Assumptions and scenarios

Warning View the most recent version.

Archived Content

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.


The data presented in this document do not constitute predictions; rather, they are part of an exercise designed to estimate how the ethnocultural diversity of the Canadian population might evolve in the future according to various scenarios of population growth. As it happens, the number of scenarios, and the assumptions underpinning them, is virtually infinite; it was therefore necessary to make choices. The choice of scenarios presented here was guided by the following considerations:

  • they had to be plausible based on the recent situation and past trends;
  • they had to reflect the inherent uncertainty of any forecasting exercise, since the future is by its very nature unknown;
  • insofar as possible, they had to be consistent with the assumptions and scenarios of other series of population projections disseminated by Statistics Canada;

Assumptions and scenarios have also been submitted to members of the scientific committee created for this project so they could comment them. When possible, assumptions were aligned with what was being proposed, at the time of this writing, on the upcoming edition of Population Projections for Canada, Provinces and Territories, 2009 to 2036 (Statistics Canada Catalogue no. 91-520). The assumptions of Population Projections for Canada, Provinces and Territories, 2009 to 2036 were also the object of consultations, in particular with provincial and territorial focal points and demographers.


A limited number of assumptions were ultimately selected (see Table 2).1 For a number of aspects of the projections, a single assumption was chosen, either because the simulated phenomenon changes little over time or because of the difficulty of creating or justifying an alternative assumption, often owing to the lack of available data. Thus, there is only one assumption for fertility differentials, parameters for assigning characteristics to newborns, mortality differentials, net non-permanent residents (NPR), emigration rates, emigration differentials, probabilities of a change in schooling level, parameters for marital status, departure of children from parental home and probabilities of a change from one religious denomination to another. On the other hand, there is more than one assumption for number of children per woman and age at maternity, life expectancy, the immigration rate and immigrants' places of origin, as well as interregional migration. The paragraphs below briefly describe, for the main components projected, the assumptions that were selected.2

Table 2 Key assumptions for projections of the diversity of the Canadian population, 2006 to 2031


Three assumptions were selected for the base risks of childbearing3 in the framework of projections of the diversity of the Canadian population, all based on the assumptions which, at the time this report was completed, were expected to be incorporated into Population Projections for Canada, Provinces and Territories, 2009 to 2036 (Catalogue no. 91-520). The three assumptions combine target figures for an average number of children per woman (a total fertility rate or TFR) and an average age at fertility. According to the medium assumption, a total fertility rate of 1.7 children per woman is reached in 2009, with an average age at fertility that is 0.4 years higher than in 2007 (when it was 29.75 years). With the low and high assumptions, the targets-respectively total fertility rates of 1.5 and 1.9 children per woman combined with an increase of 0.6 years in the average age at fertility for the low assumption and 0.2 years for the high assumption-are reached in 2013. In 2006 and 2007, age-specific fertility rates are aligned to Canadian vital statistics data.

Under the medium assumption, the choice of a target average number of children per woman that is slightly higher than in 2007 (1.70 compared to 1.66) is justified by the fact that preliminary data for 2008 and 20094 indicate a continuation of the increase in fertility observed in the past few years in Canada. Moreover, 1.7 is a number quite similar to the completed fertility rate for cohorts born in the 1970s, the most recent cohorts for which this indicator can be calculated. Also, the total fertility rate adopted for the medium assumption corresponds to a level that could be described as average when compared to the other G8 countries. Among these, three-France, the United States and the United Kingdom-have higher fertility (1.9 children per woman or more) and four-Germany, Italy, Japan and Russia-have lower fertility (1.5 children per woman or less).5 These differences among the G8 countries, like the differences among Canada's provinces,6 also guided the development of the high and low assumptions.7 The fairly rapid attainment of the targets has to do with uncertainty regarding the short-term evolution of fertility in Canada, with recent years having been marked by a sudden rise in fertility after a period of nearly ten years of stability. Also, the assumption of a continuation of the upward variation in the average age at fertility is intended to reflect the generation effect whereby fertility has occurred ever later from one generation to the next in Canada and continues to do so.

