(This essay was published in Hong Kong Economic Journal on 6 February 2013)


Discrimination can manifest itself in all aspects of life. In this essay I focus on labor market discrimination against women in Hong Kong. This form of discrimination is important in its effects because it directly impacts the economic well-being of discriminated minorities in a substantial way, affects social mobility with possible cross-generation effects, limits the economic potential of victims, leads to social discordance in society in a major way, and retards economic growth in the aggregate. In short it is a significant issue.


In Hong Kong the task of tackling discrimination is focused on the work of The Equal Opportunities Commission (EOC), a statutory body set up in 1996 to implement anti-discrimination legislation. Economists do not necessarily doubt the value of such public commissions in fighting against discrimination, but have argued that competition in the market place is an important component of the fight against labor market discrimination. Promoting openness and lowering entry barriers to enhance market competition can be effective methods to limit discrimination. The pioneering work in this area is Professor Gary Becker’s work, The Economics of Discrimination, published in 1957 based on his doctorate dissertation at the University of Chicago.


Study of Discrimination Widely Acclaimed


His work received wide acclaim immediately upon publication. Leading economists and sociologists endorsed his work. Economist Professor Melvin Reder wrote in the American Economic Review, “This is an unusual book; not only is it filled with ingenious theorizing but the implications of the theory are boldly confronted with facts.” Sociologist Professor Karl Schuessler wrote in the American Sociological Review, “The author’s solution to the problem of measuring the motive behind actual discrimination is something of a tour de force.” Becker’s work has continued to influence research on discrimination to this day. A recent study published in 2007, fifty years after the original study first appeared, concluded that Becker’s analysis and empirical predictions have withstood the tests of time.


Becker’s analysis examines the economic effects of discrimination in the market place because of race, religion, sex, color, social class, personality, sexual preference, or other non-pecuniary considerations. Discrimination is defined as a situation where a prejudiced employer is only willing to offer below market pay to an employee belonging to a discriminated minority, even though that employee is as productive as any other employee who is not discriminated against.


Discrimination acts as a “tax wedge” on the employer for indulging in his prejudices. The prejudiced employer offers above market pay to an employee who is not in the discriminated minority. Discrimination thus is economically detrimental not only to those who are discriminated against, but also to those who practice discrimination. Employers who do not discriminate will be more cost efficient and more productive. They do not have such a “tax wedge”.


By demonstrating that employer discrimination in the market place reduces their own incomes as well as those of the minority, Becker successfully challenged the Marxist view that discrimination helps the person who discriminates. He pointed out that if an employer refuses to hire a productive worker simply because of race, religion, sex and so on, that employer loses out on a valuable opportunity.


According to Becker’s analysis, discrimination will be less pervasive in more competitive industries because companies that discriminate will lose market share to companies that do not. He also presented evidence that discrimination is more pervasive in more-regulated, and therefore less-competitive, industries. The idea that discrimination is costly to the discriminator is common sense among economists today, and that is due to Becker. But this is still not well known among the general public.


The main result of Becker’s analysis is that in the short-run, less prejudiced employers earn higher profits than their more prejudiced counterparts who choose to employ more expensive workers. Since capital can move freely in the long run under perfect competition, unprejudiced employers will expand at the expense of their more prejudiced counterparts. Competition thus seems to ensure that prejudiced firms are driven out of the market in the long run. In the case of gender discrimination, pay gaps between women and men due to employer prejudice will disappear in the long run. Becker, however, also cautioned that discrimination could persist if there are factors that limit the amount of competition in the labor market or in the product market.


In Hong Kong, women represent by far the largest group of individuals in the economy identified by the EOC as potential victims of discrimination. Is discrimination against women in the labor market a serious problem here?


Table 1 shows the percentage of pay differences in monthly earnings between women and men. We use data on employees only and domestic helpers are excluded from the analysis. The average pay difference in monthly earnings between women and men among all employees in the years 1976-1996 was quite high at -34.4%. The average pay difference narrowed during 1996-2011 and was at -20.0%. This represents a narrowing of about 14.4% between the two periods. 


Table 1: Percentage Pay Differences between Women and Men: Crude and After Controlling for Schooling and Experience



Crude Differences in Monthly Earnings

Remaining Differences in Monthly Earnings After Controlling for Schooling and Experience

  All Employees Never-married Employees Ever-married Employees Never-married Employees Ever-married Employees




























































Note: Following Jacob Mincer’s pioneering work on Schooling, Experience, and Earnings (1974) experience is defined as the number of years since completion of schooling.


Earnings Reflect Return to Schooling


The fact that some men do better than women in the labor market does not, in itself, signify the presence of discrimination. It would be more surprising if such differences were not observed. The question is, why do women earn significantly lower pay as a group?


One reason for the difference in monthly earnings and hourly wage rates between women and men is that women have less schooling than men. Schooling is an important form of human capital. Professor Jacob Mincer has shown that the percentage difference in earnings between two individuals due to one additional year of schooling can be shown to be equal to the rate of return to schooling. Therefore, if the rate of return is 10% then a person with one more year of schooling will earn 10% more earnings. 


Another reason for the difference in earnings is the amount of on-the-job training an employee acquires through work experience. Earnings have generally been found to increase with years of work experience reflecting the accumulation of human capital after schooling. The amount of post-schooling investment generally declines over time as one approaches retirement because the incentive to learn work related skills and knowledge just before one retires is usually minimal. Earnings therefore rise with years of experience and decline as the employee approaches retirement.


For men, the number of years after completion of schooling is a fairly good measure of their on-the-job training, as they tend to hold full-time jobs continuously. Moreover, men tend to spend more time working in the labor market after they marry and have children. The division of labor in the family usually allows men to accumulate more on-the-job training, while the opposite often happens to married women.


