(This essay was published in Hong Kong Economic Journal on 20 February 2013)
“When I use a word,’ Humpty Dumpty said in rather a scornful tone, ‘it means just what I choose it to mean — neither more nor less.”
“The question is,” said Alice, “whether you can make words mean so many different things.”
“The question is,” said Humpty Dumpty, “which is to be master— that’s all.”
Lewis Carroll, Through the Looking-Glass, and What Alice Found There, 1871
The government has announced that a poverty line will be defined by mid-year, presumably as a first step towards the goal of alleviating poverty. There has been much lobbying on the issue from various social groups and political pressure has been mounting in Hong Kong to transfer income and benefits to the “poor” in society.
This pressure is the consequence of a confluence of various factors, which include:
(1) a general opening of the political system so that political power has become more dispersed, making it possible for organized “grass roots” political interests to become better represented;
(2) measured inequality in household income distribution, which has risen in society in the past two to three decades;
(3) rising property prices, which have widened the inequality of wealth distribution and fostered a sense of deprivation and unfairness among the have-nots;
(4) a perceived lack of upward mobility in society as economic growth has slowed;
(5) the rapid rise of an elderly population, some of whom appear to be without a means of support due to advanced age and personal circumstances;
(6) the increase in large numbers of immigrants from the Mainland with fairly low levels of skills;
(7) a general swing in popular attitude towards the left following the onset of the most severe global economic recession since the Great Depression; and
(8) a perception that the government has failed to act to assist the poor in the midst of the external shocks Hong Kong has endured.
The above list is not exhaustive but it suggests there are multiple and complex reasons why the “poor” are poor. It is not clear that we have a good understanding today of the factors that cause poverty and their relative importance. Whatever research I am aware of remains highly scattered, incomplete, and quite superficial. Society is, however, keen on tackling poverty. We are facing a situation of a lot of talk and a lot of undigested information, but very little real understanding.
Different Approaches to the Poverty Line
Will constructing a poverty line help us understand poverty better? Will it explain why the “poor” are poor? I doubt it, but the public wants this to be done and the government has committed to doing so. So what is to be done under the circumstances?
There are two approaches to defining the poverty line. The first is the American approach which is not the favorite in Hong Kong for fear that its complexity will prolong discussion and debate and delay the goal of adopting a poverty line by mid-year. The other is the European approach, which has been proposed for Hong Kong by the Hong Kong Council of Social Services and Oxfam.
The American approach was originally developed in 1963 and 1964 by Mollie Orshansky, an economist working for the Social Security Administration (SSA). It was based on devising an expenditure budget required for a poor household to have a minimally socially acceptable standard of living. It was recognized that this standard of living would be adjusted over time to reflect changing standards of living and social norms.
According to Orshansky, “Unlike some other calculations, those relating to poverty have no intrinsic value of their own. They exist only in order to help us make them disappear from the scene….With imagination, faith and hope, we might succeed in wiping out the scourge of poverty even if we don’t agree on how to measure it.” The implications of the American approach is that poverty could be eliminated or significantly reduced if we adopted the right social policies to assist the poor and with the help of a growing economy.
The European approach categorizes households by the number of members it possesses so we have 1-person households, 2-person households, 3-person households, and so on. Within each household category the median household income is first determined. The poverty line is then defined as 50% of the median income. Therefore there will be as many poverty lines as the number of household categories. Households with incomes below their category-specific poverty line are considered as being in poverty.
Poverty is then defined in relation to the income of the median household. With such a definition poverty can never be eliminated through social policies to assist the “poor”, even when median incomes rise over time in a growing economy. Poverty becomes essentially an unchanging proportion of the population at the lower end of the income distribution. Such a definition creates a “poverty class” by official definition and institutionalizes the “poor”. Social policy to combat poverty becomes a never ending attempt to transfer income and benefits to those who are below the poverty line without any expectation that such a class will disappear over time as hoped for by Mollie Orshansky.
Perhaps the most significant factor in choosing which definition to adopt for a poverty line is its implications for the future development of political life. The American approach at least holds out the promise that poverty can be eliminated in the long run and the hope that the politics of poverty does not have to be forever a zero-sum game between the haves and have-nots. The European approach essentially turns poverty into a permanent political agenda for society. A growing economy with rising standards of living for all cannot remove the eternal divisiveness of the politics of poverty. This is the approach favored by most advocacy groups in Hong Kong.
Much is at stake for the future of our city in choosing the poverty line. The nature of our society and our politics will depend on it. Unfortunately, the fortunes of many interest groups and grass roots organizations who have been the major advocates of the poverty line are heavily tied to the European approach. I do not believe the European approach can be in the interests of Hong Kong. I certainly hope it will not be adopted, but I fear it is a forgone conclusion.
If the European approach were to be adopted, then can we devise a better operational definition of the poverty line so it can be more equitable and efficient? Is there an alternative measure to the one based on household size?
Household Size Not a Reliable Anchor
The household is a changing concept. Ideally, a household should be defined around a family relationship that acts to a large extent as a single decision unit on behalf of its members. For example, measures of inequality like the Gini-coefficient should be based on a single decision unit, but invariably they are almost always constructed using the census definition. The census definition, which is used for the purpose of collecting data for a population census, is based on members living under one roof.
