(This essay was published in Hong Kong Economic Journal on 27 February 2013)
Last week I wrote about the proposed poverty line for Hong Kong and the complexities in drawing such a line. The government aims to announce a poverty line by mid-year. What I am concerned with is the basis for determining if someone is “poor”.
Household size has been the measure used in places such as Europe, but age can also be used and I believe it is a fairer measure. The difference between the two approaches can be empirically examined using data from the 2011 population census.
In Table1 we define Poverty Line A as 50% of the median income of households with the same household size (excluding domestic helpers). Using Poverty Line A we find 22% of households are below the poverty line. This breaks down to 32% of 1-person households, 26% of 2-person households, 19% of 3-person households, 17% of 4-person households, and 16% of households with 5 or more persons. As we can see, smaller households have a larger proportion of households below the poverty line.
When I apply the definition of a poor household using Poverty Line A to include the age of the head of the household, however, a strong skewing of the results occurs which, as I show below, does not hold up under a different and I believe more reasonable scenario – that of Poverty Line B which is 50% of the median income of households within the same age group (excluding domestic helpers).
Anchoring on Age Minimizes Bias
Using Poverty Line A, I calculated the percentage of households below the poverty line to be 22% among household heads aged 24 or below, 9% of those aged 25-34, 15% of those aged 35-44, 15% of those aged 45-54, 20% of those aged 55-64, 40% of those aged 65-74, and 53% of household heads aged 75 or above. These results show older households have a much larger proportion of households below the poverty line.
Now compare the results in Table2, which uses Poverty Line B. Here, we find 24% of households are below the poverty line, a number slightly higher than Poverty Line A. More importantly, the proportion of poor households is relatively constant across age cohorts; it rises only slightly with older cohorts. So 20% among households whose heads are age 24 or below are below the poverty line, 20% of those aged 25-34, 22% of those aged 35-44, 21% of those aged 45-54, 26% of those aged 55-64, 30% of those aged 65-74, and 36% of those households with heads aged 75 or above. This more even distribution is a far more sensible result as it shows that every age cohort has poverty to more or less the same extent, something one would normally expect in any society.
Applying the definition of a poor household using Poverty Line B, I calculated the percentage of households below the poverty line to be 54% among 1-person households, 29% among 2-person households, 16% among 3-person households, 11% among 4-person households, and 8% among households with 5 or more persons. The use of Poverty Line B produces a much larger measured proportion of poor small households. This is again a sensible result and is in line with almost everything we know about the incidences of poverty.
Households with a small number of members have much greater difficulty in absorbing negative shocks. Their ability to manage the associated risks is also lessened. A typical person who cannot find a spouse is probably economically unattractive as a marriage partner of choice. It is well known that most inhabitants of the caged homes observed in the past half-century in Hong Kong fall into this category; unfortunately, the allocation criteria for public housing units does not work to their advantage.
Public policies seldom focus on assisting these individuals, very often simply because they are difficult to reach. The opening of the Mainland provided them with an opportunity to get married and enter the public housing program for the first time in their lives. Some of them are now better off, but most continue to require assistance. Fortunately this is finally reaching some of them thanks to the change in their marital status after the opening of the Mainland.
It is interesting to compare the results of applying Poverty Lines A and B. Table3 provides a cross-tabulation of the poor based on the two definitions. Both definitions show 71.1% of households are classified as above the poverty line, and 17.5% of households below the line. But 6.6% of households that are above the Poverty Line A definition fall below Poverty Line B. These are mostly smaller sized households. And 4.7% of households classified above Poverty Line B are below Poverty Line A, mostly elderly persons who live in larger households.
Clearly the choice of definition to anchor the poverty line has a significant impact on the type of households society will identify as worthy of assistance. This further underscores the fact that poverty is a multi-dimensional challenge that cannot be reduced to a single dimensional indicator. But if one has to pick such an indicator, then anchoring it to age is preferable to anchoring it to household size. Common sense would certainly tell us that in each age cohort there will be poor households. The belief that the elderly are overwhelmingly poor is an artifact of bad measurement or worse, a failure to understand why measured income at one point of the life cycle is a bad indicator of whether a household is poor or not. The diagram in Figure 3 in the article published last week makes the same exact point.
Elderly Poverty Exaggerated
We can look at excess numbers and percentages of poor households under the Poverty Line A and Poverty Line B classifications in Table4. These comparisons are important because Poverty Line A, which is proposed by the Hong Kong Council of Social Services and Oxfam, would significantly underestimate the numbers and percentages of poor households among small size households, especially among 1-person households. It would also overestimate the number of poor households among the elderly population.
Table 4 shows that compared to the Poverty Line A classification, Poverty Line B classifies 41580 (or 8%) more households as poor. The number of poor households is higher among 1-person and 2-person households, but lower among households with 3 or more persons. In particular, there are 86580 (or 41%) poor households in excess among 1-person households. The number of poor households is also higher among households aged 25-64, but lower among households aged 65 or above. In particular, there are 47580 (or 62%) fewer poor households among households aged 75 or above. These are very significant differences.
