Academic Interests
Population forecasting, modelling marriage and household dynamics, mathematical demography.
Courses taught
Background
 M.Sc. in Applied Mathematics, Delft University of Technology, the Netherlands, 1977
 Ph.D. in demography, University of Utrecht, the Netherlands, 1990
 Research officer, Department of Population Statistics, Statistics Netherlands (19771981)
 Research associate, Netherlands Interdisciplinary Demographic Institute (19821990)
 Senior research associate, Statistics Norway (19901998).
Positions held
 Editor 20062011, Demographic Research
 Member, editorial board European Studies of Population
 Member, International Editorial Committee, Population, Paris
 Member 20052015, Scientific Advisory Council Netherlands Demographic Institute, The Hague
Tags:
Demography,
Economics,
Income Employment and Welfare
Publications
Citations at Google Scholar
Papers in peer reviewed journals
 Keilman, Nico (2020) Evaluating probabilistic population forecasts. Economie et Statistique. ISSN 03361454. 520521, s 49 64. https://doi.org/10.24187/ecostat.2020.520d.2033
 Keilman, Nico (2020) Modelling education and climate change. Nature Sustainability April 2020. ISSN 23989629. https://doi.org/10.1038/s4189302005158
 Keilman, Nico (2019) Mortality shifts and mortality compression in period and cohort life tables. Demographic Research 41(40), 11471196
 Borgan, Ørnulf & Keilman, Nico (2018). Do Japanese and Italian women live longer than women in Scandinavia?. European Journal of Population. ISSN 01686577. . doi: 10.1007/s1068001894682
 Linder, Dennis; Frigessi, Arnoldo; Piaserico, Stefano & Keilman, Nico (2017). Simulating the life course of psoriasis patients: the interplay between therapy intervention and marital status. Journal of the European Academy of Dermatology and Venereology.
 Keilman, Nico (2016). A combined BrassRandom Walk approach to probabilistic household forecasting: Denmark, Finland, and the Netherlands 20112041. Journal of Population Research () 127.
 Keilman, Nico (2016). Household forecasting: Preservation of age patterns. International Journal of Forecasting 32(3) 726735.
 Measures for human reproduction should be linked to both men and women (first author, with Krzysztof Tymicki and Vegard Skirbekk). International Journal of Population Research 2014. http://dx.doi.org/10.1155/2014/908385. [pdf]
 Utilising expert opinion to improve the measurement of international migration in Europe (fifth author, with Arkadiusz Wisniowski, Jakub Bijak, Solveig Christiansen, Jonathan Forster, James Raymer, Peter Smith). Journal of Official Statistics 29(4)2013, 583607, http://dx.doi.org/10.2478/jos20130041 [pdf]
 Hvor robust er det nye pensjonssystemet med hensyn til levealdersutviklingen? ("How robust is the new public pension system for life expectancy development?") (first author, with Lisa Keller). Samfunnsøkonomen 09/2013 [doc]
 Probabilistic household forecasts based on register data: The case of Denmark and Finland (second author, with Solveig Christiansen). Demographic Research 06/2013; 28:1263. [pdf]
 Cohen, Joel; Kravdal, Øystein & Keilman, Nico (2011). Childbearing impeded education more than education impeded childbearing among Norwegian women. Proceedings of the National Academy of Science of the United States of America. ISSN 00278424. 108(29), s 11830 11835 . doi: 10.1073/pnas.1107993108
 Juha Alho and Nico Keilman, On future household structure. Journal of the Royal Statistical Society Series A 173(1)2010, 117143
 Nico Keilman and Solveig Christiansen, Norwegian elderly less likely to live alone in the future. European Journal of Population, DOI 10.1007/s1068000991959
 Nico Keilman, Concern in the European Union about Low Birth Rates, European View 2008
 Nico Keilman, European demographic forecasts have not become more accurate during the past 25 years, Population and Development Review, 2008, 34(1)
 Maarten Alders, Nico Keilman and Harri Cruijsen, Assumptions for longterm stochastic population forecasts in 18 European countries, European Journal of Population, 2007, 23:3369
 Juha Alho, Maarten Alders, Harri Cruijsen, Nico Keilman, Timo Nikander and Dinh Quang Pham, New forecast: Population decline postponed in Europe, Statistical Journal of the United Nations ECE12, 2006, 23:110
 Nico Keilman, The Impact of Demographic Uncertainty on Liabilities for Public Old Age Pensions in Norway, New Zealand Population Review 31(1), 2005, 3550
 Juha Alho, Maarten Alders, Harri Cruijsen, Nico Keilman, Timo Nikander and Dinh Quang Pham, Population decline postponed in Europe, Statistical Magazine, 2005
 Nico Keilman and Dinh Quang Pham, Time series based errors and empirical errors in fertility forecasts in the Nordic countries, International Statistical Review 72(1), 2004, 518, summary
 Nico Keilman, Biodiversity: The threat of small households, Nature 421, 30 January 2003, 489490
 Nico Keilman, Dinh Quang Pham and Arve Hetland, Why population forecasts should be probabilistic  illustrated by the case of Norway, Demographic Research 615 May 2002, 409454
 Nico Keilman, TFR predictions and Brownian motion theory, Yearbook of Population Research in Finland 38, 2002, 209221
 Nico Keilman, Data quality and accuracy of United Nations population projections, 195095, Population Studies 55(2), 2001, 149164
 Nico Keilman, Uncertain population forecasts, Nature 412, 6846, 2001, 490491
 Nico Keilman and Dinh Quang Pham, Predictive intervals for agespecific fertility, European Journal of Population 16(1), 2000, 4166
 Nico Keilman, How accurate are the United Nations' world population projections?,Population and Development Review supplement to Volume 24, 1998, 1541

Keilman, Nico (2020). Evaluating probabilistic population forecasts. Economie et Statistique.
ISSN 03361454.
520521, s 49 64 . doi: https://doi.org/10.24187/ecostat.2020.520d.2033
Full text in Research Archive.
Show summary
We demonstrate how a probabilistic population forecast can be evaluated, when observations for the predicted variables become available. Statisticians have developed various scoring rules for that purpose, but there are hardly any applications in population forecasting literature. A scoring rule measures the distance between the probability distribution of the predicted variable, and the actual outcome. We use scoring rules that reward accuracy (the outcome is close to the expected value of the prediction)and sharpness (the predictive distribution has low variance, which makes it difficult to hit the target). We evaluate probabilistic population forecasts for France, the Netherlands, and Norway. For all three countries, we use results from the UPEproject ("Uncertain Population of Europe"). We inspect prediction intervals for population size in the period 20042019 and 3000 sample paths for population pyramids for the year 2010. For the Netherlands and for Norway, we compare the UPEresults with findings from the official probabilistic population forecast by Statistics Netherlands (20012019) and from a probabilistic forecast for Norway (19972019). All forecasts were computed using the cohortcomponent method and stochastically varying parameters for fertility, mortality and migration. We show that the UPEforecasts for the Netherlands and for Norway performed better than the other forecasts for these two countries. The error in the jumpoff population caused a bad score for the French forecast. We evaluate the 3000 UPEsimulations of the age and sex composition predicted for the year 2010. When normalized for population numbers in each agesex category, the predictions for the Netherlands received the best scores, except for the oldest old. The age pattern for the Norwegian score reflects the underprediction of immigration after the enlargement of the European Union in 2005.

