Subscribe Now Subscribe Today
Research Article
 

Macroeconomic Determinants of Skilled Labour Migration: The Case of Malaysia



Chuie Hong Tan and A. Solucis Santhapparaj
 
Facebook Twitter Digg Reddit Linkedin StumbleUpon E-mail
ABSTRACT

In this study, macroeconomic determinants of immigration are analysed with pooled cross-country and time series data on skilled immigration into Malaysia from 39 countries during 1998-2004. Results demonstrate clearly that both high frequency (such as differences in cyclical economic conditions) and low frequency determinants (such as differences in levels of economic development) have been important for determinants of immigration into Malaysia. Immigration is determined by the income growth of Malaysia, change in Malaysian unemployment rate and population of the source countries. Year-to-year changes in skilled immigration flows are dominated by economic fluctuation.

Services
Related Articles in ASCI
Similar Articles in this Journal
Search in Google Scholar
View Citation
Report Citation

 
  How to cite this article:

Chuie Hong Tan and A. Solucis Santhapparaj, 2007. Macroeconomic Determinants of Skilled Labour Migration: The Case of Malaysia. Journal of Applied Sciences, 7: 3015-3022.

DOI: 10.3923/jas.2007.3015.3022

URL: https://scialert.net/abstract/?doi=jas.2007.3015.3022

INTRODUCTION

Malaysia is an important destination country for skilled migrants, particularly from South Asia and East Asia. Its political stability, better infrastructure and consistent economic growth create a pull of migrants towards the country.

The microeconomic determinants of migration (at the individual and family level) are well understood, yet as Ghatek et al. (1996) observe, there are few analyses at the macroeconomic level where immigration policy is typically made. The common macroeconomic determinants used were job availability (Stark and Taylor, 1991), unemployment rate (Hatton, 1995), wage and per capita income (Greenwood, 1985; Hugo, 1998; Timmer and Williamson, 1998).

Drawing on Malaysia’s recent experience with immigration, this paper analyses both low and high frequency factors on the patterns of skilled immigration. Empirical studies of migration generally rely on either cross sectional or time series data for their observations (Todaro, 1969; Ghatek et al., 1996). Time series studies show that the flow of migration is affected by the business cycle (high frequency properties of the data) in both sending and receiving countries, while cross-country analyses highlight the importance of income levels and standards of living (low frequency properties). The approach taken here is to use time series data, thus maximizing the identifying power of our sample.

In Malaysia, the flows of the skilled migrant labour are associated essentially with the development of international business, transnational corporations and banks and the transfer of government officials and international civil servants of one type or another. According to the Malaysian Economic Report 2004/2005, there are about 1.3 million registered foreign workers as at July 2004, constituting nearly 12% of the total employment in Malaysia. Out of the total foreign workers, skilled foreign workers accounted for 3%. Foreign professional, technical and skilled workers working in Malaysia and earning RM2,500 per month and above are issued Employment Passes. In Malaysia, they are commonly referred to as expatriates. The main aim of this study is to evaluate potential magnitudes for these flows and to suggest effective policies for altering them in Malaysia.

MIGRATION AND ITS DETERMINANTS

Asia-Pacific migratory system is multipolar-that is the various receiving nations are not contiguous but widely scattered geographically. Massey (1998) identify four basic poles of reception in Asia and the Pacific: Australia, Japan and Newly Industrialized Countries (NICs) in South Asia and East Asia. Australia is the oldest and best developed in the migratory systems and like other traditional receiving countries immigrant origins have shifted markedly away from Europe and towards developing nations, mostly elsewhere in Asia and the Pacific. Japan began importing migrants during the 1970s, mainly from other nations in Asia but also from South America, where large numbers of Japanese had settled earlier in the 20th century. In the 1990s newly industrialized nations such as Korea, Taiwan and Hong Kong in the east and Singapore, Thailand and Malaysia in the south began to join the system. In the Asia Pacific migratory system, 81% of immigrants are originated in Asia, with around 7% each in the Americas and Oceania and 4% from the Middle East (Massey, 1998).

