The Canadian Safety Net for the Elderly

Summary

Recently various analysts have called attention to the apparent success of the Canadian social assistance system in reducing poverty among the elderly and have suggested that there may be lessons to be drawn from the Canadian experience that are relevant to the evolution of the U.S. Supplemental Security Income (SSI) program. This article profiles the Canadian system, compares the system to the U.S. SSI program, reviews the consequences for elderly poverty rates, assesses system costs, and then comments on pertinence of the Canadian experience to SSI policy. The Canadian minimum income guarantee for the elderly is substantially more generous than what is provided by the United States, but it is misleading to claim that the Canadian system costs only "slightly more" than the U.S. program. Such a judgment overlooks a key and costly part of the Canadian system, the Old Age Security demogrant. We estimate the total costs to Canada of providing income support for elderly persons receiving a Guaranteed Income Supplement (GIS) in 2004 to be approximately C$13.3 billion (roughly US$11.1 billion), slightly more than 1 percent of gross domestic product (GDP) and almost fourteen times the U.S. allocation for SSI and food stamps for elderly SSI recipients. The significance of this commitment is underscored when it is recognized that in 2004 Canadian GDP per capita was just 80 percent of the U.S. level. The Canadian example suggests U.S. policymakers consider better integration of SSI with basic Social Security benefits, experimenting with alternatives to restricting SSI eligibility to individuals with very few assets, and reducing barriers to program access.

Introduction

The future of the U.S. Social Security program continues to be the focus of a public debate compelled by the system's projected insolvency. Although the outcome of the political struggle is difficult to forecast, it is likely that changes made to the Social Security system to address financing will affect the nation's "safety net," or minimum income guarantee, for elderly and disabled people—the Supplemental Security Income (SSI) program. The 2001 report of the President's Commission to Strengthen Social Security acknowledged this connection and recommended that "Social Security reform plans should also encompass reforms in SSI policy, to improve retirement incomes for those persons who might not otherwise attain poverty-level income in old age" (President's Commission 2001, 136). Since the Commission issued its report, several Social Security reform proposals have included provisions aimed at providing a minimum guaranteed benefit for workers with low career earnings. Little has been done, however, to address the connection between a reformed Social Security system and the SSI program in providing income security for the most vulnerable of the elderly.

Recently, various analysts have called attention to the apparent success of the Canadian social assistance system in reducing poverty among the elderly and suggested that there may be lessons to be drawn from the Canadian experience that are relevant to SSI strategy. Timothy Smeeding and Susanna Sandstrom (2004) report, "Canada has managed to achieve much greater poverty reduction among seniors [than has the United States] while spending much less on social retirement programs than other rich countries (and slightly more than the United States)" (p. 11). They recommend considering the integrated Canadian social insurance/social assistance system "as a model for future United States OASI [Old-Age and Survivors Insurance]-SSI interactions" (p. 12).

This recommendation is based on cross-national comparisons using data collected in connection with the Luxembourg Income Study (LIS).¹ An example of these data is reproduced in Table 1. Poverty for this analysis is defined as having disposable income adjusted for family size that is less than 50 percent of the national median. When applying this standard, poverty among the elderly in the United States is the worst among the seven countries listed, and Canada typically ranks first or second.² In particular, Smeeding and his colleagues note the apparently superior performance of the Canadian system in reducing poverty rates among elderly women living alone, a growing share of all elderly persons (Smeeding with Williamson 2001, 24; Smeeding 2006a).

This article profiles the Canadian system, compares the system to the U.S. SSI program, reviews the consequences for elderly poverty rates, assesses system costs, and then comments on pertinence of the Canadian experience to SSI policy. Our core argument is that Smeeding and his colleagues are right in judging the Canadian minimum income guarantee to be substantially more generous than what is provided by the United States, but that it is misleading to claim that the Canadian system costs only "slightly more" than the U.S. program. Such a judgment overlooks a key and costly part of the Canadian system, the Old Age Security demogrant.

