Immigrant labor does not have any effects on capital investment, business closing, and offshore investment, either. Finally we discuss policy implications of the.
The Impact of Temporary Immigration of Unskilled Workers on Firm Performances: Evidences from the Korean Small-Medium Business Sector
Jai-Joon Hur* and Yongjin Nho**
Using the Korean Workplace Panel Data collected by the Korea Labor Institute, we investigate the impact of temporary unskilled migrant workers on firm profitability in the Korean small-medium business sector. We also examine the sources of the profitability impact, and whether employment of the migrant workers leads to changing capital investment behavior of the firms. We draw our theoretical model on Borjas (1995) and estimate first-difference models of panel data. Our findings are as follows. The influx of unskilled migrant workers has a positive impact on firm profitability. The immigrant labor lowers wages, and increases total employment, which substitutes for capital, leading to lower capital-labor ratio. These results indicate that the increased employment and the lowered capital-labor ratio are mediating factors to account for its effect on firm profitability. By contrast, we find no evidence of a scale effect or productivity gain of employing migrant workers. Immigrant labor does not have any effects on capital investment, business closing, and offshore investment, either. Finally we discuss policy implications of the empirical results.
I. Introduction II. Institutional Framework and Recent Patterns of Immigration in Korea III. Models of the Firm Performances Impact of Temporary Immigration 1. Theoretical Framework 2. Empirical Approaches IV. Empirical Results 1. Data and Descriptive Statistics 2. Regression Results V. Conclusions
2
I.
Introduction
The Korean immigration policy has been undergoing gradual, but significant changes over the last two decades. Traditionally the Korean government had a conservative stance toward immigration, but it has moved to relax restrictions on immigration by introducing the Industrial Trainee Scheme and then the Employment Permit Scheme, facing with urgent calls for labor shortages of SMEs and also concerned about the increasing number of illegal immigrants. In a sense it has been a process of adjusting its outdated regulations to the underlying market forces. Illegal immigrants might impose greater social costs on the host country, such as smuggling of immigrants, migration-related corruptions and human rights violations. Certainly it has not been easy for the Korean government to contain the number of immigrants just by regulation, under the condition that there have existed persistently strong demand for and supply of the cheap foreign labor. The results of the policy changes are quite substantial. The number of the unskilled migrant workers is, including the illegal, about 680,000 in 2009, increasing by about 37,000 workers annually since 2004 when the Employment Permit Scheme began to be effective. Its proportions out of the total workforce and the total wage-earners are about 2.9 percent and 4.1 percent respectively. The proportion of migrant workers is not so high, as compared to other OECD countries. However, it is not a negligible portion of workforce, especially in the unskilled labor market which is more relevant to the issue studied in this paper, because the migrant workers are mainly composed of unskilled workers in Korea. For example, its fraction out of the unskilled production workers is about 7.2 percent in 2009. It is interesting to note that the number of illegal immigrants has not been much reduced, though not increased, even after the Employment Permit Scheme was enacted. With the amnesty given to the illegal immigrants in 2003, the number of illegal immigrants was fallen by half to 154,342, but it got gradually increasing again to 223,464 in 2007. Since there are few reasons for the increasing trend of migrant workers to be reversed, the proportion of migrant workers will probably be rising afterwards, too. Taking into account the trend of lower fertility rate, higher life expectancy, and higher rate of people with postsecondary education, the proportion of immigrants is likely to be higher. Considering the increasing importance of immigrants, the economic impacts of temporary immigration in Korea merits an examination. To assess the economic impacts of migrant workers, this paper will address the issues of how the influx of migrant workers affects the firm performance of migrant-using firms, what are the mediating factors for their performance impacts, and whether the expected performance gain out of migrant workers increases capital investment of the firms and leads to a scale effect. For this objective, we will attempt to empirically investigate those issues in the Korean small-medium business sector. We will employ firm-level data analysis, for the numbers of migrant workers are varied across firms in Korea1. 1. The Korean immigration arrangements stipulate that most immigrants should be granted a temporary employment permit, conditional on being employed by a specific employer and that their transfer to another employer is not permitted without one of the good justifiable reasons specified in the law. The ceiling of migrant workers employable to a firm depends on the size of the firm, but does not exceed fifty and ten persons in manufacturing and service industry, respectively. For example, in manufacturing, five migrant workers are the maximum for firms with 1-10 native workers, and ten 3
