How Population Dynamics are Measured in Between Censuses and Projected for Future Reference Dates

2007

Date Source: 

Statistics Canada

Organizer: 

André Cyr, Hubert Denis, and Éric Caron Malenfant

Introduction

Every five years, Statistics Canada conducts a national census of population, through which a wide range of demographic data is collected on the Canadian population. The census collects information on Canadian citizens, landed immigrants and non-permanent residents who have a usual place of residence in Canada. Unlike some countries, Canada does not have a system of continuous population registration from which to obtain basic demographic data on the state and movement of the population for non-census years.
 

To fill this gap in the information system, Statistics Canada has developed a Population and Family Estimates Program that uses census data collected by Statistics Canada and administrative data provided by other government departments to estimate the Canadian population between censuses. Moreover, Statistics Canada has developed a Population Projections Program. Based on extrapolations of past trends, these projections reflect informed hypotheses of Canada’s demographic future.
 

As part of its regular work program, Demography Division continuously monitors the various components of change in the population: fertility, mortality, international and inter provincial migrations. The components of changes serve as a basis for producing the official Canadian population estimates and projections.
 

Among the demographic trends revealed by those figures, two are particularly salient: the slowing down of population growth and population ageing. Moreover, regional disparities within Canada are noticeable. In fact, Atlantic provinces, and more specifically Newfoundland-and-Labrador, have the oldest populations in the country and have recorded, for several years now, the lowest demographic growth.

Case Study Objectives

Throughout this case study, participants will use population estimates and projections produced by Statistics Canada. The method used to generate population projections presents each component of population growth (fertility, mortality and migration); many assumptions are made regarding how each component would evolve in the future. These assumptions, when combined, form a number of projection scenarios. There are typically three assumptions each on fertility, mortality and immigration and four assumptions on inter provincial migration; in combination, these assumptions generate 108 scenarios on the future course of population change.

Objectives

  1. Study the various assumptions and analyze the results of the difference projection scenarios, first focusing on the results at the national level and then for Newfoundland-and-Labrador (the participants can also investigate other provincial/territorial levels) and each time presenting the population growth and the age structure of these populations. This “sensitivity analysis” will be used to discuss the impact of each of the components of population growth on the projections.
  2. From the specific case of Newfoundland-and-Labrador, which is poised to being the first province where natural growth (residual of births minus death) is actually in decline, investigate the various assumptions needed to make that province register population growth again and stop its aging demographic process.

Background: The Methods

It is clear that the methods of demography have changed considerably with the advent of computers. Doing computational intensive techniques, like multistate life tables and hazard modeling, are now much more accessible to demographers. The constant evolution of data management systems that has improved alongside the expansion of informatic capabilities has meant that more powerful estimation techniques can be utilized.
 

What are Estimates and Projections?
 

Population Projections for Canada, Provinces and Territories
 

Population estimates and projections are similar in that they both indicate the size of the population for a given time. The distinction between them is related to their underlying concepts. Population estimates are approximate counts of the current (or recent) population, whereas population projections are expected population counts based on a defined set of assumptions for a future date.
 

Estimates

Essentially, there are two categories of estimates produced by Statistics Canada: postcensal and intercensal.
 

The first category, postcensal estimates, are produced by using data from the most recent census (adjusted for net census undercoverage) and estimates of the components of demographic change since that last census. These components include births, deaths, immigration, emigration (and its sub-components), change in non-permanent residents, and interprovincial migration. Another component, intra-provincial migration is relevant for estimates within subprovincial areas in Canada. All the elements of the production of population estimates will be discussed in this workshop.
 

The second category, Intercensal estimates, are produced every five years and reconcile previous postcensal estimates with the latest census counts adjusted for net undercoverage. This process typically takes two years after census data collection to complete (summer of 2008 for the 2006 Census).
 

Projections

Population projections produced by Statistics Canada provide the users several scenarios regarding the future size and geographical distribution of the Canadian population.
 