Canadian Social Security and Social Assistance

Canada uses "social security" as a generic term referring to a wide range of programs dealing with health, education, unemployment, and family and child assistance, as well as old age, disability, and survivors' benefits. Programs specifically providing income support for the aged, disabled, and survivors are collectively called Canada's Public Pensions System.

The Three Components

For the elderly (persons aged 65 or older), Canada's Public Pensions System has three major components. Together they provide benefits intended to "ensure a basic income to all eligible Canadians" (Human Resources and Social Development Canada (HRSDC) 2005b).³

The first component is quite similar to what is termed Social Security (the Old-Age, Survivors, and Disability Insurance program) in the United States:

• Canada Pension Plan (CPP)/Quebec Pension Plan (QPP) is a compulsory, earnings-related social insurance program providing income for retired and disabled workers and their survivors. Its benefit formula also contains significant flat-rate components for the disabled and survivors under the age of 65.

The second component has no close parallel in the U.S. system:

• Old Age Security (OAS) is a nearly universal pension financed from general revenues and paid to almost all Canadians aged 65 or older. The principal exceptions are those who do not meet residency requirements or who have very high taxable incomes.

The third component is an income-tested SSI counterpart:

• Guaranteed Income Supplement (GIS) is a non- taxable benefit paid to low- and moderate-income seniors—about one-third of the elderly population. It is a form of guaranteed annual income (benefits are reduced according to other income received). Like the OAS, the GIS is financed from general revenues.

The tabulation shows outlays on all three types of benefits for Canadian fiscal year 2005–2006 (the Canadian fiscal year runs from April 1 through March 31). Unless otherwise noted, the figures that follow are in Canadian dollars (C$); reducing amounts expressed in Canadian dollars by 15 percent to 20 percent gives approximate contemporary U.S. dollar equivalents.⁴ Outlays on public retirement benefits in Canada amount to about 4.5 percent of gross domestic product (GDP). Although direct general comparison with the United States is difficult, the combination of OASI and federal SSI payments to the elderly in the country amounts to about 3.6 percent of GDP. What is particularly striking about the Canadian public pensions system is that almost half of benefits—OAS and GIS—is financed from general revenues. In the United States less than 2 percent of costs of publicly provided pensions for the elderly is funded from general revenues.⁵

The Canada and Quebec Pension Plans. Established in 1966, the Canada Pension Plan and Quebec Pension Plan have paid benefits since late 1967. The CPP serves all parts of Canada outside of Quebec. Its parallel plan, the QPP, was in important ways a model for the CPP and has very similar contribution and benefit provisions. Covered earnings records in both systems are merged at retirement for dual contributors to calculate a single benefit. As a result, for most purposes the two programs can be considered a single system.

The Canada Pension Plan is a form of contributory, earnings-related social insurance.⁶ It pays three kinds of benefits:

1. retirement pensions;
2. disability benefits (which include benefits for disabled contributors and benefits for their dependent children); and
3. survivor benefits (which include death benefits, survivors' pensions, and children's benefits).

Only retirement pensions will be discussed here. With very few exceptions, every person in Canada aged 18 or older and aged 69 or younger with earnings must pay into the CPP. Employees and their employers each pay 4.95 percent on annual earnings between the minimum and a set maximum level ("pensionable earnings"). The minimum level is frozen at C$3,500; the maximum for 2006 was C$42,100. The self-employed pay both portions, levied on net business income. All earnings below the maximum are taxed; persons whose annual earnings fall below C$3,500 receive a refund of contributions when filing an income tax return (HRSDC 2006a).

The pensionable earnings maximum level is adjusted each January, based on increases in the average wage. The contributory period begins when the worker reaches age 18 or in January 1966 (the start of the CPP), and it continues until the worker receives a retirement pension, reaches age 70, or dies. Some parts of the contributory period are dropped out of the benefit calculation, including periods with no or lower earnings while raising children younger than age 7; low-earning months after age 65; any month when eligible for a CPP disability pension; and the lowest 15 percent of earning years in the contributory period.

