Introduction

Most analyses of regional income disparities in Canada have focused on provincial disparities. But our province of residence is not the only geographic characteristic that might influence incomes. Income levels can be influenced just as much by whether we live in a large or small city or in a rural region.

The purpose of this paper is to analyse geographic income disparities in Canada from the perspective of provinces and especially urban and rural areas. In particular, it looks at how per capita incomes vary across the urban-rural continuum—that is, how per capita incomes in large cities like Toronto and Montreal compare with medium sized cities like Halifax and Victoria, small cities like Brandon and Drummondville and with rural areas.

We are interested not only in how incomes vary across the urban-rural continuum, but also whether income disparities across city sizes and between cities and rural areas can provide us with a better understanding of provincial disparities. If incomes systematically fall with city size and between cities and rural areas, then part of the reason why provincial incomes may vary is because the populations of some provinces are concentrated in smaller cities and rural areas, while the populations of others are concentrated in larger cities. It is through this ‘urban-rural lens’ that we might better understand provincial income disparities.

There are several reasons why we expect per capita income levels to vary across urban and rural regions. First, per capita incomes may be lower in smaller cities and rural areas if a smaller proportion of their population is working relative to large urban areas. Second, workers in rural areas and smaller cities may be paid less than workers in larger urban areas. In part, this may be because workers in larger urban areas work in occupations that pay more—an occupational composition effect. It may also be that employers are able to pay higher wages in larger urban areas because of higher levels of productivity.

The analysis relies on data from the 2001 Census. Urban areas are subdivided into four size classes. The largest includes cities that have a population greater than or equal to 1,500,000—i.e., Toronto, Montreal and Vancouver (Tor-Mtl-Van). Large Urban areas are those with a population of 500,000 to 1,499,999 (e.g., Quebec City). Medium Urban areas have a population between 100,000 and 499,999 (e.g., Kingston) and Small Urban areas have a population between 10,000 and 99,999 (e.g., Red Deer). All other parts of Canada are classified as rural, which we subdivide into two types. The first is rural areas that interact with urban regions through commuting flows, what we call rural areas in the Urban Shadow. The remaining Remote Rural areas are those that do not have any interaction with cities through commuting flows. Income levels are measured by employment income per capita.

Provincial urban-rural distribution of population

In 2001, Canada’s population was overwhelmingly urban. The urban categories combined for almost 80% of Canada’s population. A further 19% lived within commuting distance of an urban area with a population of at least 10,000—the Urban Shadow. The remainder of the population (1.2%) lives in Remote Rural areas that have no interaction with urban regions with a population greater than 10,000. Rural populations that are isolated from urban areas are relatively few. Yet, for some provinces these are important. Ten percent of Saskatchewan’s population is rural, as is 5% of Newfoundland and Labrador’s.

There is considerable variation in the distribution of provincial populations across the six urban and rural classifications. None of the Atlantic Provinces or Saskatchewan have cities with a population greater than 500,000. These are also regions with large rural populations. Remote Rural and the Urban Shadow account for well over 40% of the populations of Saskatchewan and each of the Atlantic Provinces, with the exception of Nova Scotia, which nevertheless has a large rural population. This compares to a 21% share for Canada as a whole.

The remaining provinces have one or more cities that fall into the Tor-Mtl-Van or Large Urban classifications. These are the most important components of their populations. For instance, Winnipeg accounts for 60% of Manitoba’s population and Vancouver accounts for just over 50% of British Columbia’s.

In effect, Canada is made up of two types of provinces, those that are endowed with larger urban regions and those that are more rural in character. If incomes vary systematically across the urban-rural continuum, then part of the reason why we observe provincial income disparities may be because of their differing composition of urban and rural areas.

Per capita income disparities across provinces and urban and rural areas

Per capita employment incomes for provinces and the urban-rural continuum are displayed in Figures 1 and 2, respectively. As Figure 1 illustrates, considerable income disparities persisted across provinces. Per capita incomes were highest in Ontario ($18,100) and lowest in Newfoundland and Labrador ($10,100). Yet, across the urban-rural continuum, per capita incomes were even larger. Per capita incomes were highest in Large Urban areas ($18,500) and lowest in Remote Rural ($8,600). The $10,000 spread between larger urban areas and remote rural areas is greater than the $8,000 per capita income spread across provinces.

Figure 1

The final point to be drawn from Figure 2 is that incomes rise systematically with urban size and between urban and rural areas. This suggests a smaller proportion of the population in small urban and rural areas are working compared to larger urban areas and/or employment income levels are lower.