Keilman, Nico (2020). Uncertainty in population forecasts for the twentyfirst century. Annual Review of Resource Economics.
ISSN 19411340.
12, s 449 470 . doi:
10.1146/annurevresource110319114841

Keilman, Nico (2020). Évaluer les prévisions probabilistes de population. Economie et Statistique.
ISSN 03361454.
520521, s 51 68
Show summary
Résumé – Les statisticiens ont développé des règles de notation pour évaluer les prévisions probabilistes par rapport aux observations. Toutefois, on en trouve peu d’applications dans la littérature sur les prévisions de population. Une règle de notation mesure la distance entre la distribution prédictive et le résultat. Nous passons en revue les règles de notation qui privilégient l’exactitude (le résultat est proche de l’espérance de la distribution) et la précision (la distribution présente une faible variance, de sorte qu’il est difficile d’atteindre l’objectif). Nous évaluons les prévisions de population probabilistes établies pour la France, les Pays‑Bas et la Norvège. Les prévisions de la taille de la population totale des Pays‑Bas et de la Norvège ont obtenu de bons scores. L’erreur sur la population de base a engendré un mauvais score pour la prévision française. Nous évaluons aussi la prévision de la composition par âge et par sexe pour 2010. Les prédictions relatives aux Pays‑Bas ont reçu les meilleurs scores, excepté celles concernant les personnes très âgées. Pour la Norvège, le score de la structure par âge reflète la sous‑prédiction de l’immigration après l’élargissement de l’Union européenne en 2005.

Keilman, Nico & Kristoffersen, Sigve (2020). European Mortality Forecasts: Are the Targets Still Moving?, In Stefano Mazzuco & Nico Keilman (ed.),
Developments in Demographic Forecasting.
Springer Nature.
ISBN 9783030424718.
9.
s 179
 192
Full text in Research Archive.
Show summary
Many statistical agencies routinely produce population forecasts, and revise these forecasts when new data become available, or when current demographic trends indicate that an update is necessary. When the forecaster strongly revises, from one forecast round to the next one, a forecast for a certain target year (for instance the life expectancy in 2050), this indicates large uncertainty connected to mortality predictions. The aim of this chapter is to shed more light on the uncertainty in mortality forecasts, by analysing the extent to which life expectancy predictions for 2030 and 2050 were revised in subsequent rounds of population forecasts published by statistical agencies in selected countries. It updates and extends earlier work that focused on United Nations and Eurostat forecasts published between 1994 and 2004. There the conclusion was that life expectancy forecasts for 18 European countries for the year 2050 had been revised upwards systematically, by around 2 years on average during the 10year publication period. A recent analysis based on official population forecasts for Norway published in the period 1999–2018 led to the same conclusion. Here we will show that the period of upward revisions seems to have ended for some European countries.

Borgan, Ørnulf & Keilman, Nico (2019). Do Japanese and Italian Women Live Longer than Women in Scandinavia?. European Journal of Population.
ISSN 01686577.
35(1), s 87 99 . doi:
10.1007/s1068001894682
Full text in Research Archive.
Show summary
Life expectancies at birth are routinely computed from period life tables. When mortality is falling, such period life expectancies will typically underestimate real life expectancies, that is, life expectancies for birth cohorts. Hence, it becomes problematic to compare period life expectancies between countries when they have different historical mortality developments. For instance, life expectancies for countries in which the longevity improved early (like Norway and Sweden) are difficult to compare with those in countries where it improved later (like Italy and Japan). To get a fair comparison between the countries, one should consider cohort data. Since cohort life expectancies can only be computed for cohorts that were born more than a hundred years ago, in this paper we suggest that for younger cohorts one may consider the expected number of years lost up to a given age. Contrary to the results based on period data, our cohort results then indicate that Italian women may expect to lose more years than women in Norway and Sweden, while there are no indications that Japanese women will lose fewer years than women in Scandinavia. The large differences seen for period data may just be an artefact due to the distortion that period life tables imply in times of changing mortality.

Keilman, Nico (2019). Erroneous population forecasts, In Tommy Bengtsson & Nico Keilman (ed.),
Old and New Perspectives on Mortality Forecasting.
Springer Nature.
ISBN 9783030050740.
Chapter.
s 95
 111
Full text in Research Archive.
Show summary
Given a history of sizable forecasting errors, this chapter addresses the question of why demographic forecasts are uncertain. It summarizes main patterns in expost observed errors in historical population forecasts, and discusses probabilistic forecasts. Illustrations come from a probabilistic forecast for the population of Norway.

Keilman, Nico (2019). Family projection methods: A review, In Robert Schoen (ed.),
Analytical Family Demography.
Springer Nature.
ISBN 9783319932279.
12.
s 277
 301
Full text in Research Archive.

Keilman, Nico (2019). Mortality shifts and mortality compression in period and cohort life tables. Demographic Research.
ISSN 14359871.
41, s 1147 1196 . doi:
10.4054/DemRes.2019.41.40
Full text in Research Archive.
Show summary
BACKGROUND When agespecific mortality falls, period life tables give a distorted view of the life expectancy (LE) and the degree of mortality compression in birth cohorts. OBJECTIVE To derive mathematical expressions for the link between LEs and compression in period life tables on the one hand and corresponding variables in birth cohorts on the other hand. METHODS We analyse the age at death distribution (AADD) computed from the life table’s d(x)column. We derive general expressions for the moments of this distribution in a series of annual period life tables, written as functions of the moments in the AADD of cohorts. RESULTS We use data for Norwegian men and women to illustrate simple versions of the new expressions. The LE increases twice as fast across cohorts compared to what period life tables suggest under this model. Compression in Norwegian mortality, expressed in terms of decreasing variance of the AADDs, is approximately 40% slower in period than in cohort mortality. CONCLUSIONS We show how one can analyse the amount of distortion in period LEs compared to cohort LEs. In addition, we show how period compression is determined by cohort compression, together with both period and cohort LEs. CONTRIBUTION We derive new expressions that link the period AADD to the cohort AADD. Under a simple linear model, we show why compression in the period AADD often goes together with increases in the period LE.

Linder, Dennis; Frigessi, Arnoldo; Piaserico, Stefano & Keilman, Nico (2017). Simulating the life course of psoriasis patients: the interplay between therapy intervention and marital status. Journal of the European Academy of Dermatology and Venereology.
ISSN 09269959.
32(1), s 62 67 . doi:
10.1111/jdv.14567
Full text in Research Archive.
Show summary
Background Psoriasis, a chronic relapsing inflammatory disease affecting primarily the skin, shows multiple comorbidities including depression, cardiovascular diseases and other relevant conditions. Psoriasis patients experience social isolation, job loss, financial difficulties and partnership problems. Inversely, psychosocial impairments may negatively influence the disease course. Objective To explore the feasibility of a model describing the interaction of psychosocial and clinical factors over the life course of patients. Methods We considered only seven states for members of a hypothetical population: single and healthy, single and having a psoriasis flare, single and ‘cured’, coupled and healthy, coupled and having a psoriasis flare, coupled and ‘cured’, and dead. Transition probabilities between states were taken from the Norwegian Population Register for the healthy population and from epidemiological research articles. Clinical experience allowed adjustments on the assumed parameters. Results Our macromodel, which simulates the effect of therapy intervention on patients’ partnership status, yields a description of the transitions between the seven states. Treatment efficacy shows only a negligible effect on the chances of living with a partner. Conclusions Mathematical modelling of interactions between social and health variables is in principle feasible. However, complex models, comprising more variables (for instance: employment status, depression level, obesity etc.), are needed for more realistic simulations for the interactions studied. As increasing the number of variables leads to an exponential increase of the model’s state space, switching to micromodelling (representing each individual separately) may be necessary.

Keilman, Nico (2016). A combined Brassrandom walk approach to probabilistic household forecasting: Denmark, Finland, and the Netherlands, 2011–2041. Journal of Population Research.
ISSN 14432447.
34(1), s 17 43 . doi:
10.1007/s125460169175y
Show summary
Probabilistic household forecasts to 2041 are presented for Denmark, Finland, and the Netherlands. Future trends in fertility, mortality and international migration are taken from official population forecasts. Time series of shares of the population in six different household positions are modelled as random walks with drift. Brass’ relational model preserves the age patterns of the household shares. Probabilistic forecasts for households are computed by combining predictive distributions for the household shares with predictive distributions of the populations, specific for age and sex. If current trends in the three countries continue, we will witness a development towards more and smaller households, often driven by increasing numbers of persons who live alone. We can be quite certain that by 2041, there will be between two and four times as many persons aged 80 and over who live alone when compared with the situation in 2011.

