The Effect of Overtime Working Hour Restrictions …...working hours that yield the highest utility level for their work; however, women cannot work for optimum working hours as a

Lai, International Journal of Applied Economics, 8(2), September 2011, 86-106 86

The Effect of Overtime Working Hour Restrictions on Wages in Taiwan, ROC

Yu-Cheng Lai*

Shih Chien University Kaohsiung Campus

Abstract: Taiwan’s Labor Standards Law (LSL) requires employers to implement overtime working hour restrictions. These restrictions have a greater effect on women, who are not allowed to work as many overtime working hours as men. This paper examines two hypotheses: First, as the hour restrictions can encourage the efficiency of workers, it will increase wages for women in the covered sector relative to others. Second, the foreign workers importation will increase the economic freedom for employers, leading to a smaller effect of hour restrictions. This paper develops a model to measure the effect of overtime hour restrictions on wage rates. Using the Manpower Utilization Survey, this paper finds that these restrictions have an apparent effect on raising the wage rates. However, the LSL will reduce the gender wage gap via the overtime restrictions. It provides the empirical evidence that the LSL may affect the gender wage gap in Taiwan. On the other hand, the effect of the hour restrictions on wages will decline as a result of the foreign worker importation. Keywords: Hour Restrictions, Labor Standards, Overtime, Wages JEL Classification: J38, J68 1. Introduction In the early eighties, there were some criticisms of the longer working hours in Taiwan. In response to this international criticism, attempts were made to defend longer working hours as a good practice of traditional hardworking morality. However, after the late seventies, emerging public opinion and government policy favored the promotion of fundamental changes in labor policy—from growth-oriented to more welfare-oriented policies. Therefore, there was a consensus in favor of shorter working hours in order to not only cope with the international criticism but also as part of the general shift in state policy. In particular, many nations have introduced working hour restrictions for women employees in consideration of their health and safety and to provide them time to fulfill their responsibilities as wives and mothers. With regard of the disadvantages of overtime working hour restrictions, if women desire to work longer, the overtime working hour restrictions will block the chances of women earning an additional overtime premium. Moreover, an individual will choose working hours that yield the highest utility level for their work; however, women cannot work for optimum working hours as a result of overtime working hour restrictions (van Soest et al., 1990). Therefore, the shorter overtime working hours will limit the accumulation of both reputation and human capital (Cherry, 2004). 1 However, based on the loss from overtime premium and job promotion, the overtime working hour restrictions will cause some deadweight loss. Before World War Ⅱ, the first regulation for hour restrictions in Taiwan was the Factory Act (FA), drafted by the Government of Republic of China (R.O.C.) in Mainland China, and


passed earlier in 1929. The Factory Act regulates the standards hours— approximately 48 hours per week. Following the Factory Act, the Labor Standards Law (LSL), which was passed in 1984, contains the same regulation with regard to standard hours—namely, 48 hours. The LSL stipulates working hour restrictions, that is, limitations on when and how employees may work. First, it stipulates that women cannot work from ten in the night to six in the morning; exceptions are made for some occupations, for instance, managerial, technical, and health workers. Second, the LSL stipulates that the amount of overtime work cannot exceed three hours per day or forty-six hours per month in the case of men and two hours per day or twenty-four hours per month in the case of women. With regard to both the restrictions— that is, the restrictions on the amount of overtime work and those on women working at night— exceptions may be made if an agreement is obtained from either a labor union or a majority of the employees. With regard to the overtime premium in Taiwan, the LSL regulates that the employees must be paid at least 33% premium for the first two overtime hours per day (LSL article 24 point 1). Furthermore, employers need to pay at least 67% premium for overtime between two and four hours (LSL article 24 point 2). For any exception to the LSL regulation on overtime working hours, employers need to pay double for the overtime (LSL article 24 point 3). 2. The Protection from LSL Minimal changes were made to the coverage of the LSL between 1984 and 1998. During this period, industries related to manufacturing, mining, electricity, construction, transportation, and mass media were covered by the law.2 However, at the end of 1998, this coverage was extended to all industries. According to Figure 1, the manufacturing industry is the largest industry among those covered by the LSL; its proportion in the labor market was approximately 30% from 1978 to 1998. The construction industry is the second largest industry, and its proportion in the labor market is approximately 7%. Based on Figure 2, the uncovered industries include those related to commerce, finance, insurance and real estate, business services, community services (social, personal, and related), and public administration. Among these industries, those related to commerce and social services form the majority and both these industries account for 30% of the labor market. Besides the restrictions on overtime working hours and those on women working at night, the LSL comprises several provisions that mainly affect women. First, it requires that all female workers be remunerated by a firm for eight weeks of maternity leave; further, full wages are to be paid if the firm has employed the worker for at least six months. In addition to paying full wages during maternity leave, employers must also compensate the replacement workers. Moreover, employers can no longer terminate a worker’s employment owning to marriage or pregnancy.3 Second, the LSL requires that employers provide new mothers with two daily paid half-hour breaks to breast-feeding until the baby is one year old. Third, employers must transfer women to easier jobs during their pregnancy as failure to do so could adversely affect the health of either the baby or the mother. Furthermore, both the LSL and FA require that retirement and severance payment be provided to all labor. However, the LSL adds the pension fund for retirement benefits, and elderly workers should have more opportunities to enjoy these benefits rather than young ones.4 3.The Enforcement Peaks of the LSL The overtime working hour restrictions regulated by the LSL have a greater impact on women because women are not allowed to work the same number of hours that men are. In