Figure 2

Provincial disparities and urban endowments

As noted above, provinces differ considerably in their urban-rural composition. Combined with the large variation in per capita incomes that we observe across the urban-rural continuum, this implies provincial income disparities partly stem from how their populations are distributed across urban and rural areas. Some provinces may have relatively low per capita incomes because their populations are more heavily concentrated in smaller urban areas and in rural areas.

In more precise terms, part of provincial per capita income differences will be explained by their urban-rural composition when there are differences in the urban-rural composition of provinces; and employment income per capita differs significantly across urban and rural areas within each province.

Figure 3 illustrates this latter point. It plots the levels of employment income per capita across all six urban and rural classifications for several provinces and two larger regions—Atlantic Canada and Manitoba and Saskatchewan.

The first point to be drawn from Figure 3 is that for almost all provinces/regions per capita incomes fall moving from larger to smaller cities, from cities and rural areas, and from rural regions in the shadow of urban centres to more remote rural regions. The negative gradient we observe nationally is replicated across all provinces (see also Table 1).

Figure 3

The second point to be drawn from Figure 3 is that the degree of variation we observe in each urban-rural category across provinces is always smaller than the amount of variation in per capita income within each province across urban-rural classes. Per capita incomes in Medium Urban areas, for instance, range from $17,000 in Ontario to $14,000 in New Brunswick (see Table 1), while in Ontario, for example, per capita employment income ranges from $20,400 in Toronto to $8,100 in Remote Rural areas. The examples of Medium Urban areas and Ontario are not unique. In all cases, there is a higher standard deviation in per capita income across urban-rural classes within a province than there is within each urban-rural class across provinces.

Table 1: Provincial per capita incomes across the urban-rural continuum

This analysis suggests that part of the reason why we observe provincial disparities in per capita incomes is the difference in the urban and rural composition of each province. But we cannot tell from these data the degree to which urban-rural composition matters and for which provinces it matters most. To address this issue, we need to decompose the differences in provincial per capita incomes into a component that is due to ‘urban-rural composition’ and a residual component attributable to ‘other factors’. The results of this decomposition are presented in Figure 4 and Table 2.

Figure 4

For the decomposition, the ‘urban-rural composition’ component captures the influence of the urban-rural composition of provinces on their deviation from the national average, while controlling for differences in employment incomes within each rural-urban classification across provinces. Provinces whose populations are concentrated in small urban and rural areas will tend to have a negative urban-rural endowment effect, while provinces that have a relatively high proportion of their population in large cities will tend to have a positive effect.

The ‘other factors’ component might be related to province-wide labour market conditions or forces that are specific to particular classes of urban or rural areas within a province. Manitoba’s negative deviation is an example of the latter. Manitoba’s relatively low per capita income levels largely stem from Winnipeg’s low per capita income compared to its peer group of cities and because Winnipeg accounts for a higher proportion of Manitoba’s population compared to its peer group of cities.

Urban-rural endowments accounted for a significant proportion of the income disparities experienced by the four Atlantic Provinces and Saskatchewan. The case of Nova Scotia is illustrative. Per capita earned incomes in Nova Scotia are $3,616 below the national level. Of this disparity, $2,007 is due to its urban-rural composition and $1,609 is due to other factors (see Figure 4 and Table 2). Where urban-rural composition matters, provincial populations are concentrated in relatively small cities and rural areas. However, urban-rural endowments are not an important determinant of income disparities for all provinces.

For Quebec, Ontario and Manitoba, it is the performance of their large cities that matters. On the one hand, Ontario’s relative strength resulted mostly from above average incomes in Toronto, Ottawa and Hamilton compared to their peer groups of cities. On the other hand, Quebec and Manitoba’s relatively low employment incomes result from low incomes in Montreal and Winnipeg compared to their respective peer cities. The variation in the performance of these cities may be due to many factors, including their underlying industrial structure.

Alberta’s strong performance results from relatively high per capita incomes across all of its rural and urban classifications compared to their peer groups. It also results from the relatively high concentration of Alberta’s population in its Large Urban centres (Calgary and Edmonton) that outperform their peer group in terms of per capita incomes.

Table 2: Decomposition of provincial disparities

In summary, provincial disparities are closely tied to how their populations are distributed across urban and rural areas, and to the relative prosperities of their larger urban economies.

Labour market and demographic conditions

The previous section illustrated how the distribution of a province’s population across urban and rural areas can influence its overall level of per capita incomes. It would be natural to ask what underlies the strong negative gradient observed in per capita incomes from the largest urban areas to the most isolated rural regions. One possibility is that the lower per capita incomes in smaller cities and rural areas are not a reflection of lower incomes earned by workers in these places, but rather of a small proportion of people in these areas being employed because of labour market and/or demographic factors.

