour intensive production techniques, tend to be more sensitive to changes in the wage bill and expectations about relative wage cost of producing in different locations. Furthermore, they have the flexibility to transfer production facilities from one country to another in response to changing labour market conditions, in sharp contrast to the difficulties of such a move for the import-substitution MNEs which are essentially ‘location bound’. The critics argue that these two factors - the greater sensitivity to wage changes and the prowess in labour relations emanating from the ability to relocate production - make the demand for labour by export-oriented MNEs more elastic and more resistant to workers’ wage demand. Thus, under given labour supply conditions, workers employed in these ventures are likely to experience slower real wage growth compared to their counterparts in domestic-market oriented MNE affiliates and indigenous firms (Caves 1996, 132).

The defenders of MNEs contend that the wage restraint critique is based on imprecise and largely non-economic considerations. In particular, they argue that the characterisation of export-oriented MNEs as footloose ventures whose locational decisions are based largely on unit labour costs is inconsistent with the corporate behaviour in the context of the ongoing process of internationalisation of production. New communication technologies and more competitive international markets are causing NMEs to distribute their activities more aggressively across countries through global assembly and marketing networks as part of their survival strategy. In this endeavour, they have little room to take a short-term view of the host country labour market conditions. Moreover, the wage restraint hypothesis is based crucially on the rather restrictive presumption that alternative investment locations are available in abundance. This is far from the reality because low wage countries are not the same thing as good locations for investment. While labour cost is important, other factors such as the presence of strong (or potentially strong) indigenous supply capabilities, good infrastructure, political stability, and appropriate government macro-organisational policies usually figure prominently in international investors locational decisions. This is the simple reason why, despite widespread attempts to entice MNE participation in export-oriented industries, so far only a handful of countries have been able to establish themselves as locations favoured by MNEs in international production.

Given the contending views, the impact of export-oriented NMEs on real wages in DC manufacturing remains very much an empirical issue. The present study aims to contribute to the debate through an in-depth time-profile analysis of MNE behaviour in Malaysian manufacturing industry. Malaysia provides an excellent case study of the subject at hand for the following reasons. First, given the long-standing government policy in favour of FDI participation in the economy, Malaysia has a well-established and diversified foreign presence in manufacturing industry. More importantly, there has been a clear shift in the presence of MNEs in manufacturing away from import-substituting production and toward export production. Second, the labour market conditions in Malaysia also ideally suited for examining the wage setting behaviour of MNEs. Historically the government has maintained a remarkably passive attitude on labour market conditions and the wage setting behaviour of export-oriented firms has not been conditioned by minimum wage legislation. Third, Malaysia is one of a handful of developing countries for which data of reasonable quality required for an empirical investigation of this nature are available.

The study proceeds as follows. The next section sets the stage for the ensuing empirical analysis by providing an overview of MNE participation in Malysian manufacturing with emphasis on employment implications. Section 3 examines wage trends by nationality of ownership for total manufacturing and highlights the limitations of inferences drawn from such aggregate analysis. Section 4 tests the wage restraint hypothesis through the estimation of an inter-industry wage growth regression using data for forty five industries. The final section offers concluding comments.

Malaysia in among the few DCs which have successfully mobilised foreign direct investment to achieve economic growth through export-led industrialisation. The Malaysian policy emphasis on the promotion of export-oriented foreign direct investment (EOFDI) dates back to the late 1960s when specific incentives were offered to export-oriented ventures under the Industrial Incentives Act of 1968 (Lim 1992). This policy received added importance under the New Economic Policy (NEP) launched in 1970. The rich assortment of incentives offered by the Free Trade Zone (FTZ) Act, which was enacted in 1971 as a key element of NEP, included duty-free imports of raw material and capital equipment, streamlined customs formalities, subsidised infrastructure facilities and company income tax incentives. By the mid-1980s, there was a growing conviction among the Malaysian policy circles that certain elements of the ethnicity-based affirmative action policy of the NEP were inconsistent with the national economic goal of achieving greater integration of the Malaysian economy with the global economy. These policy inconsistencies were redressed and further incentives for foreign investors were introduced under the Promotion of Investment Act passed in 1986.

