Keith Banting and John Myles, eds.,
Inequality and the Fading of Redistributive Politics.
Vancouver: UBC Press, 2013.
480 pages.

Keith Banting and John Myles have assembled 17 articles on trends in inequality within Canada over the last three decades, 1980–2010, including substantial introductory and concluding articles written by themselves. This is a serious attempt to grapple with a serious problem. With apologies to authors whose articles I ignore, this review bears primarily on the articles by the editors and the one by David Green and James Townsend, who analyze explanations for wage polarization.

The editors start with two undeniable generalizations about income distribution in Canada over the three decades:

First, the distribution of market incomes grew more unequal. By one measure, the Gini coefficient of individuals’ market incomes, the increase in inequality from 1980 to 2010 was roughly 20 per cent. Taxes and transfers completely offset rising market inequality until the mid-1990s. Since then, post-tax post-transfer inequality has risen by roughly 10 per cent.

Second, post-1980 market income increases have been concentrated in the top quintile of income earners. Inflation-adjusted market incomes of those at or below the median essentially stagnated over the three decades.

Canada’s 
“neoliberal moment”

Why did the tax-transfer system become less effective over the second half of the three decades? The editors’ underlying explanation is ideological: neoliberal ideas symbolized by Reagan in the United States and Thatcher in Britain seeped into Canada. In the mid-1990s Canada experienced what they describe as a “neoliberal moment … Policy restructuring shrank programs that provided support to vulnerable Canadians, such as unemployment benefits and social assistance, and reduced the progressivity of the tax system.”

The editors admit qualifications and counterevidence: for example, trends in poverty rates, a subject obviously linked to discussions of income inequality. Measuring poverty raises thorny philosophical issues. Leaving them aside, the editors acknowledge that, by the two most widely used Canadian poverty measures, poverty either stayed constant or declined over the three decades.

The most frequently cited Canadian poverty statistic is the percentage falling below after-tax low income cutoff (LICO) thresholds. The thresholds adjust for family size, community size and inflation. The real value of what can be bought at the thresholds was last benchmarked in 1992. LICO is an absolute poverty measure: the real value of what can be purchased at threshold incomes remains constant over time. The LICO poverty rate rose in the early 1980s recession, fell in the subsequent boom, and peaked above 15 per cent in the early 1990s recession. Since the mid-1990s, the LICO poverty rate has fallen nearly in half. It rose slightly as a result of the post-2008 recession, but finished the decade at its lowest value since the mid-1970s.

The second poverty statistic measures the percentage of individuals with incomes below half the median income of all Canadians. The low income measure (LIM) is a relative measure: the thresholds rise and fall on the basis of what happens to the income of the typical Canadian with median income. LIM poverty rates fluctuated less than LICO rates, and finished the three decades at levels similar to those at the beginning.

Not surprisingly, reliance on social assistance is linked to the state of the economy. The share of Canadians receiving social assistance was about 5 per cent in the late 1970s, rose to 7 per cent in the early 1980s recession and remained constant until the early 1990s recession. In the wake of the early 1990s recession, reliance on social assistance peaked above 10 per cent. At that point most provinces abandoned policies of generous welfare access, which had been the tradition in Canada. Some provincial governments (e.g. Tories in Ontario) resorted to Sturm und Drang; others (e.g. NDP in British Columbia) relied on stealth so as not to alienate their core supporters. The lowering of post-1995 LICO poverty rates is most evident among female-headed single-parent families. From 1995 to 2010, their LICO poverty rate declined from over 50 per cent to about 20 per cent. They substituted higher market earnings for lost social assistance.

At 20 per cent, single-parent poverty is still over twice the national average, and the explanation for the decline is more complex than the previous paragraph implies. In the late 1990s Canada introduced the National Child Benefit System (NCBS), a modest negative income tax for low-income families with children. Most provinces lowered welfare benefits dollar for dollar with NCBS payments. Families received no benefit from the NCBS while on welfare, but the NCBS allowed families to exit welfare at lower earning levels. Having exited, they faced lower clawback rates on incremental earnings than when on welfare.

If we limit the discussion to social assistance, the amorphous group defined as neoliberals surely have a strong case. Before 1995, Canadian governments spent a lot on social assistance with little to show for it. Since 1995, they have designed programs to limit general use of welfare, to increase benefits for the disabled, and to provide better incentives for the poor to enter the labour market. And the LICO poverty rate has fallen to historic lows. How can this be described as a “fading of redistributive politics”? What has faded has been faith in the efficacy of generous welfare benefits combined with ease of access as means to address poverty.

