Volume 47 Issue 1, March 2021, pp. 1-17

Unique data from a 2013 Canadian survey were used to measure material deprivation. This outcome-based indicator of poverty was constructed of 17 necessities. When persons who cannot afford two or more items are considered materially deprived, material deprivation is found to affect 18.6 percent of Canadians. Of those, only 43 percent also have low income. Of Canadians with low income, only 50 percent are materially deprived. The experience of poverty-level living conditions thus regularly coincides with an income above the poverty threshold, and having low income does not guarantee material deprivation. Outcome-based poverty indicators such as material deprivation therefore offer new and relevant insights into understanding poverty and its policy solutions in Canada.

Des données exclusives provenant d’un sondage canadien de 2013 sont utilisées dans l’évaluation de la privation matérielle. Cet indicateur de pauvreté basé sur les résultats regroupe 17 biens et services de première nécessité. Si l’on considère que les personnes qui ne peuvent s’offrir deux de ces produits ou davantage sont privées sur le plan matériel, la privation matérielle touche 18,6 pour cent des Canadiens, peut-on constater. De ce groupe, seulement 43 pour cent ont aussi un faible revenu. Parmi les Canadiens ayant un faible revenu, seulement 50 pour cent sont privés sur le plan matériel. Des conditions de vie associées à une situation de pauvreté coïncident donc régulièrement avec un revenu supérieur au seuil de pauvreté, et un faible revenu n’entraîne pas nécessairement la privation matérielle. Les indicateurs de pauvreté basés sur les résultats comme la privation matérielle livrent, par conséquent, de l’information inédite et pertinente nous permettant de comprendre la pauvreté et les politiques propres à y remédier au Canada.

Poverty is a policy priority even in rich countries such as Canada (ESDC 2018; Tarasuk et al. 2019; Van den Berg et al. 2017), and understanding poverty and the effects that policies have on it requires adequate measures of poverty (Marlier and Atkinson 2010; National Research Council 1995). Defining the material dimension of poverty as the experience of adverse living conditions resulting from insufficient financial resources, we use a, for Canada, unique survey to construct an outcome-based poverty indicator called material deprivation and thereby fill a long-standing knowledge gap on poverty in Canada (Fafard St-Germain and Tarasuk 2018; Heisz and Langevin 2009; Notten 2015). Our findings corroborate a large international body of evidence showing that monetary- and outcome-based poverty indicators supplement each other and concluding that it is advisable to use both types in policy-making (Alkire 2015; Bossert, Chakravarty, and D’Ambrosio 2013; Fusco, Guio, and Marlier 2011; Nolan and Whelan 2010; Saunders and Brown 2020).

We use data from the 2013 Canadian Survey of Economic Well-being (CSEW) to develop a material deprivation indicator for Canada. This one-time survey currently provides the only data in Canada that allow for nationally and provincially representative estimates of material deprivation. We construct a deprivation scale by choosing 17 items from that survey following a validated methodology (Guio et al. 2016). These items reflect responses to questions about access to specific goods and services and also about circumstances such as whether the respondent’s household lives in a pest-free dwelling or could cover an unexpected expense of $500. The choice of the items and the construction of the index are further discussed later in this article. Notably, each of the 17 items are what we call necessities: they were rated as necessary for a minimally acceptable living standard by a majority in a provincially representative Ontario opinion survey (Matern, Mendelson, and Oliphant 2009a) or, in a few cases not covered by that survey, in surveys of other culturally close populations. We present representative statistics on material deprivation and analyze differences in the overlap between material deprivation and low income using the low income measure before taxes (LIM-BT). The LIM-BT is the only low-income indicator available in the CSEW. Earlier research for Ontario alone with different data had similar findings when using other commonly used low-income indicators (Notten 2015).

We find that 18.6 percent of Canadians are materially deprived, defined as individuals reporting that they or members of their household cannot afford two or more of the items that we labelled as necessities. Of those, 43 percent also have low income. Of all Canadians with low income, 50 percent are materially deprived. For Canadians whose characteristics are associated with an elevated risk of poverty, this translates into a very high proportion experiencing material deprivation, income poverty, or both. Material precariousness, if not outright poverty, may thus be much more widespread than low-income indicators suggest. Setting a more or less stringent deprivation threshold has a large effect on the prevalence of material deprivation, but it does not reduce the large discrepancy in overlap between both types of poverty indicators. The discrepancy arises in large part because the financial resources, needs, and living arrangements of Canadians are far more diverse than can be captured by the measurement methodologies of (leading) monetary poverty indicators. By focusing on outcomes resulting from insufficient financial resources, material deprivation indicators automatically allow for diversity in households’ circumstances (e.g., needs and resources).

These findings confirm that low-income indicators, which Canadian governments of all levels use to monitor progress on poverty reduction, may thus misdiagnose the prevalence of poverty for a sizable group of their population. The understanding of material poverty could be significantly improved by also measuring poverty using outcome-based indicators and by collecting data that offer both types of information for the same persons.

In the following section, we explain why and how outcome-based indicators supplement monetary indicators. Next, we discuss the indicators that Canadian governments use to monitor progress on poverty reduction and what they (do not) tell us. The “Data and Descriptive Statistics” section presents the CSEW data, and the “Measuring Material Deprivation” section describes the methodology for measuring material deprivation. Then, we present and analyze the results. Finally, we discuss the policy implications.

Poverty statistics play an important role in policy processes because they inform stakeholders about the evolution of poverty (level, trend, and differences between population groups), its nature (causes, consequences, and interactions), and the role of public policy in reducing poverty (Notten 2015). Research on poverty measurement assesses, among much else, to what extent indicators can reliably be used for such purposes (Guio et al. 2016; Marlier and Atkinson 2010; National Research Council 1995). A large literature covering many countries, time periods, and different types of indicators shows that there is a statistically significant but only modest association between monetary and outcome-based indicators of poverty. This literature concludes that it is advisable to use both types for policy purposes (Alkire 2015; Bossert et al. 2013; Fusco et al. 2011; Nolan and Whelan 2010; Saunders and Brown 2020).

We focus on the material dimension of poverty, defining poverty as the experience of adverse material outcomes resulting from a lack of financial resources. Monetary indicators, such as low-income indicators, operationalize this concept by measuring financial resources; outcome-based indicators, such as material deprivation indicators, measure adverse material outcomes resulting from insufficient financial resources (Guio 2009; Heisz and Langevin 2009; Notten 2015; Townsend 1979). Both types of indicators embody normative judgments on what constitutes poverty. Monetary poverty indicators judge what level of financial resources is sufficient to finance a minimum acceptable living standard. Outcome-based poverty indicators judge what living experience constitutes a sufficient level of material well-being.

The most popular monetary indicators focus on measuring a household’s income and identify a household and all its members as poor when its income falls below a pre-specified monetary amount. This poverty threshold can be derived relatively—such as Canada’s LIM, which sets the threshold at 50 percent of median income—or in an absolute fashion—such as Canada’s market basket measure (MBM), which captures the costs of a basket of necessities for a reference household in a reference period (ESDC 2018; Notten and De Neubourg 2011).

Outcome-based poverty indicators such as material deprivation rely on a series of two-tiered survey questions to assess poverty-level living conditions. A first question enquires about a factual situation (e.g., whether the household eats meat, chicken, fish, or a vegetarian equivalent at least once a day). When the answer to the first question is negative, a follow-up question asks whether this is because the household cannot afford it or because of some other reason. Households answering “no” to the first question and stating affordability as the reason in response to the second question experience so-called item deprivation.

