Hong Kong's poor need a lift - and not only on paper

Ever since the re-establishment of the Commission on Poverty, drawing an official poverty line for Hong Kong has been a high priority on the government's to-do list. On Saturday, the administration is expected to announce details of the new poverty line, long rumoured to be set at half the median household income.

The rationale for benchmarking poverty against median household income is rooted in the idea of relative poverty. Unlike more traditional notions of poverty, which are based on absolute needs (for example, food, shelter, clothing), relative poverty involves pegging one's standard of living - for better or worse - to the fortunes of the middle class. The idea is that, as society advances, the poverty threshold should increase in real terms.

The problem with a relative poverty measure is that, when it is used for the purpose of policy evaluation, it is always necessary to be mindful that the policy could alter the benchmark. This issue does not arise with an absolute poverty line, because the basic needs of a family cannot be amended by government policy (though society's concept of what these "basic needs" are may change from time to time). The government can, however, manipulate the median income. For instance, if it were to tax the middle class and transfer all the proceeds to a few tycoon families, this would reduce relative poverty in Hong Kong - not because of any change in actual poverty, but simply because the benchmark would be lowered.

Unfortunately, while this example is quite fanciful, any policy that affects the median family income in Hong Kong could count as a poverty alleviation measure under the proposed metric.

For instance, Hong Kong provides universal access to subsidised education and health care. If such access were to be means-tested - for instance, by requiring higher-income families to pay more fees - relative poverty would be reduced . This is because, under the commission's guidelines, means-testing these benefits would cause them to be added to household income. Depending on who gets the benefits, the median family income could remain the same or rise, raising the poverty threshold. But, in either case, t he income of lower-income households would increase by a margin large enough to reduce measured poverty.

This problem is inherent in the relative approach to poverty measurement. Anything that affects the median family income in Hong Kong will cause the number of people in "relative poverty" to change. As a consequence, every time the relative poverty measure goes up (or down), one has to ask whether it is because of changes among lower-income households or changes to the benchmark.