The performance indicators examined here are profitability as the final outcome variable, and labor productivity, wages, employment, labor shortage, tangible fixed asset, and capital-labor ratio as its intermediary variables. We will also examine sales, assets, business closing, and offshore investment to check if there is a scale effect of employing migrant workers. To our knowledge, there was no previous research on the firm profitability impact of temporary immigration in Korea. Since Korea has a short history of immigration, there was little opportunity to investigate the economic impact of immigration. When the influx of foreign workers is at its inception, its researches are mainly policyoriented concerning designing or redesigning its system (Böhning, 1996; Park, 1996; Yoo and Lee 2002). Since Employment Permit Scheme was introduced, the researches are extended to the labor market outcomes effect of temporary immigration (Gwon et al. 1995; Cho 2004; Yoo et al. 2004; Nho and Lee 2004; Jung 2005; Hahn and Choi 2006; Lee, Lee and Park 2007; Hur 2010a). Recently there were a couple of studies on the output effect of immigration, but at the national level (Kim and Lee 2004; Lee, Lee and Park, 2007). Kim and Lee (2004) identified, based on input-output tables, that employment of foreign workers increases the national output by KRW1.04 trillion and employment of native workers by 3,077 persons. Lee, Lee and Park (2007), employing the CGE model, found that when the percentages of foreign workers out of the total workforce are 2% and 4%, they will increase the total output by 0.379% and 0.513%, and real GDP by 0.171% and 0.177% respectively. Lee (2004) calibrated the social cost and benefit of immigration by the contingent valuation method, and identified that its benefit outweighs its cost, with its net social benefit amounting to KRW340 billion a year. The firm performance impacts of immigration are rarely examined in the previous studies in the other countries either, even if there are a fair amount of researches on its labor market outcomes effects. More or less relevant to this paper are Quispe-Agnoli and Zavodny (2002) and Howland and Nguyen (2009), which investigated empirically the impacts of immigration on capital investment, output mix and productivity for the former, and employment growth of some labor intensive industries for the latter. However, both papers examined the topics at regional industry levels, not at the firm level, unlike this paper. Therefore this paper offers a fairly unique contribution in analyzing the firm performance impact of migrant workers and as a comparative study across countries. There are two reasons for rare research on the firm performance effects of immigration. One is a theoretical prediction that the firm performance effects of immigration are closely related to labor market outcome impacts, and the other is that firm level immigration data are less available than labor market data in most migrantreceiving countries. Because it is argued that both firm performance effects and labor market impacts of immigration are “intrinsically related,” analysis of either aspect can directly imply the other (Freeman 1998). More specifically, it is predicted that benefits accruing to the employers are not only composed of the surplus which immigrants themselves provide, but also are arising out of wealth transfer between capital and labor created by immigration. Now that immigrants tend to take a small portion of the total workforce in the host countries, the redistributed wealth between
migrant workers are for those with 11-50 natives. See Appendix 1 and Appendix 2 for other sizes in manufacturing and service firms. 4
the natives, rather than the so-called immigration surplus, comprises a dominant portion of employers’ immigration benefits (Borjas 1995). The redistribution of wealth as a main source of employer benefits is attributable to the increased labor supply and consequently the lowered wages as labor market variables affected by immigration. It implies that the benefits of employers are closely related to the loss of wealth for the employees due to immigration. So we can examine either one with direct implications for the other, but the labor outcome impacts are more widely investigated, since data for the latter are more readily available (Freeman 1998). Additionally, it is predicted that the firm performance effects of immigration are indirect ones, inherently mediated by labor market outcome effects and consequently the former are more readily presumed. The argument is reinforced by the fact that employers typically champion immigration. Nevertheless an examination of the firm performance impacts of immigration is useful. Above all, it is a missing link to get an overall picture of the economic effects of immigration. Even if it is predicted theoretically that the firm performance effects and the labor market outcome effects of immigration are closely related, the separate independent empirical studies can help understand the economic impacts of immigration better, because the relationship has never been tested empirically. Empirical results are often inconsistent with its theoretical predictions, for theoretical predictions tend to be based on a restricted model with strong assumptions while all the exogenous variables are not fully controlled for in empirical models. This may be the case for research on the economic impacts of immigrations. Even for the effects of immigration on wages or employment of the natives, not to mention its firm performance effects, no strong evidences are found in the previous studies (Altonji and Card 1991, Borjas 1987, Borjas 1998, Butcher 1998, Card 1990, Grossman 1982, LaLonde and Topel 1991). Rather, studies on firm performance effects of immigrants may help fill the gap between the empirical results and the theoretical predictions for labor market outcome impacts of immigration, because inconsistency between the