Keilman, Nico (2016). Household forecasting: Preservation of age patterns. International Journal of Forecasting.
ISSN 01692070.
32(3), s 726 735 . doi:
10.1016/j.ijforecast.2015.10.007
Show summary
We formulate a time series model for household dynamics for different age groups. We model the share of the population who are in a certain household position (living alone, living with partner etc.). These household positions have very pronounced age patterns. The age profiles change slowly over time, caused by changes in home leaving behaviour of young adults, differences in survival of men and women, etc. When forecasting household positions to 2040, we want to preserve the characteristics of the age profiles. We test the LeeCarter model and the Brass Relational method using household data for the Netherlands for the period 19962010. Annual shares of the population by household position, age, and sex are modeled as Random Walks with a Drift (RWD). While the Brass model has its limitations, in our application it performs better than the LeeCarter model. The predicted age patterns based on the Brass model look more reasonable, because the Brass model is a twoparameter model, while the LeeCarter model contains only one parameter. Also, the model parameters and standard errors of the Brass model are easier to estimate than those of the LeeCarter model.

Keilman, Nico; Tymicki, Krzysztof & Skirbekk, Vegard (2014). Measures for human reproduction should be linked to both men and women. International Journal of Population Research.
ISSN 20904029.
. doi:
10.1155/2014/908385
Full text in Research Archive.
Show summary
We introduce the twosex net reproduction rate (2SNRR) and the twosex total fertility rate (2STFR)—two demographic indicators that reflect the number of children born, given age specific fertility and mortality of the adults. The main quality of these indicators is that they measure the childbearing behaviour of both women and men. The indicators have intuitive value, since they tell us to what extent adults are replaced by children. While the traditional net reproduction rate (NRR) describes general replacement trends among women only, the 2SNRR is an indicator of a population’s growth potential, irrespective of sex. We demonstrate the use of the indicators with data from Bejsce parish in Poland for the period 1800–1967 and with data from UN projections for China for future years. We discuss the consequences for our understanding of fertility trends when sex ratios deviate from normal levels.

Christiansen, Solveig Tobie Glestad & Keilman, Nico (2013). Probabilistic household forecasts based on register data  the case of Denmark and Finland. Demographic Research.
ISSN 14359871.
28, s 1263 1302 . doi:
10.4054/DemRes.2013.28.43
Show summary
BACKGROUND Household forecasts are important for public planning and for predicting consumer demand. OBJECTIVE The purpose of this paper is to compute probabilistic household forecasts for Finland and Denmark, taking advantage of unique housing register data covering the whole populations dating back to the 1980s. A major advantage is that we do not have to rely on small population samples, and we can get quite reliable estimates even for infrequent transitions. A further merit is having time series containing the population in different household positions (dependent child, living with a spouse, living in a consensual union, living alone, lone parent, living in other private household and institutional households) by age and sex. METHODS These series enable us to estimate the uncertainty in the future distribution of the population across household positions. Combining these uncertainty parameters with expected shares computed in a deterministic household forecast, we simulate 3000 sample paths for the household shares for each age and sex. These paths are then combined with 3000 simulations from a stochastic population forecast covering the same period to obtain the predicted number of households and persons in each household position by age and sex. RESULTS According to our forecasts, we expect a strong growth in the number of private households during a 30year period, of 27% in Finland and 13% in Denmark. The number of households consisting of a married couple or a person who lives alone are the most certain, and single parents and other private households are the most uncertain.

Keilman, Nico & Keller, Lisa Dahl (2013). Hvor robust er det nye pensjonssystemet med hensyn til levealdersutviklingen?. Samfunnsøkonomen.
ISSN 18905250.
(6), s 28 38

Wisniowski, Arkadiusz; Bijak, Jakub; Christiansen, Solveig Tobie Glestad; Forster, Jonathan; Keilman, Nico; Raymer, James & Smith, Peter (2013). Utilising Expert Opinion to Improve the Measurement of International Migration in Europe. Journal of Official Statistics.
ISSN 0282423X.
29(4), s 583 607 . doi:
10.2478/jos20130041

Cohen, Joel; Kravdal, Øystein & Keilman, Nico (2011). Childbearing impeded education more than education impeded childbearing among Norwegian women. Proceedings of the National Academy of Sciences of the United States of America.
ISSN 00278424.
108(29), s 11830 11835 . doi:
10.1073/pnas.1107993108

Alho, Juha & Keilman, Nico (2010). On future household structure. Journal of the Royal Statistical Society: Series A (Statistics in Society).
ISSN 09641998.
173, s 117 143
Show summary
We develop a method for computing probabilistic household forecasts which quantifies uncertainty in the future number of households of various types in a country. A probabilistic household forecast helps policy makers, planners and other forecast users in the fields of housing, energy, social security etc. in taking appropriate decisions, because some household variables are more uncertain than others. Deterministic forecasts traditionally do not quantify uncertainty.We apply the method to data from Norway.We find that predictions of future numbers of married couples, cohabiting couples and oneperson households are more certain than those of lone parents and other private households. Our method builds on an existing method for computing probabilistic population forecasts, combining such a forecast with a random breakdown of the population according to household position (single, cohabiting, living with a spouse, living alone etc.). In this application, uncertainty in the total numbers of households of different types derives primarily from random shares, rather than uncertain future population size. A similar method could be applied to obtain probabilistic forecasts for other divisions of the population, such as household size, health or disability status, region of residence and labour market status. Keywords: Household forecast; Norway; Population forecast; Probabilistic forecast; Random shares

Keilman, Nico (2010). On age structures and mortality, In Shripad Tuljapurkar; Naohiro Ogawa & Anne Gauthier (ed.),
Ageing in Advanced Industrial States.
Springer.
ISBN 9789048135523.
kap 2.
s 23
 46

Keilman, Nico & Christiansen, Solveig Tobie Glestad (2010). Norwegian Elderly Less Likely to Live Alone in the Future. European Journal of Population.
ISSN 01686577.
26(1), s 47 72 . doi:
10.1007/s1068000991959
Show summary
We analyse the future household status of elderly men and women in Norway. An important finding is that persons aged 80+ are less likely to live alone in the future, and more often with a partner. The level of mortality, the mortality sex gap, union dissolution at young and intermediate ages, and entry into and exit from institutions for the elderly are possible determinants for this new trend. We use the macro simulation program LIPRO to simulate the household dynamics in Norway to 2032, and investigate the demographic reasons for the increased likelihood of living with a partner among Norwegian elderly. Mortality plays an important role, but part of the trend is already embodied in the household structure of the current population. Keywords: Elderly, Living arrangements, Household projection, Norway, LIPRO

Alho, Juha; Cruijsen, Harri & Keilman, Nico (2008). Empirically based specification of forecast uncertainty, In Juha Alho; Svend Hougaard Jensen & Jukka Lassila (ed.),
Uncertain demographics and fiscal sustainability.
Cambridge University Press.
ISBN 9780521877404.
kap. 3.
s 34
 54

Keilman, Nico (2008). Erroneous population forecasts, I: Patrick Festy & JeanPaul Sardon (red.),
Hommage à Gérard Calot  Profession: démographe.
Institut national d'études démographiques.
ISBN 9782733240236.
.
s 237
 254

Keilman, Nico (2008). European demographic forecasts have not become more accurate over the past 25 years. Population and Development Review.
ISSN 00987921.
34(1), s 137 153
Show summary
Nowadays, demographers, population statisticians, and population forecasters have richer data, more refined theories of demographic behavior, and more sophisticated methods of analysis than they had two or three decades ago. This scientific progress should have made it easier to predict demographic behavior. But my analysis of the errors in old forecasts shows that demographic forecasts published by statistical agencies in 14 European countries have not become more accurate over the past 25 years. The findings demonstrate that scientific progress in population studies during the previous two to three decades has not kept up with less predictable demographic behavior of populations in European countries. There is no reason to be more optimistic about US Census Bureau forecasts. Population forecasts are intrinsically uncertain, and hence should be couched in probabilistic terms.