These policy initiatives - coupled with a highly favourable investment climate characterised by political stability, policy continuity, macroeconomic stability, and a long-standing open trade policy regime - set the stage for a surge of MNE participation in the Malaysian economy. By the early 1970s, Malaysia had already established a name as a favoured location for assembly activities in the global electronic industry, and the subsequent years witnessed a rapid expansion of these activities and the entry of export-oriented firms from neighbouring newly industrialising countries (NIEs) into light consumer-good industries (Ariff 1991, On 1989, Grunwald and Flamn 1985). FDI inflows have been particularly rapid since the mid-1980s as the timing of the investment liberalisation in 1985-86 coincided with economic changes in Japan and the East-Asian NIEs (the post-Plaza Accord appreciation of exchange rates, the loss of General System of Preferences (GSP) privileges and rising domestic wages) that caused many of their manufactures to seek low-cost production sites offshore. Between 1987 and 1994, FDI inflows increased by almost ten-fold and for the past four years the annual volume of FDI flowing to Malaysia has remained higher than to any of the other ASEAN countries.

Historically, MNE participation in Malaysian manufacturing was concentrated in import substitution production in areas such as food and beverages, chemical and pharmaceutical. Their involvement in export production was limited to some processing activities linked to primary product sectors. The rapid expansion of EOFDI flows since the early 1970s has brought about a dramatic transformation in the product structure of MNE participation. From about the mid-1980s, production for the domestic market has become secondary to using Malaysia as a base for manufacturing for the global market. The share of projects with an export orientation of 50 percent or more increased from 24 per cent of total approved projects in 1984-85 to over 85 per cent by the mid-1990s. This structural shift is MNE participation clearly evident in the data on the output and employment composition of foreign production reported in Table 1. For instance, the share of semiconductors and electronics alone in total value added by manufacturing amounted to over 35 per cent, up from 13 per cent in 1970. There has also been a significant increase in the relative importance of new product areas such as non-electrical machinery, consumer electronics, professional and scientific equipment. By contrast, value added shares of traditional import-substitution product sectors such as food, beverages, tobacco, iron and steel, chemicals have declined over the years.

The growing presence of NMEs in export-oriented manufacturing has dramatically transformed Malysia’s export structure which was historically characterised by a heavy reliance on a limited range of primary commodities. In the early 1970s, the share of manufactures in total merchandise exports amounted to only about 10 per cent. Since then, manufactured exports have emerged as the most dynamic element in the export structure. In 1994, with a manufacturing share of about 78 per cent, Malaysia was the developing world’s sixth largest exporter of manufactures, after the Four Dragons of East Asia and China. There are no direct estimates of the share of foreign firms in exports. However, a simple comparison of data on export and output shares of foreign firms show that they provided over three-quarters of total manufactured exports by the mid-1990s (Lall 1995). The electronics industry (which is almost completely foreign owned (Table 2)) alone contributed to over 60 per cent of total manufactured exports in 1994 (Athukorala and menon 1996).

The employment outcome of export-led industrialisation has been very impressive. By the mid-1980s, the unemployment rate in Malaysia was as high as 8 per cent. Since then it began to decline falling to only 2.8 per cent by 1995. Most of the new employment opportunities have come from the rapidly expanding manufacturing sector. The direct contribution of manufacturing to total employment increment between 1987 and 1994 was as high as 60 per cent. MNE affiliates have played a pivotal role in this impressive employment outcome (Figure 1). The share of MNEs in total manufacturing employment varied in the narrow range of 29 - 35 per cent during 1968-80. It then declined to an average level of 28 per cent in the first half of the 1980s reflecting the state-led heavy industrialisation push during this period. Since the mid-1980 the employment share has increased consistently reaching a level of over 45 per cent by 1992. With the rapid expansion of export-oriented manufacturing, the structure of manufacturing production associated with MNEs has become more labour-intensive over the years. Reflecting this development the share of MNEs in total manufacturing employment has been much faster than the increase in the share of output of these firms (Table 2). This pattern is particularly noticeable in non-metallic minerals, basic metal products, fabricated metal products and miscellaneous manufacturing.