Per capita employment income depends on several factors. They most obviously depend on the level of incomes earned by those employed. However, per capita employment incomes depend on two other factors that primarily relate to a region’s demographic and labour market conditions. First, they depend on the proportion of the population that can be employed—the working age population fifteen years and older. Those regions with a higher proportion of children will have lower per capita incomes. Hence, demographic conditions help to determine per capita incomes.

The second factor that affects per capita incomes is the proportion of the working age population that is employed. The employment rate depends both on labour market conditions facing workers and on other factors that influence the decisions of workers to enter the labour market. The lower the employment rate the lower the per capita income.

Together these factors—the employment income of workers and the employment rate—can be related to per capita income levels using the simple identity that per capita income is the product of employment income per worker, the employment rate and the share of the population that is of working age.

The values for each of these three terms, as well as per capita income, are given in Table 3 for each of the urban and rural classes. The second row gives the level of employment income per worker: it falls consistently from larger to smaller cities and between urban and rural areas. Employment incomes are lowest in Remote Rural regions. There is also a tendency for employment rates to decline with city size, although Large Urban areas have a higher employment rate than Tor-Mtl-Van. Remote Rural areas have the lowest employment rates of all. The working age to population ratio varies little across urban and rural classes. It is only in Remote Rural areas where this ratio is appreciably lower; that is, it is only in these areas where demographic conditions matter.

Table 3: Characteristics of urban and rural areas

The results presented in Table 3 suggest both variations in employment incomes per worker and labour market and demographic conditions contribute to disparities in incomes across the urban-rural continuum. Remaining to be answered is the relative contribution of these to disparities.

To simplify the analysis, we multiply the employment rate and the working age to population ratio to give us the employment to population ratio (see Table 3). As such, this measure combines both labour market and demographic effects. Since there is little variation in the demographic conditions across urban areas and between urban areas and the Urban Shadow, this term largely captures the effect of varying labour market conditions. Only in Remote Rural regions will demographic and labour market conditions both contribute to its difference from other urban-rural classifications.

Figure 5 decomposes deviations in employment income per capita into two components across urban-rural classifications. The first is the employment to population ratio, which captures the effect of deviations in the employment to population ratio from the national level on per capita incomes in each urban-rural class. The second term is employment income per worker, which captures the effect of deviations in employment income per worker from the national level on per capita incomes in each urban-rural class.

Employment income varies across urban and rural areas in Canada because workers in rural areas earn lower incomes than those in urban areas and because a smaller proportion of the population in rural areas is employed. In Tor-Mtl-Van and Large Urban areas, favourable labour market conditions raise per capita incomes. This is particularly true of Large Urban areas where just over half of their positive deviation is due to favourable labour market conditions. In contrast, labour market conditions in Small Urban areas and the Urban Shadow tend to reduce per capita incomes. Finally, in Remote Rural regions, the employment to population ratio accounts for almost half of their negative deviation. Part of this is due to unfavourable labour market conditions and part is due to demographic conditions that result in a smaller than average working age population.

Figure 5

The conclusion to be drawn is that the variation we observe in per capita incomes are exacerbated by the labour market and demographic conditions that tend to favour larger cities and disfavour smaller cities and rural areas, particularly more remote rural regions.

Conclusion

This paper demonstrates that urban-rural structure does matter. The Atlantic Provinces and Saskatchewan have below average per capita incomes to a significant extent because their populations are concentrated in smaller urban and rural areas. However, this is not true of all provinces.

Ontario’s relatively strong performance and Quebec and Manitoba’s weak performances do not stem, by and large, from a favourable or unfavourable urban-rural composition, but how incomes in their largest cities compare to their peers. Specifically, Ontario’s high per capita incomes can be traced to Toronto’s relatively high incomes, while Quebec and Manitoba’s lower per capita incomes are linked to Montreal and Winnipeg’s low incomes relative to their peers. In short, our understanding of the performance of these provincial economies depends on what underlies the relative performance of their larger cities.

How then do we interpret the strong negative income gradient from larger to small cities and between cities and rural areas? In part, this negative per capita income gradient results from relatively poor labour market conditions in smaller cities and rural areas compared to larger cities. However, this negative gradient still remains after controlling for labour market (and demographic) conditions; on average, employed workers in larger cities are paid more than those in smaller cities and rural areas.

Underlying the income gradient of employed workers are many factors, but two are probably most important. First, occupations that are higher paying may be overrepresented in larger cities compared to smaller cities and rural areas. Second, it may be that higher wages in large cities are matched by higher levels of productivity on the part of urban workers.

There are many factors that might explain the productivity advantage of locating in larger cities, which include the benefits of access to large pools of skilled labour and locating in a center with a larger number of industries. Economies that are inherent to larger cities may be very difficult to replicate in smaller cities and rural areas. Hence, part of the provincial disparities we observe may in fact be structural.

Recent feature articles