Non-coercive migration is the outcome of decisions made by utility-maximizing individuals in the context of both institutional and labour market constraints. There is clear empirical evidence that migration is positively related to regional differences in wages or per capita income and that the migration process is characterized by lags that are significant both statistically and substantively (Gould, 1979; Greenwood, 1985). These findings appear to be quite robust across time and space for many different data sets. Although migration also appears responsive to regional differences in unemployment rates, Hatton (1995) shows that the significance of these variables often comes at the expense of income or wages and vice versa, the reasons being undoubtedly related to the cyclical covariation of wages and unemployment.

Stark and Taylor (1991) argues that migration cannot simply be explained by income differences between two countries, but also by such factors as chances of secure employment, availability of capital for entrepreneurial activity and the need to manage risk over long periods. Migration decisions are made not just by individuals-they often represent family strategies to maximize income and survival chances (Hugo, 1998).

The conventional microeconomic model for explaining these patterns was developed by Sjaastad (1962) and widely used as a basis for further research such as Greenwood (1985) and Chiswick (1999). Immigrant labour supply depends on the decisions of utility-maximizing individuals and hence on their rates of return to migration as an investment, while the demand for immigrant labour depends on the decisions of profit maximizing firms and hence on its cost relative to native labour. The supply of immigrants is thus positively related to between-country differences in the relevant wage rates and negatively related to the costs of moving (defined broadly to include the costs of financing the move and the costs of adjusting to the destination country). The demand for immigrant workers is inversely related to the ratio of their wage relative to the wage of a comparable native worker and is also affected by institutional factors in the destination country (Greenwood, 1985).

A country’s macroeconomic health plays an important role in shaping immigration policy. Periods of economic distress are associated with moves toward restriction, whereas economic booms are associated with expansive policies. Analyses undertaken by Timmer and Williamson (1998) suggested that wage inequality-more than unemployment or absolute wage levels-may be the crucial factor in triggering restrictive policies. Second, immigration policy is sensitive to the volume of international flows: higher rates of immigration generally lead to restrictive policies. Third, immigration policy is associated with broader ideological currents, tending toward restriction during periods of social conformity and toward expansion during periods of support for open trade and periods of intense geopolitical conflict.

Macroeconomic determinants of migration may be derived from this basic theoretical framework. The main labour market explanatory variable for the supply of immigrant workers is the absolute difference in wages between the destination and origin countries, while for demand it is the ratio of immigrants’ to natives’ wage rates (Sjaastad, 1962). The main institutional factors affecting demand are expressed by host country restrictions (if any) on immigrant visas, while institutional factors affecting supply are dominated by restrictions on exit (and/or re-entry) in the countries of origin.

At the macroeconomic level, large differences in per capita aggregate income determine the direction of net migration flows. For two countries at the same level of development the differences in wages arise from occupation-specific labour market differences, typically stimulating migration in either direction depending on occupation. While these flows can sometimes be quite large, the two-way movement keeps aggregate net migration fairly low. For two countries at very different levels of development, however, large differences in the average wage typically reflect large differences in the same direction for most occupation and skill groups and thus stimulate a relatively large net migration flow from the low to high income country. International wage differentials are not the only factor motivating people to migrate. Massey (1998) found that most people displaced in the course of economic growth did not move hoping to reap higher lifetime earnings by relocating permanently to a foreign setting.

Additionally, the economic incentive to migrate is greater if costs of moving are relatively low and if adjustment at the destination is less costly. Moving costs can be high among the very poor with their limited access to finance, leading to the paradox that people with the most to gain sometimes seem to be least likely to migrate (Ghosh, 1992). Costs are also lower and thus, greater movements are expected between countries that share a common border. Within the country of origin, persons who speak the destination country language, or have friends and relatives already living there, would have lower costs of adjustment and thus be disproportionately represented among migrants. Adjustment costs for a migrating family can also be kept down by sequential timing of the actual move, beginning with one or two members for whom costs are relatively low so they can be in a position to facilitate the adjustment of later immigrants. A similar mechanism leads to chain migration, where immigrant flows tend to be higher between countries where there is already a large stock of origin country immigrants in the destination (Levy and Tsur, 2003).