While the average number of children and the average age at fertility are important aspects of fertility, so is its ethnocultural composition, especially in these projections. Owing to the relative stability over time of the relationships between fertility and most variables used to model it, a single assumption as to the relative risks of childbearing was selected.8 Under this assumption, the fertility gaps between the projected groups would remain as observed from 2005 to 2006 until the end of the projection period. Thus, recent immigrants and members of the Arab and South Asian visible minority groups would exhibit higher-than-average fertility, while Chinese and Koreans would have lower fertility. Since the models used take account of socioeconomic variables (see the section on methods) such as marital status and education, it is clear that the changes that will occur in the composition of the projected groups as regards these variables may affect their propensity to have children and may thus cause their fertility to shift higher or lower.


The three assumptions selected as to the base risk of dying are identical to those expected to be used in Population Projections for Canada, Provinces and Territories, 2009 to 2036. Based on the trends observed from 1981 to 2006, they all assume a continuation of the rise in life expectancy between now and 2031 and a gradual narrowing of the life expectancy gap between males and females. For males, life expectancy would reach 81.7 years under the low assumption, 83.1 years under the medium assumption and 84.5 years under the high assumption, while for women it would reach 85.4, 86.6 and 87.7 years respectively. Despite the stability and regularity of life expectancy trends in Canada since World War II, it is impossible to predict with accuracy what life expectancy will be more than twenty years into the future. For this reason, three assumptions were selected rather than a single one, for this demographic component.

The mortality gaps between the projected ethnocultural groups are likely to have an effect on the future composition of the Canadian population. Therefore, they are taken into account by means of an assumption regarding mortality differentials. Based on the observations in the Canadian census mortality follow-up study, 1991 through 2001, mortality would notably be lower for the most recent immigrants and would then increase with the time spent in Canada. It would be lower for the most educated persons, lower for visible minority persons and higher for Aboriginals, for both males and females, even when controlling for the various socioeconomic variables included in the models (see the section on methods).9 The relationships observed between mortality and the selected variables are robust, in that they are in conformity with the literature and are consistent with the estimates obtained from the National Population Health Survey.10 Accordingly, the present projections are based on only one assumption regarding mortality differentials. Unlike in the case of life expectancy, these differentials are assumed to remain constant over time and therefore to remain unchanged until 2031.


Three assumptions on annual flows of immigrants were included in Demosim. For the years 2009, 2010 and 2011, the low, medium and high assumptions correspond to the objectives of the most recent immigration plan of Citizenship and Immigration Canada, which indicates the number of persons that Canada expects to admit in the short term. Thus, from 2009 to 2011, the number of immigrants will be 240,000 with the low assumption, 252,500 with the medium assumption and 265,000 with the high assumption. Starting in 2012 and up to the end of the projection period, the annual number of immigrants is drawn from immigration rates (annual number of immigrants in proportion to the total Canadian population) of 6.0 per thousand, 7.5 per thousand and 9.0 per thousand. These rates represent approximately the lowest rate observed in Canada from 1990 to 2008 for the low assumption, the average rate observed during that period for the medium assumption and the highest rate for the high assumption.11 From 2006 to 2009, the number of immigrants actually admitted was used.

The difference between the high and low assumptions may seem sizable, especially considering that the assumptions will come into force suddenly rather than gradually. The reason for this is that immigration, unlike, say, mortality (except, of course, in the case of war or epidemic), can increase or decrease suddenly and substantially from one year to another. One need only consider the changes in the annual number of immigrants in the 1980s, when that number went from less than 100,000, to close to 200,000 in just a few years.

The composition of immigration, which also changes over time, is in turn the object of two separate assumptions. With the medium assumption, the composition of the new cohorts of immigrants by country of birth is representative of the composition of the immigration observed from 2001 to 2006. Here, approximately 62% of immigrants would come from Asian countries; China and India alone are the place of birth of more than 29% of all immigrants to Canada. While this assumption appears to closely reflect the face of immigration to Canada since the early 1990s, there have been changes in this regard in more recent years, namely since the 2006 Census. Thus, in 2007 and 2008, the immigration of persons born in China and India has fallen off, now accounting for only 24% of all immigration, and there has been an increase in the proportion of immigrants originating from the Philippines and the Americas (including the West Indies and Bermuda) in particular. Insofar as these changes could possibly signal a lasting change in immigration to Canada, and since such a change could significantly affect the future composition of the Canadian population, a second assumption is in order. Under this second assumption, immigrants' places of birth would be representative of those observed in 2007 and 2008.