Keilman, Nico; Cruijsen, Harri & Alho, Juha (2008). Changing views of future demographic trends, In Juha Alho; Svend Hougaard Jensen & Jukka Lassila (ed.),
Uncertain demographics and fiscal sustainability.
Cambridge University Press.
ISBN 9780521877404.
kap. 2.
s 11
 33

Alders, Maarten; Keilman, Nico & Cruijsen, Harri (2007). Assumptions for longterm stochastic population forecasts in 18 European countries. European Journal of Population.
ISSN 01686577.
23, s 33 69 . doi:
10.1007/s1068000691044
Show summary
The aim of the ‘Uncertain Population of Europe’ (UPE) project was to compute longterm stochastic (probabilistic) population forecasts for 18 European countries. We developed a general methodology for constructing predictive distributions for fertility, mortality and migration. The assumptions underlying stochastic population forecasts can be assessed by analysing errors in past forecasts or modelbased estimates of forecast errors, or by expert judgement. All three approaches have been used in the project. This article summarizes and discusses the results of the three approaches. It demonstrates how the––sometimes conflicting––results can be synthesized into a consistent set of assumptions about the expected levels and the uncertainty of total fertility rate, life expectancy at birth of men and women, and net migration for 18 European countries.

Keilman, Nico (2007). UK national population projections in perspective: How successful compared to those in other European countries?. Population Trends.
ISSN 03074463.
(129), s 20 30
Show summary
Compared to population forecasts of other European countries, those made in the United Kingdom during the past 30 years had somewhat larger forecast errors for fertility and smaller errors for mortality. Migration forecasts in the UK were about as accurate as the European average. After controlling for various effects such as relative data volatility both at the time a projection is made and during the period of the projection, there is no indication that recent forecasts in European countries have been more accurate than older ones. Hence population forecasts are intrinsically uncertain, and a forecast for the UK in the form of probability distributions is presented.

Alho, Juha; Alders, Maarten; Cruijsen, Harri; Keilman, Nico; Nikander, Timo & Pham, Dinh Quang (2006). New forecast: Population decline postponed in Europe. Statistical Journal of the United Nations Economic Commision for Europe ECE12.
ISSN 01678000.
23(1), s 1 10
Show summary
We present results of a probabilistic forecast for the population in 18 European countries, to 2050. Other forecasts have recently predicted a falling population size for those countries. However, there are reasons to expect higher immigration and lower mortality than the earlier forecasts did. Hence, we find that population decline is postponed in our forecast. The forecast also alerts us to the fact that many demographic developments cannot be forecasted accurately. Although ageing is certain, the extent to which this will occur is difficult to predict accurately. The number of elderly persons is very uncertain in the long run. This has major implications for all European countries in which reforms for pension systems and the provision of health care for the elderly are considered. The reforms must be robust against unexpected demographic developments.

Keilman, Nico (2006). Demographic Translation: From Period to Cohort Perspective and Back, In Guillaume Wunsch; Graziella Caselli & Jacques Vallin (ed.),
Demography: Analysis and Synthesis, A Treatise in Population Studies Volume 1.
Elsevier.
ISBN 9780127656601.
Chapter 17.
s 213
 223

Keilman, Nico (2006). Households and families, In Guillaume Wunsch; Graziella Caselli & Jacques Vallin (ed.),
Demography: Analysis and Synthesis.
Academic Press.
ISBN 0127656642.
Ch. 91 in Vol. 3.
s 457
 476
Show summary
The aim of the current chapter is to give not only a broad overview of trends in family and household developments in Europe (Section 5), but also to discuss concepts and definitions concerning the household, the family, and their members (Section 2): what constitutes a household, a family, what is a consensual union, a child, a oneparent family, a reconstituted family? Next I review various issues connected to measuring household and family developments (Section 3): de facto or de jure place of residence, measuring household and family structure at one point in time, measuring household and family dynamics over a certain period, the individual or the group as unit of measurement, the problem of longitudinal households, relationships between the events that several members of the same household may experience (e.g. a lone mother becomes a oneperson household when her last child leaves the parental home), and the representativeness of the data. The strengths and weaknesses of various data sources often used to map household and family developments are discussed in Section 4. Finally, the main trends in family and household developments in Europe after World War II are summarized in Section 5.

Keilman, Nico (2005). Erroneous population forecasts, In Nico Keilman (ed.),
Perspectives on Mortality Forecasting No. 2: Probabilistic Models.
Swedish Social Insurance Agency.
ISBN 9175003260.
kap. 1.
s 7
 26
Show summary
The purpose of this chapter is to give a broad review of the notions of population forecast errors and forecast accuracy. Why are population forecasts inaccurate? How large are the errors involved, when we analyse historical forecasts of fertility, mortality, and the age structure? Moreover, how can we compute expected errors in recent forecasts? We shall see that probabilistic population forecasts are necessary to assess the expected accuracy of a forecast, and that such probabilistic forecasts quantify expected accuracy and expected forecast errors much better than traditional deterministic forecasts do. The chapter concludes with some challenges in the field of probabilistic population forecasting.

Keilman, Nico (2005). The Impact of Demographic Uncertainty on Liabilities for Public Old Age Pensions in Norway. New Zealand Population Review.
ISSN 0111199X.
31(1), s 35 50
Show summary
The paper analyses the importance of demographic uncertainty for the net present value (NPV) of public old age pension obligations in Norway. A probabilistic population forecast is combined with a deterministic macro model for future pension expenditures. The model is applied to Norwegian data for the period 20032100. Under the current pension system, the liabilities are likely to grow by a factor of ten towards the end of the century. The demographic driving force is an assumed increase in the life expectancies of men and women by some 1316 years over the next 95 years. The results show also that longrun relative uncertainty is larger for total population than for the NPV, due to the enormous uncertainty in the number of births in the long run. In 2100, the 80 per cent prediction interval of population size is 1.5 times as wide as the median value. For the NPV, this relative uncertainty ratio is 80 per cent. Also for earlier years, the relative uncertainty in the NPV is approximately half that in population size. There is no single broad age group, in which relative uncertainty is similar to that in the NPV during the entire period.

Alho, Juha; Alders, Maarten; Cruijsen, Harri; Keilman, Nico; Nikander, Timo & Pham, Quang Dinh (2005). Befolkningsnedgang i EØSområdet utsatt. SSBMagasinet.
Show summary
FN og Eurostat har nylig spådd en befolkningsnedgang for 18 europeiske land. Nå viser resultatene fra en ny befolkningsprognose at vi har grunn til å vente større innvandring og høyere levealder enn det som FN og Eurostat har spådd. Nedgangen kommer derfor senere enn antatt, kanskje ikke i det hele tatt før 2050. For Norge forventes det et folketall på 5,75 millioner i 2050, litt høyere enn midtalternativet i SSBs befolkningsframskrivninger fra 2002.

Keilman, Nico & Pham, Quang Dinh (2005). Hvor lenge kommer vi til å leve? Levealder og aldersmønster for dødeligheten i Norge, 1900–2060. Økonomiske analyser.
ISSN 08004110.
(6), s 43 49
Show summary
Vi presenterer en analyse av levealderen og aldersmønsteret i dødeligheten for norske menn og kvinner siden 1900. Vi har brukt den såkalte LeeCartermodellen for å kunne gi anslag på dødelighetsutviklingen i framtiden. Vi forventer en fortsatt økning i levealderen, helt opp til 86 år for menn og 90 år for kvinner i 2060.

