How to Measure and Monitor Poverty? LIM vs LICO vs MBM.

The federal government has promised to launch a Canadian Poverty Reduction Strategy in the coming weeks or months on the basis of now completed consultations with Canadians and the still ongoing deliberations of an advisory committee. As part of this process, there has been discussion about which poverty or low income measure or measures should be used for the purpose of monitoring levels and trends in the incidence of poverty and gauging the impact of poverty reduction policies. At various times, there have been calls for an official Canadian poverty line, as exists in the United States and some other countries, and some have called for poverty reduction targets which would require the specification of a poverty line or lines. (See )

It should first be noted that any poverty line dividing the poor from the non poor at a given level of income and for a given household size is arbitrary and value-based. It explicitly or implicitly involves a judgement as how far below the mainstream people should fall before they are considered to be poor in terms of either their income or their ability to obtain the essentials of life. And any line must be used to tell us not just how many persons are poor at any point in time, but how far the poor fall below the poverty line. (For example, most social assistance recipients live in deep poverty, while most seniors in poverty are clustered just below the poverty line due to receipt of the Guaranteed Income Supplement to Old Age Security.)

The line should also be used to inform us how long the poor remain poor. (For example, social assistance recipients with disabilities tend to remain in low income much longer than the working poor who cycle in and out of poverty.) Finally, a useful poverty line should inform us of the incidence of poverty by age, gender, racial status and aboriginal status, disability status, economic family type, and so on, as well as by province and region.

A single poverty line as called for by some has the merit of being relatively simple and potentially easy to communicate. As well, a clear indicator showing the impact on the incidence and depth of low income of policies such as increased child and senior benefits could help build public support for a poverty reduction strategy.

However, choosing a single measure risks glossing over different concepts of poverty and overly minimizing the complexity of the issue.

Currently, Statscan provides annual data based on three different measures of low income – the LICO AT, the LIM AT and the MBM. (See CANSIM Table 206-0041 for detailed data on poverty using these measures.) LICO estimates are also presented on a pre tax basis but these are seldom used. While the three measures in use today are not described as poverty lines, they are generally used as such, and they all allow for assessment of levels and trends in a disaggregated fashion.

The LICO AT (after tax) tells us that a person or family is spending a much higher than average percentage of its income on the essentials of food, shelter and clothing (based on family size and with account taken of the size of the community in which the household resides.) The LICO line is based on 1992 living costs, so trends tell us how much progress has been made over time in terms of the ability of Canadians to purchase a basic basket of goods at 1992 spending weights..

The poverty rate in 2015 based on this measure was 9.2%, down from a high of 14.0% in 1983, but it can be questioned if poverty has really fallen so significantly. The LICO has fallen into disfavour because it does not tell us how many persons are unable to achieve a basic standard of living in terms of what Canadians are consuming today, as opposed to twenty-five years ago. For example, the LICO basket does not include the cost of internet access. The LICO does, however, give us some sense of the very long-term trend in the living standards of the poor, and tells us that there has been some absolute income growth over time among the poor.

The Market Basket Measure tells us that a household – in after tax terms, adjusted for family size – has insufficient income to purchase a modest basket of goods and services. The MBM was called for by federal and provincial ministers, and the composition of the basket was determined by government officials rather than by Statistics Canada. It has been calculated since 2002 for a reference family in a large number of communities, so it varies with the local price of housing and food. It is more than an extreme bare bones, basic needs budget insofar as it includes child care costs and the cost of a modest vehicle where transit is unavailable.

That said, there has been a lot of disagreement about the contents of the MBM basket, and many argue that it is a poor measure of the consumption gap between low income Canadians and the mainstream. MBM does not centrally view poverty as being about distance from the mainstream, bur rather as having an income which is insufficient to meet the basic needs of a low income family.

In 2015 the MBM rate at a national level was 12.1%.

The LIM measure (Low Income Measure After Tax) draws a low income line based on 50% of the income of a median household of the same number of persons. It is a purely relative measure with poverty being seen as having an income well below the norm defined as the income of a mid point Canadian family.

In 2015, the national LIM rate was 14.2%. This measure is based only on income relative to the national median income, and is not a measure of basic needs based on consumption.

LIM is very useful in terms of telling us how the bottom of the income distribution is doing compared to the broad middle-class, and how that is changing over time. It is also very useful in terms of international comparisons, telling us that the gap between the bottom and the middle is much wider today in Canada than many European countries, but that low income is much less prevalent in Canada than the United States.

