Chapter

VI Trends in Wage Inequality, 1981–2001

Author(s):
William Lee, Jorge Chan-Lau, Dora Iakova, Papa N'Diaye, Tao Wang, Ida Liu, Hong Liang, and Eswar Prasad
Published Date:
February 2004
Share
  • ShareShare
Show Summary Details
Author(s)
Dora Iakova

Income inequality in Hong Kong SAR has increased rapidly over the last twenty years (Figure 6.1). The Gini coefficient, a commonly used summary measure of household income inequality, rose from 0.451 in 1981 to 0.525 in 2001. At present, Hong Kong SAR ranks among the economies with the most uneven distributions of income in the world, even though income disparity has also widened in many other high-income economies over the last thirty years.1 Rising wage inequality has been one of the driving forces behind the increase in household-income inequality.2

Figure 6.1.Selected Regions of World: Average Gini Coefficients

Sources: Deininger and Squire (1996); and Hong Kong SAR, Census and Statistics Department.

Note: The Gini coefficient for Hong Kong SAR is estimated for unadjusted household income.

The main goal of this section is to document the evolution of wage inequality during 1981–2001.3 The analysis indicates that the adjustment of the labor market to structural shifts was accompanied by an increase in wage disparity and a rise in returns to education. In this respect, developments in wage inequality are similar to those observed in the United States during the 1980s and 1990s. At the same time, real wages in Hong Kong SAR increased significantly across the entire distribution, reflecting positive spillovers from rapid economic growth. This finding contrasts with results for the United States, where rising wage inequality since the late 1970s has been characterized by declining real wages in the lower deciles of the wage distribution.

The last two decades were characterized by rapid structural change in Hong Kong SAR. Since the mainland opened to foreign investment in 1979, manufacturing production was gradually outsourced from Hong Kong SAR to the mainland, and there was a significant shift of employment toward the service sectors. Hong Kong SAR increasingly turned into an entrepôt center for mainland-produced manufacturing exports. The share of manufacturing value added declined to 5 percent of GDP from 23 percent between 1980 and 2000. Domestic goods exports declined to 14 percent of GDP from 48 percent, while reexports increased to 99 percent of GDP from 19 percent over the same period. The stages of production outsourced to the mainland were mostly those intensive in unskilled labor, while skill-intensive services related to manufacturing remained in Hong Kong SAR (see Feenstra and Hanson, 2001).

Examining the development of wage inequality over this period of massive relocation of production is important, since it could shed light on the debate of the effect of globalization of production on relative wages. Feenstra and Hanson (1996) suggest that outsourcing of the low-skill-intensive stages of production within each industry would result in an increase in the relative demand for skilled labor. For Hong Kong SAR, the outsourcing of production to the mainland over the last two decades has been very significant. One rough measure of outsourcing is the size of outward-processing trade (involving raw materials export to the mainland for processing and a subsequent reimportation of processed goods). Such trade was nonexistent in the late 1970s, but by 2000 it accounted for 73 percent of Hong Kong SAR’s domestic exports to the mainland, 79 percent of its imports from the mainland, and 85 percent of reexports originating on the mainland. Thus, it is likely that a substantial part of within-industry skill upgrading and of the increase in wage inequality in Hong Kong SAR is related to the process of outsourcing, and this section presents evidence in support of this theory.4

Evolution of Wage Inequality

Dataset

The dataset used in this section is constructed from a representative 1 percent sample of individuals from the Hong Kong Population and By-Population Censuses conducted in 1981, 1986, 1991, 1996, and 2001. The data include information on after-tax monthly earnings from main and secondary employment, nonlabor income, industry sector of employment, occupation, education, demographics, and household characteristics for every individual. The wage variable examined is real monthly earnings from main employment. (The composite consumer price index, 1996 = 100, is used as the price deflator.) The analysis is restricted to employed people between 18 and 65 years of age.5

Summary statistics of the data are presented in Table 6.1. Median real monthly earnings increased significantly over the period 1981–2001, although the pace of growth of real earnings slowed between 1996 and 2001 in line with lower GDP growth. The share of women in the employed population increased by 10 percentage points over the period, reaching 45 percent in 2001.

