Bailey83221 (bailey83221) wrote,
Bailey83221
bailey83221

Transfers matter most: how changes in transfer systems of Canada and the United States explain the divergence in household poverty levels from 1974-1994.

The most significant and critical finding of this paper is that the most important system for explaining the divergence of relative household poverty rates between the United States and Canada from 1974 to 1994 is between the Canadian and U.S. Social Retirement systems.

International Journal of Comparative Sociology February 1, 2004


No. 1-2, Vol. 45; Pg. 87

Zuberi, Daniyal

From 1974 to 1994, Canada and the United States experienced quite substantial divergences in relative household poverty rates and inequality levels from similar starting points. Although several scholars have attempted to explain Canadian and U.S. differences in poverty and inequality levels at one point in time, none have satisfactorily explained the causes of these divergent trends. Utilizing high quality, comparable data from the Luxembourg Income Survey, my analysis of the household poverty rates demonstrates that differences in the policies and reforms of the two country's transfer systems (Definition: Payments that are made without any good or service being received in return. Much public spending goes on transfers, such as pensions and welfare benefits.) explains the divergence relative household poverty rates. The data also seriously cast doubt on other potential market, cultural, or tax system explanations. Further, by selectively removing the income from each specific category and then specific type of [Public welfare programs] and recalculating the household poverty rate, my "sensitivity-type" analysis clearly demonstrates the predominant explanatory power for differences in the structure and reforms of social insurance…income and, more specifically, social retirement benefits.

In Canada, the expansion of the Guaranteed Income Supplement (GIS) for low-income elderly families over this period provides the only plausible explanation for the dramatic reduction in the poverty rate of the elderly households relative to the United States and, perhaps somewhat surprisingly, explains most of the divergence in household poverty rates between the two countries from 1974 to 1994. As a large part of the divergence in inequality rates is also driven by reduction in household poverty rates in Canada relative to the United States, the expansion of the GIS benefits also provides a major explanation for the divergence in levels of household inequality over this period.

Introduction

Three decades ago, Canada and the United States shared almost identical relative poverty and inequality levels. Yet, despite experiencing similar macro-level social and economic transformations from 1974 to 1994, (1) the two countries have experienced diametrically opposite trends in relative household poverty. While levels of poverty increased in the United States during this period, Canada has experienced declining household poverty. Several institutional economists in the United States have utilized the comparative case of Canada to emphasize the important role of one kind of institution for explaining differences in poverty or inequality rates at one point in time. (2) These economists have presented compelling evidence that institutional differences, and not broader cultural or economic differences, explain the poverty and inequality differences between Canada and the United States in the late 1980s. These institutional differences include unionization policies and social welfare packages. Yet, despite the importance of these institutional differences for explaining differences in poverty or inequality levels at one point in time, my analysis of Luxembourg Income Survey (LIS) data on Canada and the United States over this period clearly demonstrates that it is the different ways each country has reformed their transfer programs over this period, and not other institutional differences or reforms, that comprehensively explain the divergent trends in relative household poverty rates from 1974 to 1994. My analysis utilizes harmonized LIS data to identify the relative explanatory strength of different facets of the transfer systems for explaining the divergence in poverty from 1974-1994. Surprisingly, the breakdown analysis reveals that the divergent trends can largely be explained by differences in the structure and reform of each country's Social Retirement benefits, a factor not mentioned as an explanatory factor in the previous literature. Differences in other "Social Insurance" transfers and "Means-Tested" benefits together also help explain the divergence in poverty trends, but with less power than expected. The increased effectiveness of the Canadian transfers for reducing its relative household poverty rate compared to the American system over this period has consequences for explaining divergence in inequality and possibly health outcomes and other measures of well-being between these two countries.

Literature Review: Explaining United States and Canadian Differences at One Point in Time

Past comparative cross-national Canadian-U.S. research has largely focused on explaining differences in poverty and inequality at a specific point in time. This research compellingly argues that institutional differences in unionization policies and social safety nets between Canada and the United States explain differences in union coverage rates, inequality, and poverty.

Unionization Policy Matters

Previous research has proposed that differences in poverty and inequality rates between Canada and the United States can be explained, in part, by differences in labor policies related to unionizing. This research suggests that Canada and U.S. labor policy differences, especially concerning workers' rights to organize and management's ability to block union organizing, played a critical role in observed differences in rates of union coverage between the two countries in the late 1980s. Economist W. Craig Riddell (1993) utilizes data from multiple comparable data sources to examine possible explanations for lower levels of union coverage in the United States. These potential explanations include differences in desire to unionize, changing economy and labor force, management opposition, and the legal regime (Riddell 1993: 15). Riddell's analysis of comparable social attitude data casts serious doubt on Lipset's (1990) hypothesis that the unionization gap between Canada and the United States could be explained by underlying social value differences between Canadians and Americans. Based on comparative demand side analysis, he also finds that only a small part of the unionization gap can be explained by the higher percentage of Canadian workers in the public sector. Rather, Ridden concludes that his analysis of the evidence modestly supports the hypothesis that the unionization gap can be explained by the differences in government policies and enforcement with regards to union organizing and collective bargaining as well as somewhat lower levels of management opposition in Canada (Riddell 1993: 143). In Canada, workers who desire collective representation or unionization have a much easier time organizing and joining a union than workers in the United States because labor laws do not grant nearly as much power to management to challenge unionizing efforts. What are the implications of these differences? Economist Thomas Lemieux (1993: 97) found that the union wage effects on both the dispersion as well as mean of wages are similar in the United States and Canada utilizing the 1986 Labor Market Activity Survey (LMAS) for Canada and the 1986 outgoing rotation group file of the Current Population Survey (CPS). He concludes, "differences in the pattern and extent of unionism in Canada and the United States explain 40% of the difference in wage inequality between men in the two countries" (Lemieux 1993: 97).

