Findings

View the most recent version.

The role of sector restructuring
The role of agglomeration
The role of economic diversification
The role of human capital
Other factors
Communities and their regional context
Regional differences and outliers
Conclusions
Appendix tables

The role of sector restructuring

Between 1981 and 2006 employment in agriculture declined from 437,600 to 346,400 workers and employment in other primary sectors declined from 349,400 to 330,100 workers.¹ In contrast, professional services employment, and business services employment grew approximately by a factor of 3 over the same period (from 410,000 to 1,122,000, and from 233,500 to 748,900, respectively)².

One constant in recent economic history has been the increasing value of human time (Schultz 1972). Specifically, the price of human time has been increasing relative to the price of machines. Consequently, producers of wheat and lumber and minerals and fish have substituted machines for labour. Communities dependent upon employment in these sectors have experienced a declining workforce. This pressure is expected to continue because the increasing value of human time is expected to continue. Consequently, communities that wish to maintain their workforce must find something new to export from their communities.

Not surprisingly, most communities heavily reliant on traditional sectors went through dramatic downsizing. Each column in Figure 1 represents a quintile (i.e. one-fifth of all communities)³. Most communities with the highest share of employment in primary sectors in 1981 (one-fifth had 33.5% or more of their workers employed in a primary sector) recorded a large population decline between 1981 and 2006 (Figure 1). In this group of communities, over three-quarters reported a population decline — the entire box is in the negative range for community population change in Figure 1.

Figure 1 The typical community with a low employment share in primary industries grew by 24% while the typical community with a high employment share in primary industries declined by 23% from 1981 to 2006

The opposite trend is evident for employment in more dynamic sectors, here captured by employment in distributive and producer services. Among communities with 31.6% or more of their employment in these sectors (i.e. communities in the top quintile), about three-quarters of them recorded some population growth, as opposed to almost three-quarters of the communities with less than 17.5% of employment in these sectors which recorded some population decline (Figure 2).

Figure 2 The typical community with a low employment share in distributive and producer services declined by 13% while the typical community with a higher employment share in these sectors grew by nearly 20% from 1981 to 2006

The regression model confirms these patterns (Appendix Tables A3, A4 and A5). For rural communities with some employment in agriculture, holding other factors constant (including the index of specialization), a 100% higher employment share in agriculture, was associated with a 3.7 percentage points lower population growth rate.⁴ For instance, compared to the average rural Canadian community, with 18% employment in agriculture and about 2% population increase between 1981 and 2006, a community with identical characteristics but 36% employment in agriculture (100% higher employment share), was expected to experience a population decline of almost 2% over the same period.

For rural communities, a difference in the share of community employment in other primary industries had an even larger impact on population growth. Compared to the average rural community (with about 6% employment in other primary industry and about 2% population growth), a rural community with 12% employment in other primary industries was expected to experience about a 4% population decline (holding other factors constant).

The regional context reinforces these employment effects. Specifically, employment in the agricultural sector is typically dispersed across communities within a region, while it is less so for other primary industries where the employment is likely to be concentrated in specific communities. For rural communities, each 10% difference in the share of employment in agriculture in 1981 in their region is associated with a population growth that is almost 0.6 percentage points  lower between 1981 to 2006 (other factors remaining constant). This result mainly reflects employment multiplier effects across communities. For instance, a community with a relatively diversified economy located in a highly agricultural dependent region such as the Prairies, is likely to have been affected by the restructuring of the agricultural sector in the region, even though the community's workforce was not directly employed in agriculture.

For the large majority of communities that had some employment in distributive and producer services in 1981, the community effect is generally positive and statistically significant. For instance,  compared to the average rural community (with about 18% employment in distributive services and 4% employment in producer services), a rural community with about 36% employment in distributive services or 8% employment in producer services had an expected population growth of about 5 or 3 percentage points higher, respectively (holding other factors constant).

The role of agglomeration

Historically, Canada has been characterized by sparse population settlements, many of which were developed to harvest or mine natural resources. While this industrial structure has changed progressively over time, the surge of the "knowledge economy", which is largely an urban phenomenon, has added a new dynamic to this transformation. Urban agglomerations have acquired a key role in leading regional growth. Economic density, proximity to and the size of the urban core are all key indicators of the underlying process of agglomeration that is driving population re-location.