The big problem with the LIM is that it does not take account of large differences in living costs between cities and regions. For example, no account is taken of very large differences in rents between big cities, or the high cost of food in many remote and rural communities.

There is not a great difference between the LIM and MBM measures when it comes to calculating the overall incidence of low income. Over the past five years, the LIM rate has averaged 13.5% compared to 14.2% for the MBM rate. Both rates have remained fairly constant since 2002. (The gap in 2015 – a 14.2% LIM rate compared to a 12.1% MBM rate – was unusually large.)

However, there have been some important differences over time and for some sub populations.

There is a huge difference between the LIM and MBM poverty rates for seniors (14.3% vs 5.1% in 2015.) Also, there have been big changes over time in the LIM based poverty rate for seniors. This fell from 33.1% in 1977 to a low of 3.9% in 1995, before increasing to 14.3% in 2015. The income gap between seniors and other families narrowed initially due mainly to improvements in public and private pensions, but in recent years the incomes of many seniors have been falling behind those of working age families in relative terms. This is not captured in the MBM measure. (As an aside, the LIM poverty rate for seniors would likely not rise if Old Age Security and the Guaranteed Income Supplement were to be indexed to wage growth and not just inflation.)

The apparent stability of the LIM rate over time also hides a long-term increase in the low income rate for the working age population, especially single persons, and, importantly, a major decline in the low income rate for single parent families headed by women reflecting a significant rise in participation in the labour market.

There is also a big difference between LIM and the MBM when it comes to calculating the incidence of low income in Quebec. The Quebec LIM rate is 16.2% compared to a 10.9% MBM rate. The LIM rate in Quebec is 2.0 percentage points above the national LIM rate, but the Quebec MBM rate is 1.2 percentage points below the national MBM rate. This difference is likely due to low housing costs in Quebec compared to other provinces.

The key point is that the conceptual and measurement differences between LIM and the MBM result in significant differences in rates of low income for important sub populations. It is important to have both measures to account for this complexity.

It is also important to appreciate that the drivers of the LIM rate and the MBM rate are different. The LIM rate reflects changes, not just in the incomes of low income families, but also in median incomes. The LIM rate could rise if median wages began to grow after years of stagnation, and if bottom incomes did not follow suit. By contrast, the MBM rate could fall due to increased income supports which lowered the real cost of living of the poor, even if the gap between the middle and the bottom were to grow. Both measures should register major changes in the labour market and in income transfer programs.

By way of conclusion, the LIM and the MBM are conceptually different measures, both of which provide useful and important information for analysts and policy makers. We need both to get a handle on overall low incomes and trends in different populations.


  • Michael Wolfson

    This is an excellent review. My only concern is the last sentence.

    Both the MBM and the current LIM have important shortcomings. A better approach would be to shift focus to a single new “dynamic geographic LIM” rather than continuing with a pair of flawed indicators. On the other hand, it would be far better for information on the extent and character of poverty in Canada to be regularly informed by a “dashboard” of indicators spanning areas like health, wealth, housing, and food insecurity as well as their overlaps.

    For now, though, let me focus on the MBM and LIM. There are straightforward methods to adapt the LIM for the key concern pointed out, namely its failure to reflect geographic differences in living costs across Canada. This can be addressed using the same geographic breakdowns as the MBM (if desired), but focusing more simply on the most geographically variable and straightforward components of the MBM, rents and food. And instead of enumerating all of the detailed list of commodities in the MBM, the focus could be on changes in relative prices both between geographies and over time. The overall consumption basket could then simply (albeit arbitrarily) be 33% for shelter, 20% for food, and the balance treated as the same across the country. Then this revised LIM could be updated simply, transparently and automatically as new income distribution and relative price data were produced each year.

    And it would not suffer from the same schizoid character as the MBM. The MBM is a relative measure in fits and starts, whenever it is updated (e.g. once a decade). But it is an absolute measure in between, when it is updated using the CPI to account only for price inflation. In a country as wealthy as Canada, Adam Smith’s famous quote is apt, that poverty is the inability to afford, “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”. An absolute measure of income poverty could certainly be included as one of a family of indicators in a poverty dashboard. But if there is to be a single “headline” poverty indicator in Canada, it should be a relative measure, in line with the vast majority of international comparative analyses involving OECD countries.