Table 1, in the covered sector, the overtime working hours have significantly reduced after 1986 for female workers, and reduced after 1988 for male workers. On the other hand, in the uncovered sector, the Table 2 show that the overtime working hours was free from impact of the LSL for female workers, but it will decline after 1987 for male workers. The strengthening of enforcement measures for overtime working restrictions will lead to higher compliance levels. Therefore, an increase in compliance leads to an inevitable increase in the effect of the LSL. When the LSL was first established, the legislation provided few new enforcement procedures or personnel. Moreover, few employers followed its regulations. The additional labor cost due to the LSL was low because compliance was not too high. However, the situation differed from the pre-LSL one. Furthermore, a year or two after the enactment of the LSL—namely, by 1985 or 1986—the law could be regarded as the first peak in legislation enforcement. However, the political pressures that produced the legislation soon led to a greater enforcement effort. Consequently, in 1987, an enforcement department known as the Council of Labor Affairs (CLA) was established. Thus, after the establishment of the CLA in 1987, the following year 1988— should be considered as the second peak in legislation enforcement. Since 1989, however, the construction first began to import foreign workers into the labor market in Taiwan. It may alleviate some legislation effect by import foreign workers. Moreover, enforcement was further strengthened and centralized with the enactment of the LIL in 1993. Therefore, after the enactment of the LIL in 1993, the following year, 1994, should be considered as the third peak in legislation enforcement. At the same time, after 1993, the manufacturing industry also imports foreign workers to reduce the impact of labor shortage. Prior to 1993, the CLA could only inspect firms that employed more than 100 people. In addition, it had delegated some of its power to local authorities who were not always eager to enforce the LSL. Other problems included a small number of inspectors who were insufficiently trained, an overlap of the jurisdiction of local inspection agencies and successful lobbying by local businesses against significant penalties. As a result, the CLA was unable to effectively enforce LSL inspections. Therefore, the CLA enforcement department drafted the LIL, which was passed and implemented in 1993. The LIL regulates all local inspection agencies that are now directly overseen by the central government's inspection system. This helps in avoiding effort duplication and conveys a clear message to employers about the need for LSL compliance; consequently, it reduces the effectiveness of lobbying against enforcement. The legislation Yuan of Taiwan ROC passed the amendment of the LSL in Dec. 1996, which the LSL will covered all industries in Dec 1998. However, the original covered sector will not have additional impact from the LSL. On the other hand, the original uncovered sector will have some impact from the LSL after 1998. Therefore, after the enactment of LSL in the original uncovered sector after 1998, the following years from 1999 to 2002, should be considered as the first peak in the legislation enforcement for the original uncovered sector. After 2002, the overtime hour restrictions will not have any difference between women and men. Therefore, the LSL effect on gender differences should be minor after 2002. The overtime hour restrictions regulated by the LSL have a greater impact on women because


women are not allowed to work the same number of hours that men are. In Table 1, in both the covered and uncovered sectors, the overtime working hours have significantly reduced for female workers as compared to those for male workers after the enactment of LSL in 1984. The strengthening of enforcement measures for overtime working restrictions will lead to higher compliance levels. Therefore, an increase in compliance leads to an inevitable increase in the effect of the LSL. As seen in Table 2, after the LSL enforced to all sectors in 1998, the overtime hours of women will decline for the CLA inspected sector in 1999. However, the overtime restrictions will not have gender difference after 2002. The gender difference of overtime hours in the CLA uninspected sector is larger than those in the inspected sector after 2002. The overtime restrictions constitute a more significant aspect in the Labor Standards Law in Taiwan ROC. The overtime restrictions lead to an exogenous reduction in overtime and affect on wages. The theoretical model and empirical results in this study indicate that employers will increase wage among the original covered sector after the enactment of LSL. After 1998, the empirical results in this study indicate that employers will also increase wage among the original uncovered sector in response to the LSL enforcement. 4. Theoretical Model On the basis of Hart (2004), this paper developed a rising premium schedule model to measure the effect of overtime hour restrictions on the labor market. The state of technology in the Cobb-Douglas production function is represented by A. To avoid problems of firm size, fixed capital costs and number of firms, the convenient assumptions of constant returns to employment (N) and capital (K) were adopted. The working hours (h) can represent the total working hours, and it includes both standards hours and overtime hours ( ). Using a Cobb-Douglas formulation production is given by .

OTs hhh +=αγβγβγ hNAKQ −= 1 5

αγβγβγ hNKQ −Α= 1 (1)

As seen in Equation (1), standard hours ( ) are treated as an exogenous variable, which the overtime hours ( ) will be equal to the actual minus standard hours ( ). This study also assumes that the homogeneous workers are employed for the same number of weekly paid-for overtime hours, therefore, this study writes the cost expression in (2) for an average premium rate, av. Thus,

sh

OTh shh −

NzavhhwcNC OTs ])([ ++== (2) Let the firm’s marginal overtime premium, p, increase continuously with the overtime hours, that is,

OTmhkp += 0≥m 1≥k 6

(3) Both k and m are treated as exogenously determined parameters. The total payment for overtime is given by


∫ +==OTh

OTOTOTOTOT mhkhwdhhpwwh0

2 )21()(

(4)

Substituting (3) in (4) gives

c= ])21([ 2 zmhkhhw OTOTs +++

(5)

Total cost per worker is represented by c, is the standard hourly wage, and z represents the fixed cost. Setting output price to unity, the firm aim at profit maximization can be expressed

w

cNQMax −=π

= αγβγβγ hNK −Α 1 Nzmhkhhw OTOTs ])21([ 2 +++−

(6)

This study would like to determine how the optimal choice functions respond to change in a parameter of overtime ( ). The optimal choices have to satisfy the first-order conditions. First-order conditions require

OTh0== hN ππ , which give, respectively

zmhkhhwNQ

OTOTS +++= )21( 2βγ

(7)

NwmhwkhQ

OT )( +=αγ

(8)

Based on the Equation (8), this study differences wage with respect to the overtime, it obtains the negative sign. It implies that the shorten overtime will increase wages.