Table 6.1.Summary Statistics for Wage-Analysis Sample, 1981–2001
19811986199119962001Change

1981–2001
Number of observations19,86023,11523,41728,64630,712
Share of men0.70.60.60.60.6
Share of women0.30.40.40.40.4
Median (log real wages)8.58.79.09.29.30.8
Men8.68.89.19.29.40.8
Women8.38.48.89.09.10.8
Mean (log real wages)8.68.89.19.29.40.8
Men8.78.99.29.39.50.8
Women8.38.68.99.19.20.9
Sources: Hong Kong SAR Census; and IMF staff estimates.
Sources: Hong Kong SAR Census; and IMF staff estimates.

There has been a significant shift of employment toward the service industries (Table 6.2). The share of manufacturing employment declined from 40 percent in 1981 to 12 percent in 2001, while the share of employment in the service sectors increased from 49 percent to 80 percent. Although an increasing employment share of services has been observed in most industrial countries in recent years, this shift has been much more pronounced in Hong Kong SAR.

Table 6.2.All Workers: Employment and Earnings, by Industry
19811986199119962001Change,

1981–2001
Share in Total Employment (percent)
Utilities, agriculture2.21.91.41.00.8−1.4
Manufacturing40.335.828.018.812.2−28.1
Construction8.96.36.97.97.5−1.5
Wholesale, retail, hotels/restaurants15.618.317.218.718.42.8
Import/export3.34.14.46.27.54.3
Transport, storage, and communication8.48.010.210.911.53.0
Financing, insurance, real estate, business services5.46.511.313.716.411.0
Public administration, education, health, social services11.514.514.415.116.85.4
Personal services4.34.56.27.58.84.5
Services total495664728031
Total100.0100.0100.0100.0100.0
Median Log Wage
Utilities, agriculture8.508.859.119.329.480.98
Manufacturing8.288.548.939.109.321.03
Construction8.718.789.119.169.220.51
Wholesale, retail, hotels/restaurants8.508.708.938.999.110.61
Import/export8.668.859.199.359.440.78
Transport, storage, and communication8.798.859.119.219.270.48
Financing, insurance, real estate, business services8.638.919.269.399.560.93
Public administration, education, health, social services8.799.039.309.449.650.86
Personal services8.288.398.498.238.25−0.03
Mean Log Wage
Utilities, agriculture8.598.739.169.389.540.95
Manufacturing8.368.568.929.169.411.05
Construction8.708.769.089.179.320.61
Wholesale, retail, hotels/restaurants8.578.759.019.079.160.59
Import/export8.809.039.309.449.520.72
Transport, storage, and communication8.758.909.129.239.380.63
Financing, insurance, real estate, business services8.769.049.389.509.640.88
Public administration, education, health, social services8.899.139.339.539.670.79
Personal services8.318.448.648.548.510.19
Standard Deviation of Log Wages

(multiplied by 100)
Utilities, agriculture69.774.868.971.177.37.6
Manufacturing51.055.358.463.365.414.4
Construction52.058.857.858.155.13.1
Wholesale, retail, hotels/restaurants55.353.755.856.858.73.3
Import/export65.763.861.964.365.1−0.6
Transport, storage, and communication45.948.250.655.659.313.4
Financing, insurance, real estate, business services59.863.069.473.274.314.5
Public administration, education, health, social services64.968.572.269.378.413.5
Personal services45.349.648.849.450.75.3
Sources: Hong Kong SAR Census data; and IMF staff estimates.
Sources: Hong Kong SAR Census data; and IMF staff estimates.

The median monthly wage increased in all sectors, although at different rates. Wage gains are the lowest in the construction, wholesale and retail trade, transport, storage, and communications sectors. The relative share of low-skilled occupations in these sectors is large, suggesting that wage rises have been more moderate among low-skilled workers. Wage increases in manufacturing have been larger than in any other sector, suggesting that the downsizing of labor in that sector was mostly among low-skilled workers. This is consistent with studies noting that the stages of manufacturing remaining in Hong Kong SAR are increasingly concentrated in sophisticated, high-value-added managerial and administrative services.