Banting and Myles acknowledge some of this, but it is en passant. It does not prompt them – or all other authors in this volume – to challenge their commitment to the goal, dating back to the 1960s, that antipoverty policy should be anchored on generous unconditional cash transfers, and in the limit on a guaranteed annual income. Some of the other authors are much less careful than the editors, and simply refer to post-1995 social assistance policy as an obvious example of antistate libertarian ideology at the expense of the poor.

Where the editors are on solid ground is their critique of the extent of tax reduction undertaken since 2000. It has been excessive. But here too the story is complicated. As a reason for Canada undergoing a “neoliberal moment” in the 1990s, the fiscal irresponsibility of both orders of government over the previous two decades is as important as ideology, if not more important. Looking at the sum of the accounts of Ottawa and the ten provincial governments, not once over the previous two decades had these accounts been in balance. Instead, the two levels of government engaged in mutual recriminations. Throughout the 1980s, the provinces blamed Ottawa for arbitrarily cutting the federal contribution to shared-cost programs; Ottawa accused the provinces of violating the spirit of these programs by overspending on them. Neither order of government addressed the inefficiencies in its programs.

By the mid-1990s, Canada’s net debt exceeded 100 per cent of GDP; we were the third most indebted OECD country, behind Italy and Belgium. Something had to be done. Given the large gap between taxing effort in Canada and the United States (see figure 1), it was inevitable that the emphasis would be on expenditure reductions as opposed to tax increases.

22_Figure_1

Figures 1 and 2 illustrate Canada’s taxing effort and public spending as shares of GDP. They make no reference to the kind of spending – over a third of Ottawa’s spending in the mid-1990s was on public debt interest – or the progressivity of the tax system, but they enable the reader to grasp where Canada was placed relative to other OECD countries. In the early 1990s recession, Canadian spending (figure 2) was not far below the upper quartile among OECD countries. It subsequently declined and was close to the OECD median over the second half of the decade. Post-2000 it continued to drift down as a share of GDP and roughly coincided with the bottom quartile for the final five years (admittedly, both Canada’s spending and the bottom quartile rose following the 2009 recession). Canada’s taxing effort was above the OECD median during the first decade illustrated. It drifted below the median over the second decade, finishing close to the bottom quartile of taxing effort.

How to explain wage inequality
22_Figue_2

Among the best-argued chapters is the one by David Green and James Townsend on explanations for the increasing inequality of market earnings. Their foil is the 1994 OECD Jobs Study, a major report arguing that industrial countries were experiencing long-term technological change favouring those with skills at the expense of those without, and that this was the fundamental explanation for rising earnings inequality. The OECD’s recommended policy to counter the trend was massive investment in education and skills training, combined with flexible labour markets enabling employers to hire and fire with few administrative constraints. Canada is among the countries that took this advice seriously. One piece of evidence: in the 2012 PISA comparative assessment of student school performance, Canada is the OECD country with the highest proportion of adults with postsecondary training. More on education below.

Green and Townsend are sceptical of the OECD’s explanation. Over the three decades, the ratio of mean wages among women with university degrees relative to those with high school alone was essentially constant. Among men, the ratio rose until about 2000; subsequently, it fell somewhat. How, they ask, can this be squared with the thesis of a dynamic driven by technology? Defenders of the OECD thesis counter that the increasing supply of better educated workers in the labour force offset what would otherwise have been a sharp increase in the skills premium. But, argue Green and Townsend, the shift in skills composition of the labour force occurred primarily pre-2000; it was fairly constant in the 2000s. If the OECD was right, the skills premium should have continued increasing in the 2000s. It didn’t.

The OECD’s explanation of wage inequality may be flawed, but there is clearly a large skills premium. What explains it? Green and Townsend are sceptical not only of the OECD explanation but of several others. They cast doubt on the idea that NAFTA or the massive increase of China’s presence in international trade explains much. Their preferred explanation fits within the thesis of the volume: the problem is primarily political. The OECD thesis distorted public policy. It led to unwarranted relaxation of labour legislation and a consequent decline in unionization in the private sector. It led to policies inducing firms to invest in digital technology and underinvest in machinery and equipment complementary to the skills of the less educated.