Eighty-six percent of the 6 million-plus CPP/QPP recipients collect retirement benefits, about 7 percent collect survivor benefits, and the remaining recipients are made up of disabled persons or children of the disabled. The normal retirement age is 65; retirement benefits can be taken at age 60 (with actuarial reduction) or delayed to age 70 (with actuarial increase). Maximum retirement benefits equal 25 percent of the contributor's average lifetime covered earnings (indexed by an economy-wide earnings measure). However, since virtually all Canadians aged 65 or older also receive OAS (and sometimes GIS) pensions (discussed below), the maximum retirement income from combined public sources replaces approximately 57 percent of average net earnings, about 10 percent more than the replacement rate for the U.S. Social Security system (OECD 2005, Table 4.2). Benefits are indexed to the Canadian consumer price index (CPI) and adjusted annually.⁷

Old Age Security. The OAS is the oldest component of the retirement income system. After the federal government was empowered by a constitutional amendment to operate a system of old age benefits, it set up the OAS program in 1952. OAS is paid entirely out of general revenues. All legal residents who have lived in Canada for at least 10 years after reaching age 18 are eligible for OAS at age 65. The payment amount is equivalent to about 14 percent of average wages and salaries. It is reduced based on the duration of Canadian residency. Full benefits go to residents of 40 years or more as well as to certain other persons with extended but less than continuous residence in the country.

Receipt of OAS is not conditional on retirement or income. However, the benefit is taxable, and recipients with total net income in excess of a certain amount have their benefit "clawed back" at a 15 percent rate, repaying part or all of their OAS benefit through personal income tax. The threshold where clawback begins is approximately C$62,000, and benefit loss is complete at about C$101,000. Only about 5 percent of elderly Canadians lose any of their OAS to taxes (Office of the Chief Actuary 2005, 66). Guaranteed Income Supplement. Low-income OAS pensioners are eligible for the income-tested Guaranteed Income Supplement, which is legally a part of OAS. As such, its benefits are also paid entirely out of federal general revenues. For persons entitled to a full OAS payment, the maximum monthly GIS benefit is about C$594 for a single individual and C$779 for a couple (married or common law; amounts are for March 2006). For persons not entitled to a full OAS payment, the maximum GIS payment is increased to bring the combination of OAS and GIS payment to an amount equivalent to the same total benefit that would accrue to a person with full OAS entitlement. Combined OAS and GIS benefits for a low-income couple are about the same as the highest CPP/QPP benefit paid; for a single person they are somewhat higher. (Recall that virtually all CPP/QPP beneficiaries also receive OAS, so CPP/QPP recipients have total income greater than those reliant on OAS/GIS alone.) GIS is not subject to the Canadian income tax, but benefits are reduced by 50 percent of the amount of any non-OAS income.⁸ No account is taken of assets, except for the income they generate. About one-third of Canadians older than age 65 receive both OAS and at least some GIS (Office of the Chief Actuary 2005, 93).

Both OAS and GIS benefits are price indexed, and payments are adjusted quarterly for inflation.

The Safety Net

For Canadians older than age 64, the combination of the OAS and the GIS is the income of last resort. The combined benefit paid in March 2004 is summarized for singles and couples in Table 2. (For the remainder of the article we use 2004 data to facilitate later comparison of benefits to various measures of poverty.) In addition to this level of assured income, Canadian elderly receive health care largely without charge through a system called, as in the United States, "Medicare." Government funding, mostly paid to private practitioners, covers 95 percent of hospital and doctor costs and about 70 percent of total aggregate health spending (World Health Organization (WHO) 2006, 58).

The process of applying for and receiving OAS/GIS benefits in Canada is straightforward. Human Resources and Social Development Canada (HRSDC) administers the program; payments are handled by the Canada Revenue Agency (CRA). Residents normally receive an application for CPP/QPP and OAS benefits in the mail 6 months before their 65th birthday. The form includes a preliminary notification of possible GIS eligibility and asks the recipient if he or she is interested in applying for GIS. If the response is yes, the OAS office obtains detailed information on income for the client and mails this for verification. Once the information is returned and accepted, the combined benefit is provided monthly, in most cases by direct deposit. In years subsequent to the initial award, HRSDC obtains income information from the CRA for clients who have filed a tax return. It is not necessary to visit any office at any stage during this process.

