(This essay was published in Hong Kong Economic Journal on 8 April 2015.)
Intergenerational upward mobility in the past half century in Hong Kong, measured by correlating parents’ schooling attainment to that of their children’s for those born between 1956 and 1991, has not changed once we take into account waves of emigration.
Two waves of emigration occurred, after the 1967 disturbances and ahead of the coming restoration of sovereignty over Hong Kong to China, that lowered the correlation of schooling attainment compared with peers for those born between 1961-76.
As families left, this created more educational opportunities for the children of families who stayed behind and substantially improved their upward mobility prospects in terms of schooling attainment.
These events later had two consequences. First, for the generations born in the 1980s and 1990s, there is a perception of worsening schooling opportunities compared with those born in the period 1961-1976. Second, job opportunities after 1997 have also become more competitive as some earlier emigrants and their children began to return to Hong Kong in search of better economic opportunities.
This underscores the great importance of investing in educational opportunities today to make up for lost time. Hong Kong’s real challenge is a shortage of relevant employment skills and this has detrimental effects for both economic growth and equality.
Although educational opportunities have been stable over time for each generation as a whole, they are not equal across different families with different backgrounds. An important challenge for society is how to improve the opportunities for those who are born into disadvantage.
It used to be believed that free education for all was a sufficient path to opportunity. The assumption was that the most serious disadvantage for families from poor backgrounds was the capital market imperfection problem, i.e., they could not afford the costs of education. This has been the foundation in most countries for free education policies, and very often publicly provided education. Half a century has since passed of having free education policies in place almost everywhere, but inequality has not improved. So what is the problem?
Current research tells us that although more schooling can in principle provide children with learning opportunities while at school, it does not address children’s learning time outside of school. A lot of learning takes place at home, particularly in early childhood. This explains why the disadvantages of family background cannot be offsetted by schooling alone.
Research on the U.S. experience has shown that half the variation in lifetime income inequality is determined before a child enters secondary school. This implies that an early start is needed to help children in disadvantaged families. The issue is how to identify and reach these children.
In my analysis of intergenerational mobility in schooling attainment in Hong Kong, I studied children aged 20-24 years, by which time they would have started or attained a university education, or not. I used data from the census year 2011, focusing on the group born in 1986-1991. Tables 1 and 2 present the regression coefficients of factors determining child’s schooling attainment. Table 1 is based on father’s schooling attainment and Table 2 is based on mother’s schooling attainment.
Table 1: Regression coefficients of variables determining Child’s schooling attainment in percentile rank in different census years among 20-24 years olds (Based on father’s schooling attainment)
| Census Year | Birth year | Father’s education percentile rank | Male child | No father | Arrived in past 5 years | Parents arrived in past 5 years | Private home owner | Public home owner | Public tenant |
| 1976 | 1951-56 | 0.373 | 0.1* | -3.9 | -1.9* | -1.5* | 3.6 | – | -6.6 |
| 1981 | 1956-61 | 0.319 | -0.9 | -5.6 | -7.3 | -6.1 | 2.6 | 6.1 | -2.6 |
| 1986 | 1961-66 | 0.261 | -1.4 | -5.6 | -6.2 | -4.5 | 6.4 | 10.3 | 2.9 |
| 1991 | 1966-71 | 0.255 | -1.6 | -6.2 | -5.0 | -13.5 | 5.5 | 5.5 | 2.6 |
| 1996 | 1971-76 | 0.242 | -1.5 | -6.3 | -11.2 | -7.1 | 4.7 | 4.8 | 0.8 |
| 2001 | 1976-81 | 0.294 | -1.5 | -3.6 | -12.8 | -9.2 | 6.9 | 4.4 | 0.9 |
| 2006 | 1981-86 | 0.324 | -1.5 | -5.9 | -9.9 | -8.3 | 5.9 | 1.8 | -4.8 |
| 2011 | 1986-91 | 0.307 | -0.4 | -5.1 | -10.4 | -9.3 | 6.1 | 1.8 | -5.1 |
Note: Asterisked coefficients are statistically insignificant. Schooling attainment is measured in percentile ranks.
