Presentation 16:

Population Growth, Population Estimates, and Demographic Transition

What is population growth?

• Population growth is the change in the size of the population between two dates

Can the growth be negative? Explain what does it mean to have negative growth?

• Growth can be negative, i.e. population decline

How Many ways do we have to measure population growth? Explain each way.

• The balance of the change components (births, deaths, net migrants) during the interval

between two dates

• Difference between the population sizes at the two dates

• There are two ways to measure population growth between two dates, 0 and t, i.e. in the time
interval (0,t):

• (1) Growth (0,t) = Births (0,t) – Deaths (0,t) + Inmigrants (0,t) – Outmigrants (0,t)

• (2) Growth (0,t) = Population [t] – Population [0]

• The balancing equation of population change equates (1) and (2)

• G(0,t) = P[t] – P[0] = B(0,t) – D(0,t) + I(0,t) – O(0,t)

 Rate of population Growth

• The annual growth rate of the population can be calculated using the balance of the component
rates

• Population growth rate during an interval = (crude birth rate – crude death rate) + (inward
migration rate – outward migration rate)
= rate of natural increase + net migration rate

• Growth rates are usually expressed per cent instead of per thousand

• The annual growth rate of the population can also be calculated through relating the population
change during a time interval to the total person-years lived by the population during the
interval

• The total person-years are estimated by the mid-interval population multiplied by the number
of years in the interval

𝑃[𝑡] − 𝑃[0]
𝑟(0, 𝑡) = 𝑡 × 100
𝑡 ∙ 𝑃[ ]
2
• The growth rate r in the previous formula is the average rate of growth during the interval. The
formula does not assume that the rate of population growth has been constant during the
interval

• The assumption of constant rate of growth is unrealistic if the time interval is long (e.g. 10 years,
as typical of inter-census intervals)

• However, it is sometimes useful to assume that the population does grow at a constant rate

• Under the assumption of constant rate of growth, r, the population size will grow exponentially,
that is,

𝑃[𝑡] = 𝑃[0]𝑒 𝑟𝑡
The formula for r in this case can be deduced as follows
𝑃[𝑡]
= 𝑒 𝑟𝑡
𝑃[0]
𝑃[𝑡]
By taking the (ln) of the two sides Ln(𝑃[0])=rt

𝑃[𝑡]
So 𝑟 = ln 𝑃[0] /𝑡

where r is expressed as a fraction (less than 1) not as a percentage

Doubling Time

• One way to express the rate of population growth is through asking the following hypothetical
question

• What is the time interval needed for a population to double if it continues to grow at a
specific annual rate r?

• Putting P[t] = 2P[0] in the previous formula and solving for t, we get
𝐥𝐧⁡(𝟐) 𝟎. 𝟔𝟗𝟑𝟏𝟒𝟕
𝒕= =
𝒓 𝒓

Example

• A closed population has a crude birth rate of 40 per thousand and a crude death rate of 10 per
thousand. How long would it take this population to double?

• r = 0.04 – 0.01 = 0.03

• t = (0.693147)/(0.03) = 23.1049

• It would take that population around 23 years to double

Population Momentum

• Population growth rate depends on the values of crude rates

• Therefore, it is affected by the age composition of the population

• If fertility in a hypothetical closed population with constant mortality dropped from above-
replacement to replacement fertility (net reproduction rate = 1, equivalent to TFR around 2.1), it
will continue to grow for several years

• This is called population momentum, or potential for growth

• Population momentum is the contribution of the age composition to population growth

• The growth potential for a population depends on its past history of fertility, and to a lesser
extent, mortality and migration

• Population momentum can be negative (potential for decline); this happens when a population
has very low fertility (below replacement) for a long period

• A simple estimate for population momentum can be obtained as follows

• Population momentum indicator = proportion of females under age 30 in the actual population
÷ corresponding proportion in the stationary population associated with the female life table for
that population
𝑓 𝑓 𝑓
𝑇0 −𝑇30 ∑25
0 5𝐿 𝑥
• The denominator of the ratio = 𝑓 = 𝑓
𝑇0 𝑇0

• where f denotes that the life table is for females

• A population momentum indicator larger than 1 indicates positive momentum (potential for
growth), while an indicator smaller than 1 indicates negative momentum (potential for decline)
 Presentation 17:

Population estimates, forecasts and projections

Population Estimates and Projections

• Estimates refer to the size at a past or current time point

• Forecasts refer to the expected size at a future time point

• Projections also refer to size at future time points, but under some assumptions, that might or
might not be realistic or expected

• All forecasts are projections, but not vise versa

Methods for Population Estimates and Projections

• Mathematical interpolation/extrapolation (linear; geometric; exponential)

