(This article was translated from French.)

Justin Trudeau’s government is currently conducting consultations on developing a federal poverty reduction strategy. In a process of this kind, before adopting concrete measures, it is essential to set clear objectives and establish a way to measure progress. But defining and measuring poverty is not simple.

The Europeans negotiated hard in 2010 to arrive at a common objective: to lift 20 million people out of poverty by the year 2020 (the Europe 2020 Strategy). The member states, while not agreeing on the best way to measure progress, combined three indicators of poverty. The European Union considers the following people to be poor: (1) those who live on less than 60 percent of their country’s median income; (2) those who manifest many signs of material deprivation in surveys on living conditions; (3) those who live in households where no adult works. In defining their national objectives, member states can choose one of the three indicators or they can develop their own measure based on a combination of these three.

In the United States it has also proved difficult to measure poverty, and there the official measure of poverty remains based on surveys conducted in the 1950s. This official measure is now widely considered out of date and inadequate. In parallel, since 2011, the Census Bureau has also published a more appropriate measure, the Supplemental Poverty Measure, which I discussed in a Policy Options article a few years ago. However, when the data are released each fall, it is always the less credible of these two measures, the official poverty rate, that makes headlines.

In Canada, governments have never agreed on an official measure of poverty, and Statistics Canada has not proposed one, at most working with indicators of low income. Note that here, as in the United States, when we discuss poverty we are mostly talking about income. In Canada we have barely begun to consider a measure of material deprivation like the one used in the Europe 2020 Strategy. This is not necessarily a bad thing: a lack of income is certainly the common denominator in all poverty situations. As Plume Latraverse sings, the poor have much in common, but mostly, “y ont pas d’argent” (they have no money).

All measures of poverty that are based on income revert to the same logic. Establish the threshold (a more or less arbitrary line that marks the transition from lack of poverty to poverty), then calculate the percentage of people whose income is lower than this level, adjusting for the number of people in a household (because of economies of scale, it is estimated that a four-person household’s needs are equivalent to those of two single people).

The best known poverty line is probably that of the World Bank, which puts the bar at US$1.90 per day (in purchasing power parity), based on measures of poverty that prevailed in the poorest countries in the 1980s, adjusted over time. Though this measure is useful for monitoring the evolution of deep poverty across the world — which, by the way, has decreased over the years — it clearly does not account for the situation in rich countries, where such a low absolute threshold makes no sense.

In Canada there are three types of thresholds and thus three measures of poverty. The first is Statistics Canada’s low-income cut-off (LICO), which refers to the expenditures of an average family on housing, food and clothing. This measure is still widely used, probably out of force of habit, but it has not been updated since 1992 except to adjust for inflation. It does not take into consideration, for instance, the fact that today expenditures on transportation and communications surpass those on clothing. Nor does it measure regional variations in costs (for housing, in particular). In addition, the LICO is used only in Canada, and it cannot allow international comparisons. Long dominant, this poverty measure now appears obsolete, and it is probably beyond repair. Statistics Canada has been using it less and less in its own studies.

The low-income measure (LIM), which is used much more widely, establishes the poverty threshold at a given proportion of median income, usually 50 percent (but 60 percent in the European Union). The poverty rate thus reflects the percentage of people living on less than, say, half of the median national income. There is no better measure of low income: it is simple and enables international comparisons. It is, however, sensitive to changes in the median income, which sometimes produces counterintuitive results. In a recession, for example, when the median income is flat, the poverty rate may seem to be decreasing while unemployment and economic hardship are increasing. On the whole, however, the LIM appears indispensable.

Finally, there is what is called in Canada the market basket measure (MBM). Produced since 2002, it estimates the disposable income necessary to purchase, in a given region, a basket of the goods necessary to lead a decent life. Although it is complex to calculate — because first the basket of indispensable goods has to be determined, and then the costs in each region evaluated — the MBM has the advantage that it corresponds rather well to poor people’s perceptions of their own situations. The measure captures not an abstract gap in relation to median income, but rather a very concrete inability to acquire the everyday necessities. Above all, the MBM brings forward a notion of basic needs; while the idea is not explicit in the definition of the measure, it is implicit because the market basket is defined as a necessary minimum. In Quebec, groups defending the rights of people in situations of poverty quickly saw the MBM’s potential to highlight the gap between social assistance income and what is necessary to cover basic needs.

The federal poverty reduction strategy should therefore consider two measures of poverty: the MBM, to follow the evolution of incomes from the standpoint of basic needs, and the LIM, to account for people whose incomes are below the median and to enable international comparisons. As shown in figure 1, the two measures present very different pictures of the country.

With the exception of Alberta, the LIM produces higher poverty rates than does the MBM. This discrepancy is not in itself a problem, because there is not, properly speaking, a “true” measure of poverty. One could say the two measures present the upper and lower limits of a plausible assessment of the situation. Put simply, both measures are credible and useful.

But there are a few complications and nuances that should be considered. How can we explain, for example, that in Alberta the LIM yields lower poverty rates than does the MBM? And why does poverty appear to be more widespread in Quebec than in Ontario, according to the LIM (of course, the Ontario government’s preferred measure), but less prevalent according to the MBM (the Quebec government’s measure of choice)? This is because behind the different measures of poverty, decisions — in part technical and in part political — have been made.

The inner workings of government
Keep track of who’s doing what to get federal policy made. In The Functionary.
The Functionary
Our newsletter about the public service. Nominated for a Digital Publishing Award.

The MBM, indeed, involves an elaborate calculation that is always questionable and requires periodic updates. In 2010, for example, Human Resources and Skills Development Canada, which produced the MBM at that time, announced it was revising the measure to better take into account the housing benefits received by low-income owners who did not have a mortgage — a significant category of people in some provinces. By imputing a rent to this group, the new calculation lowered the rent portion in the market basket and led to cost estimates that, for cities like Toronto and Vancouver, appeared completely unrealistic, to the point of undermining the measure’s credibility. This decision was later corrected, and the government went back to a more reasonable measure, based more heavily on rents. This technical controversy nevertheless highlighted the sensitivity of a measure like the MBM, and the need to hand its calculation over to an independent, credible, transparent advisory body, possibly based on a model like the one Michael Mendelson recently put forward for a Canadian council on inclusion and well-being.

While it appears to be more simple, calculating the LIM also involves important decisions. Here, three issues need to be taken into account.

The first is the percentage of median income that is used as the reference. When 50 percent is used (as in Canada and many other countries), the resulting thresholds end up relatively close to those of other poverty measures and are in line with conventional perceptions of poverty. The European Union, however, employs a 60 percent threshold to also include people who are at risk of poverty. At first glance this choice might seem generous because it places the bar higher. But, as Louis Maurin has noted, this decision distorts the situation and, paradoxically, downplays the problem of poverty by confounding people who have a severe lack of means with those whose incomes are modest but just barely sufficient. In Canada, for example, there are very few seniors living below the LIM at 50 percent of the national median income, but there would be many more if the threshold were set at 60 percent. In order for public policy to target people in the most severe poverty, it is undoubtedly preferable to maintain the reference line at 50 percent of the median income.

The second issue concerns the measure of the median income, in a way that is most relevant to evaluating the poverty rate. One approach is to take a median income anchored in a given year, which would then define a static target for poverty reduction. Ontario, for example, used such a fixed-in-time LIM to measure its progress in the fight against child poverty. In my opinion, this is the wrong approach. While all incomes continuously evolve, this method estimates the fate of the poorest on the basis of an arbitrarily fixed-in-time median income, for the satisfaction of seeing poverty rates go down. Another method is to take a five-year moving average for the median income, in order to attenuate fluctuations over time. Proposed by, among others, Miles Corak, this idea has merit; however, given the typical delays of several years to produce data, I believe it remains more important to have an indicator that is sensitive to change rather than to aim for stability. In short, I would opt for the simple, not massaged and transparent version of the LIM at 50 percent of the median income in the current year.

The third question is one that necessarily arises in a federation like Canada: What should be the geographical reference for the median income? Statistics Canada determines provincial LIMs on the basis of the Canadian median income, but it could just as well base them on the median income of the provinces, or even of regions. A LIM based on the Canadian median income is undoubtedly useful for comparing Canada with other countries, but to compare provinces, it leaves much to be desired. Indeed, real median incomes vary considerably from one province to another. In 2015, for instance, the median income, after taxes and transfers, for economic families and for people outside economic families, was $56,000 in Canada, compared with $48,200 in Quebec, $59,000 in Ontario and $71,300 in Alberta. On this basis, which does not correspond exactly to the way the LIM is officially calculated but follows its logic, a household would need an income above $28,000 to escape poverty, regardless of where it is in Canada. This means that a Quebec household would have to attain 58 percent of the province’s median income to exit poverty (figure 2), which is almost the European Union norm for people at risk of being poor. In Ontario, in contrast, a household that would exceed 47 percent of the provincial median income would no longer be considered poor. In Alberta, a household would have to attain only 39 percent of the province’s median income.

These disparities, produced by the use of a single Canadian median income to gauge poverty, explain why poverty appears to be greater in Quebec than in Ontario when we use the LIM, which is based on the Canadian median income, and lower when we refer to the MBM, which is based on the real cost of living in the provinces. It also explains why the LIM distinctly reduces the rate of poverty in Alberta. It all hinges on the denominator.

If the European Union used the European median income to measure poverty in the member states, almost two-thirds of Poland would be poor and there would be practically no poverty in Belgium. Using a single measure for all of Europe would confound the poverty of individuals with the relative wealth of the jurisdictions.

Canada, of course, is a country that shares taxes and transfers, and it is thus useful to have a pan-Canadian measure. But it is also a decentralized federation, whose provinces play an important social policy role. It would thus be imperative to set low-income measures at the provincial level. Some analysts have also suggested regional measures. This could be of some use, but as most relevant policy decisions are made by the provinces, a priority should be given to provincial measures.

In a recent study of low incomes among immigrants, Statistics Canada researchers Garnett Picot and Yuqian Lu open the door to this reconfiguration of the LIM and propose regionally based measures. But, for reasons that are more or less clear, they prefer to keep the current pan-Canadian LIM and adjust it for regional cost differences with the help of MBM data. This complicated approach yields poverty thresholds that are very close to those produced by the MBM itself, but lower. It is difficult to see the advantage of this procedure, when there is a simpler and more transparent solution: calculating the LIM on the basis of provincial median incomes.

The issue of how to measure poverty is therefore not entirely resolved. It remains an object of academic and normative debates, for which there is no simple answer. We should keep in mind that poverty rates constitute an imperfect construct that at best produces indicators allowing us to grasp a social challenge of the utmost importance. As much as possible, simplicity and transparency should prevail.

Photo: Shutterstock/Benoit Daoust


Do you have something to say about the article you just read? Be part of the Policy Options discussion, and send in your own submission. Here is a link on how to do it. | Souhaitez-vous réagir à cet article ? Joignez-vous aux débats d’Options politiques et soumettez-nous votre texte en suivant ces directives.

Alain Noël
Alain Noël is a professor of political science at the Université de Montréal. He is the author of Utopies provisoires: essais de politiques sociales (Québec Amérique, 2019).

You are welcome to republish this Policy Options article online or in print periodicals, under a Creative Commons/No Derivatives licence.

Creative Commons License

More like this