The combinations of component assumptions provide plausible maximum, medium, and minimum population growth levels, for each province/territory and for Canada as a whole. Projections are developed on the basis of analysis of past trends and are thought to represent reasonable alternatives for the future movement of the component values. Discussions with the provincial and territorial statistical focal points contribute to the final choices for the scenarios presented. It is not claimed, however, that the values will always remain within the range implied by the assumptions. Year-to-year fluctuations in the relevant parameters should be expected.
 

Projections are not predictions; they are calculations of future population derived from a base population and stated assumptions regarding the components of population change (fertility, mortality and migration). Projections are typically done every 5 years and use the base population of July 1 of the most current year postcensal estimate of the population.

Postcensal and Intercensal Population Estimates, Canada, Provinces and Territories

This portion will describe the methods used for calculating postcensal and intercensal estimates for the total population and for the population by age and sex at the provincial and territorial, and subprovincial levels. The data sources for each are identified, followed by a discussion of the quality of the estimates.
 

Postcensal Population Estimates, Canada, Provinces and Territories
 

Postcensal population estimates are population counts, which are derived using data from the most recent census (adjusted for net census undercoverage) and estimates of the components of population change since that census.
 

Postcensal estimates are then obtained using the component method. This method can be equated to a population accounting system, where modifications are made to the censal estimate by adding and subtracting the components of population change that occurred between July 1st and the reference date of the estimate.
 

Components can be divided into two groups, according to the characteristics of the data used: those components for which data are readily available, including births, deaths and immigration, and those that have to be estimated, including interprovincial migration, total emigration and changes in non-permanent residents (NPRs).
 

For each of the three types of postcensal estimates, preliminary, updated and final, component data have varying levels of completeness, hence the requirement of estimation procedures. In this sense, preliminary estimates, available typically three to four months after the reference date, are similar to short-term projections as they are based on preliminary counts that lead to less complete sources for component measurements and at times estimated through regression models.
 

Updated estimates are produced usually within one year, to account for availability of more complete data for at least some of the components. It takes typically two years to finalise postcensal estimates but they are considered the most accurate population numbers available until the next census.
 

Estimates of the population are produced first for each province and territory, and then summed to get an estimate of the population of Canada. Estimates are also done at the sub-provincial levels and summed up to the provincial levels.
 

The component method used in estimating total provincial/territorial populations is expressed as follows:
 

Table 1: Component Method Formula
 

Equation 1.1: P(t+i) = Pt + B(t,t+i) - D(t,t+i) + I(t,t+i) - E(t,t+i) + ΔNPR(t,t+i) + IMnet(t,t+i)
 

where for any given province and territory:

Variable Meaning
P(t+i) estimate of population at time (t+i)
Pt population at time t (censal estimate or postcensal estimate)
B(t,t+i) number of births between time t and (t+i)
D(t,t+i) number of deaths between time t and (t+i)
I(t,t+i) number of immigrants between time t and (t+i)
E(t,t+i) number of emigrants (permanent, temporary and returning emigrants) between time t and (t+i)
ΔNPR(t,t+i) change in the number of NPRs between time t and(t+i) (can be either positive or negative)
IMnet(t,t+i) net interprovincial migration between time t and (t+i)(can either be positive or negative)
(t,t+i) interval between July 1st and the reference date of the estimate
 

As we mentioned not all information used in estimating the components are available at the same time. In fact, they come to use from various sources which impacts their timeliness as well. Table 2 shows the sources and references of component data used to generate the postcensal population estimates. 
 