The average educational level of the workforce has risen at a rapid rate over the sample period (Table 6.3). Workers with a primary education or less declined from 47 percent to 16 percent of all employees, and the share of workers with a tertiary education quadrupled. The Hong Kong SAR authorities introduced compulsory primary and lower-secondary education in the early 1970s, which can partially account for the increase in the average education level. Median wage increases have been the highest for people with upper-secondary and post-secondary education, supporting the conjecture that wage increases have been higher for the more skilled.

Table 6.3.Employed Men and Women: Employment and Wages, by Education
19811986199119962001Change,

1981–2001
Average Years of Education
All employed7.58.59.310.110.83.2
Men7.78.59.19.910.62.9
Women7.38.59.510.511.03.7
Share in Total Employment (percent)
Primary and no formal4736272016−30.2
Lower-secondary19192020200.9
Upper-secondary273338404214.9
Postsecondary467651.5
Tertiary467131713.0
Median Log Wage
Primary and no formal8.48.58.88.99.00.60
Lower-secondary8.58.78.99.09.10.61
Upper-secondary8.68.89.19.29.30.70
Postsecondary9.19.29.59.59.80.67
Tertiary9.49.69.89.910.00.54
Standard Deviation of Log Wages (multiplied by 100)
Primary and no formal50.352.149.752.254.84.5
Lower-secondary49.248.749.750.953.74.5
Upper-secondary55.355.956.558.664.59.2
Postsecondary62.366.664.866.671.08.7
Tertiary73.980.185.284.585.011.0
Sources: Hong Kong SAR Census; and IMF staff estimates.
Sources: Hong Kong SAR Census; and IMF staff estimates.

Table 6.4 shows the skill intensity of different sectors, measured by the share of workers with upper-secondary education or more. Finance, insurance, real estate, and business services is the most skill-intensive sector, and the share of skilled workers (see the center panel of Table 6.4) has remained constant over time. Among the major sectors, skill upgrading has been most significant in manufacturing.6 The wage premium for workers with higher education has increased fastest in the sectors that produce or distribute tradable goods—manufacturing, transportation, and storage—suggesting that increased international trade and specialization in the production chain have benefited those with higher education.

Table 6.4.Education Attainment and Education Wage Premium, by Industry
19811986199119962001Change,

1981–2001
Average Years of Education, by Industry
Utilities, agriculture5.06.67.98.99.34.3
Manufacturing7.07.78.49.510.43.4
Construction6.67.67.98.48.61.9
Wholesale, retail, hotels/restaurants6.77.68.49.09.42.6
Import/export10.711.211.411.811.91.2
Transport, storage, and communication7.88.38.99.510.12.3
Financing, insurance, real estate, business services11.111.311.511.912.71.6
Public administration, education, health, social services9.810.410.711.411.92.1
Personal services5.88.39.510.310.85.0
Full sample7.58.59.310.110.83.2
Share of Workers with More Than 9 Years of Education, by Industry

(percent)
Utilities, agriculture183238455233.1
Manufacturing243340526035.5
Construction223231353513.0
Wholesale, retail, hotels/restaurants273641454921.6
Import/export78848586813.8
Transport, storage, and communication374248525619.3
Financing, insurance, real estate, business services84848383840.2
Public administration, education, health, social services626772747311.2
Personal services214555636746.5
Full sample354553606429.4
Correlation between changes in industry employment shares and changes in share of skilled workers−0.31−0.66−0.94−0.63
Skilled-Wage Premium, by Industry1
Utilities, agriculture3.67.28.98.111.88.2
Manufacturing4.35.36.99.09.24.9
Construction5.67.17.28.06.20.6
Wholesale, retail, hotels/restaurants4.75.06.06.87.22.5
Import/export8.97.89.29.29.90.9
Transport, storage, and communication4.46.46.38.79.14.8
Financing, insurance, real estate, business services10.411.812.713.112.72.3
Public administration, education, health, social services9.010.812.211.612.83.8
Personal services2.51.61.90.7−0.2−2.7
Full sample6.27.98.89.710.03.8
Sources: Hong Kong SAR Census; and IMF staff estimates.

