Information identified as archived is provided for reference, research or recordkeeping purposes. It is not subject to the Government of Canada Web Standards and has not been altered or updated since it was archived. Please "contact us" to request a format other than those available.
|Population Projections for Canada, Provinces and Territories
Section I : Methods and assumptions
As in previous population projections, the general projection method used here is the components method applied to the provinces and territories. This method is based on a demographic accounting system. The starting point of the projections is the age-sex distribution of the population in the base year. Survival ratios and fertility, immigration and emigration rates are applied to this population to generate the projected number of deaths, births and migrants, which are then either added or subtracted, as the case may be, to obtain the population for the following year. 1
In order to produce consistent and comparable projections for Canada, the provinces and territories simultaneously, assumptions are formulated and applied to the provinces and territories. The sum of the populations projected at this geographic level generates the population projected at the national level. Thus, the model creates a separate projection of each component at the provincial/territorial level that takes regional differences into account.
The base population for these projections is established using the official postcensal estimates of the population of Canada on July 1st, 2005 (The Daily, October 26th, 2005). These estimates are based on the 2001 Census, adjusted for net undercoverage. They are available by age, sex, by provinces and territories up to the open age group 90 years and over. However, the results of these projections are available by single year of age up to 99 years. The year-of-age distribution of the population aged 90 and over in the 2001 Census was used to distribute the population of the open age group from the estimates.
The fertility assumptions are based on a detailed analysis of trends in the completed fertility rate and total fertility rate (TFR) for Canada as a whole and then for each province and territory considered separately. They also take fertility trends in other industrialized countries into account.
As in the past, three fertility assumptions are formulated (low, medium and high), each combining an assumption regarding the distribution of age-specific rates and an assumption on how the total fertility rate will evolve until the end of the projection period. The methods used to determine the tempo and intensity of fertility are applied in the same way for all provinces and territories except Nunavut.
Rationale for assumptions
Like most industrialized countries, Canada experienced a major drop in its fertility starting in the early 1960s. This drop was all the more remarkable in that it followed the period of high fertility known as the baby boom (1946-1965). While the number of children per woman in Canada, as measured by the total fertility rate, was close to 4.0 around 1960, it dropped below the replacement level (2.1 children per woman) in 1972 and has remained below this level ever since. The downward trend has continued almost uninterrupted for forty years 2, even though the recent trend indicates that the rate has stabilized around 1.5 children per women since 2000.
Text table 1.1
The future evolution of the age structure of the Canadian population and, consequently, the future trends in aging depend primarily on the how fertility evolves. It is accordingly important to ask when this decline will cease and at what level. Are there reasons to believe that it will end soon? Has it already ended? Of course, it is impossible to predict the future fertility behaviour of Canadian women. However, it is possible to formulate various assumptions that appear plausible in light of trends in some of the determinants of fertility.
Several factors generally considered to influence fertility appear likely to play a role in maintaining downward pressure on it in the coming years: childbearing is still occurring later and later in life; the propensity to marry continues to decline while common-law unions gain in popularity; religious observance seems likely to continue its decline; and women are increasingly educated.
Furthermore, the total fertility rate is reaching very low levels in a number of other Western countries (Text table 1.1 ), showing that a fertility rate lower than currently prevails in Canada is possible. In 2002, for example, the rate was 1.32 children per woman in Russia, 1.31 in Germany and 1.26 in Spain and Italy (Sardon, 2004). Some Canadian provinces, such as Newfoundland and Labrador and British Columbia, are also experiencing a fertility rate below 1.5 children per woman.
Conversely, two closely related factors might contribute to an eventual rise in fertility: the growing proportion of immigrants from high-fertility regions of the world, and the growing proportions of persons belonging to visible minorities in Canada, whose fertility is greater than that of other women (Bélanger and Gilbert, 2003; Bélanger and Caron Malenfant, 2005). If the fertility of immigrant women and women belonging to visible minorities were to remain higher than that of other women, a continuing increase in the percentage of the population belonging to one or the other of these groups could exert upward pressure on the fertility of Canadian women.
