{"id":14011,"date":"2019-12-16T16:51:47","date_gmt":"2019-12-16T21:51:47","guid":{"rendered":"https:\/\/irpp.org\/research-studies\/whats-so-bad-about-increasing-inequality-in-canada\/"},"modified":"2022-03-31T10:08:45","modified_gmt":"2022-03-31T14:08:45","slug":"whats-so-bad-about-increasing-inequality-in-canada","status":"publish","type":"research-studies","link":"https:\/\/irpp.org\/fr\/research-studies\/whats-so-bad-about-increasing-inequality-in-canada\/","title":{"rendered":"What\u2019s So Bad about Increasing Inequality in Canada?"},"content":{"rendered":"

Ce chapitre fait partie de l\u2019ouvrage collectif\u00a0Income Inequality: The Canadian Story<\/a><\/u>. Il est le fruit de deux ann\u00e9es de collaboration entre l\u2019IRPP et le R\u00e9seau canadien de chercheurs dans le domaine du march\u00e9 du travail et des comp\u00e9tences (RCCMTC). Pas moins de 27 \u00e9minents experts y analysent les tendances en mati\u00e8re d\u2019in\u00e9galit\u00e9 au Canada, les facteurs ayant dict\u00e9 l\u2019augmentation des in\u00e9galit\u00e9s depuis le d\u00e9but des ann\u00e9es 1980 et le r\u00f4le que doivent jouer les politiques publiques pour rem\u00e9dier au probl\u00e8me.<\/em><\/p>\n

\n

In 1981, Canada\u2019s only economics textbook on economic inequality concluded confidently that \u201ceconomic inequality has remained roughly constant since the Second World War\u201d (Osberg 1981, 205). Little did its author then realize how much Canadian inequality, and the study of inequality, was about to change! The 35 years since have seen an explosion of the literature on inequality and remarkable innovations in research methodologies \u2014 both of which -undoubtedly have been driven by the changing facts of economic inequality. Nevertheless, although much has been learned in recent years about economic inequality, some important results of the early literature have slipped from view \u2014 in particular, the limitations of summary indices of inequality and the futility, in a market economy, of distinguishing inequality of opportunity from inequality of outcome. Meanwhile, the emphasis of the early literature on comparing levels of inequality has unfortunately remained. That emphasis was, in 1980-81, reasonable. When the level of inequality is fairly stable in many countries, as it then had been for several decades, thinking of economic inequality in terms of cross-sectional comparisons of societies makes sense. Such comparisons enabled researchers to ask if countries with higher levels of income inequality are less healthy, less democratic and less happy and have more crime, more conflict and less intergenerational social mobility and equality of economic opportunity than countries with lower levels of inequality. Cross-country comparisons are useful if one can depend on inequality remaining stable and if one is interested in answering a question like \u201cWhat type of society would one like to live in?\u201d Since this is an important question, and since there was a significant period during which within-country levels of income inequality did not change much, the literature on income inequality has continued to emphasize such comparisons.<\/p>\n

However, cross-sectional comparisons between countries presume that observed levels of inequality represent steady-state outcomes (that is, situations that potentially could persist into the indefinite future). But that can be true only when income growth is balanced (that is, if the rate of income growth is equal at all percentiles of the income distribution).1<\/sup>\u00a0When income growth is unbalanced, the level of inequality changes over time.<\/p>\n

Increasing inequality in a given society, and an expectation that the trend will continue, raise fundamentally different issues from a one-time change in the level of inequality. It is one thing to say that Canada is more unequal now than it was in 1980, and quite another to say that Canada is more unequal now than it was in 1980 and<\/em> Canadians can expect it to get ever more unequal each year into the indefinite future. In this chapter, therefore, I ask what the implications of ever–increasing inequality might be and whether this can possibly be a steady state.<\/p>\n