The items that constitute the material deprivation indicator are not meant to represent an exhaustive list of necessities. Rather, the intent is to identify an adequate selection of observable items that together form a scale for measuring material deprivation (Berthoud and Bryan 2011; Heisz and Langevin 2009). The items can be framed in either subsistence (absolute) or inclusive (relative) terms (Berthoud and Bryan 2011). A household is materially deprived when its score on the material deprivation scale falls below a pre-specified deprivation threshold (Alkire and Foster 2011; Guio et al. 2016).

Low-income and material deprivation indicators each measure a core aspect of the material dimension of poverty, but they do so in very different ways (Figure 1). Income is an important monetary resource; the income definitions used to construct poverty indicators cover a broad range of income sources and typically include government transfers. However, access to other monetary resources may explain why someone with low income can avoid adverse material outcomes. Savings or access to credit can help meet expenses (Brandolini, Magri, and Smeeding 2010; Rothwell and Robson 2018). Moreover, access to non-monetary resources, such as subsidized access to goods and services (Paulus, Sutherland, and Tsakloglou 2010) and employer-provided health benefits (Hajizadeh and Edmonds 2019) can reduce the need for out-of-pocket spending. Social networks can also help, either by supplementing income or by reducing the need for spending (Kalil and Ryan 2010).

There are, however, factors that explain why certain households do not have enough monetary resources to meet their needs. Conditions such as chronic illness or disability often require more resources to satisfy needs (She and Livermore 2007). Similarly, debt payment obligations reduce the finances available for current needs (Pressman and Scott 2009). Income from employment may crucially depend on significant spending such as child care and transport (Allen and Farber 2019; Beaujot, Du, and Ravanera 2013).

Figure 1 illustrates the benefits of measuring poverty by measuring material outcomes. Such indicators capture the link between insufficient financial resources and adverse material outcomes, but they do not require information about those resources, their purchasing power, or households’ specific needs and circumstances. Monetary poverty indicators, however, require additional information to adjust for such differences between households. Whereas established methods exist to accommodate for differences in purchasing power (spatial and intertemporal), demographic composition of the household, house ownership, other assets, and out-of-pocket spending on health, they require a lot of very detailed information and a host of assumptions to enable their application (National Research Council 1995). For instance, research on methods to adjust for differences in purchasing power over time shows that an Engel curve approach is no panacea to measurement challenges in the commonly used consumer price index (Emery and Guo 2020 and the references listed therein).

Figure 1: Measuring Poverty and Drivers of Material Well-Being

Source: Authors.

In practice, Canada’s monetary poverty indicators accommodate only some of these differences (Corak 2018; Experts Panel on Income Security 2018), suggesting that this could be a source of error in identifying Canada’s poor. Another well-known source of errors is the under-reporting of income (Brzozowski and Crossley 2011).

Outcome-based poverty indicators also have shortcomings. Outcome-based indicators are survey dependent and therefore depend on survey response rates. Moreover, psychological factors such as shame or adaptive preferences, for instance, may lead to under-reporting of deprivation (Breunig and McKibbin 2011; Guio 2009). Finally, the methodology for measuring material deprivation presumes that people similarly prioritize needs and preferences, a presumption that may be false for certain groups (McKay 2004).

With poverty reduction strategies underway in many of Canada’s jurisdictions and governments expressing a desire to use evidence to monitor progress, issues around poverty measurement have gained prominence in policy discussions. There is now considerably more information available on poverty than two decades ago, and most of Canada’s governments routinely use this information in policy-making. With Statistics Canada collecting a lot of it, the federal government plays a key role in making this information available to its many users, including lower levels of government.

When tracking progress on poverty reduction in Canada, monetary indicators of poverty—and especially those based on income—dominate as lead indicators for governments’ poverty reduction agendas and as primary indicators for tracking material progress. The federal government, for instance, adopted the MBM as Canada’s official poverty line in 2018, aiming to reduce the number of people with an income under it by 20 percent by 2020 and by 50 percent by 2030 (ESDC 2018). The federal government also uses many other income-based poverty indicators, including a deep income poverty rate (using 75 percent of the MBM poverty line), the average poverty gap (MBM), the LIM (which sets the poverty line at 50 percent of median income), and poverty entry and exit rates (currently based on the LIM). The MBM and LIM are also the monetary poverty indicators most commonly used by provincial and territorial governments.[1]

Only sometimes do outcome-based indicators of poverty supplement the income-based indicators in tracking material progress, and rarely do such indicators capture poverty with a conceptual intention as broad as that of income poverty indicators. The federal government, for instance, uses a measure of household food insecurity (ESDC 2018). This indicator measures moderate and severe food insecurity resulting from financial constraints and is thus a measure of severe material deprivation (Fafard St-Germain and Tarasuk 2018; Tarasuk et al. 2019). Its conceptual scope, however, is narrow, covering only one (very essential) consumption category; thus, even when a family is food secure, its financial resources may still be insufficient to meet other necessities. The same holds for two other indicators used by the federal government, unmet health needs and unmet housing needs (ESDC 2018).

A notable exception in Canada has been the material deprivation indicator used by the Ontario government during the first few years of its poverty reduction strategy (Government of Ontario 2008; Matern, Mendelson, and Oliphant 2009b). On the basis of the affordability of ten necessities, this indicator measured the prevalence of Ontarians experiencing poverty-level living conditions resulting from a lack of financial resources. The Ontario government self-financed the collection and production of the material deprivation indicator by Statistics Canada and, eventually, discontinued those efforts.

Most outcome-based indicators used to track progress on poverty reduction strategies cover aspects other than material poverty, such as child development, labour market, health, housing, education, and social inclusion. Although they do not measure poverty as such, these indicators serve important functions, such as highlighting the multidimensionality of well-being, recognizing the interrelatedness of outcomes between dimensions, and being primary indicators to monitor progress on the effects of government involvement in the provision of services in health, education, and housing. Examples of such indicators are birthweights, life expectancy, early development instrument scores, literacy and numeracy rates, chronic unemployment rate, rate of jobless families, and number of persons experiencing chronic homelessness.2

Sometimes such outcome-based indicators capture an outcome that, among other reasons, could result from insufficient financial resources. The federal indicator of unmet health needs, for instance, deems needs as unmet when they arise because of costs, waitlists, and unavailability of health care services (Statistics Canada 2019). When unmet health needs arise because the patient cannot afford to pay the costs, the underlying problem may be poverty, a gap in health insurance coverage, or both; when they result from waitlists or a lack of doctors, the underlying problem is something else. Similarly, the federal indicator of unmet housing needs includes affordability, suitability, and state of repair as criteria (ESDC 2018). Given their broader definitions, labeling such indicators as poverty indicators would be incorrect.

Despite the many advancements in data availability and their use by governments, there is still a significant information gap in Canada when it comes to measuring the material dimension of poverty as an adverse outcome resulting from not having enough financial resources.

We use the CSEW to construct a material deprivation indicator for Canada. These survey data are not publicly available but are accessible with confidentiality restrictions in Statistics Canada Research Data Centres. This one-time survey, conducted in 2013, provides the only Canadian data that can offer nationally and provincially representative estimates of material deprivation. Conducted as a sub-sample of the households in the Labour Force Survey (LFS), the CSEW offers household-level information on material deprivation, economic hardship, and income. Household refers to a person or group of persons living in the same dwelling. A responsible member of the household answers the questions on behalf of the household. Of the sampled households, 91.6 percent responded to the LFS, and 67.3 percent responded to the CSEW, resulting in a sample of 24,258 households consisting of 57,911 individuals. We excluded 1,590 individuals (2.7 percent) because of missing information on at least one of the material deprivation questions, resulting in a final sample of 56,321 individuals.