)(21

NhQhm

hw

OTOT

αγ−−−=∂∂ < 0

(9)

The working hour restrictions in Taiwan limit the amount of overtime. The overtime hour restrictions may have some effects on wage rates. The hourly wage will increase when labor production improves in response to working hour reduction (Börsch-Supan, 2002; La Jeunesse, 1999).7 The production of labor may significantly increase when employers invest more capital in order to compensate with working hours loss. Conversely, the working hour restrictions may lead to a wage decline if it does not have positive effect on labor production (Tummers and Woittiez, 1991). Bell and Hart (2003) also claim that the standard hourly wages will declines, but the overtime premium will increase as a result of higher overtime hours. Based on the optimal wage system in order to improve contract efficiency, the standard hours of trained workers pay at a rate below marginal product but overtime premium at above marginal product (Hart and Ma, 2010). Using the Manpower Utilization Survey and Human Resource Survey of Taiwan, Chen and Wang (2011) indicate that the enforcement of the new standard hours of Taiwan (84 hours for


two weeks) in January 2001 will decline the working hours, but the hours drop has gradually risen back for all employees. More importantly, the hour increase rate is higher for the low income workers. However, their study indicates that high income female workers decrease their hourly wages after the enforcement of new standard hours. In the closed economy, the reduction of working hours for females will cause pressure for their hourly wage to increase, and leading firms will need to replace women with men in production. If the substitution between female and male workers is imperfect, the rise in women's relative wages would be alleviated by the inflow of new workers. To correspond above analysis, Zveglich and Rodgers (2003) indicate that in 1989, overtime restrictions in Taiwan would have reduces female working hours and their employment by 6.1% and 0.9%, respectively. However, there is no significant effect on the wage. However, both papers (Zveglich and Rodgers, 2003; Chen and Wang, 2011) do not consider the effect importation of foreign workers, it may lead to a smaller effect of hour restrictions. Since 1989, however, the construction industry first began to import foreign workers. Therefore, the effect of foreign workers should be considered into the model. Some studies (De New and Zimmerman, 1994; Borjas, 1999, 2003) indicate that natives often fear labor market competition of foreigners, as they may induce declining wages as in the case of natives and immigrants being substitution. However, there is also the potential that they are complements, producing positive wage effect. On the other hand, the foreign workers importation may decline the labor cost for the firm in Taiwan. For the firm sought for profit maximum, employers in Taiwan will hire foreign workers to replace young women, and declining wages for those workers (Lai and Masters, 2005). Based on those previous papers, this study tries to create a clearer picture to consider a long time period (1978-2006) and effect of foreign workers. There are two hypotheses studied in this paper. First, the overtime restrictions can encourage the efficiency of workers via shorten working hours, it will increase wage for women in the covered sector relative to others. Second, the foreign workers importation will increase the economic freedom for employers, leading to a smaller effect of hour restrictions. 5. Data In order to estimate the effects of the LSL, as its primary data source, this paper used the Manpower Utilization Survey—which is an annual survey of earnings and employment data available since 1978. As seen in Table 2 in the Appendix, the robustness measures were analyzed for various age cutoffs in order to investigate the sensitivity of the results.8 Finally, the sample was restricted to full-time and part- time employees in the nonfarm private sector who were between the ages of 35 and 60, and excluded employees in the public sector, the self-employed, employers, and unpaid family workers.9 In the first stage of testing, the survey includes the annual data from 1978 to 1998 for both the LSL covered and uncovered sectors. Since the legislation was expanded to be enforced to all sectors after 1998, the empirical results can reflect the fact that legislation does not have a special effect on the covered sector after 1998. This can provide an empirical evidence to explain why the testing model sample size is restricted to the period of 1978 to 1998, and the sample size is 112,632. With regard to the LSL enforce to all sector after 1998, the sample of LSL original uncovered sector will have legislation effect after 1998. The testing model sample size in the second stage is restricted to the period of 1979 to 2006, which the sample size is 49,934 for the LSL original uncovered sector, and 129,232 for the original covered sector.


With regard to the sample selection bias, the bias may be eliminated as follows: when using the panel data to track the same individual over time (Borjas, 1999), or if the unobservable factors accompanied by sector choice change over time (Zveglich Jr. and Rodgers, 2003). Although this study lacks panel data, it pools subsamples of the working-age population for every year from 1978 to 2006. Since these subsamples are drawn from the same population over time, the sample selection bias will be minor.10 This reflects the statistical evidence that the sample selection bias may have less of an effect on the empirical results because the sample size is huge and it covers both full- and part-time employees. On the other hand, in the case of the presence of heteroskedasticity, the generalized least square (GLS) method can fix the error term from the correlation across observations (Arellano, 1987; Muhlberger, 2000). In order to adjust for heteroskedasticity, the method will be estimated by using a weighted least squares (WLS) procedure.11 6. Estimating the LSL Effects This analysis commences by classifying four demographic groups of workers by age (young and older) and gender (female and male). These groups are further subdivided by industrial sector according to the LSL coverage (covered and uncovered), where represents the difference between the covered and uncovered sectors and

cΔ

tΔ represents the change in the labor market outcomes for each demographic group after the enactment of the LSL. This paper claims that the LSL has four legislation effects— maternity/pregnancy benefits, overtime working hour restrictions, retirement/severance benefits and other legislation effects. The maternity/pregnancy benefits should only affect young women, whereas the overtime working hour restrictions will affect all women as compared to men. The retirement/severance benefits will affect elderly workers, while the other legislation effects are predicted to affect all workers. Therefore, the YoungWomentc YΔΔ in Equation (9) represents the fact that the difference-in-differences (DD) procedure for young women has three legislation effects on maternity/pregnancy benefits, overtime working hour restrictions, and other legislation effects. In Equation (11), the DD procedure for elderly women has three legislation effects on overtime working hour restrictions, retirement/severance benefits, and other legislation effects. In Equation (12), the legislation effect for young men has other legislation effects. In Equation (13), the legislation effect for elderly men will include both retirement/severance benefits and other legislation effects.