Furthermore, a number of countries have higher fertility rates than Canada. This is the case with France (1.88), New Zealand (1.95) and Australia (1.76). For a number of these countries, the current levels were reached following a reversal of a long downward trend. And in the United States, the fertility rate has been hovering around 2.0 children per woman for nearly ten years.
To the extent that these trends, which are likely to exert either upward or downward pressure on fertility, continue to develop in the coming years, the relative importance of each of them might determine the future course of fertility in Canada. Moreover, other factors might come into play that have not been considered here. In such a situation, it is appropriate to develop more than one assumption on the future course of fertility.
Assumptions on quantum and tempo
High assumption: according to the high assumption, the fertility rates by year of age observed in 2002 (the last year of data available when the assumptions were being developed) will increase linearly until they reach, in 2016, the rates for the last cohort of women for which the completed birth rate can be estimated with a low risk of error, namely the cohort of women born in 1973. Canada’s total fertility rate would go from 1.51 children per woman in 2002 to 1.69 in 2016, when it would become equal to the completed fertility rate of Canadian women born in 1973. Subsequently, fertility rates would remain constant until the end of the projection period. Thus, at the national level, the average number of children per woman under the high assumption would be comparable to the highest level reached by the total fertility rate since the end of the 1970s (1.7 in 1978 and 1979 and then from 1990 to 1992). For Nunavut, the assumption of an increase in fertility seems unlikely, since the fertility level there is already very high in relation to the other provinces and territories. For that reason, the high assumption for Nunavut is equal to the rates observed in 2002. At the provincial/territorial level, total fertility rates would range between 1.46 children per woman in Newfoundland and Labrador and 3.03 children per woman in Nunavut.
The high fertility assumption would entail a rise in fertility rates for practically all age groups, although the increase would be greater among women under 27 years of age. As a result, the age pattern of fertility would shift downward, with the mean age at childbearing estimated to drop to 28.7 years for Canada as a whole.
Medium assumption: For some ten years, fertility has been changing slowly in Canada. As in previous projection exercises, the total fertility rates under the medium assumption are set at the level at which they stood in the year of the most recent vital statistics data available when the assumptions were developed (2002) and are maintained at that level throughout the projection period. For Nunavut, the medium assumption was redefined and was obtained by calculating the average between the high assumption and the low assumption. According to this assumption, the total fertility rate for Nunavut would be 2.73 children per woman in 2016. Nunavut would thus remain the territory with the highest fertility. Newfoundland and Labrador, with a total rate of 1.31 children per woman, would continue to have the lowest fertility in Canada. For Canada as a whole, the number of children per woman would be approximately 1.51 according to this assumption, which is essentially the same level as in previous projections.
In the medium assumption, in accordance with what was assumed for the total fertility rate, the tempo of fertility in each of the geographical units would remain unchanged. As a result, the mean age at childbearing would also remain constant at 29.2 years.
Low assumption: considering that in the past decade, fertility has decreased in all provinces and territories, the total fertility rates, according to this assumption, result from extrapolating recent trends (1993-2002 3) in age-specific fertility rates at the provincial/territorial level to the year 2016. Thereafter, fertility rates are held constant until the end of the projection period. Since the linear projection of declining rates leads—sooner or later, depending on the scenario considered—to values lower than zero, it was decided to project each of the rates by year of age using asymptotic equations. This resulted in a gradual stabilization of the age structure of the fertility rates, since the rate of decrease of declining rates and the rate of increase of rising rates gradually slow with the result that each of the rates tends toward a theoretical limit without ever reaching it (with the lower limit set at zero). Also, the rate of decrease of fertility was doubled so that a total fertility rate of 1.30—the same fertility level as in the low assumption in the previous projection exercise—would be reached in 2016 at the national level. At the provincial/territorial level, the total fertility rate would range between 1.10 children per women in British Columbia and 2.43 in Nunavut.