Once eligible persons are in the system, benefits are paid monthly on the basis of past income, with over- or underpayments recouped on notification or at subsequent income tax filings. GIS benefits are offset by half of income from sources other than OAS (including the CPP/QPP). On average, a single GIS recipient in 2004 received a combined monthly sum of C$896, about 46 percent of it from GIS. The average married couple's combined payment is C$1,478, about 34 percent from GIS (HRSDC 2006b). The integration of GIS with the other systems and delivery through the income tax system leads to, from an American perspective, a very high take-up rate among eligibles.⁹

Comparison with Supplemental Security Income

The U.S. safety net for the elderly—Supplemental Security Income—is a "nationwide Federal assistance program . . . that guarantees a minimum level of income for needy aged, blind, or disabled individuals. It acts as a safety net for individuals who have little or no Social Security or other income" (Social Security Administration (SSA) 2006, i). SSI recipients living alone are categorically eligible for food stamps and Medicaid as well. As a result, when considering the safety net, it is appropriate to combine SSI and the Food Stamp Program (FSP) benefit. Because elderly recipients of means-tested benefits in both Canada and the United States receive significant medical benefits and health outcomes are similar (WHO 2006), we ignore this benefit in comparing the two safety nets.

The Federal Benefit Rate and State Supplements

In 2004 the federal SSI benefit rate (FBR) for individuals living independently was $564 per month; the rate for couples was $846. All but six states provided some type of supplement to this benefit for recipients in specific circumstances such as living in a nursing home (SSA 2005b, 5–6). Twenty-four states provided additional cash payments to single individuals and/or couples living independently. These cash supplements ranged from very little ($1.70 per month for individuals in Oregon) to quite substantial ($553 per month for couples in California). Fifty-five percent of all SSI recipients lived in states with a state cash supplement, making it appropriate to include such payments when comparing benefits between countries. Because the FSP benefit is affected by total SSI received, it is necessary to incorporate this interaction.

Chart 1 shows the net supplement paid in January 2004 to individuals and couples living alone in each of the states that supplemented the cash benefit of all elderly SSI recipients living independently. The states are ordered by the additional individual benefit adjusted for food stamp effects. The first category ("All others") covers the 27 states (including the District of Columbia) with no general cash benefit supplement; these states accounted for 44.7 percent of all elderly SSI recipients in 2004. Continuing up the ladder, the median elderly recipient ranked on the basis of total minimum benefit (FBR plus state supplement plus minimum food stamps) lived in Pennsylvania. Slightly more than one-quarter of all elderly SSI recipients lived in California; only Alaska (which is granted exceptional treatment in the FSP) had a higher net supplement. Because of an exception originally granted in 1974 and continued by subsequent legislation, California SSI recipients are ineligible for food stamps because the FSP benefit is "cashed out" in the state supplement (Arnold and Marinacci 2003). This nuance is incorporated in Chart 1 and subsequent calculations.

The Safety Nets Compared

To compare the SSI/FSP benefit with the Canadian OAS/GIS benefit, we convert the Canadian figures to U.S. dollars using the 2004 OECD Purchasing Power Parity (PPP) of C$1.23(US$1). We use the PPP recognizing that there are many problems with international comparison of real purchasing power at low income levels. Indeed, one of the advantages of relative poverty measures is that comparing the proportion of persons with incomes less than half the median—for example, across countries—allows one to avoid identifying just what half the median buys.