The data provide information on households’ ability to afford 17 different consumption items reflecting basic material or social needs (Table 1). These so-called material deprivation items form a scale that measures aspects of material deprivation, such as its prevalence. The items were selected from Statistics Canada’s Ontario Deprivation Survey (2009), as well as the European Union Statistics on Income and Living Conditions and the analysis conducted by the Eurostat Task Force on Material Deprivation (Statistics Canada 2013). Data collection follows the validated process composed of two survey questions per item described in the “Measuring Poverty” section (Guio et al. 2016).

The percentage of Canadians living in a household that cannot afford an item varies across items: 16.6 percent live in a household that cannot afford to replace worn-out furniture, and 16.5 percent are unable to cover an unexpected expense of $500, whereas only 0.6 percent cannot afford transportation to get around in their community (Table 1). Although a large majority of Canadians (71.2 percent) can afford all of these items, a significant minority cannot (28.8 percent; Figure 2); 10.2 percent of the population cannot afford one deprivation item, 5.7 percent cannot afford two items, and 3.8 percent cannot afford three items.

Income is self-reported before-tax total income. It includes the personal incomes of all household members. Respondents could give proxy answers. We follow the conventional method of adjusting income for household size (using the square root of household size) to calculate the prevalence of low income using the LIM-BT.

The income definitions more commonly used for low income in Canada reflect either after-tax income (LIM-AT) or disposable income (MBM), are based on the Canadian Income Survey (CIS), and largely draw income from tax file data. Unfortunately, the CSEW does not include the information needed to construct these indicators. Our estimates of low-income, low-income thresholds, and quintile thresholds therefore differ from those derived from the CIS. Where relevant, we discuss external evidence to assess the robustness of results.

The relation between material deprivation and income is graded, with item deprivation occurring more frequently at lower levels of income (Figure 2). Yet, some item deprivation occurs among individuals in the top three income quintiles as well, and 40 percent of individuals in the lowest quintile experience no item deprivation. This pattern signals that errors in the identification of who is poor may be substantive, regardless of whether low income or material deprivation is used to identify those who are poor.

The CSEW holds information on household characteristics (composition, age of members, rural or urban residence, homeownership, main source of income) and respondent characteristics (education level, labour force status, immigrant, Aboriginal). We use this information to disaggregate poverty indicators by household characteristics and as categorical variables in a multivariate regression with material deprivation as a dependent variable. We further use economic hardship, involving five questions recording households’ recent experiences dealing with day-to-day finances, in a cross-validation exercise to set the deprivation threshold.

Constructing the Scale

We use the material deprivation items, which are observable, to construct a scale that measures material deprivation (Berthoud and Bryan 2011, 137). We assess each item in terms of four criteria (suitability, validity, reliability, and additivity) by applying the methodology used to test the items of the European Union’s updated material deprivation indicator (Guio et al. 2016). We find that the items perform well overall, with only a few minor issues occurring for some items in some tests. Here we summarize the test results; the working paper and the Supplemental Appendix to this article offer more detail (Notten, Charest, and Heisz 2017).

The suitability criterion asserts that many in a population view the item as necessary. Meeting this criterion requires a sufficiently broad consensus that an item is an element that is necessary to attain a minimum acceptable living standard. A commonly used benchmark is that of a simple majority stipulating that at least 50 percent of a population should view the item as a necessity (Guio et al. 2016; Saunders, Naidoo, and Griffiths 2008). Another, supplementary, benchmark is that persons with lived experience, or those at high risk of living that experience, identify the item as a necessity (Matern et al. 2009b). For the suitability test, we rely on information gathered for the creation of the Ontario Material Deprivation Index in 2009 as part of a collaboration between the Daily Bread Food Bank and the Caledon Institute of Social Policy (Matern et al. 2009a). The Ontario government subsequently adopted this index as part of its indicator portfolio to monitor progress on its poverty reduction strategy. That research identified suitable deprivation items through a three-step process, starting with a survey of food bank users, followed by focus groups involving participants with lived experience of poverty, and concluding with a provincially representative opinion survey (Matern et al. 2009a). Referring to data from the third step, the items dental care, pests, footwear, and temperature were rated by more than 90 percent of Ontarians as a necessity. The items getting around, vegetables, meat, clothing, spending money, and appliances had necessity rates between 70 and 90 percent. The items furniture, gifts, friends, and hobby had necessity rates between 55 and 70 percent. Because no such data were collected for the items unexpected expense and unpaid bills, we instead report findings from similar opinion surveys in culturally close countries (Dickes, Fusco, and Marlier 2008; Guio et al. 2016; Saunders et al. 2008). For unexpected expense, necessity rates are above 70 percent in Australia, Ireland, the United Kingdom, and France. For unpaid bills, necessity rates are above 90 percent (same countries excepting Australia, for which there are no such data). In jurisdictions in which material deprivation data are used in policy processes, the regular collection of such necessity data enables continued monitoring of the suitability criterion.

Table

Table 1: Material Deprivation Questions and Incidence Rates

Table 1: Material Deprivation Questions and Incidence Rates

Item Questions % Deprived
Appliances Are you and your household able to replace or have repaired broken or damaged appliances such as a vacuum or a toaster? 7.1 (0.28)
Furniture Are you and your household able to replace worn-out furniture in your house or apartment? 16.6 (0.40)
Unexpected expense Could you and your household cover an unexpected expense today of $500 from your own resources? 16.5 (0.39)
Pay bills Are you and your household currently able to pay your bills on time? 4.9 (0.25)
Interneta (1) Do you and your household have access to the Internet at home? (2) Do you and each member of your household have regular access to the Internet during your leisure time outside your home? 1.1 (0.07)
Temperature Are you and your household able to keep your house or apartment at a comfortable temperature? 1.5 (0.13)
Getting around Are you and your household always able to get around your community, either by having a car or by taking the bus or equivalent mode of transportation? 0.6 (0.06)
Friends Are you and your household able to have friends or family over for a meal at least once a month? 3.2 (0.20)
Dental Are you and each member of your household able to get regular dental care if needed? 10.3 (0.34)
Gifts Are you and your household able to buy some small gifts for family or friends at least once a year? 2.1 (0.14)
Pests Is your house or apartment free of unwanted pests, such as cockroaches, mice or bedbugs? 0.9 (0.12)
Vegetables Do you and each member of your household eat fresh fruits and vegetables at least once a day? 2.4 (0.16)
Meat Do you and each member of your household eat meat, chicken, fish or a vegetarian equivalent at least once a day? 1.5 (0.13)
Footwear Do you and each member of your household have at least two pairs of properly fitting footwear including a pair of suitable winter footwear? 2.0 (0.13)
Clothes Do you and each adult in your household have appropriate clothes for job interviews? 2.1 (0.15)
Hobby Do you and each member of your household have a hobby or leisure activity? 4.7 (0.25)
Spending money Are all the adults in your household able to spend a small amount of money each week on themselves? 9.7 (0.33)

Notes: No. of records = 56,321. Proportions calculated from weighted counts of individuals. Standard errors are in parentheses and are calculated using 1,000 replicate bootstrap weights. For each item, respondents who answered "no" were then asked,"Is this because you cannot afford it, or for some other reason?" Respondents answering "no" to the first question and stating affordability as the reason in the second question are deprived of the item.

aFor Internet, households stating that they do not have Internet access at home are asked whether they have regular access to Internet during their leisure time outside of the home, and those responding no to this second question are asked whether this is because they cannot afford it or for some other reason. Households who lacked Internet access at home and regular access outside the home because they could not afford it are deprived.