    The MBM updates themselves, by some group of well-meaning individuals, make arbitrary judgments about the specific items to be included in the market basket. The 2010 update required almost 100 pages to describe all the myriad detailed judgments involved. This is neither simple nor transparent.

    The ”dynamic” in my reference to a “dynamic geographic LIM” refers to the idea that instead of basing the LIM on 50% of median family income in the same year, it would be based on 50% of a moving average, over the most recent 5 years (say), of this same median family size-adjusted income. This change to a moving average base for the LIM would avoid situations where, in the face of an economic downturn, the LIM-based incidence of low income could fall. Instead, with a moving average, the LIM-based poverty rate would increase, as most would expect.

    In sum, if we are to have a single “headline” income poverty indicator, it is time to revisit both the LIM and the MBM and come up with the best features of both in as simple and transparent a way possible.

  • “In sum, if we are to have a single “headline” income poverty indicator, it is time to revisit both the LIM and the MBM and come up with the best features of both in as simple and transparent a way possible.

    Poverty is not simple and no matter how much we want to try and idealize an indicator for sure- it must involve value judgements and typically these must be formed in as democratic or representative process to elevate and prevent poverty. Poverty is a local issue- it comes in many varieties and causes. We must understand these factors and bring them into the measurement process. We know from our past experiences at Statistics Canada-when pushing out some mathematical proportion like a LICO and avoiding the issue using a duck and cover strategy for constantly warning institutions not to use it as a poverty indicator is the worst of all possible acts a policy maker can do. As people in poverty were subjected to such arbitrary measures as other institutions did indeed use such measures and they did guide policy decisions.

    So first off I agree with Andrew- we need a new measure- but I disagree that any of the measures he cites are useful for actually deriving policy for addressing poverty. All three of the measures do not meet the criteria for poverty reduction- and that is to reflect a standard of living that reflects a collective effort of what the line between poverty and not poverty is at the local level and meets the needs of those in need.

    In fact it has been a war on the poor by government institutions by avoiding the issue and producing the half witted and arbitrary measures used in the past. I do not have the room to go through the critique of each- but there has been little guidance or care in government policy or statistics in place on these measures. We spend more time measures on the weekly rail car loads of gravel in the economy than we do measuring the poor at Statistics Canada- and it is insidiously covered over with pretending that any of the measures above are even remotely well thought out as being considered a useful set of constructs to measure such complicated notions as poverty. There are no easy measures- but in this age of information and local pricing- it would not be difficult to produce new information to feed into such a complicated calculation.

    I know many activists know these issues quite well but are too connected to the political processes to be this ruthless on the institutions like I am being here- but we must be when it comes to the history of measuring poverty in this country. It has been a disgrace and without a doubt a part of the problem of poverty in this country- a failure to acknowledge it with due diligence and a scientific process combined with a collective compassion. Yet somehow car loading of rail statistics get weekly counts and a massive amount of attention by Statistics Canada.

    The one issue I agree with the above authors- and I do sympathize with Andrew as he is in this position of being polite and trying to work with the institutions- it is time to pull apart the efforts of the past and come up with new measures- that reflect the interests of class, gender, race, indigenous and disability in this country. It is time for some actually existing poverty to be measured with a care and a reality that we as a country can use. Comparing to international regimes is again just a ruse to keep our institutions aligned with the current global imbalances of income and risk.

  • Michael Mendelson

    In assessing ways to measure poverty we should first ask ourselves what it is we are measuring. ‘Poverty’ is a standard of living – obviously not a good or desirable one, but nevertheless a standard of living. If so, poverty is not an amount of income: rather when we use an ‘income line’ to measure poverty we are in effect saying that anyone below that income is likely to have a poverty level standard of living. Income is therefore a measure of likelihood.

    Any income measure of poverty is going to have some false positives (households that fall below the line but whose actual standard of living is not at a poverty level) and some false negatives (households whose income is above the line but who actually do have a poverty level standard of living). A very low income line will result in few false positives but many false negatives, and vice versa for a high income line. No income line is going to have no false results, because human households are far too variable. If our objective is to count the number of households in poverty then the ideal line would be such that false positives equal false negatives so your total is accurate. If you want to identify households that warrant intervention, and you want to be as fiscally efficient as possible, then you want a lower line so as to minimize false positives. And so on – where the line should be depends upon the objectives for measuring. Given that there are many different objectives, this is an argument for several different poverty lines (remembering that regardless of which poverty line we use, we are always measuring the same underlying phenomenon).