theoretical predictions and the empirical results can be reconciled by a scale effect of immigration. Improved organizational performances may induce firms to invest more capital and demand additional labor, and they may absorb and neutralize the adversary impacts of immigration on labor market outcomes (Howland and Nguyen 2009, Quispe-Agnoli and Zavodny 2002). Methodologically this paper pays attention to the endogeneity, possibly involved in estimating the firm performance impacts of immigration. Employing migrant workers is determined through the two steps in Korea. Firstly, government determines the sector within which the firms can employ the unskilled migrant workers, and secondly, individual firms decide whether to employ the unskilled migrant workers. In both steps, the degree of labor shortage for the unskilled workers at the industry level and at the firm level may be a determining factor. This being the case, the estimates of the firm performance effects may be biased negatively, since labor-short employers who are eligible to employ unskilled migrant workers are more likely to have lower performances initially. Labor shortage, which is the single most important criterion of determining whether to employ migrant workers in Korea, is so often associated with lower wages, longer working hours, and lower safety, and these inferior working conditions, in turn, can be attributed to lower firm performances. This paper will employ panel data analysis to account for the firm-specific initial status of firm performance, assuming that it is time-invariant. The remainder of this paper is organized as follows. Section II will briefly 5
introduce the institutional framework of immigration and its recent trend in Korea, and section III will present the analytical framework of this paper based on literature review of the economic effects of immigration. Section IV will describe data used in this paper and will discuss basic statistics of the main variables and the empirical results of this study. Finally section V will draw conclusions.
II. Institutional Framework and Recent Patterns of Immigration in Korea The institutional arrangements of immigration in Korea have undergone two stages of development: the Industrial Trainee Scheme and the Employment Permit Scheme. Additionally there have been more liberal special immigration programs for the Korean ethnic groups of migrant workers, mainly composed of the Korean Chinese and the Korean Russians in the Commonwealth of Independent States.2 The Industrial Trainee Scheme was to allow the eligible firms to employ migrant workers as ‘trainees’, not employees. Its predecessor was introduced in 1991 for the Korean firms with direct investment overseas so that they could bring in and provide training for migrant workers of their overseas subsidiaries for up to one year. In 1993 the Korean government expanded the foreigner training program to the firms employing less than 300 full-time employees, which is the most frequently used definition of SMEs in Korea, in the manufacturing sector, responding to their labor shortages, and the expanded new arrangement has been officially called the Industrial Trainee Scheme. Foreign industrial trainees were basically covered by the same labor laws applied to Korean workers, but since they were ‘trainees’, they could be paid 80% of the minimum wage during the first three months. They were not eligible for severance pay even when they had worked for over a year.3 Under the Industrial Trainee Scheme, migrant workers could stay up to two years. Thereafter it was modified several times. Its coverage had been extended to the industry of coastal fishery in 1996 and also to the construction industry in 1997. The maximum duration of stay for the migrant workers was lengthened to three years (one year for training and two years for employment) in 2002. It started with a hope that it will help overcome labor shortages and also save labor costs of SMEs without undermining the native labor market. To protect the native labor market and to help SMEs save labor costs appear to be two conflicting objectives, but the Korean government attempted to balance these two objectives by restricting annual trainee quota. However, since the demand for migrant workers was far higher than the supply, unexpected things came out. Many migrant workers quit their workplaces illegally for higher wages and other better working conditions in other employers, and thus Korea saw the number of illegal immigrants increasing drastically. At that point, illegal immigrants were estimated to earn 29.9%(male) and 22.8%(female) more than the industrial trainees in 2003 (Yoo et al 2003). 2 . For more detailed information regarding the transition from Industrial Trainee Scheme to Employment Permit Scheme and related institutional changes, please refer to Hur(2010) in this volume. 3. Later trainees became eligible for severance pay by the court’s ruling that even ‘trainees’ should be paid severance pay as far as they are de facto employees. The Korean labor law stipulates that any business employing five or more workers should pay any worker employed for over a year, a month’s wage or salary for each working year, that is 8.33% of the wage or salary as severance pay, when they quit, regardless of reasons of separation. 6
Figure 1 shows the number of illegal immigrants was growing until 2002. The drastic drop of illegal immigrants in 2003 is attributed to the amnesty provided at that time, and also the number of illegal immigrants thereafter has been generally contained with a slightly growing trend until 2007 and thereafter it is decreasing. Nevertheless the proportion of illegal migrant workers to the total foreign workforce employed in Korea is 26.6% in 2009.