Figure 2 depicts wage patterns in Malaysian manufacturing by nationality of ownership during the period 1968-1992. The wage figures reported here relate to the average gross money earnings with include both regular pay and various pecuniary

benefits. Foreign firms (MNE affricates) are defined as those firms with 50 per cent or more of foreign equity ownership.

It is evident that, overall, the growth of real wages of workers employed in MNE affiliates has lagged behind that of local firms. The compound annual growth rate of MNE real wage during 1968-92 was 1.52 per cent compared to 3.78 per cent for local firms. Interestingly, the growing labour market pressure in an economy which has been approaching full employment since the late 1980s is not reflected in the MNE wage index. The compound growth rate for the period 1968-1986 (2.38 per cent) is significantly higher than that for the overall period (1.51 per cent). With the

increase in MNE share in total manufacturing employment, the time patterns of the overall wage index has tended be dominated by that of MNE index.

In the early 1970s, when MNE affiliates were involved predominantly in import-substitution production, the average wage of the workers in these firms was about 60 per cent higher than their counterparts in local firms (Figure 3). Reflecting slower growth of MNE wages this gap has tended to narrow over the time. By the late 1980s, the difference between average wages of the two ownership groups had virtually disappeared.

All in all, the aggregate wage data suggest that increased presence of MNEs in export-oriented production has coincided with a remarkable slowing down of real wage growth in the foreign-owned segment of Malaysian manufacturing. Given the continuous increase in MNE presence in the manufacturing sector, decline in MNE wages has in turn brought about a slowing down in aggregate real wage levels. This evidence is often drawn upon in the Malaysian policy debate to support the view that the reliance of EOFDI as the vehicle for rapid industrialisation lead to an unequal distribution of gains from export-led industrialisation. The problem with such interpretation is that differences observed in a comparison of average wages for the two ownership groups may simply reflect wage differentials emanating from various influences operating at the individual industry level and/or compositional changes within the aggregate, rather than possible ownership specific influences. To delineate the later it is necessary to control for the former using an appropriate econometric procedure. This is the purpose of next section.

In this section, an attempt is made to delineate the postulated wage restraining effect of export-oriented MNEs by controlling for labour market and industry characteristics which have been identified in the literature as causing systematic inter-industry variations in real wage growth. Our methodology is to estimate a fully specified inter-industry wage growth regression which incorporate foreign ownership, both separately and intellectively with export-orientation. The control variables used in the regression specification are standard to the empirical analysis of inter industry wage differentials. The regression specification is given below followed by a discussion variable specification and measurement.

The variables (with the expected signs for the regression coefficient in bracket) are

RWG Growth of real wages

FOW(+) Foreign ownership

EO (-) Export orientation

FOW*EO(-) An interaction term for foreign ownership and export orientation

INSWG Initial wages

EMG (+) Employment growth

FMSH (-) Female share in the work force

SIZE (+) Plant size

CONC (+) Industry concentration

CAPI (+) Capital intensity of production

WGSH (-) Wage share in production cost

PSE (+) Public-sector ownership

UNION (+) Union density

FOW and FOW*EO are the two variables which are directly related to the issue as hand. As noted, a widely held view of MNE affiliates involved in import-substitution production in developing countries is that they have a tendency to pay higher wages than their domestically-owned counterparts. Such difference may reflect MNEs’ willingness to pay wages that are more in line with what they pay in their home countries, and/or simply their desire to maintain an image of good corporate citizens in the host country (Lim 1977). According to this view greater foreign presence in a given industry should be reflected in higher wage growth. The critics of export-oriented MNEs, by contrast, perceive wage repression as an operational aspect of these firms (See Section 1). According to this view, these firms are not only highly very sensitive to wage increases but also they have prowess to resists wage demand in a given production location through credible threats to relocate elsewhere. To delineate the wage repression effect of EOMNEs we include a foreign presence - export orientation interaction term (FOW*EO) in the regression specification. According to the wage restraint hypothesis, the estimated coefficient on this variable is expected to be negative.