    It is also an argument for searching for ways to assess poverty more directly, rather than using an amount of income – which is a very blunt instrument. This is why Geranda Notten and I have been encouraging Canada and provinces to add a material deprivation measure as well as income poverty lines. Material deprivation measures are used in Europe and by the EU statistical agency. This kind of measure looks at what a household actually has and what they actually can do (e.g. have a friend over for dinner), as a way of assessing standard of living directly rather than indirectly through income. It is a useful supplement to income measures.

    Now, not to be too verbose, but a few words on the LICO and the LIM:

    While it is true that the LIM is commonly seen as a reasonable way to compare poverty or low income levels among countries, the LIM does not take into account the availability of public goods (or just to speak in jargon for a second – decommodification). So a country that does not have universal health care and one that does may have similar poverty levels according to the LIM, but this will not at all reflect the lived experience of people with low income, which will be incomparably harsher in the country without universal health care. In developing countries, the availability of inexpensive or free quality education would like be a critical factor – again not reflected in the comparison of LIMs.

    The LICO is based on the likelihood of a household spending more than the average % of their income on food, shelter and clothing. As a measure of likelihood it does not in fact imply that a household below the LICO spends more than the average on food, shelter and clothing. Many will but also many will not. There are also real problems in the rebasing of the LICO after 1992 to create a truly consistent series.

    I think all these measures are useful in their own way, but they should all be used with caveats.

  • Edgardo Sepulveda

    Thanks for the useful overview, Andrew.

    In addition to the monitoring and impact assessment functions that you mention, I wanted to add a third perspective; the pros/con of using any income-based measure as eligibility criteria for participation in low-income assistance/affordability programs.

    My background is in the economic regulation of utilities and three years ago I participated in CRTC telecommunications process to determine what constitutes “basic telecoms service” in Canada and what is the federal responsibility in ensuring that Canadians have access to same.

    The intervener I worked for (PIAC) advocated for a number of measures, including an ongoing subsidy payment to low-income households to increase affordability. There are a number of such programs in the telecommunications and electricity sectors in OECD countries, including USA, France, Spain, etc. I know of only one example in Canada; as I’ve mentioned in my Ontario electricity-related blogs, the Government here recently established the OESP to provide an ongoing subsidy payment to low-income households related to electricity. By the way, the Government in BC is also considering such a “lifeline” electricity subsidy payment to low-income households in that province.

    One of the key program design parameters in any such programs is eligibility criteria. This goes to idea that these programs are targeted, not universal and require “opting in”. The international experience is that countries use some combination of income-based and/or program participation criteria. The former is that a household would qualify by showing documentary proof that its income is below a certain threshold. The latter is simpler – the household only has to show that it is currently the beneficiary of another Government assistance program. The rationale for this eligibility avenue is that the household has already had to undergo some form of eligibility screening (most likely income-based) and it does not make sense for that household to have to go though a similar but separate process. The combination is that a household could prove eligibility by either showing a low income or by participating in designated programs.

    Against the advice of interveners and other experts, the Ontario Government established OESP eligibility only based on a low-income threshold, the LIM-AT (after tax). Not surprisingly, participation has been below expectations, partly because of the income-only eligibility – households are required to submit documentary proof of their income, including CRA notice of assessments for adult members of the households, etc. Plus the administration has been relatively expensive because a specialized private-sector administration firm had to be contracted to review and approve applications. It is mostly online, so that costs are relatively high but still less than those associated with eligibility for social assistance (welfare) in Ontario and elsewhere.

    The international experience would have predicted such a relatively low “take-up”. The evidence is that such programs have the highest participation rate when a combination approach is taken. For example, the USA federal “lifeline” subsidy payment uses such an approach, linking eligibility to income below 135% of the federal poverty line or participation in a dozen or so social programs. The research is that more than 95% of participating households were determined to be eligible based on being beneficiaries of another existing program (mainly “SNAP” (food stamps). By the way, as an epilogue, the Ontario Government is belatedly “looking into” adding participation in social assistance as an alternative eligibility criteria. That is, a recipient of OW/ODSP would be eligible for OESP without having to go though the OESP-specific LIM-AT income threshold process.

    In summary, without getting into the pros/cons of any of particular existing/new income measures, and the broader discussion of means-testing, I wanted to provide a cautionary tale from the utilities trenches of relying exclusively on any type of income measure as eligibility criteria for participation in low-income assistance/affordability programs.

Leave a Reply

Your email address will not be published. Required fields are marked *