YoungWomentc YΔΔ = Maternity/pregnancy benefits effect + Overtime Hour Restrictions + Other legislation effects

(10)

enelderlyWomtc YΔΔ = Overtime Hour Restrictions + Retirement/severance benefits effect + Other legislation effects

(11)

YoungMentc YΔΔ = Other legislation effects (12)

elderlyMentc YΔΔ = Retirement/severance benefits effect + Other legislation effects

(13)

However, employing the framework of DDD developed by Zveglich and Rodgers (2003), this paper claims that the comparison for groups of elderly women and men can measure the effect of restrictive overtime working hour restrictions in Equation (14). Further, it is important to note that the DDD coefficients for each year will differ every year after 1984.


enelderlyWomtc YΔΔ - = Overtime hour Restrictions effect elderlyMentc YΔΔ (14)

In Equations (15) and (16), this paper uses log value of the monthly wage ( ) as a dependent variable. Based on the questionnaire of the Manpower Utilization Survey in Taiwan, it will be the individual monthly wage adjusted by the price of 1978. However, after controlling the working hours, the model can estimate the effect for the hourly wage as a result of overtime working hour restrictions.

iWage

Overtime working hour restrictions should have a greater effect on women than on other demographic groups. This paper assumes that other factors—for instance, changes in product demand across industries— should affect women in the same manner that they affect men. Thus, in order to estimate the effect of the LSL on wage rates of women before the LSL enforce to all sectors, this paper will examine how changes in the relative differences of the labor market outcomes of women differ in the covered versus the uncovered sectors since the enactment of legislation. However, after the LSL enforce to all sectors, this paper will examine how changes in the relative differences of labor market outcomes of women differ in the original uncovered sector versus the original covered sector because the LSL original uncovered sector will be enforced after 1998. This paper employs the legislation in covered industries as the main identifying variable. Although the government exogenously determines which industries should be covered, they may be correlated with unobservable sector-specific factors, which, in turn, may be uncorrelated with the introduction of the LSL. However, the estimation of the cover dummy variable can capture the endogenous bias due to the cover-sector fixed effect (Rosenzweig and Wolpin, 1986). Individuals are only observed for the industrial sector that they actually select because the individual choice of the industrial sector may be correlated with the unobserved determinants. Before the LSL enforce to all sectors in 1998, we adopt the Cover ( cΔ ) dummy variable with a value of 1 if the employee works in an industry that is covered by the LSL (or that will be covered after 1984), and it represents the difference between the covered and uncovered sectors. After the LSL enforce to all sectors, we use Size ( cΔ ) is a dummy variable with a value of 1 if the employee works in inspected sector industry that is uncovered by the LSL which the firms is over 100 employees, and it represents the difference between the inspected and uninspected sectors. The Td variable ranges from 1 to 29, representing every year from 1979 to 2006. This can measure the trend effect, which can capture the effects of macroeconomic conditions over time. Before the LSL enforce to all sectors in 1998, Yeari dummy variables represent the difference in the situation before and after the enactment of the LSL, which can capture the effects of the macroeconomic conditions over time. This is a set of dummy variables for every year from 1985 to 1998, using the sample for every year from 1978 to 1984 as the base year. After the LSL enforce to all sectors in 1998, Yeari dummy variables represent the difference in the situation before and after the enactment of the LSL for the original uncovered sector, which can capture the effects of the macroeconomic conditions over time. This is a set of dummy variables for every year from 1999 to 2006, using the sample for every year from 1979 to 1998 as the base year.12 As seen in Equation (15) and (16), , which is the coefficient of the two-way interactions id


(Cover × Yeari or Size × Yeari , tcΔΔ ), represents the covered industry-specific or inspected-sector-specific economic conditions over time. , which is the coefficient of the

three-way interactions (Gender×Cover× or Gender×Size× ), represents the effects of the LSL on different gender groups, which are derived using the DDD method. Before the LSL enforce to all sectors, the three-way interactions (Gender×Cover× ) can measure the gender differential effect of overtime restrictions. However, after the LSL enforce to all sectors, the DDDD method will be the better than DDD method to measure the gender differential effect of overtime restrictions. It measures by the differential for the coefficients of the three-way interaction (Gender×Size× ) between the original uncovered and covered sector, which can eliminate the effects of different firms sizes.

ie

iYear iYear

iYear

iYear

More importantly, the model will add a variable to control the effect of foreign workers—Fw is the dummy variable for industries that import foreign workers. Therefore, if the regression has Fw and Gender×Fw variables, the regression may control the effect of foreign workers importation. Otherwise, it may include the effect of foreign workers importation. Table 1 in the Appendix displays the definitions of the variables as well as the descriptive statistics of the data. These equations also include the variables for human capital and other factors related to labor supply among women and their interactions with the gender dummy variables. These variables include experience, tenure, years of education, and marital status, and the independent variable of monthly working hours. Moreover, each of these variables interacts with Gender, which is a dummy variable for gender groups. Estimating the LSL Effect on Wages before the LSL Enforce to All Sectors

i

n

iii

n

iiii

n

ii

n

iii

i

uYearCoverGendereYearCoverdYearGenderc

YearbCoverGenderaCoveraTdGenderaTdaFwGednera

FwaMarGenderaMaraHourGenderaHouraTenGenderaTenaExpGenderaExpaEduGenderaEduaGenderaaWage

