Rapid responses are electronic comments to the editor. They enable our users
to debate issues raised in articles published on bmj.com. A rapid response
is first posted online. If you need the URL (web address) of an individual
response, simply click on the response headline and copy the URL from the
browser window. A proportion of responses will, after editing, be published
online and in the print journal as letters, which are indexed in PubMed.
Rapid responses are not indexed in PubMed and they are not journal articles.
The BMJ reserves the right to remove responses which are being
wilfully misrepresented as published articles or when it is brought to our
attention that a response spreads misinformation.
From March 2022, the word limit for rapid responses will be 600 words not
including references and author details. We will no longer post responses
that exceed this limit.
The word limit for letters selected from posted responses remains 300 words.
It is hard to disagree with John Appleby (BMJ 2014; 349: g5184) that population projections are often wrong. In reality, they are even worse: projections – single demographic scenarios based on predefined assumptions – are almost always wrong. Formally, the probability of getting any single projection right is infinitesimal – essentially zero – but this may not make them useless.
However, it is difficult to recognise actual demographers in the caricature painted in the feature. In real life, the demographic community has become increasingly impatient in preaching about the necessity to acknowledge the uncertainty in population forecasts at least since the seminal paper by Alho and Spencer (1985). Demographers have also been producing graphs, such as those presented in Appleby’s paper, for quite some time, also for the UK (Shaw 2007). This is nothing new.
Even the practice of official statistics is slowly changing. Official probabilistic forecasts for national population are already prepared by Statistics Netherlands, Statistics New Zealand, and are trialled by the United Nations Population Division (Gerland et al. 2014). Probabilistic methods offer users of forecasts much more information than the mere variants of alternative futures suggested in the feature, which the Office for National Statistics prepares anyway. The key challenge is how to communicate this to the users so that it helps them make prudent and robust decisions (Bijak et al. 2015).
Appleby raises two interesting points: that life expectancy has been projected to stop increasing for decades, but this is yet to occur; and that economists are frustrated by the inaccurate population projections they incorporate into their own forecasting models. However, assuming in official population projections that there is a life expectancy asymptote has vexed demographers considerably (Oeppen and Vaupel 2002). That economists are also frustrated is a red herring: as Appleby notes at the outset, the uses of demographic forecasts are often much more practical and local than national economic planning and most use much shorter horizons than 2100.
We may agree to disagree about which forecasts are better fit for their purpose: demographic or economic. Still, it is worth pointing out that what gives population forecasters some competitive advantage is that populations are more stable than economies, and a lot of information about the future is embedded in the size and structure of the population of today. It is much easier to predict the population which is already born, which explains why demographic predictions quickly get worse beyond the horizon of one generation (some 20-30 years) ahead. Still, some of the decisions may require even longer a time frame, for example those concerning the stability of pension systems.
Hence, despite the demographer presented in the feature being largely made of straw, it can still remain a useful metaphor, as it draws attention to the dangers of putting too much faith in single deterministic projections of future populations (Bijak et al. 2015). There is a need for all users to engage further in exactly how they use information about future population, and what type of uncertainty could affect their decisions, just as economists may be reflecting on whether key data is representing what users want to know about. Statistics are better for everyone if they address the right questions, rather than what is easy or analytically tractable (Hand 1994).
References
Alho J, and Spencer BD (1985) Uncertain population forecasting. Journal of the American Statistical Association 80(390): 306–314, DOI: 10.1080/01621459.1985.10478113
Bijak J, Alberts I, Alho J, Bryant J, Buettner T, Falkingham J, Forster JJ, Gerland P, King T, Keilman N, O’Hagan A, Onorante L, Owens D, Raftery A, Ševčíková H, and Smith PWF (2015) Probabilistic population forecasts for informed decision making. Forthcoming in Journal of Official Statistics; abridged version available from the NTTS conference.
Gerland P, Raftery AE, Ševčíková H, et al. (2014) World population stabilization unlikely this century. Science 346(6206): 234–237, DOI: 10.1126/science.1257469
Hand DJ (1994) Deconstructing Statistical Questions. Journal of the Royal Statistical Society A, 157(3): 317–356. DOI: 10.2307/2983526
Oeppen J, and Vaupel J (2002) Broken Limits to Life Expectancy. Science, 296(5570): 1029–1031, DOI: 10.1126/science.1069675
Competing interests:
JB is a demographer with a PhD in economics and 12 years of experience in studying population uncertainty. TK is a fellow of the Royal Statistical Society and a member of their social statistics section and education committees.