The estimated skilled-wage premium is the coefficient on years of education in a regression of log wages on years of education, gender, experience, experience squared, and recent immigrant status.

Sources: Hong Kong SAR Census; and IMF staff estimates.

The estimated skilled-wage premium is the coefficient on years of education in a regression of log wages on years of education, gender, experience, experience squared, and recent immigrant status.

Changes in Overall Wage Inequality

Wage inequality has risen substantially over the period, largely owing to rapid increases at the highest deciles of the income distribution. Table 6.5 contains different summary measures of wage dispersion—the standard deviation, the coefficient of variation, the Gini coefficient, and percentile differentials. Most measures indicate a steady increase. (One exception is the subperiod 1986 to 1991, when inequality among women declined slightly according to the first two measures.) The 90/10 percentile differential shows that the dispersion of wages increased sharply during the 1990s. By contrast, relative wages below the median (50/10 percentile differential) have remained unchanged for men and have risen for women only in the second half of the sample period. The increase in the 75/25 percentile diferential is also very modest for men. These statistics suggest that the increase in overall wage inequality has come mostly from rising wages at the top of the distribution, while relative wages in the lower half of the distribution have remained largely unchanged. By these measures, wage inequality in Hong Kong SAR has grown much faster than in, for example, the United Kingdom (Table 6.6).7

Table 6.5.Measures of Wage Inequality, 1981–2001
19811986199119962001Change,

1981–2001
Standard deviation0.580.620.640.680.730.15
Men0.550.600.640.670.690.14
Women0.550.600.590.660.730.19
Coefficient of variation *100.680.710.700.740.780.10
Men0.640.680.690.710.720.09
Women0.660.700.670.730.800.14
Gini coefficient0.340.360.380.410.43
Men0.330.360.380.400.41
Women0.320.350.360.390.43
Wage Percentile Differentials
All workers
p90/p101.41.51.41.81.90.50
p90/p500.80.80.81.01.00.28
p50/p100.60.60.60.80.80.22
p75/p250.60.70.70.80.90.31
Men
p90/p101.31.41.51.61.70.45
p90/p500.70.80.91.01.10.35
p50/p100.50.60.60.60.60.10
p75/p250.60.60.70.80.90.24
Women
p90/p101.31.51.51.72.00.66
p90/p500.70.90.90.91.10.33
p50/p100.60.60.60.80.90.33
p75/p250.50.60.70.91.00.49
Sources: Hong Kong SAR Census; and IMF staff estimates.Notes: The wage variable is the logarithm of real monthly wages. The letter “p” denotes percentile—for example,“p90” denotes the ninetieth percentile.
Sources: Hong Kong SAR Census; and IMF staff estimates.Notes: The wage variable is the logarithm of real monthly wages. The letter “p” denotes percentile—for example,“p90” denotes the ninetieth percentile.
Table 6.6.Change in Log Wage Differentials, by Percentile
All WorkersHong Kong SAR,

1981–2001
United Kingdom,1

1980–98
90–100.500.25
90–500.280.16
50–100.220.09
75–250.310.17
Sources: For Hong Kong SAR data: Hong Kong SAR Census; and IMF staff estimates. For U.K. data: Prasad (2002).

Log hourly wages.

Sources: For Hong Kong SAR data: Hong Kong SAR Census; and IMF staff estimates. For U.K. data: Prasad (2002).

Log hourly wages.

The widening of income disparity in Hong Kong SAR has been accompanied by rising real wages at all deciles of the distribution (Figure 6.2). Although wage growth has been higher at the top percentiles, cumulative real wage growth has been quite substantial even at the bottom half of the distribution. (An exception is the real wage growth for women at the bottom 25 percent during the low-growth period 1996–2001.) This is in contrast to the United States, where real wage growth in the lower deciles was negative in the 1980s and early 1990s, especially for males (see Freeman and Katz, 1994).

Figure 6.2.Real Wage Growth, by Decile

Sources: Hong Kong SAR Census; and IMF staff estimates.