Specifically, there is no natural upper bound to the real income of those at the top of the income distribution, and thus no natural upper bound to the gap between their income and that of median households. But can an ever–increasing income gap between the top 1 percent and everyone else possibly continue indefinitely? \u201cMore inequality,\u201d in the sense of increasing inequality over time, raises important questions: What sort of society are we becoming? What processes could equalize income growth rates across income classes and thereby stabilize the distribution of income? How likely are they to occur? What happens if income growth rates remain unequal?<\/p>\n

I begin by asking what can be learned about the long-run implications of higher inequality from comparisons of cross-sectional data. I then compare real income growth rates over time at the very top of the income distribution and for the bottom 90 percent and bottom 99 percent in Canada, the United States and Australia to illustrate how unique the postwar period of balanced growth (and consequent stability of the income distribution) actually was in these three countries. In recent decades, the rapidly rising share of market income of those at the very top end of the distribution has reflected a new normal: unbalanced growth. I then suggest that there is little reason to expect an equalization of market income growth rates any time soon, and I argue that unbalanced growth of incomes cannot be a long-run steady state. Unequal income growth rates imply changes in savings flows that may cumulate to changed stocks of indebtedness, financial fragility and periodic macroeconomic crises. Ever-increasing income gaps also imply increasing top-end spending on political influence and children\u2019s education, as well as ever–increasing incentives to advertise the luxury consumption that fuels envy. Greater macroeconomic, political and social instability is therefore a likely implication of more inequality over time. I conclude by suggesting that, if markets do not spontaneously auto-equilibrate, the political economy of increasing inequality will be crucial, but the outcomes of those processes are very unclear.<\/p>\n

Cross-sectional Comparisons of Inequality and Implications<\/h2>\n

By economic inequality, I mean differences among people in their command over economic resources. However, because every society has many different types of economic resources, used by many different people at different points in time, the measurement of inequality depends crucially on being specific about inequality of what<\/em> among whom<\/em>, when<\/em>. Like most of the literature, I focus in this chapter on inequality in the distribution of annual income.2<\/sup>\u00a0But to decide which <\/em>annual income concept is most appropriate to discuss, one first needs to ask why we want to know.<\/p>\n

In a market society, income flows perform dual functions. \u201cMarket income,\u201d for example, is simultaneously the payments of firms and the receipts of individuals. Firms pay individuals to motivate the supply of labour and capital to the production process, while individuals typically pool their receipts within households to enable personal utility from consumption. If one wants to think about how changes in the size, structure and organization of firms and markets are changing inequality, one therefore should start with inequality among individuals in their receipt of factor payments \u2014 that is, their individual market income before tax.<\/p>\n

Much of the literature is motivated, however, by concern about inequality in the distribution of well-being from consumption, because equity in well-being is important in a social justice sense. If \u201cinequality\u201d is to be understood in terms of potential consumption, the fact that most people live in households and share consumption with other family members implies that the appropriate annual income concept on which to focus is the total disposable \u2014 that is, after taxes and transfers \u2014 money income of households. If well-being from the consumption of disposable income is to be measured accurately, some allowance should be made for the economies of scale in consumption that are available to larger households. In affluent countries, therefore, it has become common practice to adjust household income for family size and to report the distribution of equivalent disposable income among all individuals.3<\/sup>\u00a0One can thus discuss income inequality either from a production perspective, in terms of the inequality of factor payments (individual market income), or from a consumption perspective, in terms of inequality in household receipts of purchasing power (household market income plus transfers minus taxes). In this chapter, I primarily emphasize the former perspective\u2008\u2014 that is, trends in the inequality of individuals\u2019 shares of pre-tax market income.<\/p>\n

However, an important lesson from the consumption perspective, as Heisz (in this volume, figure 2) illustrates, is that the level of inequality of disposable household income varies considerably among advanced market economies. Although all member countries of the Organisation for Economic Co-operation and Development (OECD) compete in the global marketplace and are -increasingly interconnected in trade and harmonized in market regulation, there is a broad range in levels of within-country inequality of disposable household income. Evidently, the institutional framework of market capitalism and effective participation in the modern global economy do not dictate a unique level of disposable household income inequality; social choices are possible. But since it is hard to imagine that the level of income inequality could be unconnected to other aspects of society,4<\/sup>\u00a0an important question then is, \u201cWhat exactly are the implications of more or less income inequality?\u201d<\/p>\n

