President Obama appointed Rebecca Blank—a capable, no-nonsense, PhD in economics, and a former Dean at the University of Michigan—to his new administration, and told her to answer a simple question: how should the United States measure poverty?
Diane Finley, the Conservative government’s Minister responsible for social policy, is not exactly Canada’s Becky Blank. She might certainly be capable, and she might certainly be no-nonsense, but to be honest she doesn’t seem to have a lot in common with the American academic turned policy-maker: except, of course, that in Canada Ms. Finley is ultimately the person responsible for poverty measurement.
Both women have pulled off quiet revolutions by recognizing that the poverty line should evolve over time: as the things needed to participate normally in society change, so should the poverty line.
This was evident to the first great economist, Adam Smith, who wrote some 250 years ago:
By necessaries I understand not only the commodities which are indispensably necessary for the support of life, but whatever the custom of the country renders it indecent for creditable people, even of the lowest order, to be without. A linen shirt, for example, is, strictly speaking, not a necessary of life. The Greeks and Romans lived, I suppose, very comfortably though they had no linen. But in the present times, through the greater part of Europe, a creditable day-labourer would be ashamed to appear in public without a linen shirt, the want of which would be supposed to denote that disgraceful degree of poverty which, it is presumed, nobody can well fall into without extreme bad conduct. Custom, in the same manner, has rendered leather shoes a necessary of life in England.
The official US poverty line was established in 1964, and was based upon what was considered a minimal but acceptable standard of living in the late 1950s. At that time this amounted to about one-half of the typical family’s income. With the exception of accounting for inflation it has remained unchanged, and as a result is now only about one-quarter of a typical income.
Changing this threshold has proved near impossible because the authority to do so lay in the White House: with Republicans in power, reasonable, evidence-based proposals for reform fell on deaf ears; and no Democrat wanted to take the heat of implementing statistical changes that might imply higher measured levels of poverty.
In Canada the situation has been only somewhat better. The most commonly used line to divide the poor from the not-poor reflects the consumption patterns of Canadians in the early 1990s, a time before most of us had sent our first e-mail. The “Low Income Cut-Off” continues to be the headline measure released by Statistics Canada.
A new indicator of the cost of a minimal standard of living—consisting of a nutritious diet, basic transportation needs, and adequate housing—was implemented by the federal and provincial governments in the late 1990s, but with no real plan on how it would be updated.
The ‘Market Basket Measure’ had surprisingly little influence on how we see ourselves, in large measure because for some period of time the federal government wouldn’t let us see it. The authority to release this statistic rested with the federal minister, and at one point it was necessary to resort to Access to Information requests to extract the numbers from the ministry.
Becky Blank did an end run around the quagmire of poverty measurement by having the US Census Bureau develop an entirely new indicator based upon solid academic thinking of the last two decades, and most importantly having the poverty line evolve gradually with time.
This new poverty line is defined as the level of spending on food, shelter, clothing, and utilities that two-thirds of American families are able to exceed. It is automatically updated by reflecting average spending patterns during the five most recent years. The recent Census Bureau publication of the “supplemental measure of poverty” paints a different picture of the poor than the official measure.
Diane Finley quietly did a similar thing more than a year ago.
The Market Basket Measure has been updated to reflect the realities of participating in the Canada of 2010. Some of this was forced upon the federal and provincial public servants deriving the numbers. A used Chevy Cavalier was considered the basis for calculating transportation costs in areas that don’t have public transport, but General Motors doesn’t even produce it anymore. Now the Ford Focus is used.
But more importantly what it means to live a normal life has also changed, and the government has introduced a rule for including new goods and services into the costs of a modest and basic standard of living. When as many as 70% of Canadian families in at least seven of the provinces with two-thirds of the population buy something, then that good or service is considered a necessary part of our lives. Certainly more complicated than the American rule, but typically Canadian and reflecting long-standing precedents in other areas of public policy.
On this basis the new necessities of life include: computer and internet services, cell phones, and garden supplies. But alas the experts proposed to break their own rule at the very outset. Gardening may be good for you, most Canadians may do it, but ultimately the government feels you can do without it if you have to. The same goes for cell phones.