Changes in Inequality Between and Within Groups

The rise in overall wage inequality may reflect a change in the average wages received by different groups in a society (between groups) or an increase in the dispersion of wages within those groups. The population sample is divided into groups by education and by industry of employment to examine the relative contributions of these two factors to changes in inequality over time. The evolution of the 90/10 percentile differential by group suggests that inequality within both industry and education groups has risen.8 An interesting observation is that inequality within the more educated groups is much greater than within less educated groups, suggesting that returns to unobserved skills (or ability) rise with education.

To examine changes in within-group inequality more formally, while controlling for between-group variation in observable skills, the residuals from a set of standard wage regressions are examined (Table 6.7).9 This analysis indicates that within-group inequality accounts for more than three-quarters of total inequality. The change in residual (within-group) inequality also accounts for close to three-quarters of the change in overall inequality.10 Thus, the increase in inequality in Hong Kong SAR mostly reflects growing wage dispersion within education and industry groups. One interpretation of this large increase in inequality, after accounting for the effect of formal education and experience, is that the transformation of the economy from manufacturing to a trade-intermediary and financial center has increased the return to entrepreneurial ability.

Table 6.7.Measures of Residual Wage Inequality: Percentile Differentials
19811986199119962001Change,

1981–2001
p90/p101.131.211.241.351.430.31
p90/p500.560.580.610.670.680.13
p50/p100.570.630.630.690.750.18
Sources: Hong Kong SAR Census; and IMF staff estimates.Notes: Wage residuals are from regressions of log wages on dummies for education, experience, experience squared, marital status, and gender. The letter “p” denotes percentile—for example, “p90” denotes the ninetieth percentile.
Sources: Hong Kong SAR Census; and IMF staff estimates.Notes: Wage residuals are from regressions of log wages on dummies for education, experience, experience squared, marital status, and gender. The letter “p” denotes percentile—for example, “p90” denotes the ninetieth percentile.

Accounting for Evolution of Wage Inequality

Changes in Returns to Skills

A standard human-capital-regression framework is employed to study the evolution of returns to education for employed men and women, controlling for different personal characteristics. The logarithm of monthly wages is regressed on education dummies, labor-market experience, and dummies for marital status and recent immigrant status.11 The results are reported in Table 6.8. The wage premium for upper-secondary education relative to lower-secondary education has risen from 17 percent in 1981 to 36 percent in 2001 for men. For postsecondary and tertiary education, the premiums were significantly higher and have increased faster over the period. The levels of the premiums are at the upper range of those typically found in other advanced economies, although differences in the definition of education variables make cross-country comparisons difficult. The increase in the premiums has also been very rapid. This happened despite the significant increase in workers’ average education, which implies that the increase in demand for more educated workers has persistently outpaced the increase in supply.

Table 6.8.Returns to Education and Experience Based on Human-Capital Equations
MenWomen
1981198619911996200119811986199119962001
Education Premiums1
Education level
Upper-secondary/crafts1.171.261.281.311.361.401.411.451.471.42
Postsecondary/technical institute1.671.901.941.972.152.302.302.202.092.33
Tertiary2.162.582.872.843.022.842.882.592.642.77
Returns to Experience2
Potential experience level
5 years2.73.73.74.24.91.83.12.72.63.6
15 years1.21.81.82.12.60.81.51.11.11.8
25 years−0.30.0−0.10.10.2−0.30.0−0.4−0.3−0.1
Sources: Hong Kong SAR Census; and IMF staff estimates.Note: The results are based on a regression of log wages on years of experience, years of experience squared, and dummies for education, marital status, and recent immigration status. The detailed results were reported in IMF (2002).

The premiums are relative to lower-secondary education (exponent of the ordinary-least-squares (OLS) coefficients).

Returns to experience are evaluated at specific experience levels.

Sources: Hong Kong SAR Census; and IMF staff estimates.Note: The results are based on a regression of log wages on years of experience, years of experience squared, and dummies for education, marital status, and recent immigration status. The detailed results were reported in IMF (2002).

The premiums are relative to lower-secondary education (exponent of the ordinary-least-squares (OLS) coefficients).

Returns to experience are evaluated at specific experience levels.

The returns to experience also have increased over time. One possible explanation is that practical experience has become more valuable during the period of structural change. An alternative, or complementary, explanation could be that any downward wage adjustments have been concentrated among new entrants in the labor market.