Social Safety Nets Matter

Economists Rebecca Blank and Mafia Hanratty utilize comparable survey data from the Current Population Survey on 50,000 American families and the Survey of Consumer Finances on 30,000 Canadian families' income and work behavior and find that these differences help explain the differences in poverty rates between the United States and Canada in 1986 (Blank and Hanratty 1993: 199). Blank and Hanratty standardize the income and poverty measures to compare differences between the United States and Canada in poverty rates based on an absolute poverty line. As shown in Table 1, they find that, despite the modestly lower income of Canadians relative to Americans, cross-sectional evidence demonstrates that Canada had lower-levels of family poverty (Blank and Hanratty 1993: 191).

Not only did Canada have lower levels of family poverty, Blank and Hanratty also find that Canada has a smaller poverty gap, for instance, less income would be required to bring poor families up to the poverty line (Card and Freeman 1993: 10).

To more accurately assess the impact of social policy differences, Blank and Hanratty compare poverty rates before and after governmental transfers in both the United States and Canada, with the controversial assumption that income support policies do not impact other sources of income (based on rising labor force participation rates in both countries). They find that the Canadian system was much more effective at reducing poverty through government transfer programs. These programs reduced family poverty rates by 5.7 percentage points in Canada compared to only 1.9 percentage points in the United States. The Canadian system reduced poverty among single parent families by 14.3 percentage points as compared to 5.3 percentage points in the United States (Card and Freeman 1993: 10). Subsequently, they ran simulations which estimated the impact U.S. anti-poverty programs would have on Canadians, if U.S. rules were to be applied in the United States, and the Canadian rules on Americans, if Canadian rules were to be applied in the United States. Applying Canadian program rules and benefits to U.S. data create striking results, especially for single parent families:

The poverty rates of single parent families with children would decline from 43% to 16% if the United States adopted the "mean" Canadian program, assuming Canada's participation rates were duplicated in the United States. Assuming 100% participation rates, poverty among this group would nearly disappear. The results are not very sensitive to the range of assumed labor market elasticities. (Blank and Hanratty 1993: 192)


These findings relied on some debatable assumptions.

First, Blank and Hanratty assume that the more generous and flexible benefits would not effect individuals' labor market behavior. Yet Blank and Hanratty argue that this is a reasonable assumption in light of the fact that earnings and work effort among single parents are similar in the United States and Canada, despite Canada's more generous social assistance benefits (1993: 212-214).

Second, many more Canadians take advantage of the social assistance benefits for which they are eligible relative to Americans. Participation rates, or benefit take-up, are much harder to estimate in a simulation as these are both culturally determined and contingent upon program rules. The higher the estimated participation rates, the more powerful the impact Canadian policies would have on U.S. families. Indeed, with these assumptions in place, Canadian policies would almost completely eliminate poverty among families with children (Blank and Hanratty 1993: 216-217). These results prompted them to conclude that "the principal reason for higher total income among the disadvantaged groups in Canada is higher transfer income. Poor, near poor, and single parent family earnings are no higher in Canada, but transfers are substantially higher" (Blank and Hanratty 1993: 202). Other scholars working with Luxembourg Income Survey data yielded similar results, such as Rainwater (1993).

Not Different Economic, Historical, or Other Trends

Although previous research has not focused so much on explaining the divergent trends between the United States and Canada, some scholars have utilized trend data and analysis to demonstrate that institutions in general, and not cultural or social differences between Canada and the United States, are responsible for differences in poverty. (3) Over the past 20 years, Canada and the United States have experienced similar forces of globalization along with technological, demographic, and social change, including rising rates of divorce and solo parenthood. Economists Card and Freeman describe the similarities well:

Canada and the United States are as close economically and socially as any pair of countries in the world. The two countries share similar cultural traditions and enjoy comparable living standards. Both countries have highly educated and skilled workforces, with similar industrial and occupational structures. Many of the same unions and firms operate on both sides of the border...Throughout the past century Canada and the U.S. have shared similar economic experiences. Both were major recipients of the European immigration and capital flows; more recently, both have experienced large in-flows of non-European immigrants. Both escaped the destruction of World Wars I and 1I. Both had "baby booms" in the1950s that produced comparable demographic structures. And both developed broadly similar income security and labor market regulations over the course of the twentieth century. But against this backdrop of similarity are "small differences" in policies, institutions, and economic outcomes. (Card and Freeman 1993: 1)


The two countries have also experienced remarkably similar labor force sector shifts away from manufacturing and agriculture to the service sector, as shown by Reitz (1998).

Although differences between two country's particular business cycles of booms and recessions often provides a challenge to the validity of conclusions based on cross-national comparative research, the following figure based on evidence by Anthony Atkinson, Lee Rainwater, and Timothy M. Smeeding (1995) from OECD data suggests that Canada and the United States largely share similar patterns and rates of economic growth from 1979 to 1990.