Once again, a concern for consistency with Population Projections for Canada, Provinces and Territories, 2009 to 2036 entered into the development of the assumption on base risks of emigrating. The present projections, unlike Population Projections for Canada, Provinces and Territories, 2009 to 2036, use the concept of net emigration rather than the separate components of emigration;12 however, the reference period for calculating the probabilities of leaving by age and sex is the same: 1991 to 2008. The rationale for using this reference period as the basis for the assumption selected is two-fold: we wanted to use an assumption based on medium-term observations, since the projections are also medium-term; and it was not possible to obtain a coherent data series beginning before 1991.13 The base risks of emigrating thus obtained are applied to projected populations starting in 2009, prior to which figures from Statistics Canada population estimates are used directly.

For differential emigration too there is only one assumption, since differences in emigration levels between the projected groups are sufficiently robust over time according to the Longitudinal Database.14 The assumption selected is based on data extending from 1995 to 2005 that show a stronger propensity to emigrate among recent immigrants to Canada, in particular those born in the United States, Western and Northern Europe and Far Eastern countries (primarily China, Japan and Korea). The differentials taken into account in the model are held constant over the projection period.

Internal migration

Two assumptions on internal migration were selected, each concerning the propensity to migrate, migrants' choices of destinations and the ethnocultural composition of internal migration. The medium assumption was created by combining observations regarding internal migration patterns for the periods 1995-1996, 2000-2001 and 2005-2006,15 while the assumption regarding recent migration is based solely on the period 2005-2006. The two assumptions are largely similar with respect to the composition of internal migration: each relies on the same consistent migration patterns, e.g., strong propensity of young persons to migrate to metropolitan areas; greater migration of recent immigrants compared to other immigrants; even greater migration of non-immigrants living outside their province of birth and very strong propensity of the latter to make a return migration; and lower probabilities of migrating among visible minorities, especially when they live in one of Canada's largest metropolitan areas.16

However, the volume of migration by region of origin and the choice of destination can vary considerably from one period to another. Furthermore, for some regions, internal migration is the main component of population growth, ahead of international migration and natural increase. In light of these considerations, alternative assumptions were selected. The medium assumption is based on several reference periods, an approach that offers two advantages: it reduces cyclical effects and draws on a larger sample. The recent assumption, for its part, enables us to estimate how projected populations would be affected if recent trends were to be maintained over a longer period. Based solely on observations from the 2006 Census, it also allows the effect of generation status on inter-regional migration to be taken into account. This was not possible with the medium assumption owing to the absence of information on place of birth of parents for observations prior to 2001.


Education is a key determinant of fertility, internal migration and mortality. Therefore, this socioeconomic characteristic is of great importance for the future composition of the Canadian population. A single assumption was selected regarding the probability to change level of schooling. This assumption suggests a progressive levelling-off of the trend towards an increase in the education level of the Canadian population up to 2031 and also maintains, again until the end of the projection period, the gaps between ethnocultural groups as estimated from information on education in the 2001 General Social Survey combined with data from the 2006 Census (see Methods section). Under this assumption, members of visible minority groups would generally have higher probabilities of graduating than the rest of the population. More specifically, the probabilities of obtaining a high school diploma would be higher for each visible minority group than for the rest of the population, while at higher education levels, this would hold true only for some specific visible minority groups. Aboriginals, by contrast, would continue to have lower probabilities of graduating than the rest of the population. The remarkable stability of these gaps over time justifies selecting a single assumption.17

Religious mobility (see Box 1)

These projections use a single assumption regarding mobility between religions, with such an assumption being essential to the projection of some religious groups. Under this assumption, the probabilities of leaving each religion and the choices of a new religious group would, as shown by the Ethnic Diversity Survey and a cohort-based analysis of decennial censuses from 1981 to 2001, be favourable to groups with no religion and Christians not included elsewhere.18 They would be unfavourable to all other groups, more especially Protestants and Catholics, which in the past have experienced the largest net losses of numbers through religious mobility.


With three assumptions for age-specific fertility rates, three assumptions on future life expectancy, three for the future number of immigrants, two possible formulations of the composition of immigration by country of birth and two internal migration schemas, 108 projection scenarios were possible. Of them, five were selected (see Table 3). Three are analysed in the following pages, while two are presented in the appendix only, in tabular form. Although these five scenarios are considered plausible, one should not be considered as more probable than the others, since the future is, by its very nature, unknown. Instead, readers are encouraged to consider the range of possibilities that these various scenarios offer.