About three-quarters of the communities with lower population density (below 2.1 people per square kilometre in 1981) experienced a population decline over the 1981 to 2006 period (data not shown). In contrast, about three-quarters of the communities with higher population density (above 28 people per square kilometre) experienced population growth. All the communities with population density above 1,000 people per square kilometre recorded some population growth.

The relationship between density and growth is also strong when considering the density of the region in which the community is located, regardless of the density of the actual community. Over three-quarters of the communities that were located in regions in the top quintile with more than 75 people per square kilometre in 1981 experienced population growth between 1981 and 2006, while the opposite holds for communities located in regions with less than 16 people per square kilometre (in the lowest quintile) (Figure 3). However, note the variation within each quintile of regional population density. Not all communities in low density regions suffered population decline and not all communities achieved population growth if they were in regions with a high population density.

Figure 3 The typical community in a low density region declined by 22% while the typical community in a high density region grew by almost 40% from 1981 to 2006

Communities located at a greater distance from an urban core (Census Metropolitan Area (CMA) or Census Agglomeration(CA)) grew less over the 1981 to 2006 period than communities located in proximity to an urban core. For example, about three-quarters of communities within 30 kilometres from an urban core had positive, and often large, population growth (Figure 4). A strong negative gradient is evident as we move away from an urban agglomeration. Almost three quarters of communities that were located more than 80 kilometres away from an urban agglomeration of any size experienced population decline between 1981 and 2006.

Figure 4 The typical community in close proximity to an urban centre grew 19% while the typical community beyond 80 kilometres from an urban centre declined by 18% from 1981 to 2006

Beside proximity to an urban core, the size of the urban core is also positively associated with population growth of the surrounding communities or the urban core itself. About three-quarters of communities in proximity to an agglomeration of 10,000 to 20,000 inhabitants had a population decline or a marginal growth over the 1981 to 2006 period, while the same share of communities in proximity to agglomerations over 150,000 inhabitants experienced sustained population growth (Figure 5). Thus, both distance and size of the nearest CMA/CA matter in determining demographic population growth of a community. On average, communities in the shadow of larger urban areas experienced higher growth than communities in the shadow of smaller agglomerations.

Figure 5 The typical community in the shadow of small urban centres declined by 18% while the typical community in the shadow of a large agglomeration grew by 15% from 1981 to 2006

When the relationship between population growth and agglomeration factors is assessed in a multivariate framework, the significance of these relationships is further confirmed and quantified (Appendix Tables A3, A4 and A5). In a multi-variate framework, the agglomeration process appears best captured by proximity of the community to an agglomeration and the population size of the nearest agglomeration. However, the population density of the community, itself, has a negative association with population growth, suggesting a process of urban decongestion experienced by most of the urban agglomerations across Canada.

Distance to an urban core has a statistically significant effect both when small and large agglomerations are considered. Compared to the average community of Canada (located 264 kilometres from a large agglomeration), communities that are 10% further away from a large urban core (i.e. about an additional 26 kilometres away) had an expected population growth approximately 1 percentage point lower between 1981 to 2006 (other conditions being the same) (Appendix Table A3). In contrast, a community with similar characteristics but located 130 kilometres from a large urban centre (i.e. about a 50% reduction in distance) had an expected population growth about 5 percentage points higher. Distance to smaller metro areas (with a population of less than 500 thousand people) also has a significant effect on community growth.

The size of the nearest agglomeration matters. Compared to the average community (located in proximity of an agglomeration of 150,000 people), a community located near a centre that is 100% larger (i.e. 300,000 people) had an expected population growth approximately 1 percentage point greater, from 1981 to 2006 (other factors being the same) (Appendix Table A3). Thus, communities located in the shadow of large urban centres have, on average, benefited from the process of agglomeration in the core urban region. However, once agglomeration factors are accounted for, local population density appears to have a negative effect on population change, which can plausibly be explained in terms of a decongestion process captured by this indicator.