[FIGURE 1 OMITTED]

The similarities in economic structure and trends between Canada and the United States over the past several decades increases the validity of a comparative cross-national research design for isolating the impact of institutional differences at one point in time. In combination with similar starting points, the similarity in social and economic transformations and cycles also increases the validity of drawing conclusions from testing the impact of institutions on trends in rates of poverty.

Transfers Matter Most

While past research has established that various institutional and policy differences between the United States and Canada can explain the difference after tax and transfer relative poverty levels at one point in time, my analysis of LIS data clearly demonstrates that only differences in the structure and reform of the transfer systems of each country can explain the divergent trends of these poverty levels from 1974 to 1994.

This section begins with a discussion of why relative and not absolute poverty levels are utilized for this analysis.

Second, it provides some background on the primary data source, the Luxembourg Income Survey (LIS), as well as the specific sources and limitations of LIS data on Canada and the United States for this kind of analysis.

Third, it explicitly describes the divergent trends in after tax and transfer poverty rates after 1974.

Fourth, it demonstrates that these findings clearly suggest that the explanation for this divergent trend is most likely to be found in differences in the tax and transfer systems.

Fifth, it demonstrates that these divergent trends cannot be explained by differences in the tax system, and must be attributed to differences in the structure and reform of each country's transfer system.

Finally, it reviews and casts serious doubt on other possible explanations of the divergence in poverty rates.

Relative vs. Absolute Poverty Rates

My analysis utilizes data on relative household poverty rates precisely because they allow the poverty line to change over time and provide a more consistent standard measure across advanced industrial nations. Hence, relative rather than absolute poverty rates are the more valid metric for the analysis of poverty trends if the primary concern is the comparative cross-national analysis of how policies affect households. The use of relative vs. absolute poverty measures in this paper is further supported by its concern with households that earn enough income to purchase the basket of goods necessary to prevent social exclusion (Osberg 2000). As Osberg explains, the real difference between relative and absolute poverty lines is the transparency of what determines the poverty line: relative poverty explicitly links it to income distribution, while absolute poverty lines are based on the value of a market basket of goods that is implicitly connected to the income distribution anyway (2000: 4). Yet there clearly exists a dynamic and changing basket of goods necessary for inclusion and full participation in society that changes over time and between countries. So, for this analysis the relative poverty rate represents the most conceptually useful and empirically valid measure.

Luxembourg Income Survey Data Sources and Limitations

The Luxembourg Income Survey provides the highest quality, accessible, and comparable data for cross-national comparisons (for more information see http://www.lisproject.org). Income data from national sources are collected and stored in Luxembourg, the Canada's Survey of Consumer Finances, and the U.S. March Current Population Survey in this case, and the income and demography variables "harmonized" to have the same meaning across these datasets (Smeeding and Ross 1999: 10-11). National data go through the "lissification" or harmonization process once every five years; and especially in the early years the comparable U.S. and Canadian data may be a year apart. While this process provides accessible and comparable data, it certainly does not perfectly "standardize" the data or eliminate all the problems of noise in the original datasets (Smeeding and Ross 1999: 11). Some of these noise problems may have a particularly serious impact on the accuracy of analysis of poverty. Hence, it is important to outline these limitations and potential problems and consider the impact of these drawbacks on the comparative analysis. Although both datasets have been cleaned up as much as possible, both the Canadian and U.S. datasets contain missing income data, likely biased downwards, and inaccuracies in income reporting that are more dramatic systematically at the top and the bottom of the income distributions. Both of these weaknesses can challenge the validity of findings from cross-national comparisons, particularly if one variable is systematically more biased in one direction. The greatest danger in the U.S. and Canadian comparative case is that low-income families in the United States are severely under-reporting their income, from all sources, including transfers, relative to Canadian low-income families. Yet the good must not be the enemy of the perfect, and many powerful research findings have been derived about poverty based on these two datasets, despite these potential limitations. While they are an important limitation to consider, these drawbacks likely do not fundamentally invalidate the findings of this analysis. As both Canada's Survey of Consumer Finances and America's March Consumer Population Survey share these problems of missing data and underreporting at the top and bottom, the detrimental effects are partially nullified or partially canceled out.

Past Similarity and 20 Years of Divergence

The LIS data clearly reveal that the United States and Canada experienced divergent poverty trends, from similar starting points in 1974, in the period 1974 to 1994. The following figure highlights the post-tax and transfer rates of poverty from LIS data.
As presented in Figure 2, the household poverty rate in Canada, based on median income, decreased from 15.6 percent in 1975 to 10.6 percent in 1994. In the United States, on the other hand, the household poverty rate increased somewhat over the same period from 15.8 percent in 1974 to 17.9 percent in 1994. The same divergence in trends is even more pronounced based on relative household poverty measured against the mean income.

[FIGURE 2 OMITTED]

As presented in Figure 3, the relative household poverty rate after taxes and transfers in Canada, based on mean income, decreased from 19.7 percent in 1975 to 14.4 percent in 1994. During nearly the same time period, the U.S. post tax and transfer relative household poverty rate increased from 20.4 percent in 1974 to 37.5 percent in 1994.