If we know the population size at two dates 0 and t, and want to estimate the population size at
another date u, where u can be earlier than t (interpolation) or later than t (extrapolation)

We assume a specific mathematical model of population growth; fit the model using available
data (i.e. estimate the model parameters); apply the model to estimate population size at date u

The Linear Model

𝑷[𝒕] = 𝑷[𝟎] + 𝒃𝒕
where b is the annual size of population growth, assumed to be constant
𝑷[𝒕] − 𝑷[𝟎]
𝒃=
𝒕
𝑷[𝒖] = 𝑷[𝒕] + 𝒃 ∙ (𝒖 − 𝒕)

The Geometric Model

𝑷[𝒕] = 𝑷[𝟎](𝟏 + 𝒓)𝒕

where r is the annual rate of population growth, assumed to be constant and applied only
once per year (discrete growth)

𝑷[𝒕] 𝟏/𝒕
𝒓=[ ] −𝟏
𝑷[𝟎]

𝑷[𝒖] = 𝑷[𝒕](𝟏 + 𝒓)(𝒖−𝒕)

The Exponential Model

𝑃[𝑡] = 𝑃[0]𝑒 𝑟𝑡
where r is the annual rate of population growth, assumed to be constant and applied
continuously (continuous growth)
𝑃[𝑡]
ln ( )
𝑃[0]
𝑟=
𝑡
𝑃[𝑢] = 𝑃[𝑡]𝑒 𝑟(𝑢−𝑡)
Example

The population of Sweden on 1st January 2008 was 8,416,599 and on 1st January 2009 became
8,461,554. Estimate the population size on 1st July 2020.

• Linear model:

b = 8461554 – 8416599 = 44955

P[1/7/2020] = 8461554 + (11.5)(44955) = 8978537

• Geometric model:

r = (8461554/8416599) – 1 = 0.0053

P[1/7/2020] = 8461554 (1.0053)^(11.5) = 8996122

• Exponential model:

r = ln(8461554/8416599) = 0.0053

P[1/7/2020] = 8461554 exp(11.5x0.0053) = 8996122

The logistic curve

• Geometric and exponential growth are unrealistic for long time periods, especially when the
growth rate is large, because limited resources put checks on population growth

• The logistic curve which has an upper asymptote was considered more realistic

• It is rarely used now to estimate population size; but has other applications in projecting fertility
and mortality

• A three-parameter logistic curve takes the form:

𝑎
𝑃[𝑡] =
1 + 𝑏𝑒 −𝑐𝑡
• Where a is the upper asymptote (the upper limit of population that can inhabit a certain area,
or the carrying capacity of the area),

• b and c are parameters that determine the shape of the curve. The parameters are
estimated from historical data
Time series analysis

• Advanced statistical methods for prediction using time series data have also been used in
population estimation and projection; when projecting the total population and also when
projecting change components

• They have the advantage of incorporating uncertainty through treating population growth as a
stochastic (random variable)process

• They require rich data; and are not always better than simpler techniques

Component methods

• Component methods use information on the base (initial) population and on the components of
population change during a time interval (births/fertility rates; deaths/survival ratios;
migrants/migration rates) to estimate or project population at the end of the interval

• Application of the population balancing equation is a component method applied to the total
population

𝑷[𝒕] = 𝑷[𝟎] + 𝑩(𝟎, 𝒕) − 𝑫(𝟎, 𝒕) + 𝑰(𝟎, 𝒕) − 𝑶(𝟎, 𝒕)

The Cohort Component Method

Prorating and ratio extrapolation

• These methods are used to estimate (or project) subgroups within a total population that is
already estimated or projected

• The subgroups could be age-groups, or could be regions or political or administrative units or

any other subgroups

• Prorating, or constant share method, assumes that the proportional distribution remains
constant as it was on the base date, such that
𝑃𝑖 [0]
𝑃𝑖 [𝑡] = 𝑃[𝑡] ∙
𝑃[0]
• Other assumptions, beside constant share, can be used if there are information on two previous
dates

• For example, the change in the share of the subgroup to the total population observed in the
past is extrapolated in the future, assuming a linear trend

• This is called shift-share method

Use of symptomatic or auxiliary data

• Symptomatic or auxiliary data are easy-to-obtain data on variables that are related to
population size in an area; such as number of housing units, number of licensed cars, tax files,
number of pupils enrolled in local schools, registered births or deaths, etc.

• Based on available data on the population size and these symptomatic variables (usually from
censuses), ratios are calculated or regression models are fitted

• These ratios or regressions are then applied in time periods other than the census years to
estimate population size from readily available symptomatic data