Table 3 presents the combined SSI and FSP benefit for the United States and the corresponding OAS/GIS benefit for 2004 converted to U.S. dollars. For consistency with our definition of the safety net—the minimum income guarantee—we cite for each state grouping the minimum food stamp entitlement for singles and individuals living alone and dependent wholly on the FBR plus, where available, the state supplement. However, because of housing and medical cost deductions, most elderly SSI recipients receive more than the minimum food stamp amount. To gain a more complete picture of the combined level of support, we have also included an estimate of the mean FSP benefit received by elderly SSI recipients dependent on SSI.¹⁰ The U.S. benefit for states without a state supplement is shown along with comparable calculations for Pennsylvania (the state with the median recipient; see Chart 1) and California. For singles, the Canadian benefit is from 32 percent (with average FSP benefit) to 43 percent (at minimum FSP benefit) greater than the SSI/FSP benefit in states without a supplement; for couples, from 37 percent to 53 percent. Only in California and Alaska do benefits approach Canadian levels.

Compared with payments to couples, the Canadian system treats single adults less generously than do all but the most generous of U.S. states—in general, the benefit for individuals in the United States is a larger fraction of the couples benefit than is true in the Canadian system. As shown in Table 1, the comparatively lower Canadian poverty rates for singles discussed by Timothy Smeeding and Susanna Sandstrom (2004) arise because the level of benefits for both singles and couples in Canada is much greater than in the United States, whether considered in absolute amounts gauged by purchasing power (as in Table 3) or relative to the national median income (as in Table 1).

Both the Canadian and U.S. systems provide for situations in which an eligible individual lives in the same household with a spouse (or, in Canada, a common-law partner) who is ineligible. Here, too, the Canadian system is more generous. In Canada a person aged 60–64 who is partner of a senior OAS/GIS recipient is eligible for an allowance that brings the couple's total income to a level equivalent to what they would receive if both were OAS/GIS recipients. The allowance for a pensioner's survivor younger than age 65 is about 10 percent less than the maximum OAS/GIS benefit for single persons; once reaching age 65 the survivor becomes eligible for full OAS/GIS benefits in her or his own right (Office of the Chief Actuary 2005, 43). In the United States there is no benefit for the survivors of SSI recipients and no complementary benefit for ineligible spouses. Rather, income of ineligible spouses in excess of exclusions that are allowed regular beneficiaries plus a set-aside for a "living allowance" is considered available to the SSI recipient and reduces the SSI benefit (SSA 2006, 13).

Benefits and Poverty

Perception of the consequences of the Canadian system for poverty depends on the standard employed for poverty assessment. Canadians use two methods for assessing poverty—a "low-income cutoff" (LICO) and a "low-income measure" (LIM) (Statistics Canada 2006). Poverty in the United States continues to be officially assessed using a measure originally developed in the early 1960s (Ruggles 1990). A more comprehensive measure proposed in the mid-1990s by a panel convened at the behest of Congress by the National Research Council—an arm of the National Academy of Science—has not been officially adopted (Citro and Michael 1995).

The Low-Income Cutoff and the Low-Income Measure

The low-income cutoff is constructed by determining the income level at which families on average spend 20 percentage points more of their total outlays on necessities—food, clothing, and shelter—than do the average Canadian families with the same number of persons living in the same region and urban/rural circumstance.¹¹ In focusing on relative outlays for necessities, construction of the LICO is consistent with the recommendations made for the United States by the National Research Council panel. The standard varies by community size; it is lowest in rural areas and highest in urban areas of 500,000 or more. The cutoffs are reported both before and after income taxes, but both measures include transfers. In 2004 the before-tax LICO for a single adult living in an urban area with a population from 100,000 to 499,999 was C$17,515; for two persons the cutoff was C$21,804. Thus the combined OAS/GIS benefit (see Table 2) fell about 31 percent short of the single-person cutoff and 10 percent below the two-person cutoff.¹²