Source: Canadian Survey of Economic Well-being; authors’ calculations.

Figure 2: Number of Deprivations, for Total Population and Adjusted Household Income Quintiles

Notes: Estimates of marginal quality (CV between 16.6% and 33.3%) are included; Estimates of poor quality (CV > 33.3%) are suppressed (occurring in fourth and fifth income quintiles). CV = coefficient of variation.

Source: Canadian Survey of Economic Well-being; authors’ calculations.

The validity criterion states that the item relates to other known aspects associated with poverty (Guio et al. 2016). The validity tests involve estimating three binary logistic regressions per item to assess whether it has a statistically significant relationship with three variables available in the CSEW: low income (LIM-BT) and two indicators of economic hardship. One economic hardship indicator measures the ease with which a household is able to meet its needs in terms of transportation, housing, food, clothing, and other necessary expenses (Statistics Canada 2013). A household experiences hardship when it responds “difficult” or “very difficult.” The second indicator supplements the first with four other indicators of economic hardship: whether the household asked for help from friends or family, took on debt or sold an asset, turned to charity, or experienced financial difficulty as a result of a long-term disability or health problem. We identify a household as experiencing economic hardship when it has at least one positive response out of the five questions, thereby using a broader definition of economic hardship than the first indicator. All coefficients in the 51 regressions are statistically significant at the 1 percent level.

The reliability criterion stresses that each item, and the scale, captures one concept. The first reliability test assesses whether the items form one latent concept. We estimate Cronbach’s alpha, a statistic measuring the internal consistency of the scale (Nunally and Bernstein 1978). Cronbach’s alpha is 0.83, a value that is well above the satisfactory threshold level of 0.70 (see Supplemental Appendix). After the sequential removal of each item and recalculation of alpha, its value decreases for most items while slightly increasing for Internet (0.8340) and pests (0.8346), indicating that removing these latter two items yields only a very small improvement.

The second reliability test uses a statistical model based on Item Response Theory (IRT) to assess the characteristics of each item individually and thereby also its complementarity to the other items. We estimate a unidimensional two-parameter IRT model, which offers estimates of the severity and discrimination of each deprivation item (Guio et al. 2016). Severity reflects the likelihood that a person will lack the item. We find that the items’ severity scores vary considerably, a desirable trait, because the level of material deprivation differs between materially deprived households. The severity scores of two items (Internet and pests) have a standard deviation that slightly exceeds the commonly used threshold of three standard deviations (Guio et al. 2016). Such items capture a deprivation level that affects only a very small population, a reason for considering an item’s exclusion. The discrimination parameter of the IRT model measures how well a specific item differentiates between a deprived and a non-deprived person. We use this parameter to calculate the correlation between the item and the latent variable (Cox 2008). All items are highly correlated (above 0.9) and well above the minimum threshold of 0.4 used by Guio et al. (2016).

The additivity criterion requires that persons experiencing more item deprivations are worse off than those with fewer item deprivations. We test this by assessing whether households with more deprivations have, on average, a lower income level than those with one or no deprivation (Guio et al. 2016). Using income as a proxy for financial resources, we calculate, for each possible pair of items, the average income of households with zero, one, or two deprivations, and we use t-tests to assess whether these averages are statistically significant at the 5 percent level. The income differences are statistically significant for all comparisons of single and zero deprivations (272 pairs of items and 136 t-tests). The levels between single- and double-deprived households are statistically significant in two-thirds of the tests (93 of 136 tests). Items for which the additivity test failed most frequently are pests (11), temperature (11), friends (9), and vegetables (8).

There is no standard for the number of items to retain for a scale. With few items, one risks that the measure misses deprivations experienced by certain population groups, particular contexts, or both. With many items, especially when some of them are strongly correlated, there is redundancy. Here, we retain all 17 items for the material deprivation scale because most of them perform very well on most, if not all, of the tests. We find minor issues for one of the reliability tests (Internet and pests) and one of the additivity tests (pests, temperature, friends, and vegetables). With 17 items, the number is similar to that used in material deprivation scales elsewhere. For instance, New Zealand’s index also has 17 items (Perry 2016) and the European Union’s material deprivation index has 13 indicators (Guio et al. 2016). Material deprivation measures, as do many other measures, need an updating mechanism to stay relevant. This involves a periodic review of items in the scale (every five or ten years) and other considerations, such as the costs and quality of data collection.

Constructing the Indicator

We use the deprivation scale to construct an indicator for the prevalence of material deprivation by adding the number of item deprivations per person and comparing this number to a deprivation threshold (Alkire and Foster 2011).3 The indicator thus measures the percentage of Canadians experiencing material deprivation. We focus on the prevalence because this is the most common indicator used in policy-making.

Setting the deprivation threshold is the most influential decision in constructing a material deprivation indicator (Alkire and Foster 2011). With our data, the prevalence would be 28.8 percent for a threshold of one item, 18.6 percent for two items, 12.9 percent for three items, 9.1 percent for four items, and 6.2 percent for five items.

We selected the threshold by means of an empirical cross-validation exercise, which involved using information correlated with poverty (income and economic hardship) to divide the population into a group that is more likely to be poor and a group that is less likely to be poor (see the “Setting the Deprivation Threshold through Empirical Cross-Validation” section in the Supplemental Appendix). If identified as not materially deprived for a given threshold, people in the first group are likely to be false negatives, thus representing a possible measurement error. If identified as materially deprived for a given threshold, people in the second group are likely to be false positives, also representing a possible measurement error. We compared how the size of each error changes as the threshold increases. Judging false negatives and false positives as equally problematic errors of measurement, we select a threshold of two deprivations because at that threshold the marginal change of persons in both measurement error groups is of similar size.

We performed various robustness tests to ensure that the conclusions we draw are not overly sensitive to choices in constructing a scale and indicator (see Supplemental Appendix). In the next section, we estimate and analyze all results for a three-item deprivation threshold. A three-item threshold reduces the prevalence of material deprivation, but it does not change the substantive findings regarding the characteristics of the population at risk of material deprivation or its relation with low income.

We also estimated the effect of excluding a deprivation item on the material deprivation rate for five possible deprivation thresholds. The effects of excluding an item are very small for most items. At a threshold of two deprivations, they range from a reduction in prevalence of less than half a percentage point to 3 percentage points. Excluding the two highest incidence items from the scale would even increase the disagreement between low income and material deprivation indicators, which is a key finding of this research (see the Results section and the Supplemental Appendix).

In this section, we present the first representative estimates of material deprivation in Canada and analyze to what extent persons’ material deprivation status and low-income status (LIM-BT) overlap, which is possible because the data include both types of information. We thereby offer a rare glimpse into the relation between poverty measured as having low income and poverty measured as the experience of adverse material outcomes resulting from insufficient financial resources. We thus offer evidence of what insights policy-makers, policy researchers, and other stakeholders could lack in the absence of such data.

We start by comparing the prevalence of poverty at a national level with that experienced by different types of households. The material deprivation rate is 18.6 percent (SD = 0.4) using a two-item threshold, and it would drop to 12.9 percent (SD = 0.4) using a three-item threshold (Figure 3). Deprivation varies considerably across households, ranging from 11.6 percent (SD = 0.6) among couples without children to 50.4 percent (SD = 2.3) among lone-parent households. The variation between household types is similar at the three-item threshold, albeit at a lower prevalence.

