Saturday, January 21, 2012

Moving up in life

SOCIAL mobility is the bugbear that will not go away. With the gap between the rich and poor widening, Singapore has been grappling with the issue of whether children from poor families can move up. And until a week ago, there has been no comprehensive study across generations on whether the rich tend to stay rich, and the poor remain poor.

By Cai Haoxiang


Last Thursday, the Ministry of Finance (MOF) released the results of a study that tracked the extent to which children's incomes and education are linked to that of their fathers.

The study comes after Prime Minister Lee Hsien Loong's warning last October that 'our society is stratifying'.

In his speech at the opening of the 12th Parliament, he had said: 'The children of successful people are doing better, the children of less successful people are doing less well.'

The MOF study tracked income mobility for males born between 1969 and 1978. Its study found that this cohort has enjoyed plenty of opportunities to move up in life.

The son of a father in the bottom 20 per cent of income earners, for example, has a two-thirds chance of breaking out of that income group, the study shows.

And he has a 10 per cent chance of rising all the way to the top 20 per cent of income earners here.

The children of a primary-educated father are also likely to leapfrog his low educational background by at least two rungs, rising on average to an A-level or ITE certificate. Almost half - or 45 per cent - will make it to a diploma or university education.

That compares to 56 per cent of the entire cohort who attained a diploma or university degree.

Just how mobile?

ACADEMICS welcomed the study and described it as robust, given the quality of MOF's data, the large sample size and rigorous methods used in the analysis.

Assistant Professor Irene Ng of the National University of Singapore (NUS) social work department said that equality or inequality across different generations can now be tracked using 'good data' in a 'fair' way.

Dr Ng was the academic involved in two recent studies on income mobility, which, until the MOF's study, were the only Singapore literature on the matter.

She had used data from a national survey of youths aged 15 to 29, but the data had various constraints compared to that used in the MOF study.

'As academics, my co-authors and I did not have access to large-scale data from the Department of Statistics,' she tells Insight.

Her sample size was small: 271 parent- child pairs.

The MOF's study by contrast had a sample size that was almost 150 times bigger: 39,500 father-son pairs who are Singaporeans or permanent residents.

Moreover, in the survey that Dr Ng based her study on, parents' incomes were not directly measured but depended on what the youths thought their parents earned, which could have led to errors.

By contrast, the MOF study was based on DOS data and could measure incomes with 'high precision', the paper's author Yip Chun Seng wrote.

Dr Yip, who is MOF's principal economist on fiscal policy, also noted that he had a larger sample size to work with. He was thus able to overcome the challenges faced in the previous studies.

Asked why the study was commissioned, Dr Yip tells Insight via e-mail that it was 'quite natural' for policymakers to want a better understanding of the issue.

In the paper, he wrote that high mobility is seen as desirable because it means that society offers similar chances of achieving economic success regardless of one's background.

Low mobility, on the other hand, would present a case for 'active government intervention to 'level the playing field'', he wrote.

Using the better-quality data at his disposal, Dr Yip found a higher level of income mobility than the estimate Dr Ng arrived at in her previous studies.

The measure of mobility he found was 0.22 to 0.3. That figure is called the correlation coefficient - a statistical measure of how the father's income is related to the eldest son's.

A higher figure implies lower mobility because the son's income is more closely tied to his father's.

Considering sons, rather than children of both sexes, is standard practice in such income mobility studies because life events like marriage and childbirth affect daughters' employment and thus complicate income measurements.

The MOF's figure of 0.22 to 0.3 indicates moderate to high income mobility, compared to Dr Ng's previous conclusion of lower mobility at 0.44, after adjusting for data limitations.

Figures of 0.18 for Finland, 0.19 for Canada and 0.27 for Sweden - indicating high mobility - had been estimated in a 2006 study by University of Ottawa economist Miles Corak for cohorts born between the 1940s and 1960s.

Professor Corak has also estimated a figure of 0.47 for the US and 0.50 for the UK - indicating low mobility.

But Dr Ng believes that the true level of income mobility in Singapore lies somewhere between her estimate of 0.44 and the MOF's top estimate of 0.3.

She points to two limitations of the MOF study, which were also acknowledged by Dr Yip in his paper.

Firstly, the fathers measured in the study are too old, averaging more than 50 years of age.

That could introduce some biases where only those who stayed in employment had their data recorded.

The ideal period to measure income is in the late 30s to early 40s, the same period the sons were measured, she says.

Secondly, data from fathers' incomes was tracked only over a five-year period.

But the data shows that sons' incomes are more tightly linked to their fathers' when a longer period of measurement is used, like up to 10 years in some studies.

The MOF also stated in its report that the moderately high level of mobility for the cohort it studied might not apply to future generations.

'If the limitations were adjusted for, we could get a lower indicator of mobility,' says Dr Ng.

NUS sociologist Tan Ern Ser says he was 'quite impressed' with the analysis, and that he was not surprised that there is still upward social mobility.

'But the argument is whether it is slowing down, and whether we are doing worse than other developed countries,' he says.

Class background does influence the probability of upward mobility, he notes, pointing to a finding of the study that sons of poor fathers are likely to remain poor.

The study found that the mobility measure for those at the bottom 10 per cent was up to 0.36, compared to 0.3 for those in the middle.

But Dr Yip tells Insight that the figure of 0.36 is 'not too much higher' than the average of 0.3, and low compared to an equivalent figure of 0.67 in the US for the poorest 10 per cent.

NUS sociologist Vincent Chua, meanwhile, says that the reference point 'should not be the United States, but the other class groups within Singapore itself'.

Dr Chua is of the view that 'channelling resources downwards to disadvantaged groups will help a great deal to achieve a more inclusive society'.

What next?

MORE can and likely needs to be done to keep tabs on mobility.

Dr Ng has various proposals, including tracking the impact of government policies in education and immigration on social mobility.

'You can look at different points in time when policies change, for cohorts that experienced policy changes, to see if the numbers change,' she says.

One example is the streaming policy introduced in the 1980s and the through-train programme introduced in 2004 that allows bright students to skip the O-level examinations.

That is because studies in other countries have shown that different education tracks lead to different wages, she says.

Prof Tan adds that measures on wealth and occupation can also be added to future studies.

'I think wealth - broadly speaking, the five Cs and assets like stocks, cash, properties and cars - is a much better measure of class than income,' he says.

He also suggests having data on how long it takes for the son to acquire a four-room flat and a mid-sized car as compared to his father, because they are 'the key indicators of being in the middle class'.

Asked about these suggestions, Dr Yip points out that income and educational attainment are widely-used yardsticks because they are easier to measure accurately. 'On the other hand, the obstacle to measuring wealth, occupation and others is usually data. Given that meaningful measures are harder to obtain on the latter, less is known about them,' he says via e-mail.

It will be some years before another such large-scale study is done to estimate income and educational mobility.

Dr Yip says that mobility estimates like those in his paper 'tend to be fairly stable, so there is not that much value in doing this every year'. 'The main interest is probably on younger cohorts, and we would need to wait some years to study them. This is because ages below 30 are generally regarded as too early to measure income mobility.'

The next age group up for study is likely to be those born in the 1980s. They would at most be in their early 30s today.

That suggests a wait of perhaps 10 years before the next government study of this scale. In the meantime, academics and others interested in mobility have much new data to dive into.