With fertility rates following a downward trend among the youngest women and an upward trend among older women, the application of this method generates a older fertility tempo than in 2002. During the projection period, the curve of age-specific rates would become increasing skewed toward the right (chart 1.2 ). Ultimately, the mean age at childbearing would be approximately 31.1 years in Canada as a whole.
Text table 1.2
Text table 1.3
Text table 1.2 shows in detail the total fertility rates attained in 2016 and then held constant for the provinces and territories.
Canadian males and females have one of the highest life expectancies at birth in the world (text table 1.3 ). In 2002, Canadian males were outperformed in this regard only by Icelandic, Japanese, Swedish and Swiss males; Canadian females, by Japanese, French, Icelandic, Spanish and Swiss females.
Rationale for assumptions
Projecting mortality is both simple and complicated. It is simple because the trends observed over the course of a century are clear; life expectancy at birth has almost always risen from one year to the next, even though the rate of increase has varied. It is complicated because no one knows the limits of human longevity and no one can foresee the future advances that might influence it. Indeed, the scientific literature is divided concerning future gains in life expectancy. Some researchers, who expect that life expectancy at birth will continue to rise rapidly in the coming decades, set no limit on that rise in the short run (Oeppen and Vaupel, 2003). Others, pointing out that average annual gains are slowing, argue that we are approaching the limits of the life expectancy of a population and that there will henceforth be little additional growth (Olshansky et al., 2001). Still some researchers have even recently suggested that a decrease in life expectancy at birth is not out of the question because of the growing prevalence of various health problems such as obesity (Olshansky et al., 2005). For these reasons, three assumptions for mortality were elaborated.
Method of projecting mortality
Since 1992, the method used has been based on the model proposed by Lee and Carter (1992), which is often used throughout the world to project mortality because of its relative simplicity. This model consists in parameterizing the past trend of age-specific mortality rates and then, by means of a statistical model, projecting the time series of one of the parameters. In the last two exercises, Statistics Canada projected national life expectancy using this method, then derived provincial and territorial life expectancies from it by applying ratios calculated by dividing the life expectancies of each province and territory by the national life expectancy observed in recent years. It was therefore assumed that the differences observed between the provinces and territories would hold throughout the entire projection period. The use of ratios was necessary because applying the Lee-Carter method to each of the provinces led to a future divergence in provincial/territorial life expectancies in relation to the national average, a divergence deemed less probable since it has not been observed in Canada for more than 30 years.
For this series of projections, a new method of projecting mortality was used. This method, developed by Li and Lee (2005), is used to project the future course of life expectancy for each province without generating a divergence in relation to the national average. It thus avoids having to rely on province-to-Canada ratios and uses all the data available (mortality rates by provinces and territories) to project the future course of mortality.
As in the preceding population projections, the mortality of Yukon, the Northwest Territories and Nunavut was projected using the ratios method, since in their case, random variations—sometimes sizable ones owing to the small populations involved—limited the use of the method for the provinces. Yukon and the Northwest Territories, which have a similar mortality profile, were grouped together. The mortality of Nunavut was modeled separately, since that territory has a lower life expectancy than the other two.
The high and low assumptions are based on the 95% confidence intervals of the ARIMA model used to project the mortality trend parameter for each province.
Text table 1.4 shows life expectancies at birth for Canada, the provinces and territories observed in 2002 (last year available when the assumptions were developed) and projected for 2031, by sex for the three assumptions. Chart 1.3 shows the life expectancy of Canadian males and females since 1971 and projected to 2031.
Text table 1.4
According to the medium assumption, life expectancy at birth would reach 81.9 years for Canadian males and 86.0 years for Canadian females in 2031. This is a gain of 4.7 years for males and 3.8 years for females over the period, representing an average annual gain of approximately 0.14 years and 0.12 years for males and females respectively. This assumption is slightly more optimistic than the one used in the previous population projections, which assumed levels of 80.0 and 84.0 years in 2026 for males and females respectively. Just as in 2002, Newfoundland and Labrador would be the province with the lowest life expectancy in 2031 and British Columbia the one with the highest. The gap between the province with the highest life expectancy and the one with the lowest would remain stable at around 2.7 years for males and would drop from 2.0 to 1.6 years for females.