+××+×+×+

+×++×++×

++×++×++×++×++×+++=

∑∑∑

∑

===

=

111

11716151413

121110987

6543210

(15)

where

9814

9713961295119410939928917

906895884873862851

14

1

YearbYearbYearbYearbYearbYearbYearbYearb

YearbYearbYearbYearbYearbYearbYearb ii

i

++++++++

+++++=∑=

Estimating the LSL Effect on Wage after the LSL Enforce to All Sectors


i

n

iii

n

iiii

n

ii

n

iii

i

uYearSizeGendereYearSizedYearGenderc

YearbSizeGenderaSizeaTdGenderaTdaFwGednera

FwaMarGenderaMaraHourGenderaHouraTenGenderaTenaExpGenderaExpaEduGenderaEduaGenderaaWage

+××+×+×

++×++×++×

++×++×++×++×++×+++=

∑∑∑

∑

===

=

111

11716151413

121110987

6543210

(16)

where

068057

046035024013002991

8

1

YearbYearb

YearbYearbYearbYearbYearbYearbYearb ii

i

++

+++++=∑=

7. Empirical Results As seen in Table 3, the LSL has significant positive effects on the wage rates. As seen in column (A) in Table 3, the empirical results revealed that the overtime restrictions had a positive effect, that is, an approximate increase in wages by 5% to 9% from 1985 to 1997. As seen in the column (B) in Table 3, the excluded in the results the effect of foreign workers, it reveals stronger and more significant results rather than the results included the effect of foreign workers in column (A). It approximates increase in wages by 5% to 13% from 1985 to 1998. For the results of column (A) and (B) in Table 3, there are three peaks for the effect of legislation enforcement on wages. First, the first strong results that are observed from 1985 and 1986 probably reflect the influence of the LSL, which was passed in 1984. Second, the first strong result that is observed from 1988 probably reflects the influence of the CLA, which was established in 1987. Third, the third phase of strong results that are observed from 1994 to 1998 probably reflect the influence of the LIL, which was passed in 1993 in order to improve the enforcement of the LSL. After the LSL enforce to all sectors, as seen in Table 4, the LSL has significant positive effects on the original uncovered sector compare to original covered sector. The result of column (3) in Table 4 indicates that the LSL has an approximately increase in wages by 13% in 1999. The result of column (6) in Table 4 reveals that excluded in the results the effect of foreign workers, it indicates smaller results than included in the results the effect of foreign workers in column (3). It approximates increase in wages by 12% in 1999. However, the WLS results for year of 2000 and 2001 have less significant effect in columns (3) and (6) in Table 4. After 2002, the overtime restrictions treat women and men equally. Therefore, the overtime restrictions will not have any significant gender-differential effect of wage after 2002. The positive effect on wages will provide the empirical evidence. In particularly, before the LSL covered to all sectors, the overtime restrictions have a stronger effect for covered relative to uncovered sector in 1997, with a positive effect on wage by 9% and 13% (as seen in column (A) and (B) in Table 3), respectively. After the LSL enforce to all sectors, the original LSL uncovered sector will experience the overtime restrictions, it increases about 12-13% for the original LSL uncovered sector compare to the original covered sector. However, this study indicates that the results included the effect of foreign workers have


lesser significant effect and smaller effect than the results excluded the effect of foreign workers. 8. Conclusion Based on more extensive data (1978-2006), this paper further studies the effect of overtime working hour restrictions on wages. It may differ from the findings by Zveglich and Rodgers (2003) and Chen and Wang (2011)— namely, that there are no significant effects on female wages, and negative effect on high income female wages, respectively. This paper discusses which overtime working hour restrictions may lead to the effect on wage increase for female workers. This paper claims that the LSL had a positive effect in wages for covered sector relative to uncovered sector. It also increases wage of women when the LSL original uncovered sector began covered to LSL. Therefore, the overtime restrictions not only provide better working conditions for female employees such as shorter working hours, which can improve their health as well as take care their family. The overtime restrictions would likely to decline the gender wage gap because it can increase the wage of female workers compare to male workers. However, the LSL will decline the gender wage gap via the overtime restrictions. It provides empirical evidence in Taiwan that the LSL may affect the gender wage gap. On the other hand, the foreign workers importation declines the effect of hour restrictions, leading smaller wage increases. The impact of foreign workers relieves the urgent demand for the production loss as a result of hour restrictions. Foreign workers import to Taiwan since 1989, and only by few industries. Especially, after 1993, the manufacturing industry imports the foreign workers. The importation of foreign workers slow down the raising rate of wages for women, and alleviates some wage burden for firm. However, the importation of foreign workers may worsen the gender wage gap. In order to protect working right of women as well as lower labor cost of employers, the government of Taiwan (ROC) should compensate with female workers due to importation of foreign workers.


Figure 1: The Industrial Proportion of Covered Sector (1978-1998)

Covered Sector

0

5

10

15

20

25

30

35

40

78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98

Year

Perc

entag

e

FarmMiningManufacturingElectricityConstructionTransportation

Figure 2: The Industrial Proportion of Uncovered Sector (1978-1998)

Uncovered Sector

0

5

10

15

20

25

78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98

Year

Perc

entag

e

CommerceFinanceBusinessSocial ServicePublic Adminstration

Source: Year Book of Manpower Survey Statistics, Taiwan Area, Republic of China, 1998. Note: All numbers are percentages; the numbers in the parenthesis are male and female ratios, respectively.