10 May 2015
Jakub Bijak
Associate Professor in Demography
Thomas King, Newcastle University
University of Southampton
Department of Social Statistics and Demography, University of Southampton, Southampton, SO17 1BJ
On Population Forecasts and Forecasters
It is hard to disagree with John Appleby (BMJ 2014; 349: g5184) that population projections are often wrong. In reality, they are even worse: projections – single demographic scenarios based on predefined assumptions – are almost always wrong. Formally, the probability of getting any single projection right is infinitesimal – essentially zero – but this may not make them useless.
However, it is difficult to recognise actual demographers in the caricature painted in the feature. In real life, the demographic community has become increasingly impatient in preaching about the necessity to acknowledge the uncertainty in population forecasts at least since the seminal paper by Alho and Spencer (1985). Demographers have also been producing graphs, such as those presented in Appleby’s paper, for quite some time, also for the UK (Shaw 2007). This is nothing new.
Even the practice of official statistics is slowly changing. Official probabilistic forecasts for national population are already prepared by Statistics Netherlands, Statistics New Zealand, and are trialled by the United Nations Population Division (Gerland et al. 2014). Probabilistic methods offer users of forecasts much more information than the mere variants of alternative futures suggested in the feature, which the Office for National Statistics prepares anyway. The key challenge is how to communicate this to the users so that it helps them make prudent and robust decisions (Bijak et al. 2015).
Appleby raises two interesting points: that life expectancy has been projected to stop increasing for decades, but this is yet to occur; and that economists are frustrated by the inaccurate population projections they incorporate into their own forecasting models. However, assuming in official population projections that there is a life expectancy asymptote has vexed demographers considerably (Oeppen and Vaupel 2002). That economists are also frustrated is a red herring: as Appleby notes at the outset, the uses of demographic forecasts are often much more practical and local than national economic planning and most use much shorter horizons than 2100.
We may agree to disagree about which forecasts are better fit for their purpose: demographic or economic. Still, it is worth pointing out that what gives population forecasters some competitive advantage is that populations are more stable than economies, and a lot of information about the future is embedded in the size and structure of the population of today. It is much easier to predict the population which is already born, which explains why demographic predictions quickly get worse beyond the horizon of one generation (some 20-30 years) ahead. Still, some of the decisions may require even longer a time frame, for example those concerning the stability of pension systems.
Hence, despite the demographer presented in the feature being largely made of straw, it can still remain a useful metaphor, as it draws attention to the dangers of putting too much faith in single deterministic projections of future populations (Bijak et al. 2015). There is a need for all users to engage further in exactly how they use information about future population, and what type of uncertainty could affect their decisions, just as economists may be reflecting on whether key data is representing what users want to know about. Statistics are better for everyone if they address the right questions, rather than what is easy or analytically tractable (Hand 1994).
References
Alho J, and Spencer BD (1985) Uncertain population forecasting. Journal of the American Statistical Association 80(390): 306–314, DOI: 10.1080/01621459.1985.10478113
Bijak J, Alberts I, Alho J, Bryant J, Buettner T, Falkingham J, Forster JJ, Gerland P, King T, Keilman N, O’Hagan A, Onorante L, Owens D, Raftery A, Ševčíková H, and Smith PWF (2015) Probabilistic population forecasts for informed decision making. Forthcoming in Journal of Official Statistics; abridged version available from the NTTS conference.
Gerland P, Raftery AE, Ševčíková H, et al. (2014) World population stabilization unlikely this century. Science 346(6206): 234–237, DOI: 10.1126/science.1257469
Hand DJ (1994) Deconstructing Statistical Questions. Journal of the Royal Statistical Society A, 157(3): 317–356. DOI: 10.2307/2983526
Oeppen J, and Vaupel J (2002) Broken Limits to Life Expectancy. Science, 296(5570): 1029–1031, DOI: 10.1126/science.1069675
Shaw C (2007) Fifty years of United Kingdom national population projections: how accurate have they been? Population Trends 128, 8–23. Via: http://www.ons.gov.uk/ons/rel/population-trends-rd/population-trends/no-...
Competing interests: JB is a demographer with a PhD in economics and 12 years of experience in studying population uncertainty. TK is a fellow of the Royal Statistical Society and a member of their social statistics section and education committees.