Over the past 30 years, an ever-expanding group of scholars has used cross-country comparisons to address the implications of inequality. Richard Wilkinson and Kate Pickett\u2019s 2010 book, The Spirit Level: Why Equality Is Better for Everyone<\/em>, is a particularly famous example. In it, the authors compare the level of inequality of disposable household income with a host of social indicators \u2014 average levels of health status, trust, social mobility, infant mortality, educational performance, violence, obesity, mental illness, teenage births, homicides and imprisonment \u2014 both across OECD countries and across states within the United\u2008States. They conclude that, along all these dimensions, in places where there is more income inequality there are also more social problems.<\/p>\n

However, even if more inequality might be guilty of causing these problems, proving <\/em>it rigorously is difficult. Wilkinson and Pickett use correlations and scatter plots that cannot establish causation, and they depend heavily on data from only 25 affluent countries, which exposes their work to the critique that this or that \u201coutlier\u201d might be dominating their results. As Leigh, Jencks and Smeeding (2009, 399) put it, in discussing the relationship between inequality and health, \u201ca fundamental problem is the fact that this is a field with too many theories for the number of available data points.\u201d<\/p>\n

Summary indices of inequality and their limitations<\/h3>\n

Although many studies (for example, Osberg and Smeeding 2006) have documented that the vast majority of survey respondents in all countries express a preference for less inequality, surveys rarely define the term precisely. Cross-country comparisons indicate, however, that nations differ sometimes in the share of income of those at the top of the income distribution, sometimes in the share of the middle classes and sometimes in the gap between the poor and the rest of society. Which aspect of inequality matters most? In much of the recent academic literature on inequality, there is little evidence of awareness of the complexities of index construction that early literature emphasized.<\/p>\n

Like many other recent scholars, Wilkinson and Pickett (2010) use the Gini coefficient as if it were an unambiguous measure of inequality, but Atkinson\u2019s (1970) article on the limitations of using a single number to summarize the many differences in individuals\u2019 access to economic resources remains fundamental. He noted that, among the many possible summary indices of inequality, one should accept only those that satisfy some basic ethical criteria \u2014 such as the \u201cprinciple of transfers\u201d (i.e., that a transfer of income from poor to rich should always imply an increase in any acceptable index of inequality).5<\/sup>\u00a0Within the class of such axiomatically defensible inequality indices, the Gini coefficient is more sensitive to income changes in the middle of the distribution than is the Theil index (which is low-end sensitive) or the coefficient of variation (which is high-end sensitive) (see Jenkins 1991; Osberg 1981, chap. 1; Osberg 1984, chap. 1). Atkinson (1970) showed that these indices will agree unambiguously only sometimes in their cross-country rankings of inequality. Specifically, they will all agree that country A has less inequality than country B only when it is always true, at all points in the income distribution, that the relatively poorer in country A have a bigger share of income than comparable people in country B. Atkinson also showed, however, that, in real-world comparisons, it is rare to observe such unambiguous dominance.<\/p>\n

To illustrate the ambiguities that can hide beneath simple comparisons of the Gini coefficient, I invented the example of \u201cAdanac,\u201d a mythical country in which the income distribution is very simple: the bottom four-fifths (80 percent) of the population all have the same low income \u2014 each quintile group (20 percent) of people gets 10 percent of total income. The income of those in the top 20\u2008percent is exactly six times higher, and thus they get 60 percent of total income (Osberg 1981, 14). In real-world Canada in 2011, the total income share of the poorest 20 percent of the population was actually 4.1 percent, while the share of the richest 20 percent was 47.2 percent. So these assumptions imply that in Adanac, both the extremes of the income distribution would get a substantially bigger share of national income than they do in Canada. However, the middle 60\u2008percent of Adanac, who in total have an income share of 30 percent, are considerably worse off than the middle 60 percent in real-world Canada, whose actual share in 2011 was about 48.7 percent. Which society is more unequal? I picked the numbers for Adanac so that it has approximately the same Gini coefficient as does Canada. But these are very different societies.6<\/sup>\u00a0The answer to which is more unequal depends on which part of the income distribution one thinks matters most for inequality: the top-end share, the share of the middle classes or the gap between the poor and everyone else.<\/p>\n