The low-income measure is based on income alone; this standard is set at 50 percent of median adjusted individual income. Using an equivalence scale, the adjustment—made before calculation of the median—converts each Canadian's family income into the "equivalent" income for a person living alone. The equivalence scale used by Statistics Canada for this purpose assigns a weight of 1.0 to the oldest person in the family, a weight of .4 to the second oldest regardless of age and to all other adults (persons aged 16 or older), and a weight of .3 to any other family members. Thus to find the "individual equivalent" scale for a family of two adults, one would divide whatever family income measure is used by 1 + .4 = 1.4. Once half the median is determined for individuals using this "equilivised" measure as a base, the cutoff is adjusted for larger families by multiplying by the appropriate equivalence weight. The LIM in 2004 was C$16,253 before taxes for a single adult and C$22,255 for a couple. The March 2004 OAS/GIS minimum was therefore 75 percent of the LIM for single adults and 87 percent for couples.

Development of the LIM was prompted in part by interest in international comparisons. The LIM standard approximates the measure used by the Organisation for Economic Co-operation and Development (OECD) in some recent comparative poverty studies (Förster and Mira d'Ercole 2005) and in the work of Timothy Smeeding and colleagues, previously cited. The difference between the Canadian measure and that of the OECD is that the OECD equivalence scale is simpler, giving a weight of the square root of family size to each family member. Smeeding uses the square root equivalence scale but focuses on income after taxes, as does the LICO measure that Statistics Canada (2006, 9) "prefers." In practice the choice between the Canadian and square root equivalence scales is of little consequence for most outcomes: Note that both the Canadian and square root scales assume that two-adult households require 1.4 times as much income to achieve the same living standard as a single adult living alone. For purposes of this article, we refer to the square root equivalence scale as the OECD scale, recognizing that some OECD work involves more elaborate adjustments (Förster 2005).

No distinction is drawn among adults on the basis of age in calculating poverty thresholds for the LICO, the LIM, or the OECD scale.

Direct Benefits Comparison with the United States

Table 4 considers the difference between Canada and the United States from the perspective of the poverty measures. We report U.S. figures for Pennsylvania, given that state's "median" status (see Chart 1), and we assume that the SSI recipient receives the average food stamp benefit (see Table 3). The first line of data in Table 4 shows that the Canadian basic income guarantee for the elderly substantially exceeds that provided in the United States. In the second line we report for the United States the OECD poverty standard of 50 percent of median adjusted income, expressed in terms of gross income before taxes but including cash transfers.¹³ Using the Canadian equivalence scale, the couple value is 1.4 times the single adult value. The Canadian LIM is taken from official sources and converted to U.S. dollars using the same purchasing power adjustment as that applied in Table 3. Real income is lower in Canada, and half the median—the OECD relative poverty measure—is about 3 percent lower. We also include, in line three, the official U.S. poverty measure, based on the cost in current prices of a yardstick outlay established in 1963. Having established these points of reference, data lines four and five report the maximum benefit in both countries as a percentage of (1) the 50-percent-of-median standard and (2) the U.S. poverty standard, that is, the numbers appearing in lines two and three.

Clearly, by both Canadian and U.S. standards,¹⁴ the Canadian system is significantly more generous than its American counterpart. This generosity extends beyond the minimum guarantee. Every dollar in income from social insurance above $20 reduces the SSI payment by a dollar, but in Canada GIS benefits are "taxed" at only a 50 percent rate. The 50 percent disregard leads to a high proportion of Canadian GIS recipients who also have some income from the Canadian Pension Plan or its Quebec equivalent—83 percent in 2003 (unpublished data provided by the Canada Department of Human Resources and Social Development). It is the combination of a high basic benefit with substantial disregard of other pension income that lifts so many elderly Canadians above the poverty thresholds. The fixed and small SSI disregard produces substantially less mixing: Fifty-seven percent of elderly U.S. SSI recipients had some income from Social Security in December 2004 (calculated from data in SSA (2005a, Table 8)). Social assistance generosity is more than a matter of cash or disregards: Apart from the differences in benefits and treatment of other income, differences in prevalence of poverty in Canada and in the United States may also result from differences in ease of access to the respective safety net systems. It appears to be mechanically much easier to establish OAS/GIS eligibility than it is to initiate and sustain SSI/FSP payments, and GIS has no assets test.