Export-oriented firms generally operate under greater market pressure compared to domestic-market oriented firms which enjoy both policy-induced and natural protection. This may be particularly true for exporting firms located in a small exporting economy like Malaysia given the nature of the export mix (standardised consumer goods and components in vertically integrated industries) and small share in world supply in most (if not all) of product lines (Athukorala and Riedel 1996). To the extent that MNEs tend to be relatively more concentrated in export industries, the wage restraining effect of market pressure can be reflected in slow wage growth compared to their local counterparts. For these reasons, EO is included as an additional explicator in the regression.

Among the other explanatory variables, wage at the start of the period (initial wage, INSWG) is included to allow for a widely observed regularity in the evolution of inter-industry wage structure in market oriented economies, namely the compression of inter-industry wage differentials (or, the leveling of the wage structure) over time. The leveling of the wage structure, whenever it occurs, is normally the result of large absolute increases in the industries paying lower wages in the initial period than in those paying higher wages then. If this pattern holds for the Malaysian wage structure, then the estimated coefficient of INSWAG should carry a negative sign.

Employment growth (EMG) is included to capture the impact of demand pressure in the labour market on sectoral wage growth. If labour markets are generally competitive and if each industry faces a roughly similar less-tan perfectly elastic supply curve of labour, then industries wishing to expand their labour force would have to pay higher than average wages. This reasoning suggests a positive relationship between percentage employment growth and wage growth, other things remaining unchanged.

The choice of establishment size (SIZE) and capital intensity (CAPI) is suggested by the efficiency wage literature (Brown and Medoff 1989, Oi 1990). All other things constant, large firms may pay higher wages than smaller firms, presumably because shirking is harder to monitor in larger firms. Tendency for shirking might be greater in large firms because employees’ discontent with the job is likely to correlate positively related with the firm size. By similar reasoning, one can postulate that higher capital intensity encourages firms to pay efficient wages, since it is more costly for capital intensive firms to suffer employee shirking or absenteeism. Apart from this efficiency consideration, one can also expect firms in capital intensive industries to concede to higher wage increases because wage bill would typically account for a relatively small proportion of production cost. Put simply, firms in capital intensive product lines may not be so cost conscious with respect to wages.

The share of the wage bill in total cost of production (WAGSH) is an indicator of the relative importance of wages in the firms’ profitability calculations. An inverse relationship is, therefore, expected between this variable and RWG. As noted, the capital intensity variable (CAPI) also capture the impact on wage growth of an industry’s capacity (or willingness) to concede to wage increases. However, capital intensity is only a partial measure of the willingness to pay higher wages. This is because it does not capture the relative importance of material inputs in the cost structure. This is a significant omission particularly when it come to export oriented assembly activities. This reasoning justifies the use of WGSH as an additional variable.

The sex mix of the work force (female share, FMSH) is usually considered an important explicator of industry wage differential on grounds of sex discrimination in employment, the tendency on the part of employers to pay less to women workers for the same job. Should this be the case, then the female presence in the work force has a direct negative impact on average industry wage. At the same time in industries where substitution possibilities exist between male and male workers, a high proportion of female employment may also serve to depress the level of male earnings. For these reasons, a negative coefficient is hypothesised for FMSH. It is important to note that a negative relationship between FMSH and RWG can also emanate from two other factors; the low level of unionisation found in female-intensive industries and the concentration of female employment in unskilled jobs. We control for the former effect by the inclusion of UNION as a separate explanatory variable. Unfortunately we do not have data on the skill mix of the work force to explicitly allow for the latter impact. Thus the coefficient of FMSH may be capturing the impact of both sex discrimination and skill mix.