Women’s work experience tends to become fragmented once they marry. During the childbearing and childrearing years they often withdraw from the labor force or work only part-time. They may return to the labor force after their children have grown up and are more able to care for themselves. But the number years after completion of schooling is a poor measure of the amount of on-the-job training they acquire. Conversely, never-married women (and men) can fairly reliably measure their full-time work experience by the number of years after schooling.  It is therefore natural to expect the pay differences between ever-married women and men to be greater than that between never-married women and men.


Pay Discrimination No Longer Exists


This scenario has in fact been the case in Hong Kong, but some important changes have happened in recent years. Table 1 shows the crude percentage pay differences in monthly earnings between women and men by marital status. Ever-married employees are those who are married, divorced or widowed. The average percentage pay difference among ever-married women and men in the years 1976-1996 was high at -48.1%. The corresponding figures among never-married employees were much smaller and averaged -14.2%. This is a difference of 33.9%. Clearly pay differences were very different then between ever-married and never-married employees.


However, the percentage pay differences dropped significantly in the years 1996-2011. The average percentage pay difference was around -32.4% among ever-married employees. The corresponding figures among never-married employees were not only smaller but reversed sign to average around 3.0%. This is a difference of 35.4%. The percentage pay differences have narrowed from the first period 1976-1996 to the second period 1996-2011. Women have narrowed the gap by 15.7% among ever-married employees and 17.2% among never-married employees.


What is happening here? Why have the percentage pay differences dropped? More reliable estimates are needed to determine the impact of sex discrimination on pay difference between women and men. One cannot simply rely on the crude percentage pay differences in Table 1. A better approach is to estimate the remaining percentage pay differences after controlling for years of schooling and years of work experience.


Multiple linear regression methods (多元線性回歸方法) were used to estimate these remaining pay differences. In the absence of data on the actual years of work experience, the number of years since completion of schooling was used as an approximation. This is not likely to be a good measure of actual work experience of ever-married women whose work experience is often fragmented after marriage.  But it is a good approximation of the work experience of men and never married women.


Table 1 also shows estimates of the remaining percentage pay differences between women and men for never-married and ever-married employees after controlling for years of schooling and experience. The estimates of never-married employees are the best approximate measures of any possible pay consequences of sex discrimination in the labor market.


We found that in the period 1976-1996 never-married women earned an average of about 12.8% less than never-married men per month.  During 1996-2011, their pay differences in monthly earnings had dropped to around -1.4%; representing a narrowing of 11.4%.


Based on these findings one would have to conclude that sex discrimination against women in the labor market as reflected through pay differences has probably been absent since at least 1996.


Should we also conclude that the drop in pay differences between women and men is the result of the good work of the EOC? I would hesitate to draw such a conclusion. First, 1996 is the year the EOC was established and it is also the same year the pay differences dropped. One has to be very skeptical about the effectiveness of a new institution in making any significant policy impact in the first year it starts operating. Second, pay adjustments are typically made once a year. Even if the EOC could perform miracles it could not have changed pay contracts that were set before it started to function. 


So why did the pay difference between men and women start to narrow around 1996?  The year 1996 represented a turning point in the sex composition of the labor force. More women were joining the labor force at a fast-rising pace. The number of men joining the labor force stopped growing after 1996. As a consequence in the period 1986-1996 the ratio of women to men in the labor force was stagnant –– 0.59 in 1986, 0.57 in 1991, 0.58 in 1996, but it began to accelerate upwards afterwards, reaching 0.65 in 2001, 0.74 in 2006, and 0.79 in 2011 (see Table 2).


Table 2: Number of Economically Active Men and Women in the Labor Force 1976-2011 (in 1000’s)


Number of Men

Men’s        year-on-year percentage growth

Number of Women

Women’s    year-on-year percentage growth

Ratio of Women to Men





















































Note: Domestic helpers excluded.


The rapid increase of women workers made men scarcer in the labor market. This made it costly for prejudiced employers to discriminate against women. Competition in the market place eliminated the pay difference between women and men that had existed prior to 1996.


Although the average percentage pay difference between never-married men and women was -12.8% prior to 1996, I believe only a fraction of that pay difference can be attributed to sex discrimination in the labor market. The expected marriage rate and fertility rate of single women in the period 1976-1996 was probably higher than the subsequent period 1996-2011. Since most single women in the labor force expect to become married and have children; therefore, they tend to spend less time at work and invest less time in on-the-job training to augment their labor market skills and job career. For this reason earnings of single women should still be lower than single men with the same level of experience.


How the pay difference is apportioned between lower investments in on-the-job training and labor market discrimination against women is not known.  But given that the estimated -12.8% pay difference is not a large number, the adverse effect of sex discrimination on pay cannot have been very large before 1996. Today we have reached the stage where single women are getting higher pay than men. It is no surprise that more men are staying home. The significance of these developments is unknown and has hardly been studied.


What can be concluded is that sex discrimination in the labor market is unimportant today. Given that women still continue to have disrupted careers due to marriage and fertility they will probably still lag behind men in senior positions. But the gap will continue to narrow as marriage and fertility rates continue to fall and women’s education surpass men’s. 





Gary S. Becker, The Economics of Discrimination, The University of Chicago Press, 1957.


Kerwin Kofi Charles and Jonathan Guryan, “Prejudice and the Economics of Discrimination, NBER Working Paper 13661, National Bureau of Economic Research, Cambridge, MA, December 2007.


Jacob A. Mincer, Schooling, Experience, and Earnings, Columbia University Press, 1974

Share 分享到:
Print Friendly

Leave a Reply

Your email address will not be published.