[FIGURE 3 OMITTED]

Differences in the Tax and Transfer Systems

My analysis of the LIS data clearly demonstrates the overwhelming explanatory power of differences in transfer systems and reforms, and not other kinds of institutions or explanations, for elucidating the divergence in poverty trends in Canada and the United States from 1974 to 1994.

If market, cultural, or economic differences could explain the divergence in poverty trends, then these rates must also diverge based on pre-tax and transfer income data. Yet, as shown in Figure 4, pre-tax and transfer poverty rates based on median income in Canada increased at about the same rate and level as in the United States from 1974 to 1994.

[FIGURE 4 OMITTED]

In Canada, the pre-tax and transfer relative poverty rate increased from 28.1 percent in 1975 to 35.5 percent in 1994; similarly, in the United States it increased from 28.6 percent in 1974 to 33.9 percent in 1994. Before taxes and transfers, Canada has a marginally higher rate of household poverty against median income in 1994 as compared to the United States. The following figure shows that pre-tax and transfer household poverty trends follow the similar identical pattern, as measured against mean income.

As presented in Figure 5 above, pre-tax and transfer household poverty rate rose in Canada from 30.2 percent in 1975 to 37.9 percent in 1994; similarly, the household poverty rate increased from 31 percent in 1974 to 37.5 percent in 1994. So, the evidence from the Luxembourg Income Survey clearly shows that explaining the divergence in trends must arise from differences in the tax and transfer systems.

[FIGURE 5 OMITTED]

From similar starting points, the Canadian tax and transfer system clearly has become more effective at reducing market generated relative household poverty. The following figure presents the percentage point reduction or "effectiveness" of the tax and transfer system in each country over time for reducing relative household poverty rate.

Based on a relative poverty rate against the median household income, Figure 6 above demonstrates that the Canadian system substantially increased its "effectiveness" at reducing pre-tax and transfer household poverty rates from 12.5 percentage points in 1975 to 24.7 percentage points in 1994. While also increasing its poverty reduction "effectiveness" over the same period, the United States tax and transfer system went from reducing pre-tax and transfer household poverty rates from 12.8 percentage points in 1974 to only 16 percentage points in 1994. The same pattern is affirmed in the following Figure 7, which contrasts the trends in poverty reduction effectiveness of the Canadian and U.S. tax and transfer systems over the same period, but on poverty rates generated against the mean income in each country.

[FIGURES 6-7 OMITTED]

The evidence presented in Figure 7 above shows that the Canadian tax and transfer system increased its relative household poverty effectiveness from 10.5 percentage points in 1975 to 23.5 percentage points in 1994, while the U.S. system only marginally increased from 10.6 percentage points in 1974 to 12.8 percentage points in 1994.

Difference in Transfer Systems and Reforms: The Key Explanatory Variable

Until this point, the effects of taxes and transfers have been considered together. Removing tax income from the U.S. and Canadian income data and re-calculating the relative household poverty rates after transfers over this period highlights the lack of explanatory power of tax system differences; and demonstrates that differences in the structure and reforms of the Canadian and U.S. transfer systems explain the divergent relative household poverty trends between the United States and Canada from 1974 to 1994. If the divergence in household poverty trends could be explained by differences in the structure and reform of the tax systems between the United States and Canada, then we would expect diverging trends in the "effectiveness" of the two tax systems for reducing household poverty rates over this period. However, Figure 8 presents evidence that the Canadian and U.S. tax system basically have followed a similar pattern of increasing, yet still minimal, effectiveness of reducing relative household poverty rates over this period. Although the case of the poverty rate based on the mean provides extra supporting evidence, the remaining analyses in this paper will focus on poverty rates in each country calculated against the median income. Median income calculations are less sensitive than mean income calculations to erroneous outlier income values in the data, which are somewhat problematic in both the U.S. and Canadian case.

[FIGURE 8 OMITTED]

Figure 8 shows that Canada's tax system reduced relative household poverty rate by .8 percentage points in 1975 and 2.6 percentage points in 1994. Similarly, the U.S. system reduced the relative household poverty rate .4 points in 1975 and 2.5 percentage points in 1994. So there is no way that the differences between the structure or reforms of the Canadian and U.S. tax systems can be responsible for the divergence in post-tax and transfer poverty rates from 1974 to 1994. Hence, the divergence in poverty rates must be explained by differences in the structure and reform of the respective Canadian and U.S. transfer systems.

Not Directly Unionization Policy

Despite suggestive evidence and contrary findings in the previous literature on U.S. and Canadian differences at one point in time, the above evidence strongly suggests that labor policy differences between the United States and Canada as related to unions do not directly explain the divergent trends in poverty rates between the United States and Canada from 1974 to 1994.

The correlated trends of divergent union coverage with poverty rates may initially suggest that these labor policy differences directly impact poverty rates. In both Canada and the United States, approximately 30 percent of the non-agricultural labor force were members of a union during the period 1950 to 1970. After 1970, union coverage rates in the United States began to fall steadily every year, particularly after 1975. By 1985, union coverage had declined to 20 percent of the labor force and has continued to fall (Banting et al. 1997). In sharp contrast, Canada's rate of union coverage of its workforce increased steadily after 1970 to a high of 40 percent in 1970 in 1985 before dipping slightly down to 35 percent in 1990 and then increasing again slightly through 1994 (Banting et al. 1997). After 30 years of divergence, Canada's union coverage currently is at least double that of the United States, across all economic sectors. Indeed, the divergence between Canadian and U.S. levels of union coverage is largest and growing fastest among traditionally low-union workers--including females and part-time workers--and the service sector (Ridder 1993:112-113).