Industry concentration (CONC) (measured by the four-firm concentration ratio) is included to capture the impact of market power of a given industry on wage growth. The hypothesis is that since greater market power translate into excess profits, firms in monopolistic or oligopolistic industries insulate from market pressure may pursue a policy of paying supra-competitive wages. Public-sector ownership (PSE) is included to allow for the possibility that public sector firms may be more accommodative to wage demands on political grounds. Finally union density (UNION), measured as the share of union members in total work force, is included to allow for the possibility that the activities of the unions distort the competitive working of the labour market and lead to higher wage growth in industries with active union participation.

The model is estimated using cross-sectional data for 45 industries at the five-digit level of the International Standard Industry Classification (ISIC). The dependent variable (RWG) is measured as the average compound growth of real wage during the period 1976-1995. The definition and measurement of other variables and the data sources used are explained in the Data Appendix. After satisfying ourselves (using the Wu-Hauesman test) that endogeniety does not pause a problem for any of the explanatory variables, the simple OLS was used as the estimation method. SIZE, INSWG, CAPTW are measured in natural logarithms, while the other variables are in ratios/percentage form. Thus the estimated coefficients can be estimated as elasticities.

The regression results, together with standard test statistics relating to the OLS error process, are reported in Table 3. The correlation matrix of the variables are given in Table 4, to aid the interpretation of the results. The full model is reported as Equation 1. The equation is statistically significant at the one per cent level (in terms of the standard F test) and it performs well by all diagnostic tests on the OLS error process. The coefficients on foreign ownership (FOW) and FOW*EO (and three other variable) fail to attain statistical significance. The deletion of these variables is supported by the variable deletion F-test (Equation 2) and the coefficient estimates for the reaming variables are remarkably resilient to this variable deletion. Thus, our results suggest that when other relevant influences are appropriately taken into account, neither foreign ownership (FOW) nor the MNE presence in export oriented industries (FOW*EO) are significant explicators of inter-industry differences in wage growth in Malaysian manufacturing.

There is a strong bivariate correlation between EO and FOW*EO (Table 5), which is to be expected given the strong foreign presence in export oriented industries. To test if this inter-correlation influences the regression results, we re-estimated the model by replacing EO with FOW and FOW*EO (Equation 3). The coefficients on these two variables are still statistically insignificant in this alternative specification. Moreover the specification is statistically unacceptable in terms of the variable deletion test conducted against equation 1 (which is the ‘maintained hypothesis’). Thus, our inference that when, other relevant factors are appropriately taken into account, FOW and FOW*EO are not significant explicators of inter-industry wage growth differentials is remarkably resilient to the inter-correlation between EO and FOW*EO.

Among the control variables, INWG has a negatively-signed coefficient which is significant at the one-per cent level in all specifications. Thus there is strong statistical support for the hypothesis that wage differentials have been compressed over time. This result is consistent with the general pattern observed relating to changes over time in the inter-industry distribution of wages using the coefficient of variation as the index of relative dispersion. In an analysis of annual wage data for three-digit ISIC industries Richardson and Ying (1990) have noted a continuous decline in the coefficient of variation between the mid-1970s and 1985. The data for our sample of 45 industries points to a continuation of this trend into the 1990s. The coefficient of variation of industry wages declined from declined from .47 in 1986 to 0.38 in 1995.

The coefficient of EO is statistically significant at 5 per cent or better with the expected (negative sign). Thus there is statistical support for the hypothesis that greater export orientation is related with slower real wage growth. There appears to be a compromise between achieving rapid employment growth and real wage growth under export-led industrialisation.

The coefficient of employment growth variable (EMG) is statistically significant, but carries the (unexpected) negative sign. A similar a result that has been found for Malaysia in two previous studies (Richardson and Ying 1990, p. 12 (Table 11) and Mazumdar 1993). While Richardson and Ying discard the result as ‘counter intuitive’, Mazumdar comes up with an interesting explanation of this ‘odd behaviour of wages’ (Mazumdar 1993, 371). A prevalent employment practice in Malaysian manufacturing is to attach value to senior workers both in wage policy and labour shedding. Given this practice it is highly plausible, so argues Mazumdar, that structurally weak industries with slow (or negative) employment growth tend to exhibit relatively higher wage growth compared to dynamic (mostly export oriented) industries. (Note that the real wage index is based on average annual earnings). In any case, the results for the other variables are remarkably resilient to the deletion of EMG from the in the set of regressors (See Equation 4).