At the national level, using a deprivation threshold of two items, poverty in a material deprivation sense is higher than poverty in a low-income sense; using a three-item threshold, the magnitude becomes more similar (Figures 3 and 4). The LIM-BT, the only low-income indicator available in our data, yields a higher low-income rate (15.9 percent) than leading low-income indicators in Canada (13.4 percent for the LIM-AT and 12.1 percent for the MBM in 2013). As a result of differences in data collection, particularly the self-reported nature of income in the CSEW, the LIM-BT in the CSEW (15.9 percent) is lower than that in the CIS (17.6 percent). Figures 3 and 4 nonetheless show a consistent ranking of household types across indicators in terms of prevalence, with lone-parent households and one-person households having a prevalence well above the national average and couples without children having a prevalence consistently below that average, regardless of the indicator used.

Figure 3: Material Deprivation by Household Type and Deprivation Threshold

Note: Confidence intervals at 95%, based on bootstrapped standard errors.

Source: Canadian Survey of Economic Well-being; authors’ calculations.

Figure 4: Low Income by Household Type and Low-Income Statistic

Notes: Confidence intervals at 95%, based on bootstrapped standard errors. LIM-BT = low income measure before taxes (but after government transfers); CSEW = Canadian Survey of Economic Well-being; CIS = Canadian Income Survey; LIM-AT = low income measure after government transfers and income taxes; MBM = market basket measure.

Sources: CSEW, authors’ calculations; CIS, unpublished calculation; Statistics Canada Table 11-10-0135-01, income reference year 2013.

When information on different poverty indicators comes from different samples of individuals, as is presently the case in Canada, it is challenging to get a sense of the nature of poverty. A crucial insight from this research is that the conditions of low income and material deprivation coincide only for a relatively small group of Canadians, whereas it is considerably more common for Canadians to experience only one condition. Figures 5 and 6 illustrate this insight using related but different perspectives. Viewed from the perspective of low income, the dominant perspective on poverty in Canada, only half of those with low income were also materially deprived, which also means that the other half were not (Figure 5). Viewed from the perspective of material deprivation, a new contribution to understanding poverty in Canada, only 43 percent also had low income, meaning that more than half of those who were materially deprived did not (Figure 5).

Figure 6 decomposes the population identified as poor by one or both indicators and thus offers a broader perspective of the incidence of material poverty in Canada. Only 8.0 percent of Canadians simultaneously experienced low income and material deprivation, whereas 10.6 percent experienced deprivation without having low income, and 7.9 percent had low income without experiencing deprivation. This implies that 26.5 percent of Canadians are identified as poor by one or both indicators (see also Table 3).

Persons with low income are thus not automatically experiencing material deprivation, and materially deprived persons do not necessarily have low income. Nonetheless, both types of indicators “agree” on the poverty status of 81.5 percent of the population, identifying 8.0 percent of Canadians as poor and 73.5 percent of Canadians as not poor. They disagree on the remaining 18.5 percent of Canadians, who experience either low income or material deprivation (Figure 6 and Table 3). This pattern is well established in the literature and is robust to choice of country, time period, and measurement methodology (Alkire 2015; Bossert et al. 2013; Fusco et al. 2011; Nolan and Whelan 2010). In Canada, it is, for instance, also present at the national level and by household type, albeit with considerable variation in the overall magnitude and the relative size of each poverty group (Figures 5 and 6).

Figure 5: Per Indicator: Overlap in Material Deprivation Status and Low-Income Status

Source: Canadian Survey of Economic Well-being; authors’ calculations.

Figure 6: Population: Overlap in Material Deprivation Status and Low-Income Status

Source: Canadian Survey of Economic Well-being; authors’ calculations.

Table

Table 2: Population Identified as Poor by Material Deprivation, Low Income, or Both (%)

Table 2: Population Identified as Poor by Material Deprivation, Low Income, or Both (%)

Population Deprived LIM-BT Deprived, Low Income, or Both Share of Population with That Characteristic
Canada 18.6 15.9 26.5 100.0
By household type
 One-person household 24.1 31.1 40.3 11.1
 Couple with children 18.8 13.4 25.3 36.9
 Couple without children 11.6 9.9 18.1 37.5
 Lone-parent household 50.4 43.3 60.3 4.1
 Other household 24.5 19.2 32.9 10.5
By province
 Newfoundland & Labrador 18.8 22.2 29.0 1.5
 Prince Edward Island 20.0 18.4 30.3 0.4
 Nova Scotia 25.3 19.5 34.9 2.7
 New Brunswick 22.3 22.5 33.6 2.1
 Quebec 18.3 18.0 27.8 23.4
 Ontario 19.4 15.4 26.7 38.8
 Manitoba 17.9 17.2 27.6 3.4
 Saskatchewan 14.0 15.0 22.8 3.0
 Alberta 16.3 9.9 20.4 1 1.5
 British Columbia 17.8 16.1 26.1 13.3
By area of residence
 Urban 18.2 15.6 25.8 81.5
 Rural 20.2 16.9 29.3 18.5
By dwelling tenure
 Owned 13.3 10.2 19.5 77.3
 Rented 36.7 35.2 50.5 22.7
By person’s age,y
 < 18 23.4 18.8 31.4 19.8
 18–64 18.3 12.6 23.8 65.4
 ≥ 65 13.5 26.5 31.7 14.7
By employment status
 Employed 15.7 9.2 20.3 69.8
 Unemployed 42.2 33.9 51.3 4.2
 Not in LF—able to work 20.6 29.7 37.4 24.5
 Not in LF—unable to work 55.0 50.1 69.4 1.5
By highest level of education
 Some high school 31.0 36.5 47.9 1 1.2
 High school graduate 22.9 17.2 31.5 17.4
 Some post-secondary 20.9 18.8 30.4 5.4
 Post-secondary < bachelor’s 18.3 14.2 25.7 36.0
 Bachelor’s degree 11.7 9.3 16.6 19.9
 > Bachelor’s degree 10.4 7.8 14.2 10.1
By immigrant status
 Not immigrant 17.2 14.2 24.3 75.0
 Immigrant 22.7 20.9 33.1 25.0
By Aboriginal status
 Not aboriginal 18.2 15.5 26.0 97.3
 Aboriginal 33.0 28.9 44.3 2.7
By major income source
 Employment income 16.2 9.8 21.2 79.4
 Investment & retirement income 6.2 11.3 14.9 8.3
 Government transfers 44.6 58.7 68.5 10.3
 Other income 32.5 42.2 52.1 1.9

Notes: Employment, education, and immigrant and Aboriginal status reflect that of the household reference person.The categorization by employment status includes observations for which the household reference person is aged 15 y or older and thus also includes reference persons aged 65 y and older, of which 83.8 percent are categorized as “Not in LF—able to work,” 13.4 percent as “employed,” 0.6 percent as “unemployed,” and 2.2 percent as “not in LF—unable to work.” LIM-BT = low income measure before taxes (but after government transfers); LF = labour force.

Source: Canadian Survey of Economic Well-being; authors’ calculations.