Keilman, Nico Willem (2004). Démographie des ménages et de la famille: Application aux pays développés, I: Graziella Caselli; Jacques Vallin & Guillaume Wunsch (red.),
Démographie: Analyse et synthèse. Volume VI: Population et société.
Institut National d'Etudes Démographiques.
ISBN 2733220160.
Chapitre 91.
s 345
 388

Keilman, Nico Willem & Pham, Dinh Quang (2004). Time series based errors and empirical errors in fertility forecasts in the Nordic countries. International Statistical Review.
ISSN 03067734.
72, s 5 18
Show summary
We use time series models to derive model based prediction intervals for the Total Fertility Rate (TFR) in Norway, Sweden, Finland, and Denmark up to 2050. For the short term (510 yrs), expected TFRerrors are compared with empirical forecast errors observed in historical population forecasts prepared by the statistical agencies in these countries since 1969. Mediumterm and longterm (up to 50 years) errors are compared with error patterns based on socalled naïve forecasts, i.e. forecasts that assume that recently observed TFRlevels also apply for the future.

Keilman, Nico Willem (2003). Biodiversity: The threat of small households. Nature.
ISSN 00280836.
421, s 489 490
Show summary
Many studies have suggested that the increasing global human population is having a negative effect on biodiversity. According to new work, another threat comes from the rising number of households.

Keilman, Nico Willem (2003). Pensjonskommisjonen bør ta usikkerhet i befolkningsutviklingen alvorlig. Økonomiske analyser.
ISSN 08004110.
(2), s 16 24
Show summary
Pensjonskommisjonens utredning av et nytt pensjonssystem har naturlig nok et langsiktig perspektiv, faktisk helt frem til år 2050 og enda lenger. En viktig variabel i denne sammenheng er antall eldre. Men den foreløpige rapporten som Pensjonskommisjonen la frem september 2002, inneholdt ingen analyse av den fremtidige utvikling av antall eldre. Anslagene over hvor mange eldre Norge kan ha i 2050 varierer med flere hundre tusen, og vi vet ikke hvilket av anslagene som kommer til å bli riktig � bare at noen av dem er mer sannsynlige enn andre. Det er særlig levealderen som avgjør hvor mange eldre vi vil få, og fremtidig levealder er vanskelig å beregne. Når vi vet at økningen i forventet levealder i mange tilfeller er blitt betydelig underestimert de siste førti årene, er det all grunn til å ta hensyn til demografisk usikkerhet i en ny pensjonsordning, og skape pensjonssystemer som hele tiden er fleksible for demografiske forandringer.

Keilman, Nico Willem (2003). Types of Models for Projecting Mortality, In Tommy Bengtsson & Nico Willem Keilman (ed.),
Perspectives on Mortality Forecasting.
Social Insurance Board.
ISBN 9189303245.
kapittel.
s 19
 27
Show summary
This chapter presents a systematic overview of various types of models for projecting mortality. The framework is that of population forecasts of the cohortcomponent type for the elderly at the national level. I distinguish between demographic and nondemographic models (Sections 1 and 2). Demographic models are rather mechanistic, although they involve varying degrees of judgment. In this category, I distinguish between extrapolation models and interpolation models (Sections 1.1 and 1.2). Nondemographic models are much more causal in nature than demographic ones. I differentiate between macro models (Section 2.1) and micro models (Section 2.2). Finer distinctions and further qualifications are also given.

Keilman, Nico Willem (2003). Discussion: Should fertility intentions inform fertility forecasts?, In Gregory Spencer (ed.),
The direction of fertility in the United States.
COPAFS.
ISBN 0972059601.
kapittel.
s 179
 181

Keilman, Nico Willem (2002). TFR predictions based on Brownian motion theory. Yearbook of Population Research in Finland.
ISSN 05063590.
XXXVIII, s 207 219
Show summary
In stochastic population forecasts, the predictive distribution of the TFR is of central concern. Time series models can be used to predict the TFR and its moments on the short run (up to 10 or 20 years), but on the long run (4050 years) they often result in excessively wide prediction intervals. To restrict the interval width, I model the time series of log TFRvalues as a Brownian motion with absorbing upper barrier. I analyse (i) expressions for the predictive distribution of the log of the TFR; (ii) expressions for the first and second moments of the predictive distribution of the log of the TFR. Using this model I find that the second moment of the predictive distribution for the longrun TFR in Norway is insensitive for ceiling levels beyond a threshold of approximately 3.4 children per woman. This conclusion holds for a fairly broad range of innovation variances. If the log of the TFR follows a random walk, sample paths that exceed approximately 3.4 children per woman may be rejected when simulating future fertility in Western countries. This will not have any major effect on the width of the longterm predictive distribution.

Keilman, Nico Willem; Pham, Quang Dinh & Hetland, Arve (2002). Why population forecasts should be probabilistic  illustrated by the case of Norway. Demographic Research.
ISSN 14359871.
6, s 407 454 . doi:
10.4054/DemRes.2002.6.15
Show summary
Deterministic population forecasts do not give an appropriate indication of forecast uncertainty. Forecasts should be probabilistic, so that their expected accuracy can be assessed. We review three main methods to compute probabilistic forecasts, namely time series extrapolation, analysis of historical forecast errors, and expert judgement. We illustrate, by the case of Norway up to 2050, how elements of these three methods can be combined when computing prediction intervals for a population¿s future size and agesex composition. We show the relative importance for prediction intervals of various sources of variance, and compare our results with those of the official population forecast computed by Statistics Norway.

Keilman, Nico Willem (2001). Uncertain population forecasts. Nature.
ISSN 00280836.
412, s 490 491
Show summary
Traditional population forecasts made by statistical agencies do not quantify uncertainty. But demographers and statisticians have developed methods to calculate probabilistic forecasts.

Keilman, Nico Willem (2001). Data quality and accuracy of United Nations population projections, 195095. Population Studies.
ISSN 00324728.
55(2), s 149 164
Show summary
Between 1951 and 1998, the United Nations (UN) published 16 sets of population projections for the world, its major regions, and countries. This paper reports the accuracy of the projection results. I analyse the quality of the historical data used for the base population of the projections, and for extrapolating fertility and mortality. I study also the impact this quality has had upon the accuracy of the projection results. Results and assumptions for the sets of projections are compared with corresponding estimates from the UN 1998 Revision for total fertility and life expectancy at birth, total population, and the projected age structures. The report covers seven major regions (Africa, Asia, the former USSR, Europe, Northern America, Latin America, and Oceania) and the largest ten countries of the world as of 1998 (China, India, the USA, Indonesia, Brazil, Pakistan, Russia, Japan, Bangladesh, and Nigeria).

Keilman, Nico Willem (2001). La translation démographique: des indicateurs du moment aux indicateurs de génération et réciproquement, I: Graziella Caselli; Jacques Vallin & Guillaume Wunsch (red.),
Démographie: analyse et synthèse. Volume I: La dynamique des populations.
Editions de l'Institut National d'Etudes Démographiques.
ISBN 273322011X.
s 359
 378
Show summary
Ce chapitre donne un aperçu général sur la théorie de la translation démographique. Le point de départ est une matrice de taux par âge et année de calendrier. Je donne les relations mathématiques entre l'intensité et le calendrier du moment, et ces deux indicateurs de la génération. L'objectif de cette formulation mathématique est de faire la lumière sur l'ampleur de la "distortion translationelle", autrement dit, de prédire le niveau et le calendrier de la descendance finale au vu des tendences de la fécondité du moment, et réciproquement.

Keilman, Nico Willem (2001). La traslazione demografica: dalla prospettiva di periodo a quella per generazione e viceversa, I: Graziella Caselli; Jacques Vallin & Guillaume Wunsch (red.),
Demografia: La dinamica della popolozioni.
Carocci, Roma.
ISBN 8843018833.
s 439
 461
Show summary
Ce chapitre donne un aperçu général sur la théorie de la translation démographique. Le point de départ est une matrice de taux par âge et année de calendrier. Je donne les relations mathématiques entre intensité et le calendrier du moment, et ces deux indicateurs de la génération. L'objectif de cette formulation mathématique est de faire la lumière sur l'ampleur de la "distortion translationelle", autrement dit, de prédire le niveau et le calendrier de la descendance finale au vu des tendences de la fécondité du moment, et réciproquement.