Table 2: Sources and References of Postcensal Population Estimates - Component Data
 

Component Sources and References (if applicable)
Base Population (Censal Estimate) Census population counts (Census of Canada, Catalogue no. 93F0051XIE) Data from Reverse Record Check (RRC) - Coverage, 2001 Census Technical Report: Reference Products 2001 Census , Statistics Canada, Catalogue no. 92-394-XIE
Births and Deaths Health Statistics Division, Statistics Canada, receives vital statistics on births and deaths from every Province & Territory. Data are disseminated by Demography Division, Statistics Canada (Catalogue nos. 91-002-XWE, Quarterly, Table 4)
Marriages and Divorces Health Statistics Division, STC, receives vital statistics on marriages from every Province & Territory, and receives the divorce data file from the CRDP at Justice Canada.
Immigration Citizenship and Immigration Canada (CIC) data
Total Emigration Permanent Emigrants
Data from the Canada Child Tax Benefit program (CTB) from Canada Revenue Agency (CRA), and Data from U.S. Homeland Security (formerly the U.S. Department of Justice, Immigration and Naturalization Service)

Returning Emigrants
Data from the CTB program, CRA

Temporary Emigrants
Data from Reverse Record Check (RRC)
Coverage, 2001 Census Technical Report: (Reference Products 2001 Census),
Statistics Canada, Catalogue no. 92-394-XIE)
Net Change in Non-permanent Residents CIC data on Persons holding: a student authorization; an employment authorization; a Minister’s permit. Also includes Family members of permit and authorization holders.
Interprovincial and Subprovincial Migration Data from the CTB program, CRA Data processed from Income Tax file by Small Area and Administrative Data Division (SAADD), Statistics Canada
 

Demographic Trends

As mentioned before, two important population trends have had an impact on the dynamic changes of the Canadian population, and in particular the Atlantic population; the slowing down of population growth, in some cases actual decline, and ageing. The following charts and tables reveal the extent of these changes over time and how dramatic the effects have been, especially on the province of Newfoundland-and-Labrador.


Figure 1 clearly shows that while the Rest of Canada has been growing a little slower over the last 3 decades, the Atlantic region has, in fact, been declining in population since the early to mid 90s.


Figure 1: Population Growth, Atlantic Region and ROC, 1971-2006


alt text
 

Between the 1996 and 2001 censuses the population in Canada grew by more than 1.4 million people (Table 3). However population growth was not distributed evenly across the Canadian provinces. The Atlantic region saw a decline of over 12,000 people during this period, all of which is attributed to a loss of population in the province of Newfoundland and Labrador. 
 

Table 3: Factors of Demographic Growth, 1996-2001
 

1996-2001 Canada NL PEI NS NB Atlantic Region ROC*
Total Growth 1,456,187 -26,595 2,377 9,859 2,139 -12,220 1,468,407
Natural Growth 615,866 4,397 1,772 8,505 8,148 22,822 593,044
Net International Migration 840,321 1,083 423 7,717 2,403 11,626 828,695
Net Interprovincial Migration 0 -32,075 182 -6,363 -8,412 -46,668 46,668
*ROC - Rest of Canada 
Source: Demography Division, Statistics Canada (estimates are adjusted for Census and net undercoverage)
 

Between the 1996 and 2001 censuses the population in Canada grew by more than 1.4 million people (Table 3). However population growth was not distributed evenly across the Canadian provinces. The Atlantic region saw a decline of over 12,000 people during this period, all of which is attributed to a loss of population in the province of Newfoundland and Labrador.
 

Figure 2: Natural Growth Rates, Atlantic Region and ROC, 1971-2006

alt text
 

Compared to the Rest of Canada (ROC) the rate of natural growth in the Atlantic region has fallen dramatically over the last 35 years. While the natural growth rate in the Atlantic was above the national average in 1971-72, it is now way below the national average. When each Atlantic province is measured separately, you actually see that the sharpest decline has been felt in Newfoundland (Figure 3). In fact Newfoundland, once having one of the highest natural growth rates, registered its first negative growth rate in 2005-2006, meaning that the number of deaths outnumbered the number of births in that year.
 