By recognizing that the poverty line must change as society changes Ms. Finley has introduced a much-needed reform to an important area of public policy, and on this score her bureaucrats have made her look as smart and capable as the best of them.
But the challenge for her and her American counterpart is to now make these new concepts the headline measures their statistical agencies focus upon, letting the Low Income Cut-Off and the Official US measure drift into the past. Adam Smith would be proud!
Economics for public policy 
Nice post. I always find these things fascinating on an social-epistemological level, the way social values and empirical data get intertwined in the operational definition of concepts and thresholds – and the network of statisticians that need to coordinate in order to construct the databases in accordance with these definitions. The modern management of the state is really amazing on an anthropological level.
Do you know when these new measures are to be implemented?
Nicely put. Values come into play in defining something like “poverty” and its important that policy makers be up front about them. These measures are actually being used, both in the US and in Canada, and are available to those interested enough to dig for them. But they are not given prominence by the statistical agencies, which was the concern I was raising in the final sentence of the post.
huh…. Are they deliberately resisting or just slow to pick up the tool?
I suppose these things are interconnected. It takes time to turn the bureaucratic ship, but also existing stakeholders have positions that have to be addressed. Certainly, there is a politics around poverty measurement in the US that slowed reform, but also once the poverty statistics gets incorporated into program delivery — say a transfer program bases eligibility according to the relationship between one’s income and the poverty line — then any changes in the statistics will have implications for government budgets. This is also an important part of the US story, and something that has crept into Canadian discussions particularly since some of the provincial poverty reduction strategies are geared to Low Income Cut-offs (an unfortunate and unnecessary design decision).
Yeah I can see how that could make for a complicated situation! I was thinking about something I read once about a concept called “path dependency” seems like it’s basically it.
The new U.S. Supplement Poverty Measure (SPM) improves in a number of important ways on the current US measure, but it also has some real limitations in its current form. One problem is that the new measure produces poverty rates that are actually lower in most US states and for children nationally than the current (and unquestionably outmoded) official measure. This is surprising given that the threshold for the official measure has declined substantially as a percentage of median income over the last several decades–it has gone from roughly 50 percent of median income to just under 30 percent of median–and income inequality has increased over this period.
A related problem is that the “different picture of the poor” painted by the new supplemental measure appears inconsistent in some important ways with the picture painted by direct measures of hardship, like our food insecurity measure. In particular, the measure seems to understate both child poverty and poverty in certain very poor regions of the United States, including Appalachia and the Deep South. For more on this, see my analysis at: http://www.cepr.net/index.php/data-bytes/poverty-bytes/is-child-poverty-less-of-a-problem-than-we-thought.
Finally, I don’t think we can confidently say that the new supplemental measure will change over time “as the things needed to participate normally in society change.” Unlike Canada’s Market Basket Measure, the new supplemental measure does not include any “decision rule” for deciding when something, like a computer or education-related expenses, becomes a new necessity that should be reflected in the measure. Instead, it will only respond to changes in spending on the four very basic items that are part of the threshold: food, clothing, shelter, and utilities. If history is any guide, expenditures on these items as a share of all expenditures may actually decline over time. Here, it’s worth remembering that the “new” measure is actually based on recommendations by a expert panel that met in the first half of the 1990s, a time, as you note, “before most of us sent our first e-mail.”
In my view, the US Census Bureau would be wise to also adopt something similar to Canada’s Low Income Measure (LIM), which, as I understand it, is set at half of median adjusted household income. This would ensure that we have at least one low-income measure that doesn’t decline in value over time relative to mainstream living standards.
Thank you for your thoughtful, and helpful, comments. I would make one note with reference to the recommendation in your last paragraph concerning the LIM. It is certainly the case that we should be striving for rules-based updating, rather than discretionary updating. But I am not certain the rule in the current LIM is the right one. As it stands the LIM is the line corresponding to 50% of the median income (appropriately defined at the individual level with some acceptable equivalence scale) in the current year. Now, a poverty line should evolve as the norms concerning what it takes to participate “without shame” in a society evolve, but surely this does not fluctuate on an annual basis in the way that the median income might (for example, it might decline significantly during a recession and end up indicating that poverty rates have fallen). The caveat I would introduce is to use a five year moving average of the median income as the basis for calculating the poverty line. Five years, on the basis simply that it seems like a reasonable rule of thumb.
best mc
“One problem is that the new measure produces poverty rates that are actually lower in most US states and for children nationally” I saw some of that fluctuation in the links provided, I was wondering what they ment. Especially the povery rate of people under 18… I just figured they adjusted for the fact that a teenager’s “income” is not really what mesures his or her wealth when going to school and living with parents. (am I wrong?).