We conclude that the reason why poverty rates are so low in Canada is that the Canadian system is very generous to those with few other resources, and this reduces poverty. Solvency for the self-funding part of the program is less of a problem because almost half of total pension costs comes from general revenues. If a country redistributes more and concentrates this redistribution on the elderly, the relative incomes of elderly people rise, and poverty falls. It is not "rocket science." It is arithmetic.

Looking South

If the redistribution of general revenues is concentrated more on the elderly, the arithmetic is expensive—Smeeding and Sandstrom's assertion that Canada spends "slightly more" than the United States is far from the mark. What would it in fact cost to move SSI for the elderly to something like the Canadian safety net? A first estimate is evident from the last column in Table 3: Increase SSI benefits by 25 percent for singles, 34 percent for couples, and some comparable amount for persons living with others, and disregard more Social Security income in benefit calculation. But this estimate ignores the consequent reduction in FSP benefits, and real emulation of Canada would require elimination of the SSI assets test as well. Both an increase in benefits and elimination of the assets test would raise SSI take-up. Thus for a first estimate, suppose the United States attempts to somehow simply "fill the gap" between current incomes of the elderly and the amount needed to achieve an income equal to the same percentage of median income as that achieved by Canada's current maximum benefit—that is, 75 percent for singles and 87 percent for couples (see Table 4). Depending on certain assumptions about income underreporting in the U.S. Census Bureau's Current Population Survey (CPS), filling the gap would require increasing SSI payments to the elderly by at least 110 percent and possibly as much as 150 percent.¹⁵ Refinement of this "back of the envelope" calculation requires making assumptions about state supplements, the consequences of expansion of eligibility for take-up and private saving, and the implications for related programs like Medicaid. The "possibly as much as" concept seems likely. Costs would multiply further if the same adjustments were extended to nonelderly SSI recipients.

An alternative perspective is to consider the cost of the U.S. and Canadian systems in relation to gross domestic product. Suppose the United States was to devote the same fiscal effort to establishing a minimum income for the elderly as does Canada. What would be the cost? The tabulation on the next page provides a rough cut at comparing the fiscal commitment involved in the Canadian system compared with the cost of SSI.¹⁶

Combined federal and state SSI costs for elderly recipients amounted to about $8.3 billion in calendar year 2004 (SSA 2006, Table IV.C1 and IV.C4; state supplements for blind or disabled elderly estimated by authors). Food stamp outlays for this group are not published by the U.S. Department of Agriculture. We estimate benefits accruing to elderly SSI recipients using data from the Food Stamp Program Quality Control Database for 2004 (Poikolainen and Ewell 2005) under the assumption that FSP benefits are shared equally among all members of a recipient household. A reasonable estimate is that SSI and FSP costs for the elderly amounted to roughly one-thirteenth of 1 percent of GDP.

At first glance, Canadian outlays for their safety net look comparable to those of the United States but, of course, smaller: C$5.7 billion. However, such a comparison ignores the fact that GIS comes on top of the near-universal and substantial OAS payment. Assuming that every GIS recipient receives average OAS,¹⁷ we estimate the total costs of providing income support for elderly persons receiving GIS in 2004 to be approximately C$13.3 billion, slightly more than 1 percent of GDP and almost fourteen times the U.S. allocation for the SSI and Food Stamp programs. The significance of this commitment is underscored when it is recognized that in 2004, Canadian GDP per capita (US$32,000) was just 80 percent of the U.S. level.¹⁸ In a sense even the C$13.3 billion figure is an understatement; in the absence of OAS far more Canadians would presumably be eligible for GIS. A full accounting of the commitment to providing a minimum income guarantee would include an estimate of the general reduction in GIS obligation that the presence of the OAS demogrant affords.