Agglomeration has both benefits and costs. Recently, economists have paid a great deal of attention to economic density, suggesting that growing population density is strictly correlated to a growth in productivity. For instance, a U.S. study indicates that doubling employment density increases average productivity by around six percent (Ciccone and Hall 1996). Higher employment density has also been linked to increased innovation due to external economies generated by the interactions among the skilled and experienced labour force (Jacobs 1969). These findings are the result of a variety of agglomeration economies. For instance, a firm, located in proximity to a research centre has the opportunity for face-to-face interaction with analysts, which may be only poorly substituted by long-distance communication in the early stage of development of an idea. Similarly, pooling together researchers in close proximity facilitates knowledge sharing and further development of new ideas. In a knowledge driven economy, these are increasingly important forms of agglomeration economies.

On the other hand, sociologists and urban economists have challenged the assumption that high density reduces the cost of service delivery, once a minimum threshold is reached (Ladd 1992). High density may also result in high social costs, such as long times spent commuting, environmental costs and increased crime rates, etc.

While the net economic effect of agglomeration remains a matter of debate, the forces that determined these trends in the past two decades are likely to persist in the foreseeable future. Agglomeration (i.e., urbanization) economies are now contrasting more than ever with the lack of agglomeration economies in thin rural labour markets; a challenge that rural communities have to face.

The role of economic diversification

Efforts to diversify the local economic base have been the cornerstone of local economic development programs and, according to the results supported by the model in this analysis, rightly so. Communities with a diversified industry composition at the beginning of the 1980s were able to grow faster than localities that were specialized in specific sectors, regardless of the sector of specialization.

Figure 6 shows the relationship between the index of economic specialization⁵ of a community in 1981 and the population growth experienced by that community during the following two decades. Low values of the index imply economic diversification, while higher values indicate economic specialization. In the highest quintile, that is the communities with the highest economic specialization, just over three-quarters of the communities experienced population decline.

Figure 6 The typical community with a more diversified economy grew by 11% while the typical community with a highly specialized economy declined by 22% from 1981 to 2006

Regional diversification matters. In fact, the relationship between growth and diversification is even more evident when communities are classified according to the economic specialization of the region in which they are located (Figure 7). Over three-quarters of communities in a region with a diversified economy in 1981 (a Herfindahl Index less than 0.167) had a positive population growth in the following two decades. This is opposed to the communities located in a region with a specialized economy (Herfindahl Index of 0.71 and over) where about three-quarters of the communities experienced a population decline.

Figure 7 The typical community located in a diversified regional economy grew nearly 20% while the typical community located in a specialized regional economy declined 20% from 1981 to 2006

A multivariate analysis reinforces these findings. Holding other conditions constant, communities that had more diversified economies at the beginning of the 1980s were more likely to expand their population base. Compared to the average community, a community with a specialization index that was 10% lower (i.e. a lower Herfindahl Index) could expect a population growth of about 2 percentage points higher between 1981 to 2006 (Appendix Table A3). Among all the factors considered, community population change was most sensitive to the level of community specialization (i.e. it had the largest coefficient). It should also be noted that this is the only factor, together with distance to major urban agglomerations, which showed highly consistent and significant effects across the macro-region models (Appendix Table A4).
 
Once other factors are controlled, the degree of regional economic diversification also plays a significant role in determining the population change trajectories of communities. When the entire sample is considered (Appendix Table A3), a 10% higher index of regional economic specialization is associated with about a 1.8 percentage point lower community population growth rate, other factors being the same.

This result supports the old wisdom of local development practitioners: community economic diversification is one of the main strategies for long term sustainability. Although several communities with specialized employment in specialized regions achieved high rates of population growth (note the height of the whiskers in Figures 6 and 7), the overall pattern shows that a diversified economic base and a diversified regional economy are key factors in shaping community population trajectories. Over the long-run, economic diversification is an asset that facilitates community adjustment to economic change, and increases the likelihood that the community will be able to maintain and expand its population base.