Table 2: Regression coefficients of variables determining child’s schooling attainment in percentile rank in different census years among 20-24 years olds (Based on mother’s schooling attainment)
| Census Year | Birth year | Mother’s education percentile rank | Male child | No mother | Arrived in past 5 years | Parents arrived in past 5 years | Private home owner | Public home owner | Public tenant |
| 1976 | 1951-56 | 0.379 | 0.1 | -1.6 | -16.5 | 11.5 | 3.7 | – | -7.1 |
| 1981 | 1956-61 | 0.310 | -1.0 | -2.4 | -6.8 | -5.2 | 2.7 | 5.2 | -3.7 |
| 1986 | 1961-66 | 0.258 | -1.5 | -3.2 | -5.0 | -5.2 | 6.5 | 9.4 | 2.1 |
| 1991 | 1966-71 | 0.256 | -1.5 | -2.8 | -8.8 | -11.7 | 5.3 | 5.0 | 1.6 |
| 1996 | 1971-76 | 0.252 | -1.9 | -5.7 | -10.4 | -7.4 | 5.4 | 5.2 | 1.1 |
| 2001 | 1976-81 | 0.292 | -1.7 | -3.4 | -11.9 | -11.1 | 6.8 | 4.4 | 0.6 |
| 2006 | 1981-86 | 0.317 | -1.5 | -5.2 | -7.4 | -9.9 | 5.9 | 1.7 | -4.8 |
| 2011 | 1986-91 | 0.282 | -0.4 | -2.8 | -8.9 | -9.6 | 6.0 | 1.0 | -5.9 |
Note: Same as those in Table 1.
My findings revealed strong evidence of children experiencing significant disadvantages if they were recent immigrants, had parents who were recent immigrants, and grew up in households with a single parent. These are easily identifiable family attributes for focusing preventive and remedial measures.
The growing number of children growing up in single parent households because of rising divorce rates in rich countries, like the U.S., has been found to be the single most critical factor compromising a child’s future success. In Hong Kong, divorce rates have also been rising rapidly in the past three decades. Today Hong Kong is among the top 10 in the world in terms of crude divorce rates.
The much discussed generational gap between young people born in the 1980s and 1990s and their elders, may reflect the fact that a growing proportion of them grew up in single parent households. In 1981, 1.2% of adolescent children aged 11-20 living at home had single parents (see Table 3). These were children born in the 1960s. The figure increased to 2.9% in 1991, 7.0% in 2001, and 11.7% in 2011. These are very large increases, of nearly ten-fold over 20 years. These estimates are likely to be underestimated because parents who have remarried would not be counted as a single parent in the census data.
Table 3: Percentage of children living at home with a single parent
| 1981 Census | 1991 Census | 2001 Census | 2011 Census | |
| Children age 1-5 | 0.4% | 0.6% | 2.8% | 3.9% |
| Children age 6-10 | 0.9% | 1.9% | 5.0% | 8.2% |
| Children age 11-20 | 1.2% | 2.9% | 7.0% | 11.7% |
Between 1981 and 2011, the percentages of children aged 1-10, which are the critical years of early childhood development, who were growing up in single parent households also increased tenfold. Children born in the 1980s and 1990s have had a significantly different upbringing compared to those born in the 1960s and 1970s in this important respect.
I also found children who grew up in families that were homeowners, either in the private sector or publicly subsidized sector, had significant schooling attainment advantages compared to those living in rental housing units in the private sector. This is expected because housing ownership is a very powerful proxy for additional household wealth that is not adequately measured by parents’ education and income. The reported income in household surveys refers to labor earnings and seldom includes income derived from fixed and liquid assets.
Interestingly, I also found that the schooling advantages among those living in publicly-subsidized owned flats, predominantly those living in Homeownership Scheme units, were as strong as those living in owned homes in the private sector for the census years between 1981 and 2001, but the effects weakened significantly in the 2006 and 2011 census years.
The effect of parents’ schooling attainment on children’s schooling attainment for households living in public rental housing has also changed over time. In the census years between 1986 and 2001 they were positive compared to those living in private rental housing, but in 2006 and 2011 they were negative. These would relate to the cohorts of children born between 1981 and 1991.
What explains these differences and changes in housing effects?
The problem is one of dual interaction between two factors: (1) the changing nature of resident composition in the public housing sector, and (2) higher divorce rates among the poor.