Imhoff, Evert Van & Keilman, Nico Willem (2000). On the quantum and tempo of fertility: Comment. Population and Development Review.
ISSN 00987921.
26(3), s 549 553
Show summary
In 1998, Bongaarts and Feeney proposed a modification of the conventional (period) total fertility rate. We show that their method for adjusting period fertility is not in fact capable of isolating the pure quantum effect of yeartoyear changes in fertility. Our critique focuses on two main points: 1. cohort specific changes in timing are much more complex than Bongaarts and Feeney assume; 2. the Bongaarts and Feeney method is based on unsuitable fertility measures.

Keilman, Nico Willem (2000). Anton Kuijsten, 19442000. Bevolking en Gezin.
ISSN 0772764X.
29(1), s 1 6

Keilman, Nico Willem (2000). Précision et incertitudes des prévisions nationales de population, I: Christine Wattelar & Josianne Duchêne (red.),
Le Défi de l'Incertitude: Nouvelles Approches en Perspectives et Prospective Démographique.
AcademiaBruylant/L'Harmattan, LouvainlaNeuve, Belgique.
ISBN 287209525X.
s 33
 67
Show summary
The existing literature on expost errors observed for historical population forecasts made by statistical agencies in 15 industrialized countries is reviewed. The amount and type of data in these studies (total population size, age structure, population growth rate, components of change), the level of detail (numbers of births and deaths, crude rates, total fertility rates, life expectancies), and the period covered vary considerably among the countries. Attention is given to forecast accuracy for total population size, population growth rates, fertility, mortality, the age structure, and the dependency ratio. Among the issues to be taken up are relative errors for fertility and mortality, and common patterns across the countries. On the basis of a data set from Norway (covering the period 19691989, forecasts made between 1969 and 1987) and one from the Netherlands (for the period 19501986, forecasts made between 1950 and 1980) we investigate a possible reduction in forecast errors over time. The paper concludes with suggestions for including results from expost evaluations in official population forecasts, in order to communicate forecast uncertainty to the user.

Keilman, Nico Willem & Pham, Quang Dinh (2000). Predictive intervals for agespecific fertility. European Journal of Population.
ISSN 01686577.
16(1), s 41 66 . doi: https://doi.org/10.1023/A:1006385413134
Show summary
A multivariate ARIMA model is combined with a Gamma curve to predict confidence intervals for agespecific birth rates by oneyear age groups. The method is applied to observed agespecific births in Norway between 1900 and 1995, and predictive intervals are computed for each year up to 2050. The predicted twothirds confidence intervals for Total Fertility (TF) around 2010 agree well with TFerrors in old population forecasts made by Statistics Norway. The method gives useful predictions for agespecific fertility up to the years 20202030. For later years, the intervals become too wide. Methods which do not take account of estimation errors in the ARIMAmodel coefficients underestimate the uncertainty for future TFvalues. The findings suggest that the margin between high and low fertility variants in official population forecasts for many Western countries are too narrow.

Keilman, Nico Willem (1999). Demographic Translation: From Period to Cohort Perspective and Back, In
Démographie: Analyse et synthèse. Causes et conséquences des évolutions démographiques.
Dipartimento di Scienze Demografiche, Università di Roma "La Sapienza".
s 1
 14
Show summary
When successive birth cohorts of women get their children at progressively lower ages, births that would have occurred during a certain period without changes in the timing, are now "squeezed" into a shorter period. This pushes period fertility up, and the period Total Fertility Rate (TFR) will be inflated, compared to the TFR that would have occurred without changes in the timing. In general, even when the number of children per woman is constant over successive birth cohorts, period fertility levels may vary  they are inflated in years in which childbearing is accelerated, and deflated when women postpone childbearing. Thus period fertility cannot be used as a reliable indicator for the level of cohort fertility: period fertility may be "distorted" in times of tempo changes in cohort fertility. At the same time cohort fertility cannot be fully understood without studying periods. These qualitative links between period and cohort fertility are straightforward. But the detailed interplay between period and cohort fertility, both its quantum (level) and tempo (timing) aspects, can be formalized mathematically. The resulting expressions constitute the core of what has become known as the theory of demographic translation, a term coined by Norman Ryder. This chapter gives a brief general overview of demographic translation theory. It integrates Ryder's findings from the 1960s that he applied to agespecific fertility, with more recent insights, which can be used for analysing other demographic processes, such as childlessness, first marriage, and divorce.

Keilman, Nico Willem (1999). Over tijd en kinderen: Het Noorse model. Facta.
7, s 12 16
Show summary
What are the effects of Norway's generous family policy and extended kindergarten supply for labour market participation, fertility, and time use of Norwegian women? Should the Netherlands introduce the Norwegian model?

Keilman, Nico Willem (1999). The future of social solidarity in Europe: The demographic background, In Godrun Gaarder (ed.),
Towards flexible welfare states: The future of social solidarity in Europe.
Europaprogrammet, Oslo.
Show summary
[Internet http://www.europrogram.no] Changing age structures and changing living arrangements will challenge solidarity between generations and that between the sexes in Europe in the future.

Keilman, Nico Willem (1998). Female Dominance, In Anton Kuijsten; Henk De Gans & Henk De Feijter (ed.),
The joy of demography ¿.. and other disciplines: Liber Amicorum for Dirk van de Kaa.
Thela  Thesis Publ., Amsterdam.
s 215
 228
Show summary
Pollard's twosex marriage model is estimated on the basis of Norwegian data on first marriages by age combination of the spouses, 19831992. Simulation of the marriage market in Norway for the period 19922017, based on the marriage rates estimated for 1992, shows that competition between unmarried men of different age groups is weak, and likewise for women.

Keilman, Nico Willem (1998). How Accurate Are the United Nations World Population Projections?. Population and Development Review.
ISSN 00987921.
24(Supplement), s 15 41
Show summary
Projection results for UNprojections published between 1951 and 1991 are evaluated against corresponding estimates taken from the United Nation's 1994 Revision. Errors are computed for total population size, crude birth rate, crude death rate, life expectancy at birth, and population by fiveyear age groups, for seven major regions: Africa, Asia, the former Soviet Union, Europe, Northern America, Latin America, and Oceania. Selected results for China, India, and the USA are also presented. UNprojections became more accurate over time, not only because the quality of base line population estimates improved, but also because unforeseen declines in birth rates became less important for projection errors. But there are large regional disparities in projection accuracy. Age structure projections for the former Soviet Union and for Asia show larger errors than on average. For Asia this is explained by errors in base populations. When errors caused by wrong assumptions regarding future fertility and mortality are isolated, the projected age structure of the former Soviet Union is still very inaccurate, but also that of Oceania, Northern America and Europe.

Keilman, Nico Willem (1998). Optimistische demografen, I: Piet Hermkens; René Van Rijsselt & Karin Sanders (red.),
Differentiatie en samenleving: Opstellen voor Henk Becker.
Thela Thesis, Amsterdam.
ISBN 9051704593.
s 177
 200
Show summary
A univariate ARIMAmodel is combined with a Gamma curve to predict confidence intervals for Total Fertility (TF). The method is applied to observed agespecific births in Norway between 1900 and 1995, and predictive intervals are computed for each year up to 2050. The predicted confidence intervals for TF agree well with TFerrors in old population forecasts made by Statistics Norway. The method gives useful predictions for agespecific fertility up to around 2030. For later years, the intervals become too wide. Methods which do not take account of estimation errors in the ARIMAmodel coefficients underestimate the uncertainty for future TFvalues. The findings suggest that the margin between high and low fertility variants in official population forecasts for many Western countries are too narrow.