The labor policy differences as related to unions would appear to be a promising theory for explaining the divergence in relative household poverty rates in Canada and the United States from 1974 to 1994. Higher rates of unionization have been associated with lower levels of wage inequality for both unionized and non-unionized workers, and hence lower relative poverty rates. However, if labor policy differences had mattered more to household income, the pre-tax and transfer poverty rates should have declined in Canada as compared to the United States over this period. Yet the evidence presented in the above section shows that the pre-transfer poverty rates grew at the same rate in both Canada and the United States from 1974 to 1994. Indeed, this evidence would cast doubt on hypotheses claiming that any other institutions, including higher education policy differences (see Freeman and Needels 1993), could explain the divergence in relative household poverty trends. Of course, this does not rule out the possibility that the differences in labor policy as related to unions caused the divergence in rates of union coverage of the labor force that indirectly caused differences in the political outcome of general transfer retrenchment in the United States and expansion in Canada during this period.

Going Beyond "Transfers"

The harmonized variables provided by the Luxembourg Income Survey data allow for going beyond the "transfers" argument to begin to examine, somewhat more specifically, which kinds of transfer differences matter for explaining the divergence in relative household poverty rates between Canada and the United States from 1974 to 1994. The LIS disposable income variable utilized in the above analysis emerge from summary income, tax, and benefit data where:

Disposable Household Income = (Market) Earning + Cash Property Income + Social Insurance Transfers + Means Tested Benefits + Public Pensions + Private Pensions + Other Cash Income - Mandated Employee Contribution - Income Tax

Each of these summary variables listed above represent more detailed variables; for example:

Social Insurance Transfers = Sick Pay + Disability Pay + Social Retirement Benefits + Child or Family Allowances + Unemployment Compensation + Maternity Pay + Military/Vet/War Benefits + Other Social Insurance

Removing each one of these variables and re-calculating the relative household poverty rate for each country allows for a "sensitivity" type analysis that isolates the strength and trends concerning the impact of each variable on the poverty rate.

The following analysis reveals that the differences in the systems and reforms of Social Insurance Transfers in Canada and the United States provide a major explanation for the divergence in relative household poverty rate trends from 1974 to 1994. Removing Social Insurance Transfer Income from the post-tax and transfer income data and re-calculating relative household poverty rates reveals that the Canadian system of social insurance transfers went from being less effective than American system at reducing relative household poverty in 1974 to being nearly twice as effective in 1994.

As presented in Figure 9 above, Canada's Social Insurance Transfers steadily increased its reduction of relative household poverty rates 6.51 percentage points in 1975 to 14.37 percentage points in 1994. At the same time, America's Social Insurance Transfers continued to only reduce relative household poverty between approximately 7 and 9 percentage points. The increasing gap in poverty reduction effectiveness over time between their Social Insurance Transfers over time provides a major, yet not complete, explanation for why Canada's relative household poverty rate did not increase nearly as fast as the U.S. rate over the same time period.

[FIGURE 9 OMITTED]

Divergent trends in the strength of means tested benefits for lifting families out of poverty also helps explain the differences in poverty trends between the United States and Canada. Although less powerful as an explanatory summary variable than Social Insurance Transfers, the impact of removing all means tested income transfers (means-tested cash benefits + near-cash benefits) from the disposable income on relative household poverty rate reveals a similar pattern.

Figure 10 shows that the effectiveness of Canada's means tested programs steadily increased from reducing the relative household poverty rate by 0.94 percentage points in 1975 to 2.75 points in 1994. In the United States, on the other hand, means tested programs continued to only reduce the poverty rate approximately 1 percentage point throughout this period. Although the magnitude of the difference may appear small, the differences in the changing effectiveness of Social Insurance Transfers and all Means Tested Transfer income almost completely explains the divergent relative household poverty trends between the United States and Canada from 1974 to 1994.

[FIGURE 10 OMITTED]

Going Beyond "Social Insurance Transfers"

As the differences in the income transfers provided by the Social Insurance Transfers summary variable explain most of the divergence in poverty rate trends between the United States and Canada from 1974 to 1994, the following analysis examines the impact of removing each individual variable and recalculating the poverty rate. The "Social Insurance Transfers" variable includes Sick Pay, Disability Pay, Social Retirement Benefits, Child or Family Allowances, Unemployment Compensation, Maternity Pay, Military/Vet/War Benefits. and Other Social Insurance. Unfortunately, as these categories of transfers alone often are not significant enough to lift low-income households above the poverty line, it is difficult to isolate the impact of each type of transfer as well as if a more sensitive measure was used, such as the Sen-Shorrocks-Thon poverty intensity measure that Osberg proposes (Osberg 2000). Despite these limitations, this analysis reveals some strong and surprising results.

Although Unemployment Compensation, Child and Family Allowance, Other Social Insurance, and Military/Vet/War Benefits differences have a small role in driving the divergence of relative poverty rates, the most important explanatory factor for explaining the divergence in poverty rates is the differences in the Social Retirement Benefits.