The results suggest that greater presence of female workers in the work force leads to higher (not lower) real wage growth. This result, which runs counter to our theoretical postulate, seems plausible in the Malaysian context (Richardson and Ying 1990, On 1989). The recent rapid expansion in employment in Malaysia has come largely from industries characterised by a large proportion of Women workers in the labour force. And there is evidence that most of these dynamic product lines, in particular, assembly activities in the electronics industry, specifically requires skills specific to female workers. At the same time as the economy reaches virtual full employment, female labour has become a relatively more scarce factor compared to male labour. This is because the reliance on labour migration is predominantly an option available only for meeting labour shortages relating to male workers (Pillai 1995). The increased participation of females in higher education would also have resulted in a contraction of the supply of unskilled female workers. These labour marker developments imply that the supply of female labour has become relatively less elastic compared to that of male labour. Thus a negative coefficient on FMSH is highly plausible. This results is also consistent with the findings of Richardson and Ying (1990, p. 68) on the behaviour of gender related wage differentials. On an examination of male-female earnings ratio for broad occupational categories in five industries with a large representation of women workers, they found that, across the board, gender-related wage differentials had moved in favour of female workers.

Of the reaming variable, capital intensity (CAPI) carries a statistically significant and positive coefficient supporting the hypothesis that, other things remaining equal, capital intensive industries tend to experience higher wage growth. In the overall regression, the firm size variable (SIZE) has the expected (positive) sign, but fails to achieve statistical significance because of the inter-correlation with CAPI. In a specification which exclude CAPI, the coefficient of SIZE attains statistical significance (Equation 5). Finally, there is some weak statistical support for the hypothesis that the involvement of trade unions in wage setting contributes to inter industry wage differentials.

In this paper we have taken a critical look at the proposition that export-oriented growth achieved through the involvement of MNEs leads to the suppression of real wage growth in developing countries, leading to an unequal distribution of gains from the internationalisation of production. When allowance is made for various other determinants of manufacturing wage growth at the individual industry level, there is little evidence to suggest that MNEs use their prowess to restrain real wage growth. Our results are consistent with the pro-MNE view that under the current process of internationalisation of production it is not correct to consider export-oriented MNE affiliates as foot-loose ventures.

How can one explain the relatively slow wage growth of foreign firms compared to that of domestic firms (and the consequent narrowing of the wage gap between MNE affiliates and local firms) which is clearly visible in available aggregate wage data (Section 3)? Our regression analysis comes up with two possible explanations. First, there is strong evidence that greater export orientation is associated with slower wage growth. This would have been reflected in a relatively slow growth of MNE wages because, since the mid-1980s, there has been a palpable shift in the structure of foreign presence in Malaysian manufacturing, away from home-market oriented activities and towards export-oriented production. Second, there has been a strong process of wage convergence under way in which low wage industries (which are generally characterised by greater indigenous ownership) are catching up with high- wage industries as part of the process of rapid economic transformation. Third, the increase in labour intensity of production as part of the process of greater internationalisation production may have constrained relative wage growth of MNE affiliates, but our result on this is inconclusive because of the overlap between the capital intensity and establishment size variables used in the regression analysis.

The data series used in this study have been compiled from the following publications:

(1) Bank Ngara Malaysia, Quarterly Bulletin, (various), Kuala Lumpur

Data on employment and wages for total manufacturing for the period 1968-70 used in Section 3 are compiled from (2). The price deflator used for constructing the real wage index is the consumer price index obtained from (1). Wages include both regular wage/salary and other pecuniary benefits paid by the employer. A firm is classified as foreign (a MNE affiliate) if more than 50% of equity capital is foreign owned. The data series are reported in Appendix Tabel A-1.

The regression analysis cover 45 industries at the 5-digit level of the International Standard Industry Classification for which employment and wage data are available in (3) for the period 1976-1995. These industries account for over 80 per cent of total output in Malaysian manufacturing (excluding petroleum refining).