Insert [ Figure 1] around here.
Enacted in August, 2003, and put into effect in August, 2004, the Employment Permit Scheme was conceived as a response to the criticism about the Industrial Trainee Scheme. It maintained the attribute of temporary immigration with the maximum period of stay being three years, but in this case as an employee, not as a trainee. The migrant workers have the right of not being discriminated in terms of wages. They are protected by the minimum wage rule, and they are eligible for severance pay. It is determined annually by ‘Migrant Worker Policy Committee’ which sectors are entitled to employ migrant workers, and the scope of sectors allowed has been changed just slightly. The decision-making is based upon the degree of labor shortage in the sector and also the degree of skill requirements to be satisfied. The Employment Permit Scheme has a similar coverage of employers as before: firms employing less than 300 full-time employees in the manufacturing sector, the agriculture and stockbreeding industry, the construction industry, and the business of collecting and selling of recyclable materials, refrigerated storage business and hotel business in the service sector. Employers in the service sector such as restaurant business, housekeeping, cleaning, and nursing are allowed to employ migrant workers under the system, but only those migrant workers with the Korean ethnicity. Under the Employment Permit Scheme, an employment permit is issued for a migrant worker under the condition that he/she has been hired by a specific employer. More specifically, a permit of employing migrant workers is provided first to the employer, when he/she fails to find a suitable native worker despite his/her effort through the public Employment Service Center. Then the Employment Service Center refers to the employer as three-fold foreign job applicants as demanded through jobbroking activities, and only those migrant workers selected by the employer can get their employment permit. Thereafter the migrant workers can still get a permit of changing employer up to three times in the case of justifiable reasons stated by the law such as the temporary or permanent closure or bankruptcy of the establishment, the employer’s refusal to renew the employment contract or contract termination, and failure to pay wages or violence. Any employer should purchase departure guarantee insurance or lump-sum trust for its migrant workers, and that migrant workers themselves should buy insurance or trust for their return cost. The special program for the Korean-ethnic foreigners is more liberal than the program for the non-Korean ethnic group. According to the Working Visit Permit Scheme, introduced in 2007, Korean-Chinese and Korean-Russians of the formers Soviet Union Republics are granted five year visa, and once entered they can stay up 7
to five years. They can be admitted before being employed and freely to enter and exit the country during the period of stay. They are also free to change their employer, and their scopes of employment allowed are wider than non-Korean ethnic migrant workers, including not only manufacturing, construction, agriculture and stockbreeding, but also the service sector such as restaurant business, nursing, cleaning, housekeeping, social welfare services, and business support services. With the increasingly liberalized modifications of immigration arrangements, the number of migrant workers grew by 260,000 persons between 2005 and 2007, as is shown in Figure 1. The recent growth is mainly due to the increasing inflow of Korean-ethnic immigrants. Around one third of the total immigrants are female, and the composition of immigrants by gender is generally stable over time.4 The percentage of migrant workers including professional, non-professional, and illegal workers in the total employment is 2.9% in 2009, same as in 2008. The distribution of migrant workers across sectors is not available officially. The information summarized in Figure 2 and Figure 3 was constructed from official data and estimates based on the initial jobs at the time of entry, ethnic groups, the proportion of illegal immigrants, and the sectoral distribution of illegal immigrants in 2003 by Lee (2008). The proportion of migrant workers to the total workforce within a sector is highest in the housekeeping service, accounting for 16.3% of the total workforce within the sector in 2008, though the sector is not included in this study. The construction industry and the manufacturing industry have the second and third highest proportions of migrant workers, explaining 9.1% and 6.0% of the total workforces within the sectors in 2008 respectively. On the other hand, the greatest proportion of migrant workers by sector is employed in the manufacturing sector. The percentage of foreign workers employed in the manufacturing sector peaked at 47.3% of the total foreign workforce in 2005 and declined to 36.7% in 2008. The growth of migrant workers in the manufacturing sector has been relatively slow, as compared to other sectors. The second highest proportion of migrant workers is working in the construction industry. The proportion of migrant workers in the construction industry, which witnessed fast growth, to the total migrant workers was 26.0% in 2008.