Further insights can be gained by assessing the degree to which the likelihood of material deprivation and low income varies by the characteristics of a household (Table 2). One key insight is that characteristics associated with an elevated likelihood of material deprivation are often the same as those associated with an elevated likelihood of low income.4 For instance, of persons living in lone-parent households, 50.4 percent are materially deprived and 43.3 percent have low income. Such poverty rates are well above the national average and well above that of other household types. Low-income and material deprivation indicators thus corroborate that lone-parent households in Canada are materially vulnerable. Other characteristics with a higher likelihood of poverty for both indicators are living in rented housing; receiving government transfers as their main source of income; and living in a household in which the reference person is unemployed or unable to work, had only some high school education, or is Aboriginal or an immigrant. The reverse also often holds in the sense that characteristics associated with a lower likelihood of material deprivation are also often the same as those associated with a lower likelihood of low income. Such characteristics are living as a couple without children, living in owned housing, having investment and retirement income as the main source of income, and living in a household whose reference person has a higher level of education or is employed.

Moreover, statistics concerning the overlap between low income and material deprivation additionally offer the crucial insight that poverty, or at least material precariousness, among groups with an elevated likelihood of poverty may be even more widespread than low-income indicators suggest (Table 3). Of persons living in lone-parent households, for instance, 60.3 percent experience at least one condition, whereas 33.4 percent experience both at the same time. The percentages of other population groups with very high rates of experiencing at least one condition are those living alone (40.3 percent); receiving government transfers or other income as their main source of income (68.5 percent and 52.1 percent, respectively); living in rented housing (50.5 percent); and living in a household in which the reference person is unemployed (51.3 percent), unable to work (69.4 percent), had only some high school education (47.9 percent), or is Aboriginal (44.3 percent). Such high incidence rates not only amplify the scope of the problem but could also change trade-offs in policy and program design options, such as eligibility criteria and tapering of benefits (see the “Discussion and Conclusion” section).

Sometimes, it appears that low-income and material deprivation indicators disagree about the likelihood of poverty for certain population groups (Table 2). For instance, persons aged 65 years and older have a low risk of poverty, according to the material deprivation indicator (13.5 percent vs. the national average of 18.6 percent), whereas they have an elevated risk of poverty according to the LIM-BT indicator (26.5 percent vs. the national average of 15.9 percent). In total, the LIM-BT and material deprivation indicators disagree about risk levels for three provinces (Newfoundland and Labrador, Saskatchewan, and Alberta), two age groups (adults aged 18–64 y and adults aged 65 y and older), and four other socio-demographic characteristics (couples with children or a household reference person who is not in the labour force and able to work, a high school graduate, or not an immigrant).5

Owing to the lack of data, we cannot assess whether these findings are robust to alternative low-income thresholds. Although it is possible that certain disagreements in relative risk levels persist (see Notten 2015 for evidence on Ontario), we suspect levels of disagreement are smaller between material deprivation and the (more preferable) low-income indicators such as the LIM-AT and the MBM. We would expect smaller geographic disagreements in risk level between material deprivation and the MBM because the MBM accounts (to some extent) for geographic differences in purchasing power, whereas the LIM indicators do not. We also expect smaller differences in the relative poverty risk characteristics between the material deprivation indicator and LIM-AT because the LIM-BT takes only government transfers into account, whereas the LIM-AT additionally includes the redistributive effects of income taxes. In the case of the population aged 65 years and older, for instance, the 2013 low-income rates for the LIM-AT are somewhat below the national average (11.6 percent vs. 13.4 percent), and they are considerably below the national average for the MBM (4.2 percent vs. 12.1 percent; Statistics Canada 2021).

Table

Table 3: Overlap in Material Deprivation and Low Income: Population Identified as Poor by One or Both (% of Total Population)

Table 3: Overlap in Material Deprivation and Low Income: Population Identified as Poor by One or Both (% of Total Population)

Population Deprived, Low Income, or Both Both Deprived But Not LIM-BT LIM-BT But Not Deprived
Canada 26.5 8.0 10.6 7.9
By household type
 One-person household 40.3 14.9 9.2 16.2
 Couple with children 25.3 6.9 11.9 6.5
 Couple without children 19:1 3.4 8.2 6.5
 Lone-parent household 60.3 33.4 17.0 9.9
 Other household 32.9 10.8 13.7 8.4
By province
 Newfoundland & Labrador 29.0 12.0 6.8 10.1
 Prince Edward Island 30.3 8.IE 11.9 10.2
 Nova Scotia 34.9 9.9 15.4 9.6
 New Brunswick 33.6 11.2 11.0 11.3
 Quebec 27.8 8.5 9.8 9.4
 Ontario 26.7 8.0 11.4 7.3
 Manitoba 27.6 7.4 10.5 9.7
 Saskatchewan 22.8 6.2 7.8 8.8
 Alberta 20.4 5.8 10.5 4.1
 British Columbia 26.1 7.8 10.0 8.3
By area of residence
 Urban 25.8 8.0 10.2 7.6
 Rural 29.3 7.8 12.4 9.2
By dwelling tenure
 Owned 19.5 4.0 9.3 6.2
 Rented 50.5 21.5 15.2 13.7
By person’s age,y
 < 18 31.4 10.8 12.5 8.0
 18–64 23.8 7.0 11.3 5.6
 ≥ 65 31.7 8.4 5.1 18.1
By employment status
 Employed 20.3 4.6 11.1 4.6
 Unemployed 51.3 24.7 17.4 9.1
 Not in LF—able to work 37.4 13.0 7.6 16.8
 Not in LF—unable to work 69.4 35.7 19.3E 14.4
By highest level of education
 Some high school 47.9 19.6 11.4 16.9
 High school graduate 31.5 8.5 14.4 8.7
 Some post-secondary 30.4 9.3 11.6 9.5
 Post-secondary < bachelor’s 25.7 6.9 11.4 7.3
 Bachelor’s degree 16.6 4.5 7.3 4.9
 > Bachelor’s degree 14.2 4.0 6.4 3.8
By immigrant status
 Not immigrant 24.3 7.1 10.1 7.1
 Immigrant 33.1 10.4 12.3 10.4
By Aboriginal status
 Not Aboriginal 26.0 7.7 10.5 7.8
 Aboriginal 44.3 17.6 15.4 11.3
By major income source
 Employment income 21.2 4.8 11.4 5.0
 Investment & retirement income 14.9 2.6E 3.6E 8.7
 Government transfers 68.5 34.8 9.8 23.9
 Other income 52.1 22.6 9.9E 19.6

Notes: Employment, education, and immigrant and Aboriginal status reflect that of the household reference person. Percentages may not add up due to rounding. Cells marked with an E indicate estimates of marginal quality (coefficient of variation between 16.6% and 33.3%). Table A.13 in the Supplemental Appendix also provides statistics reflecting the overlap as a percentage of the poor population. LIM-BT = low income measure before taxes (but after government transfers); LF = labour force.

Source: Canadian Survey on Economic Well-being; authors’ calculations.

We developed a measure of material deprivation for Canada and provided the first representative estimates of material deprivation, using the one-time CSEW. We compared, for the same individuals, poverty status using the material deprivation and LIM-BT indicators. The analysis yielded two important findings for measuring poverty in Canada:

  1. The relationship between having low income and experiencing poverty-level living conditions is not as tight as is often presumed, even though our findings confirm that both conditions are associated with each other.

  2. Poverty or near poverty may be more widespread than the prevalence of low income suggests, and this particularly affects population groups that have characteristics that are associated with an elevated risk of poverty.