Ambiguities of measurement have become particularly relevant for Canada in recent years, since the increase in the Gini coefficient of equivalent household disposable income after 2000 has been much smaller than the increase in the 1990s. However, this flattening of the upward trend in the Gini coefficient masks a trend to the greater polarization of incomes in Canada. Since 2000, the relative advantages of the middle class have declined \u2014 that is, the ratio of the average income of the second, third and fourth quintiles to that of the first quintile has fallen. As Beach (in this volume, 171) documents, there has been a \u201cmarked decline in the proportion of workers who receive middle-class earnings\u201d in -Canada. Because the incomes of many in the middle class have converged downward toward the incomes of the poor, a perverse sort of equalization of incomes has occurred among the bottom 80 percent. At the same time, the income share of the top 1 percent has continued to increase, but this is obscured in the Gini coefficient, which adds up the absolute value of the income differences among all possible pairs of people. In the calculation of the Gini coefficient, lessened inequality among the bottom 80 percent \u2014 that is, the \u201cdeclining middle class\u201d\u2008\u2014 is combined with, and offsets, rising inequality at the top end of the income distribution. Most Canadians might think that a declining middle class combined with an increasing share of income for the top 1 percent would amount to more inequality, not the same level of inequality \u2014 but that is not how the Gini coefficient is calculated.<\/p>\n

Inequality of opportunity and intergenerational inheritance<\/h3>\n

Although cross-sectional evidence on many of the social implications of higher levels of income inequality might not be rigorously conclusive, recent literature unambiguously confirms old perceptions about the connection between more inequality of outcome and more inequality of opportunity. As Brunori, Ferreira and Peragine (2013, 7) conclude, using cross-country regressions, \u201cCountries with a higher degree of income inequality are also characterised by greater inequality of opportunity\u2026Less unequal countries are also those that have a higher degree of intergenerational mobility.\u201d Corak (2004, 2013) and many others have made the same point.<\/p>\n

The idea that economic inequality accumulates and deepens over generations is hardly new. Indeed, in a market economy in which parents can control their personal expenditures on the human capital of their own children, it is inevitable that the inequality of income of adults will influence the inequality of opportunity of children. As Alfred Marshall (1913, 562) remarked, \u201cThe professional classes especially, while generally eager to save some capital for<\/em> their children are even more on the alert for opportunities of investing it in<\/em> them\u201d (emphasis in original), while the children of the working classes \u201cgo to their graves with undeveloped abilities and faculties.\u201d Marshall insisted especially that \u201cthis evil is cumulative,\u201d and generations of sociologists have since confirmed the intergenerational transmission of socio-economic status.<\/p>\n

In economics, the best-known neoclassical formalization is the Becker and Tomes (1979) parental altruism model of intergenerational bequest. In this model of human capital acquisition and financial bequest, the unequal distribution of income among cohorts of parents is due partly to their own unequal inheritances from the previous generation, which, in turn, enables unequal transfers to their own children. Inequality in parental income thus has an \u201cincome effect\u201d on inequality of opportunity because more parental income enables more disparity in the \u201cenrichment expenditures\u201d (Corak 2013, 91) that increase the skills of advantaged children. Increasing inequality among parents also has a \u201cprice effect\u201d in that the widening of the income differential between the \u201csuccess\u201d and \u201cfailure\u201d of their children implies greater incentives for parents to invest private resources in their children\u2019s human capital. In general, the whole notion of equality or inequality of opportunity makes no sense at all in a one-generation model of human behaviour. However, as soon as the human capital model is extended to consider two or more generations, inequality of outcome in one generation inevitably generates inequality of opportunity in the next generation, and a strict distinction between inequality of outcome and inequality of opportunity becomes untenable.<\/p>\n