Table 1: The Statistics for Overtime Working Hours before 1998 Covered Sector Uncovered Sector

______________________________ _______________________________ Year Women Men Women Men (1) (2) (3) (4)

1978 2.7583 4.5290 4.0338 4.4568 1979 2.8578 4.2750 4.5509 4.7068 1980 2.4171 3.5813 3.5326 3.6841 1981 1.5623 2.5109 3.9047 3.9044 1982 1.4291 2.0936 2.9342 2.8214 1983 1.9152 2.4481 3.4436 3.2121 1984 1.5463 2.3213 3.0225 3.2369 1985 1.7995 2.3527 3.6073 3.5232 1986 1.9167 2.9293 3.7934 3.7089 1987 1.6068 2.4840 4.1695 3.4126 1988 1.3285 2.2528 2.9479 2.5875 1989 1.0578 1.6920 2.7023 2.5398 1990 0.8687 1.5911 2.2978 2.3126 1991 0.8827 1.5209 2.6540 2.4755 1992 0.9400 1.8117 1.9968 2.0820 1993 0.8730 1.8343 2.5500 2.4301 1994 0.7338 1.3089 1.8677 2.0447 1995 0.5571 1.0105 1.9011 1.9961 1996 0.6059 1.0230 1.8295 1.9884 1997 0.4501 0.8896 1.6007 1.8234 1998 0.4577 0.7695 1.5347 1.8609 Source: The table is calculated from the data of Manpower utility survey from 1979 to 1998.


Table 2: The Statistics for Overtime Working Hours in the Original Uncovered Sector Inspected Sector Un-inspected Sector

______________________________ _______________________________ Year Women Men Women Men (1) (2) (3) (4)

1987 2.0743 2.7833 2.6523 3.1905 1988 1.8423 1.9829 2.3984 2.8360 1989 1.4323 1.8792 2.0525 2.4502 1990 0.8225 1.2587 1.7589 2.0157 1991 0.9105 1.0617 1.5847 1.9346 1992 0.8015 1.3845 1.6148 1.8206 1993 0.7630 1.3435 1.5161 1.9835 1994 1.1709 1.6739 1.7235 2.0474 1995 0.5536 1.1625 1.4028 1.5588 1996 0.6970 1.1065 1.3364 1.3252 1997 0.8253 1.0091 1.2880 1.3458 1998 0.6429 0.9126 1.1403 1.2232 1999 0.5317 0.9958 1.1444 1.1336 2000 0.5848 0.9263 1.1729 1.1143 2001 2.7228 3.2583 3.8207 3.7347 2002 2.5889 3.1176 3.6726 3.6748 2003 2.3915 2.9959 3.6865 3.7225 2004 2.6103 3.2956 3.8084 4.0530 2005 2.7360 3.3074 3.8733 4.3414 2006 2.9898 3.8817 3.7253 4.2427 Source: The table is calculated from the data of Manpower utility survey from 1987 to 2006.


Table 3: The Effect of Overtime Working Hour Restrictions on Wages Before 1999

(A) (B) _________________________________ _______________________________

(1) (2) (3) (4)

a.***statistically significant at the 1% level. **statistically significant at the 5% level. * statistically significant at the 10% level.

1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 Adj. R sq. F value Sample Size

0.0692 0.0553 0.0159 0.0679 0.0277 0.0146 0.0148 0.0525 0.0403 0.0332 0.0542 0.0671 0.0903 0.0310 0.5870 3264.6 163009

(0.0282)** (0.0306)* (0.0274) (0.0277)** (0.0250) (0.0260) (0.0235) (0.0239)** (0.0220)* (0.0213) (0.0212)*** (0.0207)*** (0.0205)*** (0.0196)

0.0686 0.0547 0.0139 0.0680 0.0276 0.0145 0.0177 0.0563 0.0796 0.0734 0.0923 0.1052 0.1281 0.0695 0.5875 3181.3 163009

(0.0288)** (0.0295)* (0.0277) (0.0274)** (0.0247) (0.0267) (0.0242) (0.0241)** (0.0251)*** (0.0243)*** (0.0239)*** (0.0237)*** (0.0232)*** (0.0228)***

b. As seen in columns (1) and (3) of Table 3, the coefficients of iYearCoverGender ×× can estimate the legislation effect, and it

calculated from Equation (15). The columns (2) and (4) are standard deviation. c. The column (A) are the results include the effect of foreign workers, and the column (B) are the results exclude the effect of foreign workers.


Table 4: The Effect of Overtime Working Hour Restrictions on Wages after 1998 (A) (B) ________________________ ___________________________________ (1) (2) (3) (4) (5) (6)

1999 2000 2001 2002 2003 2004 2005 2006 Adj. R sq. F value Sample Size

0.0435 (0.0445) 0.0238 (0.0439) -0.0011 (0.0408) -0.0229 (0.0425) -0.0893 (0.0467)* -0.0671 (0.0411) -0.0276 (0.0380) -0.0600 (0.0373) 0.5197 1150.55 49,934

-0.0826 (0.0208)*** -0.0630 (0.0215)*** -0.0809 (0.0212)*** -0.0672 (0.0221)*** -0.0844 (0.0231)*** -0.0384 (0.0215)* -0.0531 (0.0222)** -0.0778 (0.0210)*** 0.5904 3963.62 129,232