The problem with the census definition is that some households may have members who are excluded because they do not live under the same roof even though they form one decision unit on matters relating to family welfare. On the other hand those who live under one-roof may not constitute a single decision unit.
There is of course not much we can do about the data we have to work with. One might even believe the census definition is good enough. But if public policy is going to be based on a poverty line anchored on household size then we should consider whether this is a reliable anchor.
We know a household has a life cycle and its size changes over time. Incomes and expenditures change over the life cycle. More importantly the composition of its members also changes and there are adults and children, working and non-working members, and so on to make households rather non-homogeneous. Household incomes and expenditures in any particular year fail to provide good summary measures for comparing the welfare of different households. When we talk about whether a household is poor, we care not only if poverty occurs in one particular year but whether it lasts long enough to impact on standards of living, upward social mobility, and the intergenerational transmission of economic conditions.
The decision to leave or join the household depends on many factors. Some of these factors are internal to the development of the household, e.g., birth, growing up, and the ageing of members. Some factors are external, e.g., public policies that transfer resources to a household conditional on membership composition, either implicitly or explicitly. Some external factors are recurrent, e.g., economic booms and busts that give rise to incentives for some household members to either leave or join the household, temporarily or permanently. The factors that can affect the size and composition of a household are numerous.
This makes it worrisome to anchor the definition of the “poverty line” on household size. Current proposals reported in the media do exactly this. The critical issue here is that household size is a decision outcome that is affected by numerous factors. It is not a robust way to construct a poverty line. In particular, it is vulnerable to two types of adverse effects. First, it introduces an element of arbitrariness in defining whether a household is in poverty. Second, it encourages opportunistic behavior to take advantage of benefits tied to the poverty line through manipulating the rules of the system.
Let me give two examples.
First, consider two households that both have two adults and one child. Assume all the adults have the exact same employment and earnings so the household incomes of these two households are identical. “Household Naughty” decides to report that a non-working grandparent lives in the same household. With one additional member “Household Naughty” falls below the poverty line. This is opportunistic gaming behavior. Assume “Household Honest” does not engage in such behavior and therefore stays above the poverty line. Giving benefits to “Household Naughty” is arbitrary and is unfair to “Household Honest”.
Another example would be two households that are identical in every respect. In “Household A” both parents decide to work and pay the grandmother to look after their children. In “Household B” one parent stays at home to look after the children while the other parent works for longer hours on the job. Suppose the income in “Household A” is the same as “Household B” after deducting the transfer to the grandmother. However, the measured household income is higher for “Household A” than “Household B”. If “Household B” is deemed to fall below the poverty line and “Household A” is not, then it is arbitrary and unfair for “Household A”.
Although these examples are constructed ones, it would not be difficult to uncover many real life cases of arbitrariness and unfairness once we anchor the poverty line on a factor, like household size, that can be manipulated and influenced by other factors.
Age is a Logical Anchor
To avoid this problem it is far better to anchor the poverty line on a factor that is independent and not affected by other variables. A good and simple anchor would be the age of the household head. Age is not a variable that can be easily manipulated and it is independent of internal and external factors. It would also not be difficult to define the poverty line as 50% of the median household income of those households that are in the same age group.
More importantly, age is a natural proxy for a household’s life cycle. Household size tends to be highly correlated with age, exhibiting a hump-shaped profile (see Figure 1), rising with a peak at around age 52 and then declining continuously. Before the age of 25 and after the age of 85 the profile is choppy due to the small number of households in the sample, although average household size has a spike at age 20, reflecting the presence of young households with children that have decided to leave their parents’ household to form their own.
Household income also tends to be highly correlated with age – also exhibiting a hump-shaped profile (see Figure 2) that is relatively flat between ages 30 to 60. There are many factors contributing to this phenomenon.
First, adult members in older households tend to have less schooling resulting in lower earnings. Second, the effect of women’s labor force participation behavior adds another complication. Households in their 40s tend to have more children and lower female labor force participation rates than those in their 30s. Households in their 50s have higher female labor force participation rates as their children have grown and they have fewer childcare demands. These age related cohort factors imply that it is more appropriate to compare household income inequalities among households in the same age cohort rather than by household size.
Moreover, younger households save a larger fraction of their income than older households so comparing income across age could give a misleading indication of income inequality. Since savings change over the lifecycle, a better indicator of income inequality must consider income at different times of life.
Consider for simplicity two individuals both earning $5000 per month. Person A is 25 years old and Person B is 40 years old (see Figure 3). A direct comparison of their earnings would lead one to believe both are equally poor. But this is obviously not the case. Since the age-earnings profile for Person A is higher than that of Person B, therefore the former is the wealthier person.
Or compare Person A who earns $8000 per month at age 50 with Person C who also earns $8000 per month but is 65 years old. One would have to conclude that Person C is the wealthier person because his age-earnings profile is above that of Person A.
Anchoring the poverty line on age provides a fairer yardstick in identifying those who are truly in poverty. It also avoids the arbitrariness inherent in comparing households who are at different points of their household life cycle. Age determines to a large extent household size, but it is not influenced by household size and other factors.
Next week I will examine in detail the different breakdowns in those considered poor using Hong Kong data on household size and on the age of the head of the household, and how the former leaves out key groups who are indeed poor. If we are to adopt the 50% median cutoff to determine poverty, there are fairer ways to do this than the examples seen overseas.