Canada's unemployment compensation income consistently lifted more households out of relative poverty as compared to the United States and showed a slight, but inconsistent, upward trend over this period.

[FIGURE 11-12 OMITTED]

Canada's child and family allowances transfer income pulled a small, but steadily increasing, percentage of households out of poverty over this period. At the same time, income from child and family allowances continued to not pull any households above the poverty line in the United States over this period.

The transfer income generated from "Other Social Insurance" also lifted a very small, but increasing percentage of families out of poverty in Canada over this period. In the United States, "Other Social Insurance" transfer income began to raise about half a percentage of households out of poverty after around 1991.

Not surprisingly, removing Military/Vet/War Benefits, Sick Pay transfer income, Disability Pay Benefits, and Maternity Pay transfer income individually had no effect on poverty rates in either Canada or the United States.

[FIGURE 13 OMITTED]

The most significant and critical finding of this paper is that the most important transfer system differences for explaining the divergence of relative household poverty rates between the United States and Canada from 1974 to 1994 is between the Canadian and U.S. Social Retirement systems.

Figure 14 above demonstrates that Canada's Social Retirement transfers only reduced the relative poverty rate 4.39 percentage points of households above the relative poverty line in 1974; yet increased rapidly to reduce the relative poverty rate 11.05 percentage points above the relative poverty line in 1994. In 1974, the U.S. Social Retirement benefits reduced the household poverty rate more than in Canada--reducing the rate by 7.2 percentage points. Yet, by 1994, the U.S. Social Retirement benefits were having significantly less impact on the household poverty rate than in Canada, and continued to only reduce the relative household poverty rate by 6.51 percentage points. The divergence in the effectiveness of Social Retirement Transfer Income over time largely explains the divergence in relative household poverty rates between Canada and the United States from 1974 to 1994.

[FIGURE 14 OMITTED]

What Policy Reform Explains the Divergence in Effectiveness of Retirement Benefits?

Canada's expansion of its Guaranteed Income Supplement (GIS) for the elderly explains the improved effectiveness of Canada's "Social Retirement Transfer Benefits" for reducing Canada's overall rate of household poverty relative to the United States from 1974 to 1994. Hence, increases in this targeted "negative income tax" style supplement to Canada's universal Old Age Security (OAS) and Canada and Quebec Pension Plans (C/QPP) benefits largely explains the divergence of Canadian and U.S. household poverty rates over this period. First introduced in 1966, the GIS provides an income-tested benefit to low-income elderly in order to supplement their retirement benefits, with assets excluded and each dollar of income reducing benefits only 50 cents (Myles and Pierson 1997). Through the 1970s and 1980s, Canada,s GIS was expanded in stages and the value of the benefits increased relative to mean wages; at the same time the value of OAS and C/QPP benefits stagnated during the 1980s (Myles and Pierson 1997). A simplified tax form also increased the take-up rates of the GIS to over 90 percent of eligible low-income elderly (Myles and Pierson 1997). In 1989, the Canadian government reduced OAS benefits, but only to those elderly with incomes higher than $ 51,765 (CAD) so that all benefits disappeared for those elderly with approximately $ 89,000 (CAD) per year (Myles and Pierson 1997). The expansion of the GIS benefit to elderly families had quite a dramatic effect on the poverty rates of elderly households over this period. According to the analysis of LIS data by Smeeding, Torrey, and Rainwater (1993), the elderly poverty rate in Canada was higher than in the United States in the mid-1970s; yet by the mid-1980s, only 7 percent of Canadian elderly were living in poverty compared to 22 percent of American elderly (Myles and Pierson 1997).

Implications for Explaining Divergent Trends in Inequality

The above analysis demonstrates that differences in the structure and reform of primarily U.S. and Canadian "social insurance transfers." in particular social retirement transfers and the expansion of the Canadian Guaranteed Income Supplement (GIS) largely explain the divergent trends in relative household poverty from 1974 to 1994, which also helps to explain divergent trends in household inequality levels experienced by the two countries over this period. A closer examination of changes in the rates of inequality suggests that the divergence in levels of inequality is primarily driven by the improved position of Canadian families at the tenth percentile of the income distribution as compared to the median household.

Three decades ago, Canada and the United States shared similar levels of household inequality, as measured by LIS data. With households made equivalent through the use of a square root of family size factor, Canadian and American households at the 90th percentile of the income distribution had approximately five times the amount of household income compared to the income of households at the tenth percentile. In the United States, this "P90/P10" ratio (one of the standard measures of inequality) increased steadily from 5.16 in 1974 to 6.42 in 1994. The U.S. Gini measure of household inequality also increased from 32.3 in 1974 to 37.5 in 1994. In sharp contrast, Canada's "P90/P10" ratio of inequality dropped steadily from 5.01 in 1974 to 3.93 in 1994. So by 1994, those households in the 90th percentile of the income distribution in Canada were receiving less than four times the income of those at the tenth percentile. The Canadian Gini measure of household inequality also dropped over this period from 32.3 in 1971 to 28.6 in 1994.