Lyngstad, Jan; Keilman, Nico Willem; Bojer, Hilde & Thomsen, Ib (1998). Poverty and Economic Inequality: Concepts, Measures, and Methods, In Nico Willem Keilman; Jan Lyngstad; Hilde Bojer & Ib O Thomsen (ed.),
Poverty and Economic Inequality in Industrialized Western Societies.
Scandinavian University Press.
ISBN 8200224937.
I Introduction.
s 3
 31
Show summary
This chapter introduces the various concepts, measures, and methods used in the contributions to this book, a book which approaches the study of poverty and economic inequality in industrialized Western countries from various points of view. We start by presenting two general approaches to welfare research and some of the general problems connected to the analysis of inequality as well as the analysis of poverty. We then address some problems of special relevance for the study of inequality (inequality measures and their properties) and for investigating poverty (poverty definitions and poverty measures). At the end we discuss income concepts and data sources, units of analysis, household definitions, equivalence scales, and problems related to longitudinal studies and to analyses of the impact of redistributional policies.
View all works in Cristin

Mazzuco, Stefano & Keilman, Nico (ed.) (2020). Developments in Demographic Forecasting.
Springer Nature.
ISBN 9783030424718.
261 s.
Show summary
This open access book presents new developments in the field of demographic forecasting, covering both mortality, fertility and migration. For each component emerging methods to forecast them are presented. Moreover, instruments for forecasting evaluation are provided. Bayesian models, nonparametric models, cohort approaches, elicitation of expert opinion, evaluation of probabilistic forecasts are some of the topics covered in the book. In addition, the book is accompanied by complementary material on the web allowing readers to practice with some of the ideas exposed in the book. Readers are encouraged to use this material to apply the new methods to their own data. The book is an important read for demographers, applied statisticians, as well as other social scientists interested or active in the field of population forecasting. Professional population forecasters in statistical agencies will find useful new ideas in various chapters.

Bengtsson, Tommy & Keilman, Nico (ed.) (2019). Old and New Perspectives on Mortality Forecasting.
Springer Nature.
ISBN 9783030050740.
350 s.
Show summary
More than 10 years have gone since the Swedish Social Insurance Agency published a series of booklets entitled Perspectives on Mortality Forecasting. Five volumes appeared in print, on different topics that all are relevant for anyone faced with the task of computing a forecast of mortality in future years. Each volume contained the papers presented in a series of workshops organized by the Stockholm Committee on Mortality Forecasting during the years 2002–2007. The current volume contains reprints of the contributions to the five booklets. Each part of this book corresponds to one original booklet. The field of mortality forecasting is in continuous change. The purpose of the current volume is to track this development by showing the reader what the main issues were some 10–15 years ago, together with an update. Therefore, the book starts with an introductory chapter, which summarizes recent new insights on the following topics: – The need for accurate mortality forecasting today for systems for pension, health care, and elderly care – Determinants and dynamics of life expectancy – Causes of death and lifestyle factors, such as smoking and obesity – Cohort and period perspectives – Compression of mortality – Joint forecasting of mortality in similar populations – From scenarios to stochastic modelling – The way conditions in early life affect mortality in later life – The increasing gap in life expectancy with respect to position in the income distribution Some of the material in the introductory chapter relates to only one of the five sections, while other items in the list above cut across themes. The last item brings up a new topic of mortality forecasting, which was not dealt with in any of the workshops.

Keilman, Nico (ed.) (2005). Perspectives on Mortality Forecasting No. 2: Probabilistic Models.
Swedish Social Insurance Agency.
ISBN 9175003260.
79 s.
Show summary
This is the second volume in a series presenting papers from workshops on mortality organized by the Stockholm Committee on Mortality Forecasting. The chapters focus on probabilistic (also labeled as stochastic) forecasts, in other words forecasts in which uncertainty has been quantified. Given a history of sizable forecasting errors, the first paper, by Nico Keilman, addresses the question of why demographic forecasts are uncertain. In the second paper Juha Alho outlines the statistical background of uncertain events and forecasts of these. In the third paper Maarten Alders and Joop de Beer sketch the approach taken by Statistics Netherlands in their stochastic forecast of mortality, while in the fourth paper Shripad Tuljapurkar presents a model for mortality analysis and forecasting that has proven to be feasible for probabilistic forecasts. He gives illustrations of US and Swedish mortality, and discusses also possible implications of uncertain mortality for future pension expenditures.

Bengtsson, Tommy & Keilman, Nico Willem (ed.) (2003). Perspectives on Mortality Forecasting.
Social Insurance Board.
ISBN 9189303245.
97 s.
Show summary
Mortality projections are an essential input for projections of the financial development of pension schemes. Governments and insurance companies all over the world rely on good mortality projections for efficient administration of their pension commitments. But the historical record shows that, during recent decades, demographers have continually underestimated improvements in life expectancy for persons 60 and older. The demographic models used in projecting mortality are usually based on statistical modeling of historical data. The question is, is it possible to improve forecast accuracy of mortality models, and if so, what do demographers need to do to achieve this result? This is the question that provided the impetus for forming the Stockholm Committee on Mortality Forecasting. It brings together a broad range of disciplines, and the work will be reported in the series �Social Insurance Studies�. The committee�s first volume addresses general issues. Why are we interested in improving mortality projections? Which forecast models are in use? How can we forecast life expectancy? How would an epidemiologist improve existing models? How can we use historical experience in getting forecasting right? Which projection models are currently in use in Finland, Norway, and Sweden? Key words: mortality forecasting, pensions, social insurance, statistics, epidemiology, history, Nordic countries

Keilman, Nico Willem; Lyngstad, Jan; Bojer, Hilde & Thomsen, Ib O (ed.) (1998). Poverty and Economic Inequality in Industrialized Western Societies.
Scandinavian University Press.
ISBN 8200224937.
View all works in Cristin

Keilman, Nico (2020). A probabilistic forecast for the population of Norway, In
Norway’s 2020 population projections: National level results, methods and assumptions.
Statistics Norway.
ISBN 9788258711497.
chapter.
s 177
 182

Keilman, Nico (2020). Modelling education and climate change. Nature Sustainability.
ISSN 23989629.
s 497 498 . doi: https://doi.org/10.1038/s4189302005158
Show summary
The Intergovernmental Panel on Climate Change (IPCC) is preparing its Sixth Assessment Report, to be published in 2022. The IPCC provides policy makers with the latest scientific insights on global climate change, its consequences and risks, and suggests possibilities for adaptation and mitigation. Because the future is unknown, the IPCC uses various models and scenarios for exploring how greenhouse gas emissions and other factors of climate change might develop, determined by natural science variables and socioeconomic variables. Here I argue that not only do the scenarios fail to cover a large enough range of possible futures, we do not know how probable they are, and that confuses policy makers.

Keilman, Nico & Mazzuco, Stefano (2020). Introduction, In Stefano Mazzuco & Nico Keilman (ed.),
Developments in Demographic Forecasting.
Springer Nature.
ISBN 9783030424718.
Chapter 1.
s 1
 20
Full text in Research Archive.
Show summary
Future trends in population size, age structure, regional distribution, and other demographic variables are important for a wide range of planning situations. Hence, many statistical agencies and independent researchers compute demographic forecasts at various levels of detail. The primary aim of this book is to sketch new developments in the field of demographic forecasting. This chapter addresses various issues taken up by the authors of this volume. We discuss deterministic and probabilistic approaches to forecast uncertainty, Bayesian and frequentist perspectives, the role of experts compared to purely data driven methods, and ways to communicate forecast results to the users.

Bengtsson, Tommy; Keilman, Nico; Alho, Juha; Christensen, Kaare; Palmer, Edward & Vaupel, James W. (2019). Introduction, In Tommy Bengtsson & Nico Keilman (ed.),
Old and New Perspectives on Mortality Forecasting.
Springer Nature.
ISBN 9783030050740.
Chapter 1.
s 1
 19

Keilman, Nico (2019). Probabilistic household and living arrangement forecasts.
Show summary
First, we discuss the shortcomings of deterministic forecasting models, and argue why probabilistic household and living arrangement forecasts are necessary for informed decisionmaking. Information from probabilistic forecasts allows policy makers, planners, and other forecast users in the fields of housing, energy, social security etc. to take appropriate decisions, because some household variables are more difficult to predict, and hence more uncertain, than others. It also guides them once actual developments start to deviate from the most likely path. New actions or updated plans are unnecessary as long as developments are likely to remain close to the expected future. Next, we review probabilistic household forecasts published since the turn of the century. An important issue is how to evaluate, expost facto, the accuracy of these probabilistic forecasts. We introduce the notion of a scoring function, the general idea of which is that a forecast that predicts the actual outcome with high probability should receive a better score than one that predicts the same outcome with lower probability. Scoring functions are useful for evaluating both interval forecasts (given in the form of prediction intervals with prespecified coverage probability), and forecasts available in the form of a full probability distribution, either analytically or as a sample. Finally, we give empirical results for scoring functions applied to the first known probabilistic household forecasts.