We showed that these findings are robust to moderate changes in the measurement methodology, such as a higher deprivation threshold and the exclusion of a deprivation item from the scale. Owing to data constraints, our analysis is limited in two respects. First, when constructing the deprivation scale, we could only test the items that are available in the CSEW. Second, the only low-income indicator in the data (LIM-BT) relies on self-reported income data and only partially accounts for redistribution, and it does not account for geographic differences in the costs of living. Other research on material deprivation in Canada and elsewhere strongly suggests that neither of these limitations will affect the main findings of this research (Guio 2009; Heisz and Langevin 2009; Notten 2015; Perry 2016; Saunders et al. 2008). Having fewer or additional items in a scale affects estimates regarding the prevalence and severity of material deprivation, but not its presence. Similarly, being able to compare a person’s material deprivation status with a better measure of that person’s low-income status would still yield sizable groups for which the measures would not overlap.

These findings have implications for public policy in Canada, particularly in relation to measuring the evolution of poverty, identifying the risk levels of specific target groups, and the role that public policies (could) play in reducing poverty. Our first finding indicates that there is considerable uncertainty around the accuracy of monetary poverty indicators to identify households’ specific circumstances. With such indicators playing a prominent role in governments’ poverty reduction strategies, this translates into an argument for governments to observe multiple indicators and consider material deprivation in their reporting on poverty reduction.

Our second finding implies that policy-makers may underestimate the poverty levels, or at least the material precariousness, of population groups known to have an elevated risk of poverty. Policy-makers often specifically monitor the evolution of poverty for such groups by means of disaggregated poverty statistics (e.g., single-parent households, Indigenous Peoples), and they examine effects of various policy designs on such groups. Knowledge that poverty or near poverty may be more widespread than having low income may thus strengthen arguments favoring less stringent eligibility criteria to qualify for government transfers, subsidies, or services. Likewise, such knowledge may also reinforce arguments against a strong tapering of income-tested benefits of programs (such as the Canada Workers Benefit for workers receiving low income from employment). Given that material deprivation indicators measure a latent concept by means of a limited number of observable items, this information is generally not suited to contexts of program delivery (e.g., the screening of applicants for program participation or programs aiming to reduce specific item deprivations, such as pest infestations or deprivation of food items).

Both findings yield a third policy implication, namely that the estimations on which governments rely to assess the effects and costs of public policies in reducing poverty may have unknown biases because they only partially identify the intended target group. For instance, when measuring the poverty reduction associated with a program such as the Canada Child Benefit, low-income indicators do not count the effects of the transfer among materially deprived households whose incomes are above the poverty threshold as poverty reduction. At the same time, they count all effects for households with a low income as poverty reduction. There is also a bias on the cost side, because public spending on materially deprived households with incomes above the low-income threshold does not count as spending toward poverty reduction, whereas spending on low-income households that do not experience material deprivation is counted as such. These biases may also affect comparisons of policy options, for instance comparisons of the effects of child benefits and subsidized daycare or pharma care. Low-income indicators tend to account for government transfers while underestimating or ignoring the poverty reduction effect of subsidies. Canada’s low-income indicators count all received income support until a household’s income reaches the poverty threshold (Figure 1). When it comes to subsidies, the LIM does not account for daycare expenses at all, whereas the MBM incorporates an estimate of daycare expenses through its definition of disposable income (this estimate could underestimate actual expenses because it is drawn from the household’s tax file).

Currently, we can only speculate about the scope and magnitude of such biases, but, given the relatively large gaps in overlap between low-income and materially deprived households, they could be substantive. Ultimately, an overreliance on monetary poverty indicators could influence the type and generosity of policy choices made in Canada’s jurisdictions. The dynamics of the policy-making process may lead to different policy outcomes when knowledge that a significant group of Canadians is experiencing material poverty without having low income becomes more commonplace.

We thus conclude that this gap in understanding the material dimension of poverty and its policy solutions in Canada could be significantly reduced by measuring poverty using outcome-based indicators and by collecting data that preferably, and, if not, at least regularly, offer both types of information for the same persons.

For the measurement of material deprivation, future efforts in Canada would preferably start with participatory research that re-evaluates the ongoing suitability of items (such as those used here) and identifies possible new items (along the lines described in Matern et al. 2009a, 2009b). A subsequent step would involve the collection and testing of material deprivation data, as discussed for the European Union in Guio et al. (2016), which also includes the type of analysis performed for this research. The possibility of developing various deprivation scales (e.g., for the population as whole and for specific population groups) could also be studied because international experiences suggest their relevance elsewhere (e.g., children in Europe [Guio et al. 2018], pensioners in the United Kingdom [Kotecha, Arthur, and Coutinho 2013]). Although such steps are essential in an evidence-based decision-making process and can offer considerable guidance to decision-makers, there always remains a gray area of consequential methodological choices for which the evidence is inconclusive (e.g., which items to include [or not], the number of items in the scale, and what deprivation threshold or thresholds to select). Because such final choices ultimately involve value judgments, they are thus best left to political decision-makers (Notten and Kaplan 2020).

Acknowledgements

We thank Anne-Catherine Guio, Dave Gordon, Elden Fahmy, Marco Pomati, Hector Najera Catalan, and Shailen Nandy for their methodological advice. We also thank four anonymous reviewers, Andrew Heisz, Charles Plante, and the participants in the 2017 and 2018 sessions of the annual conference of the Canadian Economics Association in Antigonish (30 May 2017) and Montreal (1 June 2018) for their comments.

Notes

1 Newfoundland and Labrador uses the poverty rate based on the low income cut-offs as its headline indicator (Government of Newfoundland and Labrador 2014).

2 For a summary of indicators used to monitor poverty reduction strategies in four provinces, see AuCoin, Hills, and Notten (2015, 53–54).

3 Calculating at the level of individuals is a preferred practice in poverty measurement. Because the income and material deprivation information is measured at the household or family level, this practise implies that everyone in the same household or family has the same poverty status. See Section 4.2 in Notten et al. (2017) for more explanation and references regarding the methodological choices made to create the indicator (e.g., measure, weights).

4 We mention only characteristics that are statistically significant and sizable when controlling for other characteristics in two multivariate regressions using the material deprivation and low-income indicators as dependent variables (Notten, Charest, and Heisz 2017). See also the Supplemental Appendix.

5 We assess this by comparing the national poverty rate with that of a group with a particular characteristic (point estimates and confidence interval), using the material deprivation rates as calculated with both two- and three-item thresholds.