A closer examination of changes in the rates of inequality suggests that the divergence in levels of inequality is primarily driven by the improved position of Canadian families at the tenth percentile of the income distribution as compared to the median household and the somewhat worse position of U.S. households at the tenth percentile compared to the median household than with differences in growing inequality at the top of the household income distribution. After taxes and transfers, the ratio of incomes of households at the 90th percentile of the income distribution compared to the median income only declined ever so slightly in Canada from 1.90 in 1971 to 1.85 in 1994, while in the United States it increased from 1.90 in 1974 to 2.19 in 1994, a change of +.29. Yet the ratio of median household income in Canada to the income received by those at the tenth percentile dramatically declined over the same period from 2.63 in 1971 to 2.13 in 1994, or a change of -.50. In the United States, the ratio of median household income to the income of households at the tenth percentile increased steadily from 2.71 in 1974 to 2.93 in 1994, a change of +.22. So, the majority of the increasing divergence in inequality levels between Canada and the United States over this period is explained by the growing income gap between those households at the bottom or tenth percentile in the United States (compared to a shrinking gap in Canada) and not increasing inequality between those at the top or 90th percentile compared to the mean (which marginally increased in both countries). While the wealthy got wealthier over this period in both countries, the poor in Canada marginally improved their relative position while they became worse off in the United States.

As the divergence in inequality rates over this period is largely explained by the improved position of those families at the tenth percentile as compared to the median in Canada as compared to the United States over this period, it appears that the differences in the structure and reform of the transfer systems, in particular "social insurance" transfers, provide a critical explanation for the divergent trends in household inequality experienced by the United States and Canada from 1974 to 1994. The increasingly generous and effective transfer system in Canada, especially the Social Retirement Benefit system or the expansion of the Guaranteed Income Supplement (GIS), played a critical role in improving the relative position of households at the low-end of the income distribution, as compared to the United States. By lifting an increasing percentage of households out of poverty, the Canadian transfer system has also significantly contributed to the overall reduction of household income inequality in Canada during a period in which similar economic and social changes have resulted in an increase in household inequality levels in the United States. Hence, the differences in the reforms of the tax and transfer system arguably play a more important role than differences in education wage differentials and unionization for explaining the diverging inequality trends between Canada and the United States.

Conclusion

From 1974 to 1994, Canada and the United States experienced quite substantial divergences in relative household poverty rates and inequality levels from similar starting points. Although several scholars have attempted to explain Canadian and U.S. differences in poverty and inequality levels at one point in time, none have satisfactorily explained what caused these divergent trends. Utilizing high quality comparable Luxembourg Income Survey data, my analysis of the household poverty data demonstrates that differences in the policy and reforms of the two countries' transfer systems must explain the divergence relative household poverty rates. The data also seriously cast doubt on other potential market, cultural, or tax system explanations. Further, by selectively removing the income from each specific category and then the specific type of transfer income and re-calculating the household poverty rate, my analysis clearly demonstrates the predominant explanatory power for differences in the structure and reforms of social insurance transfers income and, more specifically, social retirement benefits. In Canada, the expansion of the Guaranteed Income Supplement (GIS) for low-income elderly families over this period provides the only plausible explanation for the dramatic reduction in the poverty rate of the elderly households relative to the United States witnessed in the LIS data and, perhaps somewhat surprisingly, explains most of the divergence in household poverty rates between the two countries from 1974 to 1994. As a large part of the divergence in inequality rates is also driven by reduction in household poverty rates in Canada relative to the United States, the expansion of the GIS benefits also provides a major explanation for the divergence in levels of household inequality over this period.

Table 1.

Cross-Sectional Evidence on Poverty Rates 1986



Canada United States

Single-Parent Families 32.3% 45.3%

Two-Parent Families 5.2% 6.8%



Source: Blank and Hanratty (1993: 191).

NOTES

(1.) Changes in the Canadian and U.S. social welfare systems as well as the lack of "lissified" Canadian income data for 1997 limited this analysis from 1974 to 1994.

(2.) Many sociologists tend to underplay or ignore the critical role played by government policy and labor market institutions of advanced industrial countries in shaping social stratification and social mobility. All too often, social phenomenon such as rising inequality, urban poverty, and economic hardship are examined without an explicit examination of the kind of government policies that have contributed to their formation or crystallization. For example, many stratification sociologists focus on the impact of educational attainment or family background on social inequality and mobility (e.g., Blau and Duncan 1967). Other sociologists focus on the role of organizational ascription on inequality (see Baron and Newman 1990; Kanter 1977). Yet these theories miss the embeddedness of social stratification in a country's policy regime that can have a dramatic impact on both outcomes and opportunity.

Based on degree of worker de-commodification, Esping-Andersen classified both the Canada and the United States as neoliberal welfare states (1990). Yet classifying both of these countries in the same category obscures significant differences between U.S. and Canadian government policies and institutions. The findings of past empirical studies and this paper suggest that these policy differences, although marginal as compared to the differences between the United States and Sweden's policy regimes, play a significant role in the divergence of Canada and the United States along particular socioeconomic indicators over the past 30 years.

(3.) Despite sharing many similarities, Canada and the United States also differ socially and culturally in many ways, with the roots of these differences extending back to the early in the colonial era (see Lipset 1990). For example, while differences in gun control policies may be partially responsible for differences in violent crime rates between the two countries, the effects of policy and culture are extremely difficult to isolate because violent crime rates have always been lower in Canada than in the United States. Yet differences in the two countries' poverty rates, inequality, unionization, social welfare policies and health systems have emerged since the early 1970s. Hence Canada-U.S. comparative analysis allows for a greater understanding of the link between policy differences and social stratification.