Insert [Figure 2] and [Figure 3] around here.
III. Models of the Firm Performances Impact of Temporary Immigration 1. Theoretical Framework To investigate the firm performance impact of immigration, we draw our analytical framework upon the previous theoretical models of its labor market outcome impact well documented in literature, especially Borjas (1995) and Friedberg and Hunt (1995). The determining factor of the economic impact of migrant workers in the 4. According to Lee (2008), the percentage of female migrant workers was 32.9% in 2004, 30.6% in 2005, 31.1% in 2006, 32.1% in 2007 and 33.5% in 2008, respectively. 8
models lies in the extent of substitutability between migrant workers and the other production factors. The influx of migrant workers increases the supply of labor of the same kind, dropping its wages. It will lower the price of production factors with which it is perfect substitutes, and raise the price of production factors with which it is complements. For example, the wage rate of unskilled native workers as perfect substitute of unskilled migrant workers will be adversely affected by the influx of unskilled migrant workers. On the other hand, the wage rate of skilled workers, as long as it is complement with the unskilled immigrants, will be raised. The impact of migrant workers on the price of its imperfect substitutes is ambiguous because the substitution effect competes with scale effects. The influx of unskilled migrant labor and the subsequent lowered wages of the unskilled labor will reduce the demand for its substitute factors, but at the same time it will increase the demand for all inputs as the optimal output level becomes higher. Therefore, the wage of the unskilled native labor may be not much adversely affected by the influx of unskilled immigrants when the scale effect is large. This was conjectured to be a good reason for the weakly negative impacts of the unskilled immigrants estimated in the previous empirical studies (Howland and Nguyen 2009). The relation between the unskilled immigrants and capital, which is an important point of interest in this paper, is likely to be characterized as partially substitutes, thus allowing us to predict that the impact of migrant workers on capital returns is ambiguous. The different economic impacts of migrant workers across production factors imply that wealth transfer may occur among the owners of the production factors. Economic benefits accrue to the advantaged groups due not only to the benefits directly provided by immigrants, but also to wealth transfers from the adversely affected groups (Borjas 1995, Freeman 1998). Separating wealth transfer from direct benefits named as immigration surplus, Borjas (1995) claims that the wealth transfer is much greater than the immigration surplus in the United States. Distinction between wealth transfer and immigration surplus is especially important in this paper, for the firm performance impact of migrant workers can be accounted for more plausibly in terms of both direct benefits and the redistribution of wealth due to their use. To illustrate the firm performance impact of migrant workers, let’s set up an aggregate production function Q at the national level as follows. It is assumed that the production function has three inputs, capital K, skilled labor Ls and the unskilled labor Lu. The unskilled labor is composed of unskilled native workers Du and the unskilled migrant workers Fu. It is also assumed the unskilled natives and the unskilled immigrants are substitutes, and that the skilled natives and the unskilled immigrants are complements. The price of the output is the numeraire and assumed to be unitary. Returns to capital and wages for the skilled and unskilled workers are indicated by r, ws and wu respectively. All the outputs are assumed to be distributed within the nation. ,
(1)
where
.