Fafard St-Germain, A.A., and V. Tarasuk. 2018. “Prioritization of the Essentials in the Spending Patterns of Canadian Households Experiencing Food Insecurity.” Public Health Nutrition 21(11):206578. https://doi.org/10.1017/S1368980018000472. Google Scholar
Fusco, A., A.-C. Guio, and E. Marlier. 2011. “Income Poverty and Material Deprivation in European Countries.” 2010 Edition. Luxembourg: Publications of the European Union. https://doi.org/10.2785/1144. Google Scholar
Government of Newfoundland and Labrador. 2014. “Newfoundland and Labrador: Poverty Reduction Strategy—Progress Report.” At https://www.cssd.gov.nl.ca/poverty/pdf/prs_progress_report.pdf. Google Scholar
Government of Ontario. 2008. “Growing Stronger Together. Ontario’s Poverty Reduction Plan.” Toronto: Government of Ontario. Google Scholar
Guio, A.-C. 2009. “What Can Be Learned from Deprivation Indicators in Europe.” Luxembourg: Office for Official Publications of the European Communities. At https://ec.europa.eu/eurostat/web/products-statistical-working-papers/-/KS-RA-09-007. Google Scholar
Guio, A.-C., D. Gordon, E. Marlier, H. Najera, and M. Pomati. 2018. “Towards an EU Measure of Child Deprivation.” Child Indicators Research 11(3):83560. https://doi.org/10.1007/s12187-017-9491-6. Google Scholar
Guio, A.-C., E. Marlier, D. Gordon, E. Fahmy, S. Nandy, and M. Pomati. 2016. “Improving the Measurement of Material Deprivation at the European Union Level.” Journal of European Social Policy 26(3):219333. https://doi.org/10.1177/0958928716642947. Google Scholar
Hajizadeh, M., and S. Edmonds. 2019. “Universal Pharmacare in Canada: A Prescription for Equity in Healthcare.” International Journal of Health Policy and Management 9(3):9195. https://doi.org/10.15171/ijhpm.2019.93. Google Scholar
Heisz, A., and M. Langevin. 2009. “Material Deprivation in Household Panel Surveys: International Evidence and Lessons for Canada.” In Statistiques sociales, pauvreté et exclusion sociale: hommage à Paul Bernard, ed. G. Fréchet, D. Gauvreau, and J. Poirier, 26977. Montreal: Les Presses de l’Université de Montréal. Google Scholar
Kalil, A., and R.M. Ryan. 2010. “Mothers’ Economic Conditions and Sources of Support in Fragile Families.” Future of Children 20(2):3961. https://doi.org/10.1353/foc.2010.0009. Google Scholar
Kotecha, M., S. Arthur, and S. Coutinho. 2013. “Understanding the Relationship between Pensioner Poverty and Material Deprivation.” Research Report 827. Sheffield, UK: Department for Work and Pensions. At https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/197675/rrep827.pdf. Google Scholar
Marlier, E., and A.B. Atkinson. 2010. “Indicators of Poverty and Social Exclusion in a Global Context.” Journal of Policy Analysis and Management 29(2):285304. https://doi.org/10.1002/pam.20492. Google Scholar
Matern, R., M. Mendelson, and M. Oliphant; Daily Bread Food Bank and Caledon Institute of Social Policy. 2009a. Developing a Deprivation Index: The Research Process. Ottawa: Caledon Institute of Social Policy. Google Scholar
Matern, R., M. Mendelson, M. Oliphant; Daily Bread Food Bank and Caledon Institute of Social Policy. 2009b. Testing the Validity of the Ontario Deprivation Index. Ottawa: Caledon Institute of Social Policy. Google Scholar
McKay, S. 2004. “Poverty or Preference: What Do ‘Consensual Deprivation Indicators’ Really Mean?Fiscal Studies 25(2):20123. https://doi.org/10.1111/j.1475-5890.2004.tb00102.x. Google Scholar
National Research Council. 1995. Measuring Poverty: A New Approach. Washington, DC: National Academies Press. https://doi.org/10.17226/4759 Google Scholar
Nolan, B., and C.T. Whelan. 2010. “Using Non-Monetary Deprivation Indicators to Analyze Poverty and Social Exclusion: Lessons from Europe?Journal of Policy Analysis and Management 29(2):30525. https://doi.org/10.1002/pam.20493. Google Scholar
Notten, G. 2015. “Child Poverty in Ontario: The Value Added of Material Deprivation Indicators for Comparative Policy Analysis in North America.” Journal of Comparative Policy Analysis: Research and Practice 17(5):53351. https://doi.org/10.1080/13876988.2015.1044244. Google Scholar
Notten, G., J. Charest, and A. Heisz. 2017. “Material Deprivation in Canada.” Working Paper 1715E, Department of Economics, University of Ottawa, Ottawa. At https://ruor.uottawa.ca/handle/10393/40339. Google Scholar
Notten, G., and C. De Neubourg. 2011. “Monitoring Absolute and Relative Poverty: ‘Not Enough’ Is Not the Same as ‘Much Less’.” Review of Income and Wealth 57(2):24769. https://doi.org/10.1111/j.1475-4991.2011.00443.x. Google Scholar
Notten, G., and J. Kaplan. 2020. “A Cross-Validation Method for Narrowing the Range of Poverty Thresholds.” Manuscript submitted for publication, last modified 15 July 2020, Microsoft Word file. Google Scholar
Nunally, J.C., and I.H. Bernstein. 1978. Psychometric Theory. New York: McGraw-Hill. Google Scholar
Paulus, A., H. Sutherland, and P. Tsakloglou. 2010. “The Distributional Impact of In-Kind Public Benefits in European Countries.” Journal of Policy Analysis and Management 29(2):24366. https://doi.org/10.1002/pam.20490. Google Scholar
Perry, B. 2016. “The Material Wellbeing of New Zealand Households: Trends and Relativities Using Non-Income Measures, with International Comparisons.” Wellington, New Zealand: Ministry of Social Development. At https://www.msd.govt.nz/documents/about-msd-and-our-work/publications-resources/monitoring/household-income-report/2016/2016-non-incomes-report.docx. Google Scholar
Pressman, S., and R. Scott. 2009. “Consumer Debt and the Measurement of Poverty and Inequality in the US*.” Review of Social Economy 67(2):12748. https://doi.org/10.1080/00346760802578890. Google Scholar
Rothwell, D., and J. Robson. 2018. “The Prevalence and Composition of Asset Poverty in Canada: 1999, 2005, and 2012.” International Journal of Social Welfare 27(1):1727. https://doi.org/10.1111/ijsw.12275. Google Scholar
Saunders, P., and J.E. Brown. 2020. “Child Poverty, Deprivation and Well-Being: Evidence for Australia.” Child Indicators Research 13(1):118. https://doi.org/10.1007/s12187-019-09643-5. Google Scholar
Saunders, P., Y. Naidoo, and M. Griffiths. 2008. “Towards New Indicators of Disadvantage: Deprivation and Social Exclusion in Australia.” Australian Journal of Social Issues 43(2):17594. https://doi.org/10.1002/j.1839-4655.2008.tb00097.x. Google Scholar
She, P., and G.A. Livermore. 2007. “Material Hardship, Poverty, and Disability among Working-Age Adults.” Social Science Quarterly 88(4):97089. Google Scholar
Statistics Canada. 2013. “Microdata User Guide, Canadian Survey of Economic Well-Being, 2013.” Document available upon request. Google Scholar
Statistics Canada. 2019. “Reasons for Self-Reported Unmet Needs for Health Care Services, Household Population Aged 15 and Over, Canada.” https://doi.org/10.25318/1310069201-eng. Google Scholar
Statistics Canada. 2021. “Low Income Statistics by Age, Sex and Economic Family Type.” Table 11-10-0135-01. At https://www150.statcan.gc.ca/t1/tbl1/en/tv.action?pid=1110013501. Google Scholar
Tarasuk, V., N. Li, N. Dachner, and A. Mitchell. 2019. “Household Food Insecurity in Ontario During a Period of Poverty Reduction, 2005–2014.” Canadian Public Policy/Analyse de politiques 45(1):93104. https://doi.org/10.3138/cpp.2018-054. LinkGoogle Scholar
Townsend, P. 1979. Poverty in the United Kingdom: A Survey of Household Resources and Standards of Living. Berkeley: University of California Press. Google Scholar
Van den Berg, A., C. Plante, H. Raïq, C. Proulx, and S. Faustmann. 2017. Combating Poverty: Quebec’s Pursuit of a Distinctive Welfare State. Vol. 53. Toronto: University of Toronto Press. Google Scholar