I am not sure that Green and Townsend have identified the most important avenues whereby policy has exacerbated earnings inequality, but I agree with them that politics matter. Elsewhere in this issue, Tom Courchene makes a convincing case that conventional wisdom among economists in favour of a free-floating exchange rate has been severely damaging to the manufacturing sector, which provides many middle-class Canadians with near-median-income jobs.

Over the last two decades the exchange rate between the Canadian and U.S. dollars has been closely tied to the price of oil. The loonie is a petrocurrency. As oil prices rose post-2000, so did the loonie: from the low 60 cent range to parity by 2010. This induced major shutdowns of Canadian manufacturing plants and relocation to the United States. Furthermore, China’s impact on Canadian wages cannot be measured solely by reference to the magnitude of foreign imports in trade-exposed sectors. The Chinese demand for resources – oil and gas in particular – is part of the present Alberta-dominated federal government’s motivation for favouring oil and gas exports from western Canada, which in turn exacerbate the impact of an overvalued loonie on manufacturing jobs.

Immigrants, Aboriginals and social trust

The chapter by Keith Banting, Stuart Soroka and Edward Koning deals with multiculturalism and redistribution. They note the “surprisingly positive” survey evidence among Canadians toward immigration and the paucity of evidence that large-scale immigration has lowered Canadians’ faith in redistributive politics. They contrast Canadian attitudes with the depressing results of Robert Putnam with respect to American public opinion. In a major oft-cited study, Putnam found that ethnic diversity in U.S. communities led to an erosion of interpersonal trust among everyone. Both immigrant and old-stock Americans “hunker down” in their respective ethnic communities and lose faith in government social policy. Elsewhere in this issue I review the recent book on immigration by Paul Collier, who also refers to Putnam’s work. His emphasis is on the European experience, which in general has been much less optimistic than that in North America.

The key reason for Canadian exceptionalism is probably Canada’s strict requirements for immigration, by international standards, in terms of education and ability to speak either French or English. However, Canada has not entirely avoided the European and U.S. syndrome of ethnic ghettos dominated by poorly integrated descendants of ethnic minorities. The Caribbean-dominated Jane-Finch neighbourhood in the north end of Toronto cannot be described as high-trust. The growth of militant Islam, symbolized by women covering themselves in public, has created a backlash in Quebec. In 2013, inspired by French precedents, the Parti Québécois government proposed a Charter of Quebec Values, supported by a combination of secular cosmopolitan feminists and traditional Québécois de souche. The new Liberal government will also legislate in this area, although its measures are expected to be less sweeping than the PQ’s.

Banting and colleagues also include in their chapter a discussion of attitudes toward Aboriginals among a random sample of Canadians. Those who thought “most Aboriginals are on welfare” were much less likely to endorse the statement that “refusing welfare to single parents is unfair to their children” and more likely to support the alternative proposition that “giving welfare to single parents encourages irresponsible behaviour.” Their conclusion from survey analysis is that negative views of Aboriginal reliance on welfare has a “toxic effect” on attitudes toward the entire welfare state.

Clearly implied in this chapter is that those who supported any of the statements critical of Aboriginal use of welfare are probably guilty of racist attitudes. Some undoubtedly are racists, but some respondents probably had a more sophisticated view of social assistance and Aboriginal policy than those who composed the questionnaire. And some who gave responses favourable to generous access to welfare analysis may well have suppressed their critical thoughts and opted for the politically correct answer to avoid any stigma of racism. In sum, basing conclusions about the willingness of Canadians to maintain a generous welfare state on such questions is dubious.

A glaring omission

There is no chapter in this volume on the role of K–12 education as policy to reduce inequality. Symbolic is the first table (at p. 5), which compares trends in public “social expenditure” over the last half century among a sample of OECD countries. For reasons not explained, the data exclude education. Green and Townsend indirectly criticize public policy emphasis on postsecondary education because of its link to their foil, the OECD Jobs Strategy. Robin Boadway and Katherine Cuff briefly discuss the role of postsecondary education, but not primary or secondary. Rianne Mahon discusses the contrast between Quebec and Ontario with respect to early childhood education but does not analyze its impact on equality.

Here are a few basic statistics on the link between education, employment and earnings, according to the 2011 census. For those without high school certification only 34 per cent were employed, and median earnings (for those with at least some earnings) were only $16,000. For those with high school but no further education, the employment rate jumped by more than 25 percentage points and median earnings rose to $23,000. To achieve what Canadians consider middle-class incomes required some form of postsecondary training. For those with a trades certificate, college diploma or university degree, median earnings were $40,000.

The OECD’s Programme for International Student Assessment (PISA) has become the best-known exercise in making international comparisons of school systems. In each “round,” PISA assesses student performance at age 15 in three subjects: mathematics, science and reading. In the latest round, conducted in 2012, mathematics was the focus. Canadian schools slipped somewhat in rankings relative to 2009 but remained well above average. (Not discussed here are detailed PISA results by province. There are large differences among them.)

Educational outcomes vary according to the supply of education services and the demand by parents and their children for formal education. On the supply side, a fundamental distinction is between the services provided by schools on the one hand and by a student’s family on the other. Separating the contribution of schools from that of families (in terms of both supply and demand) is not easy. PISA makes a serious attempt to do so via construction of an index of social, economic and cultural status of students’ families. An intuitive appreciation of the PISA strategy is afforded by the social gradients illustrated in figure 3.

22_Figure_3

The figure illustrates social gradients for all students living in OECD countries and for those in Canada, in the United States and in five selected European countries. For each of these eight populations, students whose school performance has been assessed also provide information about their families. This permits construction of a national social-economic-cultural index of students’ families. The two end points in the figure for each gradient are the average index values among the bottom and top quarters of students ranked by this index with their corresponding average mathematics scores. The line connecting them is the social gradient. The gradient’s slope is a measure of the increase or decrease in expected mathematics score for a unit increase or decrease in students’ family status. (The increase in expected mathematics score arising from a one-point increase in the social-economic-cultural index appears in parentheses in the legend.) The “flatter” the gradient (the smaller the slope), the more successful is the school system in offsetting social disadvantage. By this measure, Canada has a relatively egalitarian school system.

The gradients furnish three measures of a country’s school system: (1) average performance among socially disadvantaged students, (2) average performance among socially advantaged students, and (3) a measure of the ability of the school system to overcome students’ disadvantage due to low social-economic-cultural family status.

Among the countries illustrated, Canada enjoys the highest top-quarter index value and second highest bottom-quarter index value. Its bottom-quarter mathematics score is in a virtual tie with Finland; its top-quarter score ranks third behind Germany and France. The slope of the Canadian gradient (32.7) is the “flattest” of those illustrated, and is well below the OECD average (39.3). At the other end of school system equality is France. It enjoys the second highest top-quarter mathematics score, but is tied with United States for lowest bottom-quarter score. The result is that France has the “steepest” gradient (58.1), nearly twice Canada’s. While French students from socially advantaged families perform well, the French system, like the American, is weak in offsetting social disadvantage. Disproportionately important among the bottom quarter are many second- and third-generation Muslim immigrants from Africa.

Finally, education policy should be central to Aboriginal policy. The youngest census cohort for which it is reasonable to expect high school completion is ages 20–24. Among those in this cohort who identified in the 2011 census as North American Indian or First Nation, only 60 per cent had completed high school. This compares with 90 per cent among non-Aboriginals. These statistics should lead us to explore what is going on within schools. Figure 4 displays results for reading, by grade, in a sample of reserve schools in British Columbia. The pattern is typical for Aboriginal children in schools across Canada. Among those children who attend preschool and kindergarten, roughly two thirds are reading “at grade” level. (Not all students attend preschool.) The share reading at grade level falls below half in early primary grades and drops to 20 per cent by grade 8. The at-grade share rises to 30 per cent by grade 12, as a result of large numbers of academically weak students dropping out.

While these results are hardly encouraging, B.C. is in fact faring dramatically better than the other provinces in terms of Aboriginal education – particularly with respect to those living on-reserve. The worst Aboriginal education outcomes exist in the prairie provinces. PISA does not record student ethnic identities, but the fact that Manitoba’s bottom-quarter mathematics scores are tied for lowest across provinces and that the slope of its gradient is second highest is probably closely related to the disproportionate presence of Aboriginals among Manitoba’s bottom-quarter students.

All of which implies that inequalities matter but understanding them is not easy. Banting and Myles deserve credit for making a determined effort.