By differentiating equation (1) with respect to the unskilled labor Fu, we can get a simple depiction of the firm performance impact of migrant workers. We assume here that the amount of capital K and supplies of the skilled workers Ls and unskilled native workers Du are fixed and thus inelastic for their prices. 9
, we obtain
Since
(2)
=
The left-hand side indicates the amount of changes in capital return resulting from the influx of unskilled migrant workers. Therefore equation (2) implies that capital return is the sum of the benefits from the unskilled migrant labor, which was referred to as immigration surplus by Borjas (1995), and losses of the unskilled native labor, minus benefits of the skilled labor due to the influx of unskilled migrant workers. As the wages of the unskilled workers are affected negatively by inflow of immigrants,
is smaller than zero. Thus the first term of equation (2) implies that , which will be shared by capital
the natives as a whole gain the amount
and the skilled labor. The size of immigration surplus is determined by the wage drops of the unskilled labor affected by the influx of migrant labor and the number of migrant workers employed. Thus the size is proportional to the elasticity of factor price for unskilled labor. The second term in the right-hand side of equation (2) is the amount of wealth transferred from the unskilled native workers to the capital owners and the skilled workers. The unskilled native workers lose the amount of . In general, the number of the unskilled natives Du is much greater than that of the unskilled immigrants Fu. Hence the amount of wealth transferred is likely to be much greater than immigration surplus, implying that the impact of migrant workers on firm performance arises mainly out of wealth transfer from the unskilled native workers. The third term in the right-hand side of equation (2) is the amount of economic impact on the skilled workers. Since wages of the skilled workers will be raised with is greater than zero, thereby the inflow of migrant workers by assumption, leading to positive benefits to the skilled workers and imposing higher cost on their employers. However, the cost is likely to be smaller than the positive benefits transferred from the unskilled native workers, because the number of the skilled workers is much smaller than that of the unskilled labor in the small-medium business sector in Korea. Hence, equation (2) indicates that the profit gain of firms arises from cost saving due to the lowered wage of the unskilled labor and it is proportional to the elasticity of factor price for unskilled labor. Firms’ profit gain hinges critically on how much the influx of migrant workers will lower wages of the unskilled labor. If labor shortage 10
has been really large in the Korean SMEs as alleged, then migrant workers are likely to have lowered wages of the unskilled labor considerably, providing a large profit gain. Since the annual average growth rate of minimum wage was 10.2% from 2000 to 2009, probably the sharp increase in the minimum wage may have constrained the wage impact of immigration. Up to this point, we have excluded international trade and capital mobility, assuming that we were in a closed economy. In the standard Heckscher-Ohlin model, factor price equalization occurs in an open economy. In this case, immigration of unskilled workers will increase production of more unskilled-labor-intensive goods, but factor prices will remain unchanged. In an open economy, the adjustment occurs through traded goods: immigration will cause the country to export more (or import less) the unskilled-labor-intensive goods (Friedberg and Hunt 1995). Once the country is perfectly specialized, the impact of migrant workers will have similar effects as those of the closed economy. Capital mobility is an alternative mode of employing migrant workers by going overseas rather than bringing them in. Taking international capital mobility into account, we need to relax the assumption of capital being fixed in the model. When employers undergo shortage of unskilled labor, they may resort to relocation overseas, instead of employing the unskilled labor at high wage rates. If capital is perfectly mobile, factor price equalization should occur. However, capital mobility across countries is not perfectly free, since, above all, there are service industries which have a low mobility of capital. The offshore investment of the Korean companies has been steadily increasing over time, and one of the main objectives of offshore investment by many Korean SMEs was to take advantage of low-wage overseas. If immigration provides incentives for the employers not to go offshore by lowering wage rates, then new capital investments overseas can be contained and capital may even return home, increasing capital in the Korean small-medium business sector. Capital inflow will increase supply of capital and lower returns to capital, but it will raise output level. 2. Empirical Approaches This paper will estimate the following profit function at the firm level. (3)
π=
An individual firm can adjust the level of employment and capital, not the prices of the production factors in this model, since capital cost and wages of the skilled and the unskilled labor are given to a price-taking individual firm. However, we assume that wages can be varied across firms, depending on whether a firm is within the sector eligible for the unskilled migrant workers or not. We consider a vector of other factors X which may influence the production function. Taking derivative of equation (3) with respect to the unskilled immigrants, then we can get If the production function is concave, this equation implies that its output level will be higher when the wage of the unskilled labor wu gets lowered by the influx of migrant workers. Individual firms are likely to employ more unskilled workers with their lower wages. The skilled labor will also be employed more, because it is complementary with the unskilled migrant labor. On the other 11
hand, the impact on the capital level used will be ambiguous, because capital is imperfectly substitutable or complementary with the unskilled labor. Approximating equation (3) with a linear function, we can obtain a linear regression equation as follows. It is assumed that capital return is a time-invariant variable, since information on capital cost r is not available here. Even if it varies over time, it will get into the error term. (4)
= μ i + α1
it
+ α2
+ α3
+ α4Log(w)it + βX it + ε it,
where L=Fu+Du+Ls. Since we expect that immigration have a positive impact on firm profitability through lowering labor cost, the wages, the number of the native workers employed, and capital-labor ratio may absorb some of the immigration impact. So we will examine the firm performance impact of immigration both with and without the mediated variables. The estimation procedure is a part of identifying the mediating effects of wages, the number of the native workers employed, and capital-labor ratio. In the procedure we will also estimate regression models with labor productivity, wage, the number of the native workers employed, and capital-labor ratio as their dependent variables. Additionally we will check the impacts of employing migrant workers on capital investment and output such as sales, asset, offshore investment and temporary or permanent closing to see if there is a scale effect. Estimating these models, we may encounter a reverse or dual causality between firm performance and migrant worker employment. The number of migrant workers employed can vary across and within sectors. The variability of migrant workers across sectors is reasonably assumed to be exogenous, even though the labor shortage at the sector level is involved in determining the eligibility of the sector. In contrast, the number of migrant workers employed across individual firms within the sector allowed to employ migrant workers can be affected by the performance of individual firms, since labor shortage is probably correlated with firm performance. We will account for the time-invariant variable to affect the number of temporary immigrants by employing a fixed-effect model of panel data. Because we use tworound panel data, we can cross out the time-invariant fixed individual effect by first differencing. However, the fixed effect model does not control the change of firm performance over time. Therefore, if the number of migrant workers employed depends on the change of firm performance, the endogeneity problem will remain unsolved. Also it should be noted that the first-difference model is vulnerable to measurement error of the independent variables. However, the problem will not be severe in this model, since the independent variables are more or less objective variables, which are less prone to measurement errors. Another remaining issue is concerned with crowding-out effect. The impact of immigration on wage, which is a major potential mediator between immigration and firm performance, occurs in its labor market, rather than at an individual firm level. Also unskilled migrant workers may crowd out unskilled native workers across sectors or among firms, thereby equalizing wage levels of the unskilled workers. However, labor, especially foreign labor, is not perfectly movable across sectors due to skill and regulatory barriers. Anyway if wage-equalizing forces operate, then the economic benefits out of the unskilled immigration may pass over to the non-users of 12
migrant workers, yielding bias toward zero for the estimates of firm performance impacts of immigration.
IV. Empirical Results 1. Data and descriptive statistics The data set used in this paper is the Korean Workplace Panel Survey, which has been collected by Korea Labor Institute every other year since 2006. The panel data set contains one year earlier information of human resources, employment relations, industrial relations, and working conditions at the establishment level. Currently the data set has been collected up to two round panels in 2006 and 2008. The respondents are HR specialist, IR specialist, and employee representative of union or joint consultation committee in each establishment. The sample size in the first round panel comprises 1,740 establishments with 30 and more permanent employees in the private sector, which cover all the industry excluding agriculture, fisheries and mining. Among them, 100 establishments in the public sector are included, which will not be used in this paper. The sample size of the second round becomes 1744 establishments, including 125 establishments in the public sector. Out of the observations of the first round, 1301 establishments in the private sector are maintained in the second round panel after dropping out due to temporary or permanent closing, relocation and declined response. About 62.6% of the remaining establishments are SMEs with less than 300 employees. This paper will analyze the SME samples where a part of the companies are eligible to employ migrant workers. Here excluded are also the multiple establishment firms, because financial performances information at the firm level and employment information at the establishment level do not match each other in those firms. Finally we used 255 observations, after deleting the establishments which have missing values for the main variables used in the current models. The large loss of sample size is due to low availability of financial performance information in the sample. The data contain information on the number of the unskilled migrant workers and various types of native workers in each round panel. The percentage proportion of migrant workers out of the total workforce in each firm is used for the variable of migrant workers. Here the number of the total workforce includes the permanent workers and the directly employed contingent workers such as temporary workers and part-timers, but excludes the dispatched workers and the onsite contract workers. The data also provide information on the number of vacancy for the majority occupation workers in an establishment, by which the survey indicates the occupation that comprises the greatest portion of employees in the establishment. We obtain the percentage of labor shortage rate by taking the ratio of the number of vacancy over the number of the incumbent workers in each establishment, multiplied by 100. We use operating income per capita, sales, and sales per capita as indicators for firms’ profitability, output and labor productivity respectively. We use tangible fixed asset as an indicator for capital, and so tangible fixed asset per capita for the variable of capital-labor ratio. We also check other types of assets such as non-investment and total assets at the end of the years as indicators for capital investment, and temporary or permanent closing and offshore investment as its alternative indicators. Temporary 13