In 1976, 48.3% of the households living in public rental housing units were in the top 50% of the household income distribution of the total population (see Table 4). In 2011 this had fallen to 25.0%. Public tenant households are significantly poorer compared to 35 years ago. This is not the case among private housing tenant households. In 1976 some 46.1% of them were in the top 50% of the household income distribution of the total population, by 2001 they had increased to 60.7%. Private housing tenants are now a much wealthier group.
Table 4: Percentage distribution of households by housing accommodation type by income deciles, 1976–2011
| Public Tenants | Private Tenants | Public Homeowners | Private Homeowners | |||||
| 1976 | 2011 | 1976 | 2011 | 1981 | 2011 | 1981 | 2011 | |
| Bottom 30% | 28.3 | 48.1 | 33.2 | 19.8 | 4.6 | 20.9 | 22.3 | 21.5 |
| Bottom 50% | 51.7 | 75.0 | 53.9 | 39.3 | 14.8 | 43.4 | 39.5 | 37.0 |
| Top 50% | 48.3 | 25.0 | 46.1 | 60.7 | 85.2 | 56.6 | 60.5 | 63.0 |
| Top 30% | 24.6 | 7.1 | 27.7 | 42.0 | 59.1 | 28.4 | 41.4 | 44.2 |
A fall-off in household wealth can also be found among households living in government subsidized homeownership flats. In 1976, some 85.2% of them were in the top 50% of the household income distribution of the total population, but by 2011 this had fallen to 56.6%. These homeowners became relatively less well off probably because their housing wealth had not appreciated by as much as those in the private sector and so they were less able to invest more in their children.
In almost every country, we have observed that the poor are more likely to become divorced. Hong Kong is no exception. The number of divorced men and women has risen rapidly over time. In 1981 there were 12,580 and 11,160 divorced men and women. By 1991 this had risen to 21,700 and 28,920, by 2001 to 57,800 and 94,200, and by 2011 to 94,920 and 179,400.
Table 5: Number of married and divorced men and women by housing accommodation (in thousands)
| Public Renter | Private Renter | Public Owner | Private Owner | All | |||
| Married | Male | 1981 | 334 | 369 | 7 | 296 | 1,005 |
| 1991 | 474 | 255 | 101 | 467 | 1,298 | ||
| 2001 | 522 | 269 | 282 | 580 | 1,652 | ||
| 2011 | 513 | 306 | 305 | 680 | 1,803 | ||
| Female | 1981 | 342 | 324 | 7 | 306 | 979 | |
| 1991 | 464 | 206 | 103 | 476 | 1,249 | ||
| 2001 | 480 | 242 | 278 | 570 | 1,570 | ||
| 2011 | 490 | 290 | 303 | 681 | 1,763 | ||
| Divorced | Male | 1981 | 4 | 6 | – | 3 | 13 |
| 1991 | 8 | 6 | 1 | 6 | 22 | ||
| 2001 | 22 | 17 | 6 | 14 | 58 | ||
| 2011 | 42 | 22 | 11 | 21 | 95 | ||
| Female | 1981 | 3 | 5 | 0 | 3 | 11 | |
| 1991 | 9 | 7 | 2 | 11 | 29 | ||
| 2001 | 33 | 25 | 11 | 25 | 94 | ||
| 2011 | 79 | 35 | 23 | 42 | 179 |
What is particularly noteworthy is that the divorced men and women are heavily concentrated in public rental housing. In 2011, about 28% of married men and women were living in public rental housing, and 44% of divorced men and women. In other words some one-fifth of the households living in public rental housing estates in 2011 were divorced. This compares with only one-twentieth in 1981 – a four-fold increase.
Hong Kong’s public housing estates are transforming into areas of growing poverty, with more divorced households. They are increasingly poor neighborhoods for motivating our children to higher aspirations. I have grave doubts about the wisdom of continuing the development of more public rental housing units to solve our shortage of housing units. A far better solution is to build subsidized homes for ownership so that families have a stake to stay together. By keeping families together, we prevent more children from falling into a state of disadvantage that will be detrimental to their future upward social mobility. Why foster and concentrate the poor and the divorced in publicly subsidized ghettos? Poor children deserve a better deal. A city of homeowners is also less politically divided.