Table 3 Composition of selected scenarios for projections of the diversity of the Canadian population, 2006 to 2031

The reference scenario, a low-growth scenario and a high-growth scenario are analysed in this document. The reference scenario combines medium fertility, life expectancy, immigration, an immigration composition representative of what was observed from 2001 to 2006, and medium internal migration. It serves to estimate what the population would be if the recent situation and trends were to continue in the coming years. The low-growth scenario differs from the reference scenario in that it instead combines low assumption on fertility, life expectancy and immigration, whereas by contrast, the high-growth scenario instead assumes high fertility, high life expectancy and high immigration. The latter two scenarios reflect uncertainty as to the future evolution of the Canadian population, and they shed light on what would happen if recent trends were to shift in such a way as to either slow or accelerate the growth of the population. Clearly, such a shift would affect both the size of the population and its ethnocultural makeup.

The two scenarios whose results are merely appended are both identical to the reference scenario except for one component, namely internal migration with the alternative internal migration scenario, and the composition of immigration with the alternative immigration scenario. The alternative internal migration scenario reproduces the inter-regional migratory patterns observed from 2005 to 2006 rather than those for 1995-1996, 2000-2001 and 2005-2006, while the alternative immigration scenario assumes that immigrants' countries of birth that will be added during the simulation will be representative of what was observed in 2007 and 2008 instead of 2001 to 2006. These two variants of the reference scenario yield two distinct portraits of the Canadian population of tomorrow, which illustrate the extent to which both its composition is sensitive to regional and international migratory phenomena.


  1. It should be noted that when observed data were available for reference periods subsequent to the starting point for projections, they were used directly or indirectly as alignment targets for the parameters for the applicable years (age-specific fertility rates for 2006 and 2007, number of immigrants for 2006 to 2009, increase in the number of non-permanent residents from 2006 to 2009, mortality rates by age and sex in 2006 and net emigration rates from 2006 to 2009). In this case, the assumptions described on the following pages began to be applied only in the year following the last year of observation.
  2. This description is complemented by the section that describes the methods used to project the components.
  3. The distinction between base risks and relative risks is explained in the section on methods.
  4. Compiled for selected provinces.
  5. See Population Reference Bureau (2009)
  6. At the provincial level, fertility gaps are observed similar to those exhibited by the G8 countries, with fertility being equal to or greater than 1.9 children per women in the Prairie Provinces and lower, at approximately 1.5 children per woman, in Newfoundland and Labrador, Nova Scotia, New Brunswick and British Columbia. Quebec, Ontario and Prince Edward Island are closer to the national average (Statistics Canada (2009)).
  7. Other reasons may also be cited to justify these alternative assumptions. As regards a possible increase in fertility, note the recent increase in the total fertility rate, not only in Canada but also in several other G8 countries. On the other hand, various social changes often associated with declining fertility have probably not run their course. Among the main manifestations of these changes is the increase of common-law unions as a form of conjugality, the increased education of women and the fall-off in religious attendance. However, it is not outside the realm of possibility that the relationship between fertility and these phenomena will change over time.
  8. The results of the analyses that were conducted in order to make these projections are quite similar to those obtained by Caron Malenfant and Bélanger (2006) and Bélanger and Gilbert (2003).
  9. Also see Wilkins (2008).
  10. This is the source that was used to develop the relative risks for Population projections of visible minority groups, Canada, provinces and regions: 2001 to 2017. Despite its limited sample, it too showed that immigrants' mortality is lower during the first years after they settle in Canada and that it tends, with time spent in Canada, to converge with that of the rest of the population.
  11. We say "approximately" because in reality, the average was 7.6 per thousand with extremes of 5.8 per thousand (in 1998) and 9.0 per thousand (in 1992 and 1993).
  12. Net emigration is the number of emigrants minus returning emigrants plus net temporary emigration.
  13. Before 1991, one of the components of net emigration was not estimated, namely the net number of persons temporarily abroad.
  14. They are also consistent with the observations from the Reverse Record Check survey, which were used in developing the assumption used for Population Projections of Visible Minority Groups, Canada 2001 to 2017.
  15. It should be noted that a constant geographic structure-the one for 2006-was applied to the database that was used to create the medium assumption.
  16. See, for example, Dion and Coulombe (2008).
  17. See Spielauer (2009).
  18. These are Christians other than Catholics, Protestants and Orthodox Christians. This group  includes persons who indicate "Christian", as well as those who indicate "Apostolic," "Born-again Christian" and "Evangelical" (Statistics Canada, 2003 (1), op. cit.) This group stood out between 1991 and 2001 by the sizable increase in its population.
Date modified: