Thursday, July 17 There's no bit of data so without content that it can't prove two opposing points of view. No sooner had Statistics Canada released the results of an important new study tracking the numbers of the poor in Canada than both left and right had embraced it for their own purposes. The research suggests that, far from a static underclass of the permanently destitute, the ranks of those whose incomes fall below StatsCan's Low Income Cut-Off (LICO) -- about $32,000 for a family of four in a large city - - undergo a high degree of turnover from year to year. The agency has only been collecting data for the Survey of Labour and Income Dynamics for two years, but in that time about a million people crossed the LICO in either direction: about a 25 per cent churn rate. This conforms with similar research in other countries.

Aha, said the right, proof that the numbers of the poor have been grossly overstated. Where the official "poverty rate" counts 17 per cent of Canadians as being poor, only about half of these, or 8.5 per cent of those surveyed, remained poor from one year to the other.

Aha, said the left, that means the poor are not trapped in their present state, either by their own sloth or by the welfare system. The whole campaign to reform social assistance has been a cruel waste of time.

In fact the study proves neither theory. No one ever said that everyone on welfare is stuck there: only that the prospect of a dollar-for-dollar loss of benefits is a pretty stiff disincentive to those who might be tempted to earn a little income on their own. As for whether a single year's "snapshot" overstates the incidence of poverty, who can say? Granted, it would be more revealing to calculate and compare incomes over a person's entire lifetime, so as not to count as impoverished those who are but temporarily poor: university students, say, or businessmen who had a bad year. But whether a lifetime income approach would raise or lower the poverty rate is not immediately obvious.

What the study does confirm, yet again, is the limitations of LICO as a measure of poverty. Certainly it makes hash of year-to-year comparisons: it's not the same people in both groups. As recent U.S. research on income dynamics has shown, average incomes for those who were counted among the poorest 20 per cent of the population at the beginning of the 1980s rose dramatically by the end of the decade, even as the average income of the bottom fifth in any given year was dropping.

Other research, notably Prof. Christopher Sarlo's work for the Fraser Institute, has pointed out some of the gaps in the LICO data. Its definition of income does not include benefits in kind, for example, nor capital gains -- nor, obviously, unreported income. Sarlo notes that, of all households below the LICO in 1990, almost one in five owned their homes, mortgage-free.

Nearly all, 97 per cent, had colour televisions. Some 53 per cent owned a car -- the same rate as for those above the poverty line. And so on.

But the real problem is more fundamental: LICO is not actually a measure of poverty at all, in the dictionary sense of "absolute deprivation," but rather of inequality -- or relative poverty, if you like. StatsCan starts by calculating the share of the median family's income that is spent on food, shelter and clothing -- 35 per cent, in 1992. It then simply adds 20 percentage points to that figure. LICO is set at the point at which the same expenditure would consume 55 per cent of household income.

As median family income rises, then, so does the definition of poverty.

Indeed, today's low income threshold is equivalent in real terms to a middle class income in the 1950s. By way of contrast, Sarlo has constructed a "basic needs" index, similar to the market basket provinces use to set welfare rates, which tries to measure how much it actually costs to pay for the essentials.

This inevitably involves some relativistic judgments of its own. But it at least provides a consistent benchmark against which to measure our progress as a society.

This is not a mere statistical quibble, or an attempt to define away the poor.

It raises important questions of policy. Relative poverty may be as worthy of concern as absolute poverty -- or it may not. Which should be our priority? It's no good saying both. For what if the two conflict? What if the number of people in relative poverty is rising, while the number in absolute poverty is declining? Are we getting ahead, or falling behind?

For more on this, and on some surprising new international evidence rating Canada's poverty performance against other rich countries, tune in next time.