or permanent closing is identified at the time of the second round survey. Since the temporarily or permanently closed establishments drop out of the second round survey, this information allows just for a cross-sectional data analysis. The amount of offshore investment is not available in the data, but a Likert-type of response is available. We construct a dummy variable indicating the existence of offshore investment from the response. We get a wage variable by dividing the total labor cost by the number of employees. Since the wage level is surprisingly high in some observations, we delete the case whose value for the wage variable is greater than mean plus 2*standard deviation. The wage variable is defined as an average labor cost and thus its value depends on labor composition as well as non-wage labor cost. So here we employ two other representative wages: the starting annual earnings for newcomers with high school diploma and college diploma. The two wage variables represent the wages for the unskilled workers and the skilled ones. In Korea, recently high school graduates are more likely to get unskilled jobs and wage levels in SMEs tend to be compressed, the starting annual earnings of high school graduate will be a good indicator for the wage of the unskilled native workers, who compete with the unskilled migrant workers. This paper will also control for some characteristics and contexts of firms whose information is available in the data set. Those control variables are unionization, firm age, the degree of changes in output demand, and the degree of competition in its domestic output market. Since firm age will disappear in a fixed effect model of panel data, we plug in the squared term of firm age, instead. Unionization is indicated by a dummy variable of whether to be unionized or not, and firm age square is computed by taking square of the year passed since its establishment. Both the degree of changes in output demand and the degree of competition are subjectively measured Likert-type variables.
Insert
around here.
Table 2 presents summary statistics of the variables used in this paper. To obtain a brief description of the firm performance impacts of employing migrant workers, we compare the statistics between the users and the non-users of migrant workers by year. The fractions of firms which employ migrant workers are 17.6% in 2005 and 23.1% in 2007. Not being reported in the table, the users of migrant workers are more heavily distributed in the manufacturing sector. The average proportions of migrant workers for the users are around 9.94% and 10.04% respectively in the corresponding two years. The financial state of the migrant worker users is worse than those of the non-users. The users of migrant workers have lower levels of assets, and yield much lower sales, sales per capita, and operating income per capita, as compared to their counterparts. Especially in 2007, the amounts of sales, sales per capita, and operating income of the user firms are less than half of the non-users. However, note that capital and capitallabor ratio of the user firms are also much lower than those of the non-users, so that the lower performances of the users may be partially attributed to their lower capitallabor ratios. All these basic statistics may also indicate that lower performers are more likely to employ migrant workers, raising the issue that the number of migrant 14
workers employed can be determined endogenously. The basic statistics show that the user firms are a little more inclined to invest overseas. However, industry effects might be involved here, since the service sector firms are in general less likely to invest overseas. If we make the comparison within the manufacturing sector, then their average values between the user firms and the non-user firms are reversed: 0.26 and 0.19 in 2005, and 0.30 and 0.24 in 2007, respectively. The basic statistics also indicate that the user firms are more prone to business closing. The rate of business closing of the users and the non-users since the first round survey are 0.082 and 0.062 respectively. By contrast, the descriptive statistics indicate that the firms using migrant workers may perform better with respect to employment. Employment levels of the user firms are greater than those of the non-users. It is also worthy of noting that the labor shortage ratio of the user firms become lower in 2007 while it was higher in 2005. It is an interesting result because it was alleged that initially the user firms have suffered from higher labor shortage. The greater size of employment in the user firms seems related to the lower labor cost identified in those firms. We can see from the descriptive statistics that the differentials of average labor cost between the users and the non-users are around KRW8.35 million in 2005 and KRW7.43 million in 2007. However, wage levels for the representative workers of high school graduates and college graduates are just slightly different between the users and non-users, indicating that the differentials of average labor cost may be arising out of the difference in labor composition and/or non-wage labor cost.
Insert
around here.
2. Regression Results The regression results are presented as follows. The estimates of the impact of migrant workers on profitability are reported first, followed by its estimated impacts on the intermediary variables of labor productivity, labor market outcomes (wage, employment, labor shortage ratio), capital-labor ratio and capital, which will help identify the mediators of the profitability effect. Finally, the estimates of the impact which migrant workers have on output and capital investment variables will be presented. Table 3 shows the profitability impacts of employing migrant workers. The regression models are first difference models of panel data, controlling for the fixed individual effect. The logarithmic value of operating income per capita is employed as the main dependent variable for the models. Because operating income per capita can have negative values, the logarithm of Z with its value smaller than 1 was smoothed out in this way; -log(-Z), if Z