The tabulation below, which compares fiscal commitment for 2004, shows a problem with international comparisons of benefit systems—failure to consider context. At first glance the benefit cost of the Canadian GIS is similar to U.S. outlays for SSI. But virtually every poor elderly Canadian citizen gets OAS, for which there is no U.S. equivalent. It is true that OAS is not means-tested (aside from the high-end clawback). But GIS is clearly calibrated within the context of OAS (indeed, it is delivered by the same agency), and it would be misleading not to include both when describing the Canadian minimum income guarantee. To report, as Smeeding and Williamson (2001) have, "a country like Canada has a very efficient income-related lower tier benefit which produces a low poverty rate for a modest level of social expenditure" is to miss the OAS point and surely to mislead readers about the success of targeting in the Canadian system and the costs of its emulation.

Review of the Canadian system does suggest some changes in the SSI and the Food Stamp programs that would marginally facilitate access and ensure that the lowest-income elderly receive at least the minimum benefit. Ideally, the FSP benefit would be evaluated by local SSA offices in conjunction with SSI benefit review, unifying the application process and reducing FSP administrative costs for states. While at present, SSI applicants can obtain FSP applications at the SSA office and can file initial applications there, establishing and sustaining benefits requires visits to the agency administering FSP benefits, even for households made up entirely of SSI recipients and therefore categorically eligible. Considerable experimentation with joint processing of SSI and FSP applications has occurred. The results suggest that simplification of FSP benefit calculation and integration with SSI would raise take-up and reduce administrative costs for states (Food and Nutrition Service 2000; Food Research and Action Center 2004).

It is clear that establishing eligibility for the GIS in Canada is simplified by the absence of asset tests; this permits integration with the internal revenue system and the other benefits programs that are conditioned only on income. Although inclusion of the value of some assets in needs assessment is viewed as important in the United States, it might be possible to shift asset assessment from the present all-or-nothing eligibility determination to a sort of quasi-annuitization in which a fraction of the value of counted assets is treated as income in determining the SSI benefit (see Davies, Rupp, and Strand 2004). Such a procedure could have favorable consequences for savings behavior of the near elderly and the elderly themselves, especially if a portion of assets was disregarded altogether in benefits computation. The potential costs of such a change are, of course, difficult to evaluate; Rupp, Strand, and Davies (2001) estimate that completely eliminating the SSI assets test would have increased annual costs by 7.7 percent in 1991.

Both the Canadian OAS/GIS and the U.S. SSI/FSP systems guarantee minimum incomes that are below the OECD poverty standards, yet poverty rates are much higher in the United States. In part this is because the gap between the maximum system payment and the poverty standard is much greater in the United States (see Table 4). However, the effect of the Canadian system is also enhanced by the substantial disregard of Canadian/Quebec Pension Plan income in calculation of the GIS benefit. This allows the GIS benefit to fill the gap between the poverty line and the combination of pension and OAS benefits for persons with modest pension entitlement. The GIS 50 percent disregard is proportional and is therefore not affected by inflation. In contrast, for SSI only the first $20 of any monthly Social Security payments is disregarded; beyond this level every dollar of income from Social Security counts against the SSI benefit. The $20 disregard has not changed since the inception of the SSI program in 1974. Had the $20 disregard been held constant in real terms, the current value would be approximately four times greater. Regardless of the poverty standard employed, emulation of the Canadian disregard policy would reduce poverty among the elderly in the United States. Nevertheless, such changes would be costly, because every $1 increase in income disregarded in benefit calculation raises costs by at least $1—and more if such changes raise program take-up (Davies and others 2001/2002).

In this discussion we have ignored differences in health care available to the elderly in Canada and the United States, arguing that access and outcomes are roughly comparable. Although outcomes may be similar, costs are not. Various recent analyses have demonstrated that while overall spending by the United States on social welfare is comparable to that of other countries, the United States is unique in the share of such outlays devoted to health (Adema and Ladaique 2005). Thus if means could be found to increase the efficiency of health care delivery in the United States to match that of other countries without sacrificing quality, resources might be freed to target the remaining income poverty among older citizens and to address the concerns voiced by the President's Commission to Strengthen Social Security (Garfinkle, Rainwater, and Smeeding 2006).