According to the medium assumption, in 2056, life expectancy at birth for Canadian males would reach 85.0 years and for Canadian females it would reach 88.6 years.
The low assumption would put the life expectancy of Canadian males at 81.1 years and that of Canadian females at 85.3 years in 2031. The gains would therefore be more modest, with life expectancy growing only 3.9 years for males and 3.1 years for females over the next three decades. The high assumption would put life expectancy at birth for Canadian males at 82.6 years and for Canadian females at 86.6 years in 2031. According to this assumption, life expectancy would increase by 5.4 years for males and 4.4 years for females over the same period.
All scenarios developed assume a reduction of the gap — in favour of females — between the life expectancy of males and females, extending the trend observed in Canada since 1979. For example, according to the medium assumption, this gap would go from 5.0 years in 2002 to 4.1 years in 2031 for Canada.
Text table 1.5 shows a few international comparisons with the projections made by other statistical agencies of developed countries that currently have a life expectancy at birth similar to that of Canada and a few Canadian provinces. It compares the medium assumption to the one formulated by these countries for their respective projections. As may be seen, the medium assumption is consistent with those of these countries, with the positive or negative gaps generally being maintained over the projection period. For example, French females would have a life expectancy at birth of 88.3 years in 2030, compared to 86.0 for Canadian females; in 2002, the life expectancy of these two groups was respectively 82.9 years and 82.2 years. On the other hand, American females would still, in 2025, have a life expectancy at birth lower than that of Canadian females (83.6 years compared to 85.4 years respectively), maintaining the gaps observed in 2002.
International immigration is playing an increasingly important role in Canadian population growth. In Canada, as in many European countries, fertility has been below the replacement level for more than three decades. Thus, population growth cannot be maintained without relatively sustained immigration.
Rationale for assumptions
Canada might increasingly look to immigration to lessen some consequences of the inevitable aging of its population. In 2011, the first baby-boom cohorts will reach age 65, and the growth of the working age population will slow and might possibly become negative during the following decade unless there are fairly high levels of immigration. At the same time, the number and proportion of persons aged 65 years and over will increase at an accelerated rate, thus exerting increased pressure on the public pension and health care systems. Canada is one of the countries where the baby-boom was the most pronounced. Today, this is leading to a rapid aging of its working age population, and it will greatly increase the number and proportion of persons withdrawing from the labour market after 2010. Increased productivity, an extension of the retirement age or a rise in the participation rate could also reduce some of the effects of the expected aging and stagnation of the labour force.
Text table 1.5
According to some authors, international immigration is one of the most difficult components to project (O’Neill et al, 2001; Lutz et al., 2004; Howe and Jackson, 2005), especially since there is no convincing and generalized theory, owing to the importance of national policies in this area and the special considerations of each country. Although immigration is relatively stable at relatively high levels for some fifteen years in Canada, the past evolution of annual immigrant contingents shows that a reversal of trend in this area can be both sudden and sizable. In 1985, Canada received 84,000 immigrants, and seven years later, in 1992, it received 255,000. It thus appears necessary to propose more than one assumption on how this component will evolve in the future. As in the past, this population projection exercise proposes three such assumptions, using Citizenship and Immigration Canada data.
According to the assumptions of the current projections, and unlike those in past projections, the annual number of immigrants is assumed to evolve in relation with population growth by assuming a constant immigration rate until 2031 rather than a constant number of immigrants (chart 1.4 ). In the short run, the proposed levels are justified by the relative stability observed since 1990 and by indications of the most recent immigration plan by Citizenship and Immigration Canada that sets the objectives for 2006 at a number of immigrants between 225,000 and 255,000.