0.1261 (0.0491)*** 0.0868 (0.0489)* 0.0798 (0.0460)* 0.0443 (0.0480) -0.0049 (0.0521) -0.0287 (0.0464) 0.0255 (0.0440) 0.0178 (0.0428)

0.0405 (0.0449) 0.0213 (0.0446) -0.0034 (0.0420) -0.0305 (0.0434) -0.0980 (0.0434)** -0.0755 (0.0417)* -0.0390 (0.0390) -0.0670 (0.0380)* 0.5220 1114 49,934

-0.0823 (0.0212)*** -0.0625 (0.0215)*** -0.0808 (0.0210)*** -0.0723 (0.0219)*** -0.0862 (0.0228)*** -0.0411 (0.0214)* -0.0542 (0.0221)** -0.0784 (0.0212)*** 0.5914 3818 129,232

0.1228 (0.0496)** 0.0838 (0.0495)* 0.0774 (0.0469)* 0.0418 (0.0486) -0.0118 (0.0490) -0.0344 (0.0468) 0.0152 (0.0448) 0.0085 (0.0435)

a. Inside the parenthesis is standard deviation, ***statistically significant at the 1% level. **statistically significant at the 5% level. * statistically significant at the 10% level.

b. The columns (1), (2), (4) and (5) are the coefficients of iYearSizeGender ×× calculated from Equation (16), it can estimate the

DDD legislation effect. The columns (1) and (4) are calculated from the sample of original uncovered sector, and the columns (2) and (5) are calculated from the sample of original covered sector. c. The column (A) are the results include the effect of foreign workers, and the column (B) are the results exclude the effect of foreign workers. d. Both columns (3) and (6) are calculated from the differences from original uncovered and covered sectors. The column (3) will be equal to the column (1) minus (2), and column (6) will be equal to column (4) minus (5).


References Arrigo, G. L. 1985. “Economic and Political Control of Women workers in Multinational Electronics Factories in Taiwan: Martial Law Coercion and World Market Uncertainty,” Contemporary Marxism, 11, 77-95. Arellano, M. 1987. “Computing Robust Standard Errors for Within-groups Estimators,” Oxford Bulletin of Economics and Statistics, 49(4), 431-434. Bell, D. N. F. and R. A. Hart. 2003. “Wages, Hours, and Overtime Premia: Evidence from the British Labor Market,” Industrial & Labor Relations Review, 56(3), 470-480. Blackburn, M. L. 1997. “Misspecified Skedastic Functions in Grouped-Data Models,” Economics Letters, 55, 1-8. Borjas, G. 2003. “The Labor Demand Curve is Downward Sloping: Reexamining the Impact of Immigration on the Labor Market,” Quarterly Journal of Economics, 118(1), 1335-1374. Borjas, G. 1999. “The Economics Analysis of Immigration,” Handbook of Labor Economics, eds. Ashenfelter O. and Card D. Amsterdam:3(a): 1697-1760. Börsch-Supan, A. 2002. Reduction of working time: Does it decrease unemployment? Arbeitspapier 03-02, MEA—Mannheimer Foschungsinstitut Ökonomie and

Demographischer Wandel, Universität Mannheim. Cheery, M. 2004. “Are salaried workers compensated for overtime hours?” Journal of Labor research, 25(3), 485-494. Chen, L. H. and W. C. Wang. 2011. “The impact of the overtime policy reform–evidence from the low-paid workers in Taiwan,” Applied Economics, 43(1), 2011, 75 – 90. De New, J. P. and K. F. Zimmerman. 1994. “Native wage impacts of foreign labor: a random effects panel analysis,” Journal of Population Economics, 7(2), 177-192. Evans, D. S. 1987. “Tests of Alternative Theories of Firm Growth,” Journal of Political Economy, 95(4), 657-674. Hart, R. A. and Y. Ma. 2010. “Wage–Hours Contracts, Overtime Working and Premium Pay,” Labour Economics, 17(1), 170-179. Hart, R. A. 2004. The Economics of Overtime Working, Cambridge, Cambridge University Press, UK. Hart, R. A., J. R. Malley and R. J. Ruffell. 1996. “What Shapes are Overtime Premium Schedules? Some Evidence from Japan, the UK, and the US,” Economics Letters, 53, 97-102. Lai, Y. C. and S. Masters. 2005. “The Effects of Maternity and Pregnancy Benefits on Women’sWages and Employment in Taiwan 1984-1996,” Industrial and Labor Relation Review, 58(2), 274-281.


La Jeunesse, R. 1999. “Toward an Efficiency Week”, Challenge, 42(1), 92-109. Muhlberger, P. 2000.”Moral Reasoning Effects on Political Participation,” Political Psychology, 21(4), 667-695. Rosenzweig, M. R. and K. I. Wolpin. 1986.“Evaluating the Effects of Optimally Distributed Public Programs: Child Health and Family Planning Interventions,” American Economic Review, 76(3), 470-482. van Soest, A., I. Woittiez and A. Kapteyn. 1990. “Labor Supply, Income Taxes, and Hours Restrictions in the Netherlands.” Journal of Human Resources, 25(3), 517-558. Tummers M. P. and I. Woittiez. 1991. “A Simultaneous Wage and Labor Supply Model with Hours Restrictions.” Journal of Human Resources, 26(3), 393-423. Zveglich Jr. J. E. and Y. M. Ridgers. 2003. “The Impact of Protective Measures for Female Workers,” Journal of Labor Economics, 21(3), 533-555.


Appendix Table 1: Variable Definitions and Data Statistics Variables Means Std. Dev.