The widespread availability of cross-national data has had the important result of establishing that social choices can be made about the level of income inequality.7<\/sup>\u00a0When the level of income inequality could be assumed to be stable, one could use such cross-country evidence to help answer questions such as \u201cWhat sort of society would one like to live in?\u201d But that was then, and this is now. The level of inequality can remain constant only if incomes throughout the income distribution are growing at the same rate. In Canada, the United States and Australia, this has not been true since the 1980s. Instead, higher rates of income growth at the top end of the distribution are compounding on ever-higher base incomes, and these countries now must ask if this increasing inequality will stabilize, and what the implications are if it does not.<\/p>\n

Increasing Inequality over Time: Canada, the United States and Australia<\/h2>\n

For many years, the United States has had considerably more income inequality than the OECD average, while Australia and Canada have been somewhat closer to, but still above, the mean level of household income inequality in advanced market economies. In all three countries, moreover, the Gini coefficient of inequality in equivalent annual net household income has been trending up, albeit with pauses corresponding to periods of decline in the unemployment rate. In this chapter, however, I focus on the changing real incomes of the top 1\u2008-percent and the bottom 99 percent and 90 percent, for three main reasons.<\/p>\n

First, a summary index of inequality does not indicate which<\/em> parts of the income distribution have changed. In the United States and Canada, there has been remarkably little change in real incomes over the past 30 years for much of the income distribution \u2014 the real incomes of the 20th, 40th, median, 60th and 80th percentiles of the family income distribution have been quite flat over time.8<\/sup><\/p>\n

Second, the absolute size of recent changes in the income share of the top 1 percent dwarfs the magnitude of both shifts historically observed and those now occurring elsewhere in the income distribution. Prior to 1980, studying income distribution was sometimes described as being \u201cabout as interesting as watching grass grow\u201d (Salverda, Nolan and Smeeding 2009, 4; see also Aaron 1978) because changes in income share were small; between 1951 and 1981, for example, the income share of the middle 20 percent of Canada\u2019s income distribution fluctuated by 0.6 percent (see Podoluk 1968; Statistics Canada 1998). Since 1981, however, the gain in income share of Canada\u2019s top 1 percent has been more than 10 times larger than this, while the share of the top 1 percent in the United\u2008States grew from 10.8 percent in 1982 to 22.5 percent in 2012.9<\/sup><\/p>\n

Third, the differential in trend growth rates of real income between the top 1 percent and everyone else has been consistently large for more than 25 years, and there is no obvious reason to expect income growth rates to equalize any time soon. As Alvaredo et al. (2013, 3) put it, \u201cMost of the action has been at the very top,\u201d a finding echoed by Morelli, Smeeding and Thompson (2014, 79): \u201cThe rise in the top end has driven much of the distribution in the United States.\u201d10<\/sup><\/p>\n

\"\"<\/a><\/p>\n

Figure 1 shows that the overall trend in the top 1 percent share has followed a similar U-shaped pattern in Canada and the United States (and in the UK and Australia) over the past century: very high in the 1920s and 1930s, declining in the 1940s and early 1950s, stabilizing in the 1960s and 1970s and increasing strongly since the early 1980s (see also Alvaredo et al. 2012; Heisz, figure 6, in this volume, and Atkinson and Piketty 2007).<\/p>\n

In Australia \u2014 unlike in Canada and the United States, where the median real wage and the average income of the bottom 90 percent have stagnated \u2014 middle-class incomes have risen appreciably in recent years. Greenville, Pobke and Rogers (2013) argue that, since 1988, the longest resource boom in -Australian economic history has produced significant increases in employment, hours of work and the hourly real wage for the middle quintiles of the income distribution. Relying on survey data from the Household Expenditure Survey, they note that \u201crising inequality in Australia is also driven by the 99 percent, not just the 1 percent\u201d (2013, 9).<\/p>\n