Between-country differences in the level of economic development change only slowly over time and, together with country-specific (cultural and social) factors affecting the costs of moving and adjustment, establish the long run supply and demand for immigrant labour in the destination country. However, changes in wages and unemployment associated with economic cycles affect the optimal timing of migration by influencing rates of return in the relatively short run (Todaro, 1969). During periods of perceived labour shortage in the destination, upward pressure on domestic wages reduces the relative wage of immigrants even as it raises their absolute wage level and raises the rates of return to emigration from all source countries. Thus destination-country demand for immigrant labour is generally pro-cyclical, shifting upward with rising economic activity and declining unemployment. The immigrant labour supply is also responsive to conditions in the country of origin, although these (counter cyclical) effects tend to be somewhat smaller. Indeed, if low economic growth and high unemployment in the country of origin raises the wage differences with other countries, people are unlikely to move due to financial difficulty.

While labour market conditions go a long way toward explaining migration incentives for both employers and employees, actual immigration flows are also affected by the non-market restrictions associated with public policy. Visa conditions or entry quotas in the destination country are barriers limiting the effective demand for immigrant labour, while exit restrictions in the origin country limit the effective supply. As with any restricted market, the underlying supply and demand relationships remain latent even if non-market barriers are strictly enforced. Even in extreme cases where net flows appear to be dominated by political restrictions, economic incentives affect the costs of moving in an underground, black market, or otherwise illegal immigration as well as pressure for changes in immigration policy (Straubhauer, 1993).

By the 1980s, several Asian Tigers had joined the ranks of wealthy, industrialized nations. In addition to Japan, which had become one of the world’s dominant economic powers, Taiwan, South Korea, Hong Kong, Singapore, Thailand and Malaysia achieved stunning rates of economic growth during the 1970s and by the 1980s they too had become intensive in capital but poor in skilled labour (Hugo 1995). Like the countries of southern Europe during the 1970s, many switched from exporting to importing labour, while others continued simultaneously to import and export skilled workers.

MATERIALS AND METHODS

This study is concerned with the determinants of skilled migration into Malaysia from 39 countries during the period 1998-2004. The dependent variable Mi is the number of skilled immigrants to Malaysia from country i in year t. By pooling time series data for 39 countries, we can discriminate between the effects of migration determinants associated with low frequency phenomena (such as differences in levels of economic development) and those associated with relatively high frequency phenomena (such as differences in cyclical economic conditions).

The following data series were obtained for each origin country i:

= Skilled immigration from country i to Malaysia at year t
= GDP per capita of country i in year t
= Literacy rate in country i in year t
= Population of country i in year t
= Malaysian unemployment rate year t
Dt = Year dummies

Data for all variables were obtained entirely from published sources. is measured as annual skilled immigrants from each origin country i to Malaysia at year t. This information is collected from the Malaysia Immigration Department. Immigrants to Malaysia from ASEAN countries and also countries in which more than 100 skilled force working in Malaysia are considered in the analysis. is GDP per capita of country i in year t. The data is retrieved from ASEAN Statistical Yearbook (Anonymous, 2002) and The World Bank Group (http://www.devdata.worldbank.org). The measure for is the literacy rate and is population (in thousands) for each origin country i. Both information are gathered from the World Bank Group. is the unemployment rate for Malaysia, published by the Numerical Information Book (Statistics Handbook), Malaysia, (Anonymous, 2001a) and key statistics in the website http://www.statistics.gov.my. Dt is year dummies, which is used to capture macroeconomic factors that affect Malaysia’s immigration.