Wage: Log Value of the Monthly Wage Gender:

=1: Women =0: Men

Dummy variables for the following years: Year85 Year86 Year87 Year88 Year89 Year90 Year91 Year92 Year93

Year94 Year95 Year96




Year06 Td:Time Trend Cover

=1: Industry covered by the LSL. =0: Industry uncovered by the LSL.

Size

=1: firm inspected by the CLA. =0: firm uninspected by the CLA

Edu: Years of schooling. Exp: Potential experience in years.

=Age - ED - 6. Ten: Tenure years. Mar=1: Married Women

=0:Single Women Hour: Log Value of the Monthly Working hours Fw=1:The industry that import foreign workers =0: Others

8.9701 0.2995 0.0384 0.0375 0.0425 0.0432 0.0428 0.0425 0.0412 0.0427 0.0455 0.0500 0.0533 0.0521 0.0555 0.0589 0.0590 0.0608 0.0093 0.0097 0.0099 0.0100 0.0102 0.0106 18.0238 0.7817 0.2207 10.3755 15.9211 4.2077 0.5136 3.5393 0.3839

(0.5485) (0.4581) (0.1921) (0.1899) (0.2017) (0.2033) (0.2024) (0.2018) (0.1986) (0.2022) (0.2083) (0.2181) (0.2247) (0.2222) (0.2289) (0.2355) (0.2355) (0.2390) (0.0959) (0.0982) (0.0989) (0.0997) (0.1003) (0.1024) (7.6870)) (0.4131) (0.4147) (3.7367) ( 11.7066) (4.5365) (0.4998) (0.8971) (0.4863)


Appendix Table 2: The DDD Estimators for Various Age Cutoffs 35-59 40-59

_____________________________________ ______________________________________ Coef. Std. Dev. Coef. Std. Dev.

85 86 87 88 89 90 91 92 93 94 95 96 97 98

0.0685 0.0750 0.0037 0.0662 -0.0013 0.0145 -0.0064 0.0573 0.0576 0.0570 0.0556 0.0542 0.0801 0.0102

(0.0368)* (0.0400)* (0.0355) (0.0374)* (0.0320) (0.0336) (0.0294) (0.0303)* (0.0283)** (0.0271)** (0.0270)** (0.0262)** (0.0257)*** (0.0245)

0.1079 0.0806 0.0483 0.0767 0.0052 0.0110 -0.019 0.0568 0.0550 0.0648 0.0925 0.0823 0.0983 0.0226

(0.0473)** (0.0478)* (0.0455) (0.0450)* (0.0397) (0.0446) (0.0378) (0.0393) (0.0373) (0.0342) (0.0343)*** (0.0333)** (0.0318)*** (0.0304)

a. ***statistically significant at the 1% level. **statistically significant at the 5% level. * statistically significant at the 10% level.

b. Inside the Table is the coefficients of iYearCoverGender ×× , it can estimate the legislation effect.

αβ hNL =

Acknowledgment I would like to thank the Center for Survey Research, Academia Sinica for providing me the data of the Manpower Utility Survey. Footnotes * Department of Finance, Shih Chien University Kaohsiung Campus, 372 Minchuing Second Road, Kaohsiung 806, Taiwan; E-mail: [email protected]; Tel.: + 886 7 3326801; Fax: + 886 7 5371841. 1 The case will occur if working hours constitute an investment into either workers’ human capital or their reputation within the firm. If workers accumulate either more skills or gain a reputation for being a hardworking by working more overtime hours, then they will be compensated in the future by being paid higher salaries (Cherry, 2004). 2 The industries pertaining to social and personal services comprise both covered and uncovered sectors, including the mass media and other industries. Before 1998, only the mass media was a covered sector. 3 Before the enactment of the LSL in 1984, some firms only hired young single women and forced them to resign when they got married or became pregnant (Arrigo, 1985). 4 The LSL also deals with pension funds. Employers must contribute 2-15% of their employees’ wages to pension funds. 5 The labor supply will be equal to . The βγ can represent the returns to workers ( 10 << βγ ) and the βγ−1 will indicate the returns to capital, andαγ will be equivalent to the returns to working hours. 6 Hart et al. (1996) investigate that a minimum premium (k) of 1.25 and a highest premium (s) of 1.32 for Japan. 7 By working shorter hours, employees have more time to improve their health, invest in their training and they are more rested and alert during working time and therefore make fewer errors. Shortening working hours are also argued to improve staff recruitment, retention, motivation and commitment, through the offer of preferential working conditions in comparison with other employers (La Jeunesse, 1999).


8 As seen in Table 2 in the Appendix, the robustness tests use samples with various age cutoffs samples ranging from 30-59, 40-59 years. 9 Farm labor will be less affected by the LSL than labor in other sectors. Self-employed workers have flexible working hours. Public sector workers have been excluded from this study because the government is not subject to the same economic pressures as private sector workers. 10 On the other hand, if results for the OLS and MLE procedures are similar, indicating that the sample selection bias will not be a problem in the empirical test (Evan, 1987). In unreported statistics, both results for the OLS and MLE procedures are similar, indicating that the sample selection bias will not affect the empirical results in this study. 11 Based on the Blackburn (1997) method, the weights are determined by using the following procedures. First, the equations are estimated via the OLS method. Second, the square residuals from these models are regressed against a constant term. Finally, the square roots of the inverse of the predicted values from the second-stage OLS regression are used as weights in the final set of estimates. 12 The manpower utility survey in 1978 does not provide the information of firm size. Therefore, this paper uses the data from 1979.

The Effect of Overtime Working Hour Restrictions …...working hours that yield the highest utility level for their work; however, women cannot work for optimum working hours as a

Documents