Long-term differentials in income growth rates<\/h3>\n

Nevertheless, in Canada, the United States and Australia, the further up the income pyramid one goes, the faster the rate of increase of income. Figure 2 compares the long-term compound annual growth rate of real taxable income for various segments of the income distribution.11<\/sup>\u00a0Australia stands out for the positive (+1.13) growth rate of average real taxable annual income among the bottom 90 percent of tax units. However, all three countries share the pattern of unequal growth, with an accelerating rate of increase in real incomes at the top. Since the gulf between income groups will continue to widen as long as incomes at the top end grow faster than the incomes of everyone else, all three countries face the problem of unbalanced growth \u2014 income inequality will continue to increase until either bottom-end incomes grow much faster or top-end incomes grow much slower.<\/p>\n

\"\"<\/a><\/p>\n

The income share of the top 1 percent is the ratio of the total income of the top 1 percent to the total income of all persons (the bottom 99 percent plus the top 1 percent). A ratio changes over time if the rate of growth of the -numerator differs from the growth rate of the denominator. In Canada and the United States, stagnant average real incomes of the bottom 90 percent contrast with the strong growth of incomes at the top, but middle-class income growth is not, by itself, necessarily enough to prevent rising inequality. The bottom 90 percent of -Australians did get some<\/em> increase in average real income: 1.13 percent growth per year compounds, over 28 years, to a 32 percent real increase, which is much preferable to the cumulative real gains of the bottom 90 percent of -Canadians (2.1\u2008percent) and Americans (\u2013\u20091.5 percent). Nevertheless, the differential <\/em>in income growth rates across income groups still produced a similar trend of rising income shares at the very top.<\/p>\n

Although the income share<\/em> of the top 1 percent declined from the late 1930s to the mid-1970s in all three countries (see Heisz, in this volume, figure 6, for Canada and the United States), this was not due to an actual decline in the real incomes of the top 1 percent. Rather, the drop in their income share<\/em> was driven by the more rapid growth of real incomes of the other 99\u2008percent of the income distribution. The income levels<\/em> of the top 1 percent in real, local dollar terms did not decline prior to 1980; they just grew slowly (Osberg 2014, 16). Since the early 1980s, however, the incomes of the top 1 percent have grown dramatically \u2014 an upward trend to which there is no obvious upper bound. As figure 2 illustrates, income growth rates have been even larger the further up the income distribution one cares to look. Lemieux and Riddell (in this volume, 112<\/sup>) note that, \u201cbecause there are not as many people (relative to population) with extremely high incomes in Canada as there are in the United States, fractile-based comparisons of incomes at the top refer to people with substantially different levels of incomes.\u201d The key point, however, is that increasing shares <\/em>of taxable income held by the top 1\u2008percent of the distribution since the early 1980s have been driven by imbalances in growth rates across the income distribution.<\/p>\n

Balanced and unbalanced income growth over time in Canada and the United States<\/h3>\n

To show the changes over time in relative income growth rates underlying the changes in income shares, figure 3 plots the 10-year compound rate of real growth of average income of the top 1 percent, bottom 90 percent and bottom 99 percent in Canada since the 1920s, while figure 4 does the same for the United States. Figure 3 shows a 30-year period in Canada (from the early 1950s to the early 1980s) of rough equality in income growth rates between the top and the bottom ends of the distribution, sandwiched between long periods of unbalanced growth. The 1953-85 period of balanced growth meant stability in the distribution of income. However, from 1939 to 1952 the much more rapid growth of incomes at the lower end produced greater equality of income, while over the past 30 years, the more rapid growth of incomes at the top has produced greater inequality.<\/p>\n

\"\"<\/a><\/p>\n

\"\"<\/a><\/p>\n

In the United States, income growth rates for the three income groups were nearly identical from 1967 to 1982 and quite similar from 1952 to 1967 \u2014 in total, a roughly 30-year period. However, figure 4 also shows dramatic differences in income growth rates in the 1940s and again since 1980. As in Canada, the income of those at the bottom end grew much more strongly than the income of those at the top end in the 1940s, and US income inequality lessened dramatically, but the past 30 years have been dominated by the opposite dynamic.12<\/p>\n

The rapid growth of real incomes of the bottom 90 percent in Canada and the United States after 1940 started from a situation in which<\/p>\n