Changes in Relative Wages Versus Changes in Relative Unemployment

It is often argued that labor market rigidities in continental Europe have constrained relative wage changes, and adjustment to structural shifts has taken place through higher unemployment for the unskilled.12 In the United States and the United Kingdom, which have relatively flexible labor markets, the adjustment has been accomplished mostly through a decline in relative wages and less so through differential unemployment rates. Hong Kong SAR’s labor market is practically free of institutional constraints—there are no minimum wages and no unemployment insurance; less than 2 percent of the labor force is unionized; income taxes are low; and labor legislation is very limited. Therefore, one would expect the adjustment process to be similar to that in the United States.

In the period of rapid economic growth (before 1997), unemployment among both skilled and unskilled workers was very low (Figure 6.3). This supports the hypothesis that flexible labor markets in Hong Kong SAR have allowed shifts in the relative demand for skilled labor to be accommodated through relative wage changes. However, during the prolonged cyclical downturn since 1997, unemployment among unskilled workers has increased much faster than among skilled workers. This was a period of significant deflation, and it is possible that partial downward nominal rigidity of wages has prevented full adjustment through changes in the price of labor.

Figure 6.3.Unemployment Rate

(In percent)

Sources: Hong Kong SAR, Census and Statistics Department; and IMF staff calculations.

Note: Skilled workers are defined as those with at least an upper-secondary education.

Effects of Sectoral Shifts

Both skill-biased technical change and increased outsourcing to the mainland could have contributed to the increase in demand for skilled workers within industries. One way to differentiate between these explanations is to compare the changes in relative demand for skilled workers in industries that experienced a greater degree of outsourcing to changes in relative demand in industries with less outsourcing. Assuming that skill-biased technological change has been similar for all sectors, such a comparison would reveal whether outsourcing has contributed to the increase in wage inequality.13 A standard decomposition of the growth of the share of skilled employment can be used to illustrate the importance of relative demand shifts within industries versus the reallocation of labor to industries with a higher share of skilled workers as follows:

where the index i denotes industries; j denotes the type of worker (skilled or unskilled); αjit = Ejit/Eit is the group j share of employment in industry i in year t; and Ei = (Eit + Eit–1)/2. The first term of equation (1) is the change in the aggregate share of skilled workers attributable to changes in employment shares between industries that use different proportions of skilled workers. The second term captures within-industry increases in the share of skilled workers. A similar decomposition is performed to analyze changes in the aggregate-wage-bill share of skilled workers.14

Table 6.9 presents the decomposition of the growth in the share of skilled employment in total employment and the decomposition of the skilled-wage-bill share for all manufacturing and nonmanufacturing industries. Between 1986 and 2001, the increase in the share of skilled employment has been much greater in manufacturing than in nonmanufac-turing industries. The within-industry component of the increase in skilled employment accounts for practically all of the increase for the manufacturing industries. The between-industry component is much more significant for the nonmanufacturing industries, which supports the conjecture that outsourcing is linked to greater-than-average within-industry skill upgrading.

Table 6.9.Decomposition of Increase in Share of Skilled Workers:(In percent)
Employment
ManufacturingNonmanufacturing
BetweenWithinTotalBetweenWithinTotal
1981–860.248.008.232.288.0010.28
1986–910.297.127.411.923.815.73
1991–960.8011.3612.170.803.033.82
1996–20010.327.407.721.411.743.15
Wage Bill
1981–860.289.109.382.249.1811.41
1986–910.2810.7411.021.544.355.89
1991–960.8611.0411.900.523.433.95
1996–20010.115.865.971.202.203.40
Sources: Hong Kong SAR Census; and IMF staff estimates.Notes: Skilled workers are defined as those with more than nine years of education. All manufacturing industries were grouped into six categories.
Sources: Hong Kong SAR Census; and IMF staff estimates.Notes: Skilled workers are defined as those with more than nine years of education. All manufacturing industries were grouped into six categories.