Table 1: Number and percentage of skilled immigrants to Malaysia by region/country and year
Source: Malaysia Immigration Department in which the data is collected in April 2005. SEA: Brunei, Cambodia, Indonesia, Myanmar, Philippines, Singapore, Thailand and Vietnam. SA: Bangladesh, India, Mauritius, Nepal, Pakistan and Sri Lanka. EA: China, Japan, Korea and Taiwan. EU: Austria, Belgium, Denmark, Finland, France, Germany, Ireland, Italy, Netherlands, Norway, Russia, Sweden, Switzerland and United Kingdom. ME: Egypt, Iran and Iraq

For the remainder of this article, the superscript M indicates Malaysia and i indicates any of the various origin countries in the sample. The following additional variables were constructed from the above data:

Table 1 shows that while the share of skilled immigration into Malaysia from South Asia, East Asia and Middle East increases by 7.02, 1.70 and 0.25% respectively, skilled immigrants from Europe, USA, Canada, Australia and New Zealand continue to decrease to less than 20% after year 1998. The recent wooing of IT experts from India to Malaysia’s Multimedia Super Corridor (MSC) has indeed drawn a number of immigrants from South Asia. Out of the 1,100 MSC-status companies, 45 are from India and the majority of them are focused on software development activities.

Figure 1 shows a more detailed picture of migration flows into Malaysia during the period covered by this study. Beginning with the effects of high frequency phenomena, time series data for skilled immigration from all 39 countries (Mt = ΣtMij), alongside a time series for the Malaysian growth rate, gMt (Fig. 1). After the 1998 recession the number of skilled immigrants is reduced by 50% in year 2004. This could be due to the supply of local skilled and trained manpower has increased as a result of a capacity expansion in education and training institution in Malaysia.

The two variables are synchronous for the whole period; for example, the rate of unemployment changes inversely with number of immigrants.

Fig. 1: Skilled immigration and Malaysian economic growth rate. Source: Malaysia Immigration Department (Anonymous, 1993, 2001, 2002, 2003)

Fig. 2: Skilled immigration and Malaysian unemployment rate. Source: Malaysia Immigration Department, Anonymous (2001a) and key statistics in the website http://www.statistics.gov.my

Although Malaysia’s growth and unemployment rates continue to move together over the cycle, immigration flows lag these rates by a year or two. Thus, for example, the 1998 recession is not reflected by decrease in immigration until 1999. Despite the economic crisis in 1998, employment growth was impressive due to the quick recovery of the economy, particularly the manufacturing sector. Unemployment was also contained at a low rate due to the implementation of effective labour market measures. Malaysia is able to maintain the unemployment rate of 3.1% in 2000 and it is expected to reduced to 2.7% in 2005 (Anonymous, 2001b). However, the rate of unemployment was 3.5% in 2004 (Fig. 2).

Overall effects of low frequency factors on migration into Malaysia during period of year 1998-2004 may be seen in Fig. 3. In each of these figures the average annual skilled immigration into Malaysia from country i, Mi, is plotted against the ratio of Malaysia’s to country i’s average per capita income, (YM/Yi). The data are plotted for all 39 countries in the sample.

With no Malaysian restrictions on entry and active recruitment of skilled workers, Fig. 3 indicates a negative relationship between relative per capita income and migration. Countries substantially poorer than Malaysia (e.g., India, Indonesia and Philippines), had high average migration. Similarly, countries like Japan, Singapore, United Kingdom, USA, Taiwan and Republic of Korea, which had higher income also contribute skilled migrants to Malaysia. After India and Japan, a high number of skilled migrants come from China which shares a comparable income with Malaysia.

Among immigrants to Malaysia during this later period, considerations of kinship and community also seem to have been an important factor. An established community of earlier immigrants from the same origin country can not only help with employment; it can also provide amenities to short run migrants for whom large investments in host country consumption activities are not attractive. Similarly, reunification with close relatives who previously migrated to Malaysia might be the implementation of an earlier migration decision based on long run factors.

Fig. 3: Average skilled immigration and average income ratio: 1998-2004. Source: Malaysia Immigration Department, Anonymous (1993, 2001c, 2002, 2003)

Differences in the level of economic development generate wage differences that are large relative to both cyclical variations and the cost of moving. A long run immigration decision is thus relatively unaffected by short run differences, although its implementation can be timed to take advantage of favourable employment opportunities in the destination country. Coefficients of the push and pull factors thus display symmetry when the relevant wage differentials are much smaller and more sensitive to cyclical variations.