The role of human capital

A question that has attracted a great deal of attention has been the role of human capital in fostering local development. The focus on human capital has been boosted by the evidence of the steady process of knowledge intensification across all the sectors of the Canadian economy. For example, during the 1990s, higher skills occupations such as managerial and professional occupations have grown 36% and 17% respectively, in contrast to lower skills occupations which have grown generally less than 10% (Alasia and Magnusson 2005). It is not surprising then to observe that human capital indicators had a strong and positive association with long-term population growth of a community. 

The relationship between community population growth and local human capital is shown in Figure 8 and the relationship between community population growth and regional human capital is shown in Figure 9. The share of individuals with some post-secondary education in 1981 is used here as a proxy of human capital. For both graphs a clear positive gradient is evident, with communities in the bottom quintile of the human capital distribution experiencing mainly population decline over the two decades.

Figure 8 The typical community with a lower share of post-secondary graduates declined by 10% while the typical community with a higher share of post-secondary graduates grew by 16% from 1981 to 2006

Figure 9 The typical community located in a region with a lower share of post-secondary graduates declined by 12% while the typical community located in a region with a higher share of post-secondary graduates grew by 15% from 1981 to 2006

Human capital has been typically concentrated in urban regions. Hence, it could appear that human capital is proxying the effect of agglomerations. Nonetheless, when the relationship between population change and human capital is assessed in a multivariate framework, after controlling for agglomeration factors and other local and regional characteristics, human capital turns out to have an independent, distinctive and significant effect on population growth (Appendix Tables A3, A4 and A5).

Holding other factors constant, compared to the average Canadian community (32% of individuals with some post-secondary education in 1981), a community with an incidence of higher education that was 10% higher (that is about 35% of individuals with post secondary education) had about a 0.5 percentage point higher rate of population change over the 1981 to 2006 period.

A second relevant finding is that, along with the local human capital factor, the level of human capital in the region in which the community is located had a distinctive and significant effect on the growth perspective of a locality. In fact, for both rural and urban communities, being located in a region with a higher level of human capital had a larger effect on local growth than the human capital endowment of the community itself (Appendix Table A3). Other conditions being the same, a 10% higher share of individuals with some post-secondary educational attainment in the region was associated with between a 1 and 3.6 percentage point higher community population growth rate over the following 25 years.

These findings are consistent with other research on population changes, showing that a high concentration of human capital was associated with higher regional population growth. For instance, Glaeser (2005) shows that the number of colleges per capita in metropolitan areas is a good predictor of population growth. Metropolitan areas with twice as many colleges in 1940, compared to peer areas, witnessed four percent faster population growth per decade after 1970.

These findings suggest that, other conditions being the same, higher growth rates are associated with higher skill levels. As human capital becomes more important for firms operating in any sector, there are further incentives for firms to locate in regions where this input is abundant (Alasia and Magnusson 2005).

Other factors

The multivariate framework used to assess patterns of community population change includes some additional socio-economic and demographic variables. For example, the community labour force participation rate (as defined in Appendix Table A1) is positively associated with population growth, although only statistically significant in a few cases. A significant association would suggest that communities with a higher share of the population with jobs in 1981 had a higher population growth in subsequent periods.

Demographic characteristics did have some impact on the long term perspective of growth. For urban communities, a higher share of children (under 15 years of age) in the population in 1981 was associated with a higher subsequent population growth rate. For rural communities, a higher share of seniors in the population was associated with a high population growth rate after 1981.

Communities that attracted more young adults and more early retirees in the 5 years before 1981 reported higher population growth in the 1981 to 2006 period. The capacity to attract young and senior people appears to be a sign of robust demographic dynamics for a community.

Finally, urban communities with Aboriginals as a higher share of the population in 1981 grew less in the subsequent period whereas rural communities showed the opposite pattern – rural communities with a higher share of Aboriginals in 1981 grew more following 1981.

Communities and their regional context

A point that should be emphasized is that community population trajectories can be determined by local as well as regional characteristics. This analysis pays specific attention to this local/regional dimension by using a set of spatially-lagged indicators which, for each community, calculates the level of the characteristic for the region surrounding the community. These variables indicate an important distinction between community and regional effects (details are presented in Box 4).

For instance, a community may have a relatively small pool of human capital, but at the same time it may be located in a region with high levels of human capital, which can facilitate the community's capacity to stabilize demographic trends. Similarly, a community that has a relatively low share of employment in primary sectors may be located in a region with a high share of employment in these sectors. A typical example could be a small town in the Prairies surrounded by farming communities. Also, in this case, the regional context is likely to have a strong influence on community trends.

The results of the modeling analysis support this view. Although these findings should be considered as only a first step in this direction, they appear promising. In several cases, the effect of the regional indicator reinforces the community effect. This is the case of agricultural employment, human capital and, in some cases, economic specialization.

A key insight that should be emphasized has implications for both research and community development practice. Both analysts and decision-makers involved in community development should focus on communities and on the regional milieu of a given community.

Regional differences and outliers

The relationship between community population trends and explanatory variables described in the previous sections appears fairly stable across macro-regions of Canada (Box 2 defines macro-regions). There are some differences that should be noted. To investigate these differences, the same regression model applied to all Canadian communities was used for the communities within each of the five macro-regions of Canada.

For each macro-region, namely Atlantic, Quebec, Ontario, Manitoba / Saskatchewan, and Alberta / British Columbia, the statistically significant relationships between dependent and explanatory variables are similar to the patterns observed for the country as a whole; even though the magnitude of the coefficients vary to some extent, these appear also relatively consistent (Appendix Table A4). Restructuring and agglomeration forces are shifting population from rural and relatively remote areas, dominated by traditional sectors, to the vicinity of urban agglomerations dominated by dynamic sectors. However, agglomeration forces appear particularly relevant for the Manitoba / Saskatchewan macro-region.

Moreover, the model describes the behaviour of the communities of the four western provinces remarkably well. For these macro-regions the model explains 67% and 73%, respectively, of the observed variation in population growth. The estimate for Alberta / British Columbia shows a particularly high coefficient for community and regional educational attainment, suggesting that the critical mass of human capital in the early 1980s has been a key factor associated with the growth of the following 25 years. Overall, the results for the four western provinces suggest that the ongoing process of agglomeration might have a greater bearing in determining future population trends in this part of the country, as compared to other macro-regions such as Ontario and Quebec which have already achieved significant population densities in certain areas.

For the other macro-regions, the model does not fit as well as for the four western provinces, although the strength of the results is generally good for this type of a cross-section analysis.

Conclusions

Over the 1981 to 2006 period, population growth across Canadian communities was remarkably uneven. The economic restructuring of the Canadian economy has been paralleled by a significant spatial restructuring of population patterns, whose key feature was a steady process of agglomeration of population and employment in and around urban centres.

Rural depopulation trends raise concerns about the future viability of many rural communities. Population decline reduces the density of economic activities and poses further challenges to the economic sustainability of many communities. For small settlements, further downsizing makes it difficult to retain, let alone expand, basic services in the community and for the services that are retained, delivery costs may increase to unbearable levels. In the long run, this pattern of decline may threaten the quality of life of the population residing in these areas.

This bulletin discusses the factors associated with community population changes. Community leaders can use this information to conduct an assessment of their community and regional characteristics and develop realistic but pro-active initiatives for population stabilization that account for and build on local and regional assets. The main findings can be summarized as follows.

• Communities that had a higher share of employment in primary sectors and a poorly diversified economic base faced a steady population decline.
  • As capital intensification of primary sectors is likely to persist in the future, this trend is also likely to persist; rural communities need to find new commodities or services to export in order to maintain their employment and population base.
• Both proximity and size of the nearest urban agglomeration have an effect on community population growth.
  • Overall, population shifted from lower to higher density regions, although within the higher density regions growth was higher outside the urban core. The communities that had faster growth were those in proximity to urban areas, and among these, those located closer to larger urban agglomerations grew the fastest.
• Communities that started with a higher concentration of human capital had an advantage in terms of population dynamics.
  • Evidence indicates that skilled labour has concentrated in urban agglomerations over the past two decades. Expanding and improving local human capital to respond to the needs of a knowledge-intensive economy may remain a key strategy for any type of community.
• For each community, the regional context matters in determining local trajectories of growth.
  • This might appear trivial but the regional dimension in community-level analysis is in some cases overlooked. Instead, these results have implications for governance, development initiatives and research. Analysts should focus on communities in their regional context, such as rural communities in a rural region versus those in an urban region. The challenges and opportunities for similar communities in different regional contexts are substantially different.
• The relationship between community population change and the variables associated with population change appears to be stronger in western Canada.
  • Here, economic density and agglomeration size is still relatively modest in comparative terms. Larger agglomerations in western Canada appear to be far from congestion thresholds.
• The descriptive analysis shows that there are success stories within each group of communities.
  • Regardless of how we have classified communities, we (almost) always find some communities with population growth and some communities with population decline. There is a wide range in the size of community population change. Thus, some communities have "succeeded" and some communities have "not succeeded" with each of the groups we have portrayed.

Besides identifying broad forces of change (and with a focus on the average community in each group), this bulletin also highlights the variation of population growth performance (i.e. there are specific factors in these communities that we are not taking into account). Community development practitioners and researchers alike may have a lot to learn from the specific factors in these communities in order to determine whether these specific experiences can be replicated in another community.

Appendix tables

Appendix Table A1 The variables used in this study

Appendix Table A2 Descriptive statistics

Appendix Table A3 Population growth model: all communities, rural and urban, 1981 to 2006

Appendix Table A4 Population growth model: macro-regions, 1981 to 2006

Appendix Table A5 Population growth model: weighted regression, 1981 to 2006

---

Notes

1. Statistics Canada. CANSIM Table 282-0008. Labour Force Survey estimates (LFS), by North American Industry Classification System (NAICS), sex and age group. Other primary sectors include forestry, fishing, mining, oil and gas.
2. Statistics Canada. CANSIM Table 282-0008. Labour Force Survey estimates (LFS), by North American Industry Classification System (NAICS), sex and age group. "Professional services" include professional, scientific and technical services and "business services" include business, building and other support services.
3. In each of the charts, communities are grouped into quintiles – with one-fifth of the communities in each group. In Figure 1, communities are ranked by the share or percent of their 1981 employment in primary sectors. In 1981, the highest quintile (i.e. the one-fifth of communities with the highest share of employment in primary sectors) reported 33.5% or more of their workforce employed in primary sectors. Over three-quarters of these communities lost population between 1981 and 2006. This can be read from Figure 1 because the top of the "box" shows a population change (by looking at the vertical axis) of less than zero. See Box 3 to learn "How to read a box plot." One-half of the communities are within the box, one-quarter are above the box and one-quarter are below the box. The "whiskers" above and below the box show the range in outcomes (i.e. the range in the population growth rate from 1981 to 2006) for communities within each quintile. One conclusion that holds for every variable is that, regardless of the quintile class, some communities "succeed" (i.e. have higher outcomes) and some communities fail to succeed (i.e. have lower outcomes).
4. As explained in Box 4, the β coefficient for each variable in Appendix Tables A3, A4 and A5 is approximately equal to the "percentage point change" in the expected rate of population growth due a 1% change in the level of the given variable. In the third column of Appendix Table A3, the β coefficient for the share of agricultural employment in the community is -0.037. Thus, a rural community would have an expected population growth rate that is 0.037 percentage points lower compared to a community with a 1% higher share of employment in agriculture. For example, compared to the average rural community with an agricultural employment share of 18.27% (Appendix Table A2), a community with a 1% higher share would have 18.45% employed in agriculture (i.e. 18.27 plus 1% of 18.27). This community would have an expected population growth rate that was lower by 0.037 percentage points. Or, as indicated in the text above, a community with a 100% higher agricultural employment share (say, comparing the average community with an agricultural employment share of 18.27% with a community with a 100% higher share (i.e. 36.54%)) would have an expected population growth rate that was 3.7 percentage points lower. Again, looking at the average rural community population growth rate of 1.8% (Appendix Table A2), this community would have an expected population growth rate of -1.9% (i.e. 1.8% minus 3.7 percentage points).
5. The index of economic specialization used in this analysis is the Herfindahl Index (see Appendix Table A1).