Different Types of Population Projections and Their Applications

Population projections are the demographer’s crystal ball, but a far more disciplined one than the name suggests. They are calculated estimates of how many people will live in a country, state, or district at some future date, and they sit behind almost every long-term decision a government makes, from building schools and hospitals to designing pension systems and electricity grids. But not all projections are made for the same purpose. Some look at the whole nation, others zoom into a single district. Some look forward to 2050, others reach back to estimate what a population looked like a century ago. Understanding these categories is the first step to reading projection reports intelligently.

What population projections actually do 🔗

A population projection is a calculation, not a prediction. It tells us what the future population would look like if certain assumptions about fertility, mortality, and migration hold true. The U.S. Census Bureau distinguishes estimates from projections by noting that estimates describe the past and present using existing data, while projections rest on assumptions about future demographic trends. That distinction matters because every projection is only as reliable as the assumptions feeding into it.

Projections are classified along three useful axes: the geographic area they cover, the direction in time they move, and the level of demographic change they assume. A single dataset, such as the Population Projections for India and States, 2011-2036 released by the National Commission on Population, usually combines all three.

Total versus regional projections 🔗

The first split is geographic. Total projections cover an entire country, providing an aggregate view of the population at the national level. Regional projections focus on smaller units such as states, districts, cities, or even specific ethnic and socioeconomic groups. Both are essential, but they answer very different planning questions.

Total projections and national planning 🔗

Total projections are what international agencies like the United Nations publish in their biennial World Population Prospects. They tell us, for example, that India’s population is expected to keep growing for nearly four more decades before slowly declining. According to the UN’s latest demographic projections, India is expected to reach around 1.7 billion people by 2060 before easing back toward 1.5 billion by the end of the century. Numbers at this scale shape macro decisions: food security strategy, defence planning, international climate commitments, and the trajectory of the overall economy.

At the national level, the Office of the Registrar General and Census Commissioner has been preparing projections for the government since 1958. Its most recent report estimates that India’s population will grow by 25 percent between 2011 and 2036, raising population density from 368 to 462 persons per square kilometre and pushing the share of people aged 60 and above from 8.4 percent to 15 percent. Those national-level numbers feed directly into central schemes for healthcare, social security, and labour-market policy.

Regional projections and local planning 🔗

Regional projections drill down to a far more useful granularity. The same 2020 report from the National Commission on Population breaks total figures into state-wise estimates, and the contrasts are striking. The National Capital Territory of Delhi is projected to grow by 98 percent between 2011 and 2036, while Himachal Pradesh is expected to grow by just 6 percent and Tamil Nadu by 8 percent. The five southern states of Andhra Pradesh, Karnataka, Kerala, Tamil Nadu, and Telangana together are projected to contribute only nine percent of total national population growth in that period.

These differences matter for local governance. A municipal corporation in Delhi is dealing with a fundamentally different problem (where to put the next 10 million residents) than a hill district in Himachal Pradesh (managing services for a stable or declining population). Regional projections are what allow state finance commissions, urban development authorities, and district planning committees to size their investments correctly.

Regional projections also include specialised cuts: urban versus rural splits, projections for tribal or scheduled-caste populations, or estimates for specific religious communities. India’s urban population, for instance, is projected to rise from 31.8 percent in 2011 to 38.6 percent in 2036, with urban growth accounting for nearly three-fourths of the total population increase. That single insight changes everything about how cities are planned.

Forward versus backward projections 🔗

The second classification has to do with the direction in time. Forward projections are the familiar kind: they take a base-year population and roll it forward using assumptions about births, deaths, and migration. Backward projections do the opposite, reconstructing what a population looked like at earlier points in time.

Forward projections 🔗

Forward projections dominate public discussion. Almost everything labelled as “the population of India in 2050” is a forward projection. These typically use the cohort-component method, which the MEASURE Evaluation training materials describe as the most widely used demographic tool because it tracks births, deaths, and migration separately for each age and sex group as cohorts move through time. The method is data-hungry but produces the kind of detail that planners need: not just total numbers but age structures, sex ratios, dependency ratios, and so on.

Forward projections answer questions like: how many primary-school seats will India need in 2030? How will the working-age population (15-59 years) change between now and 2036? The Indian projections show this share rising from 60.7 percent to roughly 65 percent over that period, suggesting a continued demographic dividend that has direct implications for jobs, skilling, and labour-force participation.

Backward projections 🔗

Backward projections are less famous but methodologically rich. They are used to estimate past populations when historical records are missing, incomplete, or unreliable, and to correct or validate earlier census counts. Demographers have developed techniques using generalised inverses of the Leslie matrix to project an age-sex distribution backward in time, essentially running the cohort-component method in reverse.