The medium assumption is that the immigration rate would reach 7.0 per thousand in 2010 (five years after the start of the projections). It is framed by a high assumption according to which the rate would reach 8.5 per thousand and a low assumption whereby the rate would decline to 5.5 per thousand, with these levels also to be reached in 2010. Between 2005 (base year) and 2010, immigration rates are interpolated linearly between these levels and the observed rate in 2004, in order to ensure a smooth transition between the projected numbers and estimates at the start of the period.
After 2010, the immigration rates under each assumption remain constant until 2031. As a result, the range in the number of immigrants between the two extreme assumptions tends to widen, which is consistent with the idea that uncertainty increases with the length of time covered by the projection. According to these assumptions, the number of immigrants would, in 2031, be approximately 204,000, 280,000 and 364,000 respectively.
In 2031, the last baby-boomers will have reached age 65 and the size of the working age population should fluctuate much less. In the very long run, it is therefore assumed that the immigrant numbers reached in 2031 according to each scenario will remain constant until 2056, the time horizon for the projections made for Canada as a whole.
The provincial distribution of immigrants is projected on the basis of the rates observed in 2003 and 2004 (chart 1.5 ), adjusted for certain provincial particularities to better reflect recent agreements on provincial immigration objectives. It is held constant over the entire projection period. This distribution assumes that approximately 90% of the immigrants would settle in Ontario, Quebec and British Columbia, as it has been the case over the last 20 years. Text table 1.6 shows the number of immigrants received between 1981 and 2031.
Text table 1.6
In the past, the age-sex distribution of immigrants was equal to the average of the four most recent years and was held constant for the entire projection period. In light of the small variations observed in the past, this assumption of stability has been maintained for the current analysis. However, the distribution has been updated and the four most recent years available (2001-2004) are used to calculate it.
The age distribution of immigrants varies from one province to another, partly because of variations in the proportions of immigrants of different classes. For Quebec, Ontario, Alberta and British Columbia, where the great majority of immigrants settle, the distribution for each province is used. On the other hand, because of the small number of immigrants settling in the Atlantic provinces, Manitoba, Saskatchewan and the territories, a single age distribution is used for all these regions.
Emigrants are Canadian citizens or landed immigrants who have left Canada to settle permanently in a foreign country. Their number is estimated primarily by using Canadian or U.S. administrative data.
In this projection exercise, the assumption adopted takes the average emigration rates by age, sex and province observed over the last five years (from 1997-1998 to 2001-2002) and holds them constant over the entire projection period. The number of emigrants generated by the projection model would therefore increase from year to year concomitantly with the growth of the Canadian population, going from approximately 45,000 persons in 2004-2005 to approximately 55,000 in 2031.
Returning emigrants are Canadian citizens or landed immigrants who emigrated from Canada and return to settle there. Based on administrative data, their number is estimated at approximately 18,000 persons annually since 1997. The vast majority (approximately 90%) of these returning emigrants are concentrated in the four large Canadian provinces of Ontario, British Columbia, Quebec and Alberta.
The assumption adopted for this projection exercise is based on the relationship that exists between returning emigrants and past emigration, since the Canadians likely to return to Canada being those who emigrated in the past. Thus in these projections, the annual number of returning Canadians is obtained by applying a return migration rate to the number of emigrants generated by the model. The rate used is 38% of emigrants. This percentage is based on the most recent estimates available concerning both the number of returning Canadians and the number of persons emigrating from Canada.
The distribution by age, sex and province is obtained from estimates of the population as of July 1st, 2004 and is held constant over the entire projection period. 4
Persons temporarily abroad
These persons include Canadian citizens or landed immigrants who are living abroad temporarily and no longer have a usual place of residence in Canada. The data on persons temporarily abroad comes from the Census’s Reverse Record Check.
The data available on this component yield a net figure, rather than a number. That figure is the result of two flows, namely persons leaving (temporarily) and returning to Canada. During the recent period, this net figure was estimated at 26,000 annually. The great majority of losses were registered in Ontario, British Columbia, Quebec and Alberta.