Real wage growth (RWG) series is derived using average annual earring indices compiled from data from total employment and total earrings from (3) and the consumer price index from (1). Employment growth (EMG) is estimated from employment data from (3). Both RWG and EMG are annual compound growth rates estimated over the period 1976-1995 by fitting a logarithmic trend line.

Export orientation (EO) is measured using a dummy variable which take value 1 for export oriented industries and 0 otherwise. The ideal measure of EO is of course the average export propensity (export-output ratio) estimated at individual industry level. Average export propensities can be calculated by linking production data (from (2)) with trade data (from the UN trade data system). But these estimates tend to overstate export propensity ( and the magnitude of the error varies over time) for two reasons. First, output data incomplete in coverage; in the survey some industries (at the 5-digit ISIC level) are fully covered while establishments with less than 10 full-time workers are excluded in most industries. Second, production data are in producers’ prices, while export data also include costs of transportation to border and other costs incurred by intermediaries such as wholesalers. To identify export oriented industries we use the classification used by the Malaysian Department Statistics in compiling the production index This categorisation is consistent with the general pattern observable in terns of export-output ratios estimated linking trade and industrial production data as explained above. We believe that the use of a dummy variable to represent export orientation is not unsatisfactory because in Malaysia, like in other small export-oriented economies there is a clear dichotomy between domestic-market oriented and export-oriented firms. Export-oriented industries generally export an overwhelming proportion (80 per cent or more) of their output, and there are no industries whose activities significantly span both the export and domestic markets.

Union density (UNION) is defined as the fraction of work force registered as members of trade unions. The data series used is the average of the estimates for 1976, 1986 and 1992 compiled from (5). FMSH is measured in the same way using data from (4). In each data series, industries for which data are not available at the required 5-digit level are represented using the related 3-digit level estimates.

The industry concentration (CONC) is measured using the four-establishment concentration ratio. Foreign ownership (FOW) is measures using the share of foreign firms (defined as those with 50% or more foreign equity ownership) in total employment. (Results are invariant to the use of output or value added share as alternative measures). Plant size (SIZE) is the average plant size (sales volume derived by the number of reporting firms). Capital intensity (CAPI): is measured as fixed assets per worker. The data related to 1988, approximate middle year of the study period. We also used estimates for 1992 (the latest year) as well as the average for the two years, to find that the results are resilient to the choice of a particular year. All the variables listed in this paragraph are compiled from unpublished returns to (2) for 1988 provided by the Department of Statistics. 1888 is the earliest year for which data are readily available in computer readable form and we consider it as an appropriate representative year for the purpose at hand. In a comparison between 1988 and 1992 we found no significant rank reversal for any of these measures (and the results are invariant to the use of 1992 data).

_____________________________________________________________________

_____________________________________________________________________

Pillai, Patrick (1995), "Malaysia", Special Issue on Labour Migration in Asia, ASEAN Economic Bulletin, 12(2), 221-36.

ISIC International Standard Industry Classification.

(Dependent variable RWG, Sample size 45)

# t-ratios of individual coefficients are given in brackets. The test statistics are: FVD = Variable deletion (F) test conducted against equation 1, RESET = Ramsey test for functional form misspecification, JBN = Jarque-Berra test for the normality of residuals, and HET = White test for heteroskedasticity. RESET and HET statistics are based on the F-distribution, while the JBN statistic is based on the - distribution. Statistical significance is denoted as: * = 10 %; ** = 5 % and *** = 1 %.

__________________________________________________________________________________________________

Figure 2: Real Wages in Malaysian Manufacturing by Nationality of Ownership (1970=100)^#

Note:

# Annual compound growth rates estimated by fitting a logarithmic trend

line are given below. (All growth rates are statistically significant at the

one-percent level in terms of the standard t-test)

1968-86 1968-92

Source: See Data Appendix.

Figure 3: MNE Affiliate - Local Firm Nominal Wage Ratio, 1968-92

Source: See Data Appendix.