Figure 3: Natural Growth Rates, Atlantic Provinces, 1971-2006

alt text
 

How can we explain such a decline? First of all, fertility has been lower in the Atlantic region since the beginning of the mid-1980’s (Table 5). Clearly Newfoundland has the lowest total fertility rates (TFR) in the country. Similarly, the Atlantic Provinces, with the exception of PEI, have lower rates than the national level. 
 

Table 5: Total Fertility Rate per Woman3, Canada, Provinces and Territories, 1971-2004
 

Province / Territory4 1971 1981 1991 2001 2004
Canada5 2.13 1.65 1.70 1,51 1,53
Newfoundland and Labrador 1.44 1,24 1,30
Prince Edward Island 2.86 1.88 1.85 1,47 1,53
Nova Scotia 2.46 1.62 1.58 1,36 1,40
New Brunswick 2.62 1.67 1.55 1,38 1,40
Quebec 1.84 1.57 1.65 1,47 1,58
Ontario 2.15 1.58 1.66 1,51 1,50
Manitoba 2.49 1.82 1.97 1,80 1,77
Saskatchewan 2.67 2.11 2.04 1.80 1,86
Alberta 2.35 1.85 1.89 1,65 1,74
British Columbia 2.05 1.63 1.68 1,38 1,39
Yukon Territory 3.08 2.04 2.15 1,56 1,67
Northwest Territories 4.57 2.84 2.44 1,82 2,03
Nunavut 3.52 3,03 2,96
Source: Canadian Vital Statistics - Birth Database,, CANSIM (table 102-4505), Demography Division, Population Estimates Program, Statistics Canada

Since any population which has a fertility below the replacement level (2.1 children per woman). Hence, the more a population is below that level, the fastest is its ageing process. Posting the lowest fertility rates for two decades, it is thus not a surprise to see the populations of the Atlantic provinces getting older faster than anywhere else in the country.
 

With fertility rates being equal, the older population, which possesses a lower number of women in child-bearing age will generate less children and thus, lower natality rates. Moreover, with mortality rates being equal, an older population will also record more deaths. In other words, when a population is in an ageing process mode, it generates less and less births, more and more deaths. This will be reflected in decreasing natural growth rates.
 

As we saw it before, interprovincial migration net losses amplify the demographic decline and ageing of the Atlantic provinces because the young people are migrating elsewhere, likely to seek better labour conditions, and will often have children elsewhere in the country. Since the beginning of the 1970’s, the Atlantic region recorded net losses of near 150,000 persons (Figure 4). These losses would probably have been larger if one had taken into account the children that this population had.
 

Figure 4: Population Exchanges Between Atlantic Region and ROC, 1971-2006

alt text
 

Even if it increased in the recent years, international immigration cannot offset the demographic decline and aging caused by persistent low fertility, population aging and interprovincial migration. More then 75% of the rush of new migrants (immigrants) to Canada flock to the large urban centres of Toronto, Montreal and Vancouver. While this component helps to keep up the growth rate of the country and of some provinces (Ontario and BC particularly), its effects are not homogeneous and does not have a big impact on the Atlantic region demographic dynamics.
 

Clearly a number of factors have influenced the current demographic situation in Canada, the Atlantic region and in Newfoundland and Labrador. How these emerging trends impact the long term projections of the population in Canada and the Atlantic region is what this case study will try to address.

Demographic Projections

When population projections are contemplated, a number of different trends and historical patterns are studied in an attempt to better project a reasonable or informed outcome of the demographic picture of the country. The projections are established using the components method. The last sets of projections begin with the population by age, sex and province/territory of residence on July 1st, 2005. Then, on an annual basis, births and immigrants are added to this number. and deaths and emigrants are subtracted. The model also takes into account non-permanent residents and interprovincial migrations. The calculations are made at the level of the provinces and territories and then the results for Canada as a whole are obtained by summing the data for each geographic unit.
 