Conclusions

This section finds that between 1981 and 2001, real wages in Hong Kong SAR have increased substantially across the wage distribution while wage disparity has risen. Wage inequality within industry groups has increased as industry structures have shifted toward specialization in higher-value-added services. Wage premiums for higher education are large and have generally increased over the period despite an increase in the supply of educated workers, pointing to a significant shift in the relative demand for educated labor. Earnings inequality is greater among more educated groups and has increased over time, suggesting that returns to unobservable skills have increased at higher levels of education.

Both outsourcing and skill-biased technological change could have led to the observed within-industry skill upgrading and increase in skill premiums. It was not the goal of this section to determine the relative contribution of these two influences. The fact that within-industry relative demand shifts were more significant in manufacturing than in other industries, however, supports the conjecture that outsourcing has been an important cause of the observed changes.

The empirical analysis suggests that policies to increase the skill level of the labor force would be most effective in addressing structural imbalances in the labor market. Enrollment in upper- and postsecondary education and government spending on education are much lower in Hong Kong SAR than in OECD countries. Continuing economic integration with the mainland would likely lead to a further increase in the relative demand for skilled labor and rising relative returns to education and other skills. An increase in the education level of the labor force, by itself, may not make the distribution of income more equal, but it will increase equality of opportunity and contribute to sustainable economic growth, which historically has led to improvement of living standards even for the poorest.

1

The World Bank’s World Development Report 1995 ranks Hong Kong SAR as having the highest rate of income inequality among high-income economies.

2

Wages from main employment accounted for 89 percent of total household income in both 1981 and 2001 (based on a representative 1 percent sample from the Population Census).

3

There are few studies of wage inequality in Hong Kong SAR. Suen (1995) examines the effect of a change in the industrial composition on the variance of wages over the period 1976–91. Liu (1997) provides a broad overview of household and wage inequality over the same period. In a recent paper, Hsieh and Woo (2000) attempt to quantify the impact of outsourcing on the shift in relative demand (and, therefore, on relative wages) for skilled workers in manufacturing over the period 1976–96.

4

Other explanations for rising wage disparity offered by the theoretical literature include skill-biased technological change, reductions in labor-market rigidities (such as minimum wages, strong unions, restrictive labor laws, and heavily progressive taxation of income), and growing trade with developing countries. See Wolff (2000) for a review of the literature.

5

Hours worked are not reported in the data, so part-time workers are also included. The earnings data are top-coded. To eliminate outliers and top-coded observations, the bottom 1.5 percent and the top one-half of 1 percent of the wage distribution have been dropped for the purpose of this analysis. Sensitivity analysis has been performed by keeping all data points and adjusting the top-coded observations by assuming a Pareto distribution of the tail. Only the first three summary measures of wage inequality in Table 6.5 were affected by that adjustment.

6

Personal services, which has registered the largest increase in the share of highly educated labor, is a sector with a large and growing share of immigrant labor, which commands very low wages independent of educational qualifications. Total employment in utilities and agriculture is very small, and developments in this sector are just shown for completeness.

7

As has been well documented in the literature, 1980–98 was a period of rapid increase in wage inequality in the United Kingdom (see Prasad, 2002 and the references therein).

8

The detailed results for industry groups and education-attainment groups are reported in IMF (2002).

9

Log wages are regressed on years of experience; experience squared; and dummies for education, gender, and marital status.

10

Separate regressions of wages on a group of education dummies, industry dummies, and gender dummies were estimated. The regression on education dummies reduced the residual inequality the most, relative to overall wage inequality, indicating that inequality between education categories is the largest contributor to between-group inequality.

11

Potential labor-market experience has been imputed from the individual’s age and approximate years of education, based on educational attainment. A second-order polynomial of this variable is included in the regressions.

12

See Davis (1998). Prasad (2004) reviews the literature and does a case study of the German labor market.

13

Acemoglu (2003) suggests that it may be difficult to differentiate empirically between these two explanations, since they may both be present and reinforce each other. For example, more active international trade in certain sectors may induce sector-specific, skill-biased technological change. The magnitude of employment shifts in Hong Kong SAR suggests, however, that the production-outsourcing effect dominates possible technological-change effects.

14

Hsieh and Woo (2000) uses the same technique to isolate the effect of outsourcing. The analysis in this section finds that their results hold when the time period is extended to 2001.

    Other Resources Citing This Publication