The assumption adopted for this projection exercise consists in holding this figure constant at 26,000 per year until 2006, the next census year. Subsequently this net figure will be reduced linearly to obtain, in 2011, the average figure since 1991, namely 22,000. The distribution by age, sex and province is obtained from recent data on population estimates, and it is held constant over the entire projection period.
Non-permanent residents include the following persons: persons residing in Canada who claim refugee status, persons residing in Canada who hold a student visa, a work permit or a ministerial permit, as well as all dependents of these residents who are born outside Canada and reside in Canada. The data on non-permanent residents come from Citizenship and Immigration Canada.
The assumption adopted for this projection exercise holds the number of non-permanent residents constant over the entire projection period at a level equal to 390,000 persons per year, the level for the last year for which data are available (2005). This level is very close to the average for the last three years. Thus, a zero balance is assumed between persons entering this population and those leaving it. The distribution by age, sex and province is also held constant over the entire projection period (Text table 1.7 ).
Text table 1.7
Interprovincial migration is very volatile, with large fluctuations in space and time. While there are some general trends, such as the westward movement of population, the most attractive destinations for migrants have not always been the same. Fluctuations can take place very quickly. From one year to the next, traditionally positive net migration for a province can turn negative, and vice versa. With the projected decline in natural increase, interprovincial migration is certain to become a more important factor in population growth for several provinces. Hence it seems vital to develop more than one interprovincial migration scenario.
Rationale for assumptions
Conventionally, assumptions are based on past migration patterns. This approach produces plausible assumptions. As was the case in previous projection exercises, the assumptions used here are based on past trends and take the most recent data into account. In addition, the provincial and territorial statistical agencies were consulted, and they provided comments on the proposed assumptions for each component and in particular on interprovincial migration patterns.
Analysis of the data for the 1971-2003 period leads to the following
observations. Over the entire period, Alberta and British Columbia gained
the most from interprovincial migration. Their interprovincial net migration
was positive 70% of the time, averaging 11,200 and 15,000 per
year respectively. Quebec, on the other hand, has had negative net migration
every year since 1971 and has lost the most population through interprovincial
migration (an average of
Ontario’s situation is more complicated. Its net migration was positive 50% of the time, and over the entire period, it gained an average of 3,500 people a year. For the three Maritime provinces, net migration is sometimes positive (more often for Prince Edward Island) and sometimes negative (more often for New Brunswick). While their net migration is sometimes large relative to their population size, over the long haul, the gains and losses nearly offset each other. The three territories also have migration losses more often than gains (60% or 70% of the time). The internal migration assumptions must take account of both these long-term migration patterns, which are partly due to structural effects, and the high volatility of interprovincial migration.
A number of factors account for the sometimes abrupt changes in interprovincial net migration. Some are of a cyclical nature (economic recessions, for example), and others are single events (moratorium on cod fishing, referendums in Quebec), though their effects can last over a number of years. Migration patterns in the 1970s and early 1980s were influenced by the political situation in Quebec, the oil shocks and a serious economic recession. Canada’s population was also much younger, and a larger proportion was likely to migrate in response to an unfavourable situation. More recently, migration patterns seem to have changed for several provinces.
Interprovincial migration assumptions
Since there is considerable uncertainty concerning future interprovincial migration patterns, we developed four assumptions. The projection model is based on age- and origin-specific out-migration rates for each province of origin. Those rates are estimated from data for each of the three reference periods selected and are fixed for the entire projection period. Origin-destination matrix is then used to distribute out-migrants between the other provinces and territories.
The central-west assumption is based on migration patterns observed between 1996 and 2000. This period is characterised by the growth of the information technology sector that favorised the Ontario economy and by slow economic growth in Bristish Columbia. It favours both Ontario and Alberta. It is also the most favourable assumption for Saskatchewan even though the province had losses averaging 4,200 people during the period. On the other hand, it is the least favourable assumption for the Atlantic provinces, Quebec, British Columbia and the three territories.