To make the projection model operational, assumptions on each of the components of population growth must be developed. Thus, for example, the projections released in 2005 are based on three assumptions regarding fertility, immigration and mortality, four assumptions on interprovincial migration and a single assumption for non-permanent residents and each of the components of total emigration. A limited number of combinations of assumptions are then selected to create projection scenarios that are plausible in light of past trends. Following each population census, new projections replace the previous ones. The most recent edition of catalogue numbers 91-520-XPE, 91-520-XIE and 91-520-SCB include projections based on the 2001 Census.
 

For population projections a number of alternative scenarios are investigated and the impact on population growth is assessed. Assumptions on mortality, fertility, immigration, emigration and interprovincial migration are made.
 

Statistics Canada has published seven sets of population projections for Canada, provinces and territories since 1974, with the last report in 2005. The projections issued on a regular basis ensure consistent and comparable method and results at the national and provincial/territorial level. This report contains a range of projections to the year 2031. It also describes the methodology and the assumptions and provides a brief analysis of the results.
 

It should be emphasized that these projections are not predictions. Rather, they represent an attempt to establish plausible scenarios based on stated components assumptions, which are subject to varying degrees of uncertainty. As such, they are valuable statistical information for planners, policymakers, and the public at large interested in the future course of demographic change and related issues.
 

Table 6: Selected Scenarios for the Population Projections
 

Scenario Fertility Life Expectancy Immigration Interprovincial Migration Type of Scenario
A- Medium growth – medium interprovincial migration Medium Medium Medium Medium Scenario 3 - projections 2005-2031
B- Medium growth – interprovincial migration recent trends Medium Medium Medium Recent Trends Scenario 2 - projections 2005-2031
C- Medium growth –west coast interprovincial migration Medium Medium Medium West Coast Scenario 4 - projections 2005-2031
D- Medium growth – central-west interprovincial migration Medium Medium Medium Central West Scenario 5 - projections 2005-2031
E- No interprovincial migration Medium Medium Medium 0 Special Scenario - SSC 2007
F- High immigration Medium Medium High Medium Sensitivity analysis - projections 2005-2031
G- Low immigration Medium Medium Low Medium Sensitivity analysis - projections 2005-2031
H - 1 million immigrants per year Medium Medium 1 million per year Medium Special Scenario - SSC 2007
I- No immigration Medium Medium 0 Medium Special Scenario - SSC 2007
J- High fertility rates High Medium Medium Medium Sensitivity analysis - projections 2005-2031
K- Low fertility rates Low Medium Medium Medium Sensitivity analysis - projections 2005-2031
L- Fertility at 2.1 for NFL TFR = 2.1 for NFL Medium Medium Medium Special Scenario - SSC 2007
M - High life expectancy Medium High Medium Medium Sensitivity analysis - projections 2005-2031
N - Low life expectancy Medium Low Medium Medium Sensitivity analysis - projections 2005-2031
O - Fertility at 2.1 for Canada TFR = 2.1 for Canada Medium Medium Medium Special Scenario - SSC 2007


Notes on scenarios’ assumptions:

  1. A detailed description of all scenarios’ assumptions, excluding scenarios E, H, I,L and O, can be found in “Population Projections for Canadal Provinces and Territories, 2005-2031” (Catalogue no 91-520- XIE)
  2. For scenarios E, H, I, L and O, medium assumptions for each components of demographic change are similar to the medium assumptions of all other scenarios.
  3. For scenario E, “no interprovincial migration” means that net interprovincial migration of each provinces and territories are equal to 0 for the whole projection period.
  4. For scenario H, “1 million immigrants per year” means that the number of immigrants will be set at one million, at the Canada level, for the first projection year and will stay at that level for the rest of the projection. Distribution of immigrants amongst provinces and territories will be the same as the one measured in recent years.
  5. For scenario I, “no immigration” means that no immigrants will be added to the Canadian population, starting from the first year of the projection.
  6. For scenario L, “fertility at 2.1 for NFL” means that the Newfoundland-and-Labrador TFR will be set at 2.1 (replacement level of 2.1 children per woman) for the whole projection period.
  7. For scenario L, “fertility at 2.1 for Canada” means that the Canadian TFR will be set at 2.1 (replacement level of 2.1 children per woman) for the whole projection period. Variations between provinces and territories will be the same as the one mesured in recent years.