Poverty is not concentrated among the elderly. Most old people here have lived through one of the fastest periods of economic growth in Hong Kong when old age assistance programs were not generous. They would have been stupid and reckless to not save for old age retirement. Many have children who provide some support for their parents. The belief that many of our elderly are living in poverty is based on a comparison of current household income without recognizing that income varies over the life cycle of the household. This is just not good analysis.
Poverty appears in every age cohort. These are households that have low education, have experienced bad luck, are often recent immigrants, and in most cases have failed to build a family that would allow them to pool human resources and hedge against negative economic shocks. These are the households society should help. These are the households a poverty line should identify. Poverty Line A fails to do this well.
I fear tackling poverty without understanding its causes will not effectively alleviate poverty and often fail to help those that truly deserve assistance. This will be especially true if we adopt a poverty line that misses many of those who could be helped. This will also be true if various transfer programs are poorly designed and have perverse incentive effects that help many people who are not genuinely poor. The Great Society program introduced in the US in the 1960s was the most ambitious undertaking to alleviate poverty known in any society, yet income inequality in America today has hardly improved after half a century of effort.
I have often heard suggestions that Hong Kong should actively introduce many of the social welfare programs developed in advanced nations half a century ago, including their operational frameworks and designs. Defining the poverty line as 50% of the median income of households of the same size is an example of this importation of a foreign program. But we often forget that these operational designs were devised long ago based on now-outdated research.
At the time, the primary concern of economists trying to measure income inequality among households was the heterogeneity of household sizes and the composition of adults and children. Economic theory back then was basically a theory of the behavior of individuals, who were viewed as undifferentiated atoms. The fact that these atoms were of different ages and sexes, most of them with blood relations, was an inconvenience about which economists had little to say, apart from classifying a woman to be on average equivalent to (1+x) man, an adult to be on average equivalent to (1+y) child, and so on. So a woman became on average equivalent to (1+x) man and an adult became on average equivalent to (1+y) child, and so on. Since households looked so different from each other, any meaningful attempt to compare them would need a device that converted men and women, adults and children into some equivalent unit, and in addition ignore the relationships they had with one another.
Setting the poverty line at 50% of the median income of households of the same size was intended to be such a simplifying contraption that would hold constant the number of members. But it ignored other differences among members. The sole purpose was to make some sensible comparison of what are highly heterogeneous households. A popular contraption used by many governments and researchers, including the Hong Kong government, is to include taxes and various transfers in computing the Gini-coefficient measure of net household income inequality. This is a similar attempt to make difficult comparisons somewhat sensible.
However, studies of poverty have long advanced beyond the understanding achieved over half a century ago. Since the pioneering work of Professor Gary Becker of the University of Chicago, economists have made huge advances in understanding the family – the division of labor within the family, marriage decisions, fertility decisions, savings behavior, retirement choices, divorce and separation decisions, family background on the opportunities of children through schooling, health and other human capital investments, and inequality and intergenerational inequality. These studies cast a long shadow of doubt over the appropriateness of defining poverty using Poverty Line A.
Target Assistance of the Genuine Poor
Unfortunately, the anti-poverty programs in most societies have not taken on board these new research ideas because of inertia and resistance from vested interests. Switching from a definition similar to Poverty Line A to one similar to Poverty Line B would create both losers and winners, as we have seen from our numbers in Table 3. As always, the political influence of the losers is far more powerful than the potential winners. The former have become politically well organized and have their entrenched advocates. Newer and improved definitions that can help alleviate poverty better end up not being adopted. This is the tyranny of the status quo.
Hong Kong has an opportunity to make a fresh start. The future framework for the poverty debate is due to be set by government this year. If Hong Kong is to adopt a poverty line, it should not follow the blind lead of old frameworks and designs. Poverty Line B is a huge improvement over Poverty Line A, but it is important to point out that setting the poverty line using either of these definitions will not eradicate poverty over time. As I pointed out last week, both definitions are based on a rigid definition of poverty – that of people falling below a 50% median line – that cannot be changed by economic growth or policy programs. This concept of “poor” means poverty cannot be removed or even reduced.
In the long term, Hong Kong needs policies that have a good understanding of how poor households are formed and what incentives and interventions can help them exit poverty. Passion has taken society to this stage. If genuine progress is to be made, then cool heads and apolitical analysis is vital, for otherwise, we would have learned nothing from the greatest failure in humankind’s attempt to alleviate poverty – that of America’s Great Society Program.
Gary S. Becker, “A Theory of the Allocation of Time”, The Economic Journal, Vol. 75, No. 299. (Sept. 1965), pp. 493-517.
Gary S. Becker, “Human Capital and the Personal Distribution of Income: An Analytical Approach”, Woytinsky Lecture no.1. Ann Arbor: University of Michigan, Institute of Public Administration (1967).
Gary S Becker, A Treatise on the Family, Harvard University Press, 1981.