The basic idea that the threshold is to change overtime does make sense to me however.
While I can live with the MBM or the LIM for practical purposes, I think we should be aware of some of the conceptual issues and their implications. Both of these measures (and the LICOs as well) are essentially ‘pick an income’ measures, whereby we find some methodology to pick an income level and adjust it for family composition, and then call everyone above the level not-poor and everyone below that level poor. In reality, we all know from personal experience that there are vastly different levels of standard of living between households of very similar incomes. Some of these differences have to do with annual income versus wealth, access to less expensive ways of meeting household needs (such as lucking in to a decent but cheap apartment) and some has to do with household skills and time for things like meal preparation. So when we pick an income we are really claiming that households above this line have a higher probability of not being poor; while those below the line have a higher probability of being poor — and the probabilities increase with the distance from the line. It would be most useful to be able to observe living standards directly as at least once to see how accurately the income lines really do approximate poverty and what the probabilities are really like. (Tangentially I think this is possible and an empirical study could be designed which would pass the kind of standards of quality commonly expected in, for example, health research.) As well, the issue of family size adjustment remains significant for me. I think this is one of the issues affecting the measurement of poverty among the elderly. Also, I can see no reason to think that the adjustments are the same everywhere when one of the main expenditure items affecting household costs is rent, and we know rent is vastly different in different places in the country. I wonder what the US measure has to say about this issue.
Thanks for these insights Michael. I certainly agree.
We should think of these poverty lines, in the first instance, as a communication device. It is rare that our well-being changes discretely if we are on one side of a line or not. But they do have value as a basis for comparisons over time: is the poverty rate according to a particular definition getting worse or better? Using them in this way may be more appropriate than using them as indicators of overall well-being, and this is all the more reason to be certain that they are appropriately updated—which was the major message of this post.
It is also certainly correct to point that income is not a comprehensive or the only appropriate measure of well-being.
best, mc.
I was sort of hoping to inspire some US replies – we shall see. My basic point is a little different than ‘income is not a measure of well-being.’ It is that the concept of ‘poverty’ is a concept of a standard of living, and that we only use income as a convenient way to assess standard of living indirectly. (And it is very easy to show that poverty is not an income concept since if it were it would make no sense at all to say of two households of the same structure and same income that one may be poor and the other not.) I know this was not exactly what you were talking about, but I could not resist making a comment. Shouldn’t we try and design a proper double blind study to test some of our measures of poverty, rather than just blathering about it. Why not some good empirical research to test whether the measures of poverty do in fact what we would ordinarily in our society consider poverty? (OK I know I am off topic now…)
Re: Geographic differences in housing costs: The new US measure does attempt to adjust for this (and also for differences in housing status, i.e., renting, owning with mortgage, owning without mortgage). But this creates some new issues. As I found in this paper, http://www.cepr.net/index.php/publications/reports/a-modern-framework-for-measuring-poverty-and-basic-economic-security, it seems to weaken the association between state-level poverty rates and both state-level food insecurity rates and state-level education-health index numbers. Part of the problem may be that low housing costs in very poor places in the US (like Appalachia or the Delta region) reflect a lack of amenities in those places. Thus, somebody living at the housing-cost-adjusted poverty line in the Mississippi Delta may have a lower standard of living than somebody living at the housing-cost-adjusted line in Minnesota.
This and other problems that one runs into when trying to develop a “single best” poverty measure have led me in the direction of thinking that the best approach is a pluralist one: instead of a single income-only measure, we should use three or four measures, which taken together paint a picture of how many people have an “unacceptably low standard of living.” Right now, I think the UK’s new set of poverty measures–which include a deprivation index–provide the best working model of this approach, although improvements could be made on what they’ve done.