Figure 5 is an excerpt from the current population projections, specifically one projection scenario that is often used as the reference scenario because of the low stakes of the assumptions. According to this scenario, the Atlantic population would still register a lower growth rate as compared to the rest of Canada. While some rebounding is anticipated in population growth, the long term trend still forecasts a negative growth rate by the end of the projection period.
 

Figure 5: Population Growth Rate, Atlantic Region and the Rest of Canada 

1971-2006 (demographic estimate) 2007-2031 (demographic projection)

 

alt text
 

Consequently; the demographic weight of the Atlantic region will become significantly less important as it will not keep pace with the growth measured in the rest of the country (Figure 6). While it was close to 10% at the beginning of the 1970’s, the demographic weight of the Atlantic region would be slightly over 6% in 2031.
 

Figure 6: Demographic Weight of Atlantic Region, 1971-2031
 

alt text
 


1

Difference between the number of births and deaths by unit of 1,000 persons.


2

These are the number of birth minus the number of death measured in the region for each 1000 population units.


3

Total fertility rate is an estimate of the average number of live births a female can be expected to have in her lifetime, based on the age-specific fertility rates (ASFR) of a given year. The total fertility rate (TFR) is equal to the sum of the ASFR.


4

The geographic distribution of live births in this table is based on the mother’s usual place of residence.

Research Question: 

From Statistics Canada’s population estimates and projections, participants to the Case Study will try to find ways to offset those demographic trends. In other words, they will be asked to answer this type of questions:

  • Which component of demographic change has the strongest impact on population dynamics?
  • To which level, fertility rates must be raised in order to stop the demographic aging process?
  • Does immigration, at the current levels, can offset the working-age population decline and the population aging process?
  • What is the real impact of immigration on population growth and aging?
  • What about the other components of demographic change (mortality, emigration)? Do they have any impact on the demographic dynamics of the Canadian population?

Variables: 

For this case study, we will provide a number of Excel spreadsheets of the various population projection models released by Statistics Canada and a few other models that may be of interest for this analysis.


As a companion source of information. we will include a number of tables on characteristics needed to project the population for Newfoundland and Labrador. Most of these tables are available through CANSIM or STC publications.

  • Newfoundland and Labrador Population by age and sex, 1971 to 2005 (CANSIM 051-0001)
  • Births by sex and by age of mother, Newfoundland and Labrador 1991 to 2004 (CANSIM only has data for 2000 to 2004, so an internal Excel spreadsheet will be provided)
  • Deaths by sex and age group, Newfoundland and Labrador 1972 to 2005 (CANSIM 051-0002)
  • Interprovincial In-migrants by age and sex, Newfoundland and Labrador 1971 to 2003 (CANSIM 051-0012)
  • Interprovincial out-migrants by age and sex, Newfoundland and Labrador 1971 to 2003 (CANSIM 051-0012)
  • Immigrants and NPRs by age and sex, Newfoundland and Labrador 1971 to 2003 (CANSIM 051-0011)
  • Emigrants, returning Canadians and net persons temporarily abroad by age and sex,Newfoundland and Labrador 1971 to 2003 (CANSIM 051-0011)

 

References: 

Statistics Canada, Population and family estimation methods at Statistics Canada, Catalogue no. 91-528, Ottawa, April 2003. 


Statistics Canada, Population projections for Canada, provinces and territories, 2005-2031, Catalogue no. 91-520, Ottawa, December 2006.


Statistics Canada, Report on the demographic situation in Canada 2003-2004, Catalogue no.91-209, Ottawa, June 2006.


Guillemette, Yvan et William B.P. Robson, No Elixir of Youth: Immigration Cannot Keep Canada Young, CD Howe Institute Backgrounder # 96, septembre 2006.