West coast assumption
The west coast assumption, based on data for the 1988-1996 period, makes British Columbia the most common destination for interprovincial migrants, as the province’s average annual gain through interprovincial migration was 34,000 during the period. Economically, the period was marked by the recession of the early 1990s, which was particularly hard on the manufacturing sector and the central provinces’ net migration figures. British Columbia was less seriously affected, benefiting from the steady expansion of emerging Asia-Pacific economies.
Under this assumption, Alberta continues to have migration gains even though it is the least favourable of the four assumptions for the province. On the other hand, of the three reference periods used in developing the assumptions, this is the one in which Ontario, Manitoba and Saskatchewan suffer their largest losses. This assumption is also generally favourable for the Atlantic provinces and Yukon. Although it is not the best scenario for Newfoundland and Labrador, the latter’s losses during the period are of the same magnitude as its losses during the period associated with the most favourable assumption. Quebec’s average losses during this period were about 10,000.
Recent trends assumption
The third assumption is based on the most recent interprovincial migration trends available at the time of the study (for the 2000-2003 period). In a number of respects, the migration patterns seen in recent years form a new pattern, different from what had been observed since 1971. Quebec’s net migration, though still negative, has been improving rapidly since 1999-2000, and in 2002-2003, it was better than it had been at any other time during the period considered. The same is true for Newfoundland and Labrador, Nova Scotia and New Brunswick, whose net migration figures have been improving. For all of them, the 2002-2003 figures are the best they have seen since 1991-1992 (1993-1994 for New Brunswick). The figures for Manitoba, Saskatchewan, British Columbia and the territories are also becoming more favourable. These changes have cut into Alberta’s net gains, which in 2002-2003 were only half what they were three years earlier, and especially Ontario’s gains, which plunged in just three years from 22,400 in 1999-2000 to a mere 600 in 2002-2003. Preliminary data for 2003-2004 give no hint of a reversal of what could be a new trend, though it is too early to say for certain.
Compared with the net migration figures for the two other reference periods considered, the figures for this period are most favourable for Newfoundland and Labrador, Quebec, Manitoba, the Northwest Territories and Nunavut. For all other regions, the figures for this period are between the figures for the other two periods concerned. Nevertheless, Ontario and especially Alberta remain the most attractive destinations for interprovincial migrants.
A fourth migration assumption was developed by averaging the “recent trends” and “west coast” assumptions. This new, “medium” assumption covers a larger number of short-term and long-term eventualities, as interprovincial migration patterns have been unstable over the last two decades.
Projected net migration
Net migration data for the provinces are presented in charts 1.6 , 1.7 , 1.8 and 1.9 . The charts clearly show the differences in net migration that can arise when constant rates are applied to population structures that change over time.
Just as water poured into a system of interconnecting vessels will tend to distribute itself evenly across all the vessels, the high population growth in provinces that traditionally gain through interprovincial migration will inevitably produce proportionally more out-migrants, whereas the relatively low (if not negative) growth in provinces that traditionally have net migration losses will inevitably produce proportionally fewer out-migrants. Consequently, the net migration gains of high-growth provinces tend to decrease, leading to smaller losses or even gains for the low-growth provinces.
Thus, even though all of the Atlantic provinces except Prince Edward Island have net migration losses for the three reference periods considered, the net migration figures obtained by applying the out-migration rates to the projected populations up to 2031 improve over the course of the projection period, eventually becoming net gains for each province prior to 2031. Conversely, Ontario would have a net loss in 2031 under the four assumptions, even though it had large net gains during the reference periods for central-west assumption (1996-2000) and “recent trends” assumption (2000-2003).
Under the four assumptions, Quebec would continue to have negative net migration up to 2031, but the losses in 2031 would be much smaller than the losses during the corresponding reference period. The same is true for Manitoba. Saskatchewan’s situation would improve as well under all of the assumptions, and one of them would produce a net gain before 2031. On the other hand, British Columbia’s rapid population growth would lead to relatively large net losses under assumptions central-west and “recent trends”.