EMPIRICAL RESULTS

Table 2 reports regression results from pooled cross section and time series data for the sub-sample of 39 countries.

Table 2: Determinants of Malaysian skilled immigration from 1998-2004. Dependent variable: No. of skilled immigration from country i year t
(a): OLS (no fixed effects); (b): Fixed effects; ***: Significant at 1% level; **: Significant at 5% level; *: Significant at 10% level

(1)

The estimated model is where, c, ρ, β, γ, δ, φ and λ are parameters to be estimated and εit is the error term. For fixed effects specification we model εit = ωi + υit, where the ωi are the fixed effexts and υ is assumed to be the error term.

Low frequency phenomena are measured by variables representing differences in average income, YM/Yi, literacy rate, lit and population of source countries, Pit. High frequency phenomena are measured by the annual rates of growth of per capita GDP for Malaysia, gMt and for the various origin countries, git and also the change in Malaysian unemployment rate, Δuit. A one-period lagged dependent variable controls for the effects of implementation delays of staggered migration and a one-period lagged of Malaysian economic growth rate.

The results of the first two columns (Model 1) in Table 2 suggest that both long and short run market forces influence immigration into Malaysia during 1998-2004. The OLS estimates in column (a) suggest that while Malaysia’s growth rate is a significant determinant of skilled immigration into Malaysia, none of the relative per capita incomes and the growth rates of the other origin country variables have any significance. Volume of population of the origin countries is highly significant, as is the lagged dependent variable. Change in Malaysian unemployment rate has a positive coefficient and is significant at 0.1 level. This suggests that while long run factors may have dominated migration decisions, the pace of Malaysian economic activity influenced the timing of their implementation to coincide with favourable employment prospects. Thus the push represented by the mass population of the origin countries may dominate for the sources of skilled immigrant flows, but the pull of favourable economic conditions in Malaysia may be the dominant influence on the timing of immigration events.

The fixed effects specification in column (b) reinforces the inferences drawn from Equation a. The coefficients of Malaysian income growth rate as well as the lagged dependent variable are significant. The introduction of country-specific dummy variables reduces the coefficient of population, Pit to insignificance. This suggests that population either dominates or is highly correlated with other relevant low frequency phenomena, both observed and unobserved. In contrast, literacy rate of the origin countries becomes significant and has a positive coefficient. The higher level of literacy rate of the origin country will generally produces more number of skilled workers. This will result a greater volume of skilled immigrants received from Malaysia from the country, such as Japan, Singapore and United Kingdom. Malaysian change in unemployment rate shows a positive coefficient and the significance is increased to 0.05 level.

A one-period lags of gMt is introduced to allow for a richer lag structure in terms of the Malaysian income growth variable, again for specifications with and without fixed effects. The results are shown in Model 2, the last two columns. The lagged for Malaysian income growth, gMt-1 has a positive sign and is statistically significant in both models with and without fixed effects. The coefficient of the variable change in the rate of Malaysian unemployment rate over the previous year shows a positive sign and is also highly significant for both models. This suggests that the effects of gM on skilled immigration to Malaysia since 1998 have been long lasting, together with the favourable employment prospects, a fact consistent with our earlier discussion and with Fig. 1 and 2.

In the fixed effects, the coefficients of each country for both models are all negative. This implies that the number of skilled immigrants into Malaysia is decreasing during the period 1998-2004. Malaysia is in the process of reducing the dependence on skilled foreign labours. With more than 40% of the people below the age of 25 and a literacy rate of 95.1%, Malaysia’s pool of local knowledge workers is expected to grow in coming years. The importance of education as a tool to ensure a quality labour force is underscored by the fact that the number of public and private universities had been increasing and established to around 25 in 2004 from 9 public universities in 1996.

CONCLUSIONS

By pooling cross-country and time series data on skilled immigration into Malaysia from 39 countries during 1998-2004, macroeconomic determinants of immigration has been analysed. The results in this study demonstrate clearly that both high and low frequency determinants have been important for determinants of immigration into Malaysia. While year-to year or short run changes in skilled immigration flows tend to be dominated by cyclical economic conditions, long run trends are determined by income growth of the receiving country, Malaysia. Besides maintaining a political stability, its good prospect in employment market also draws the skilled migrants to Malaysia. Countries with higher volume of population show a significant determinant in the outflow of skilled migration into Malaysia. Only by pooling the data can this distinction be appreciated.