A widely used variant is the reverse survival method, which the research published in Demographic Research evaluates as a way to estimate past fertility from population data collected in a single census or survey. This is invaluable for countries where birth registration is patchy. If you know how many 10-year-olds there are today, and you know how children of that cohort have survived, you can work backwards to estimate how many babies were born a decade ago, even if no one was recording births at the time.

Backward projections also play a role in historical demography, environmental history, and ecological reconstruction, helping researchers understand long-run trends in human numbers, fertility regimes, and even forest populations.

High, medium, and low projections 🔗

The third classification is about uncertainty. Because future fertility, mortality, and migration cannot be known with certainty, demographers usually publish multiple variants built on different assumptions. The most common are the high, medium, and low projections.

The medium variant 🔗

The medium projection is the one most agencies treat as the “central” scenario. It assumes a continuation of recently observed demographic trends and is often called the most likely outcome. The Population Reference Bureau explains that the UN’s medium variant assumes a steady growth in the use of family planning that produces fertility declines similar to those seen in other countries that have already undergone the demographic transition. For most policy and academic purposes, when a single number is quoted, it usually comes from this medium scenario.

The high and low variants 🔗

The high and low variants are built around the medium one. According to the UN Population Division’s definition of projection scenarios, the high variant assumes that total fertility will run 0.5 births per woman above the medium variant, while the low variant assumes it will run 0.5 births per woman below. So if the medium scenario projects a total fertility rate of 1.85 children per woman, the high scenario uses 2.35 and the low scenario uses 1.35. Mortality and migration assumptions are typically held the same across the three variants, isolating the effect of fertility.

This half-child gap might sound trivial, but compounded over decades it produces dramatically different futures. A high-variant world has hundreds of millions more people than a low-variant world by 2100. That uncertainty is precisely the point: presenting three variants forces policymakers to think about a range of plausible futures rather than a single false certainty.

Matching variants to development context 🔗

The interpretation of these variants depends on the country’s stage in the demographic transition. A high variant is most relevant for countries still in the early phase, where fertility decline could stall. A medium variant suits fast-developing countries like India, where family planning and health services are driving steady declines but the trajectory is not yet locked in. A low variant is most useful for countries that have already crossed below replacement fertility, where the question becomes how quickly the population will shrink and age. India sits squarely in the medium-variant zone today, with its total fertility rate having already declined from 2.34 in 2011-15 to a projected 1.72 by 2031-35, well below the replacement level of 2.1.

Why these distinctions matter 🔗

Mixing up these categories leads to bad decisions. Treating a national total as a guide for district-level investment misses huge regional variation. Quoting only the medium variant without acknowledging the high and low bands creates a false sense of certainty. Ignoring backward projections means losing tools for validating questionable historical counts. The UN’s own World Population Prospects 2024 stresses that good projections are critical for designing policies in macroeconomic planning, social protection, national security, and the environment, areas where the time horizon is too long for guesswork.

The newer probabilistic projections, which combine the cohort-component method with statistical models of uncertainty, are now refining all three classifications further. But the basic framework, total versus regional, forward versus backward, high versus medium versus low, remains the working vocabulary of population studies.

What do you think? If you were a planner in your home state, which of these projection variants would you rely on most heavily, and why? And do you think India’s official projections should be revised more frequently than once every census cycle, given how quickly fertility patterns are changing?

References

1. https://www.census.gov/data/academy/resources/one-pagers/population-projections.html
2. https://nhm.gov.in/New_Updates_2018/Report_Population_Projection_2019.pdf
3. https://population.un.org/wpp/
4. https://ourworldindata.org/data-insights/india-china-europe-and-the-united-states-are-on-very-different-population-paths
5. https://ruralindiaonline.org/en/library/resource/population-projections-for-india-and-states-2011-2036/
6. https://www.measureevaluation.org/resources/training/online-courses-and-resources/non-certificate-courses-and-mini-tutorials/population-analysis-for-planners/lesson-8.html
7. https://www.sciencedirect.com/science/article/pii/B9780123946805500158
8. https://www.demographic-research.org/articles/volume/31/9
9. https://www.prb.org/resources/understanding-population-projections-assumptions-behind-the-numbers/
10. https://population.un.org/wpp/definition-of-projection-scenarios
11. https://www.un.org/development/desa/pd/sites/www.un.org.development.desa.pd/files/undesa_pd_2024_wpp_2024_advance_unedited_0.pdf