REFERENCES

Atkinson, Anthony B., Lee Rainwater, and Timothy M. Smeeling. 1995. Income Distribution in OECD Countries: Evidence from the Luxembourg Income Study. Paris: Organization for Economic Cooperation and Development.

Banting, Keith, George Hohberg, and Richard Simeon, eds. 1997. Degrees of Freedom: Canada and the United States in a Changing World. Montreal: McGill Queen's University Press.

Baron, James N. and Andrew E. Newman. 1990. "For What It's Worth: Organizations, Occupations, and the Value of Work Done by Women and Non-Whites." American Sociological Review 55(April): 155-175.

Blackburn, McKinley L. and David Bloom. 1993. "The Distribution of Family Income: Measuring and Explaining Changes in the 1980s for Canada and the United States." In Small Differences that Matter: Labor Markets and Income Maintenance in Canada and the United States, edited by David Card and Richard B. Freeman. Chicago: University of Chicago Press.

Blank, Rebecca M. and Maria J. Hanratty. 1993. "Responding to Need: A Comparison of Social Safety Nets in Canada and the United States." In Small Differences that Matter: Labor Markets and Income Maintenance in Canada and the United States, edited by David Card and Richard B. Freeman. Chicago: University of Chicago Press.

Blau, Peter and O.D. Duncan. 1967. The American Occupational Structure. New York: Howe and Wiley.

Card, David and Richard B. Freeman eds. 1993. Small Differences that Matter: Labor Markets and Income Maintenance in Canada and the United States. Chicago: University of Chicago Press.

Card, David and W. Craig Riddell. 1993. "A Comparative Analysis of Unemployment in Canada and the U.S." In Small Differences that Matter: Labor Markets and Income Maintenance in Canada and the United States, edited by David Card and Richard B. Freeman. Chicago: University of Chicago Press.

Esping-Andersen, Gosta. 1990. The Three Worlds of Welfare Capitalism. Princeton: Princeton University Press.

Freeman, Richard B. and Karen Needels. 1993. "'Skill Differentials in Canada in an Era of Rising Labor Market Inequality." In Small Differences

that Matter: Labor Markets and Income Maintenance in Canada and the United States, edited by David Card and Richard B. Freeman. Chicago: University of Chicago Press.

Kanter, Rosabeth Moss. 1977. Men and Women of the Corporation. New York: Basic Books.

Lemieux, Thomas. 1993. "Unions and Wage Inequality in Canada and the United States." In Small Differences that Matter: Labor Markets and Income Maintenance in Canada and the United States, edited by David Card and Richard B. Freeman. Chicago: University of Chicago Press.

Lipset, Seymour Martin. 1990. The Continental Divide: The Values and Institutions of the United States and Canada. New York: Routledge.

Lipset, Seymour Martin. 1996. American Exceptionalism: A Double-Edged Sword. New York: W.W. Norton.

Luxembourg Income Survey. 2001. Available at (http://www.lisproject.org/).

Myles, John and Paul Pierson. 1997. "Friedman's Revenge: The Reform of Liberal Welfare States in Canada and the United States." Politics and Society 25(4): 443-472.

Myles, John and Jill Quadango. 1994. "The Polities of Income Security for the Elderly in Canada and the United States: Explaining the Difference." In Economic Security for the Elderly: North American Perspectives, edited by Theodore Marmor and Timothy Smeeding. Washington, DC: The Urban Institute.

Osberg, Lars. 2000. Poverty in Canada and the USA: Measurement, Trends, and Implications. Presidential Address to the Canadian Economies Association, June 3, Vancouver.

Rainwater, Lee. 1993. "The Social Wage in the Income Packaging of Working Parents." Luxembourg Income Survey Working Paper Series: No. 89. April. Available at (http://www.lisproject.org/publications/).

Reitz, Jeffrey G. 1998. Warmth of Welcome: The Social Causes of Economic Success for Immigrants in Different Nations and Cities. Boulder: Westview Press.

Riddell, W. Craig. 1993. "Unionization in Canada and the United States: A Tale of Two Countries." In Small Differences that Matter: Labor Markets and Income Maintenance in Canada and the United States, edited by David Card and Richard B. Freeman. Chicago: University of Chicago Press.

Smeeding, Timothy, Barbara Torrey, and Lee Rainwater. 1993. "Going to Extremes: An International Perspective on the Economic Status of the U.S. Aged." Working Paper #87, Luxembourg Income Survey. Available at (http://www.lisproject.org/publications/).

Smeeding, Timothy and Katherine Ross. 1999. "Social Protection for the Poor in the Developed World: Evidence from the LIS." Working Paper No. 204, Luxembourg Income Survey. Available at (http://www.lisproject.org/publications/).

Wilson, William Julius. 1987. The Truly Disadvantaged: The Inner-City, the Underclass, and Public Policy. Chicago: University of Chicago Press.

Daniyal Zuberi, Sociology and Social Policy PhD Program, Kennedy School of Government, Harvard University, 79 JFK St., Cambridge, MA 02138, USA (danzuberi@earthlink.net).
Subscribe
  • Post a new comment

    Error

    Comments allowed for friends only

    Anonymous comments are disabled in this journal

    default userpic

    Your IP address will be recorded 

  • 0 comments