Apart from economic development in the both sending and receiving countries, two other policy instruments for increasing skilled migration are possible: entry relaxation in the receiving country and attractive incentives to skilled immigrants. Skilled migration inflows have been decreasing in responsive to short run economic fluctuations. However, market pressure for skilled migration from low income countries can be expected to persist or even increase, raising substantially the number of skilled immigrants to Malaysia.

REFERENCES
1:  Anonymous, 1993. World Development Report 1993 World Bank. Oxford University Press, London.

2:  Anonymous, 2001. Numerical information Book. Statistics Handbook. Government of Malaysia.

3:  Anonymous, 2001. 8th Malaysia Plan 2001-2005. Malaysia National Printing Bhd., Kuala Lumpur, Malaysia, pp: 92.

4:  Anonymous, 2001. World development indicators 2001. World Bank, Washington ,DC.

5:  Anonymous, 2002. ASEAN statistical yearbook 2002. ASEAN Secretariat, Jakarta.

6:  Anonymous, 2003. World Development Report 2003 World Bank. Oxford University Press, London.

7:  Anonymous, 2005. Economic report 2004/2005. Ministry of Finance Malaysia, Government of Malaysia, pp: 218.

8:  Chiswick, B.R., 1999. Immigration policy and immigrant quality: Are immigrants favorably self-selected? Am. Econ. Rev., 89: 181-185.

9:  Ghatek, S., P. Levine and S.W. Price, 1996. Migration theories and evidence: An assessment. J. Econ. Surveys, 10: 159-198.
Direct Link  |  

10:  Ghosh, B., 1992. Migration-development linkages: Some specific issues and practical policy measures. Int. Migration, 30: 423-456.
Direct Link  |  

11:  Gould, J.D., 1979. European inter-continental emigration 1815-1914: Patterns and causes. J. Eur. Econ. History, 8: 593-679.

12:  Greenwood, M.J., 1985. Human migration: Theory, models and empirical studies. J. Regional Sci., 25: 521-544.

13:  Hatton, T.J., 1995. A model of UK emigration, 1870-1913. Rev. Econ. Stat., 77: 407-415.
Direct Link  |  

14:  Hugo, G., 1995. Illegal Migration in Asia. In: : The Cambridge Survey of World Migration, Robin, C. (Ed.). Cambridge University Press, Cambridge, pp: 397-402.

15:  Hugo, G., 1998. The demographic underpinnings of current and future international migration in Asia. Asian Pacific Migrat. J., 7: 1-25.
Direct Link  |  

16:  Levy, A. and Y. Tsur, 2003. Migration Timing: Expected Returns, Risk Aversion and Assimilation Costs, Economic Growth, Inequality and Migration. Edward Elgar Publishing Limited, UK., pp.188.

17:  Massey, D.S., 1998. International migration and ecoomic cooperation and development in comparative perspective. Popul. Dev. Rev., 24: 383-414.

18:  Sjaastad, L.A., 1962. The costs and returns of human migration. J. Politic. Econ., 70: 80-93.

19:  Stark, O. and J.E. Taylor, 1991. Migration incentives, migration types: The role of relative deprivation. Econ. J., 101: 1163-1178.
Direct Link  |  

20:  Straubhauer, T., 1993. Migration pressure. Int. Migrat., 31: 5-38.
Direct Link  |  

21:  Timmer, A.S. and J.G. Williamson, 1998. Immigration policy prior to the 1930s: Labour markets, policy interactions and globalization backlash. Popul. Dev. Rev., 24: 739-771.
Direct Link  |  

22:  Todaro, M.P., 1969. A model of labor migration and urban unemployment in less developed countries. Am. Econ. Rev., 59: 138-148.

©  2021 Science Alert. All Rights Reserved