Keilman, Nico (2019). Probabilistic household and living arrangement forecasts.
Show summary
First, we discuss the shortcomings of deterministic forecasting models, and argue why probabilistic household and living arrangement forecasts are necessary for informed decisionmaking. Probabilistic forecasts inform the user of the uncertainty around expected household trends. As opposed to a deterministic forecast, which predicts only a few numbers (when projection variants are used) for a certain year, a probabilistic forecast tells us how likely it is that the number of households for a given year in the future will be within a certain range. Information of this kind allows policy makers, planners, and other forecast users in the fields of housing, energy, social security etc. to take appropriate decisions, because some household variables are more difficult to predict, and hence more uncertain, than others. It also guides them once actual developments start to deviate from the most likely path. New actions or updated plans are unnecessary as long as developments are likely to remain close to the expected future. Next, we review probabilistic household forecasts published since the turn of the century. An important issue is how to evaluate, expost facto, the accuracy of these probabilistic forecasts. We introduce the notion of a scoring function, the general idea of which is that a forecast that predicts the actual outcome with high probability should receive a better score than one that predicts the same outcome with lower probability. Scoring functions are useful for evaluating both interval forecasts (given in the form of prediction intervals with prespecified coverage probability), and forecasts available in the form of a full probability distribution, either analytically or as a sample. Finally, we give empirical results for scoring functions applied to the first known probabilistic household forecasts.

Keilman, Nico (2018). Eldrebølgen avlyst? Bedre termometer, men pasienten er fortsatt syk. Forskning.no.
ISSN 1891635X.

Keilman, Nico (2018). Forventet levealder og komprimering av dødelighet: perioder og kohorter.
Show summary
Forventet levealder beregnet for et bestemt kalenderår kan gi et misvisende bilde av den virkelige levealderen. Jeg viser hvordan periodelevealderen endrer seg når aldersmønsteret for dødelighet i kohorter endrer seg. Jeg analyserer også komprimering av dødelighet rundt forventet levealder i kohorter og perioder. Jeg viser hvorfor det er stadig sterkere komprimering i periodedødeligheten. Funnene er basert på periode og kohortdødelighet fra Human Mortality Database for 10 land, inkludert Norge.

Keilman, Nico (2018). Increasing (but insufficient?) optimism about future life expectancy. NIUSSP.
Show summary
In June 2018, Statistics Norway (2018) published its latest population forecast. The agency expects that the lifespan of men will increase to 87 years in 2050. This is six years more than the current value of 81 years. In a forecast published in 1993, the life expectancy prediction for 2050 was eight years lower than in today's population forecast. For women, the life expectancy prediction for 2050 increased by "only" four and a half years  from 84.5 years in the 1993 forecast to 89 years now. Demographers and other social scientists, in Norway as well as in other developed countries, have systematically underestimated the rate of increase of the life expectancy. Governments should be concerned about stronger ageing than official forecasts suggest.

Keilman, Nico (2018). Morgendagens eldre: større sjanse for å bo med partner og mindre sjanse for å bo alene.

Keilman, Nico (2018). Mortality shifts and mortality compression in period and cohort life tables.

Keilman, Nico (2018). Psoriasispasientenes livsløp  påvirker behandling deres parforhold?. BestPractice Dermatologi.
9(28), s 18 19 . doi: https://www.epaper.dk/bpnoderma/derma/derma28no_feb_2018/?offline=1
Show summary
Vår analyse viser at matematisk modellering av samspillet mellom sosiale og helsevariabler i prinsippet er mulig. Vår modell er en makromodell som simulerer livsløpet til en gruppe pasienter, som befinner seg i ulike tilstander i løpet av livet. Imidlertid trenger vi mer komplekse modeller som omfatter flere variabler (for eksempel sysselsettingsstatus, depresjon, fedme etc.) for mer realistiske simuleringer av interaksjoner som studeres. Flere variabler medfører mer komplekse simuleringsmodeller. I slike tilfeller kan det være nødvendig å bytte fra en makromodell til en mikromodell. I en slik modell simuleres livsløpet til hver enkelt person.

Keilman, Nico (2018). Training Course on Demographic Analysis and Population Projections, Pristina, Kosovo, 47 December 2018.

Keilman, Nico (2018). Økende optimisme om forventet levealder. Forskning.no.
ISSN 1891635X.
. doi: https://forskning.no/aldringstatistikkhelse/okendeoptimismeomforventetlevealder/1213385

Keilman, Nico; Pham, Quang Dinh & Syse, Astri (2018). Mortality shifts and mortality compression: The case of Norway, 19002060. Discussion papers. No. 884.
Show summary
Historically, official Norwegian mortality projections computed by Statistics Norway have consistently underpredicted life expectancy. The projected age distribution of deaths may be used to check if the official mortality projections are plausible. The aim of the paper is to verify whether the projections predict a continuation of the ongoing compression in mortality and of the steady upward shift in the ages at which people die. We use official period data on observed (19002015) and projected (2016 2060) sex and agespecific mortality to estimate the age distribution of life table deaths. We analyse trends in life expectancy at birth, modal and median ages at death, and standard deviation of the age distribution at ages > 30. The historical shifts towards longer longevity are projected to continue into the future. The projections suggest a steady increase in the modal and the median age at death for men and women towards values between 90 and 94 years in 2060. At present these ages are in the range 8390 years. Simultaneously, deaths become more concentrated around the mean, as the standard deviation of the age distribution is projected to fall continuously. Statistics Norway’s projection methodology is capable of tracking ongoing processes of mortality shifts towards higher ages and a compression of mortality around the modal and mean ages. Mortality projections could potentially benefit from including assessments of the age distribution of deaths.

Syse, Astri; Pham, Quang Dinh & Keilman, Nico (2018). Dødelighet og levealder, I: Stefan Leknes; Sturla Andreas Kise Løkken; Astri Syse & Marianne Tønnessen (red.),
Befolkningsframskrivingene 2018: Modeller, forutsetninger og resultater.
Statistics Norway.
ISBN 9788253797670.
Chapter 5.
s 57
 86

Kastrati, Avni; Uka, Sanije; Sojeva, Arijeta & Keilman, Nico (2017). Kosovo Population Projection 20172061.

Keilman, Nico (2017). Period and cohort life expectancy, mortality compression, and age at death distribution.
Show summary
We analyse the age at death distribution (AADD) of the life table, conventionally known as its d(x)column. We derive general analytical expressions for the moments of this distribution in a period life table, written as functions of the moments in a cohort life table. The first moment is the life expectancy, while the second moment reflects compression of the age at death. The expressions are partly based on an empirical regularity that we found for Norway in observed mortality data for the years 19002015, and projected mortality trends until 2100. Using the formula for the first period moment of the AADD, we derive the conditions under which cohort life expectancy increases faster than period life expectancy. We also find expressions for the period life expectancy in the year a birth cohort reaches an age equal to its own life expectancy, and for the gap between the period life expectancy in a certain year and the cohort life expectancy for the cohort born that year. Furthermore, we establish a relation between the period life expectancy in a certain year t, and the lag λ that leads to an equally large cohort life expectancy for a cohort born in year t – λ. This is the number of years it takes a period life expectancy to reach the current level of cohort life expectancy. Finally, using formulae for the second moment of the AADD, we derive expressions for lags and gaps in the standard deviations of the period and cohort AADDs. The latter measures are useful for describing trends in the compression of mortality. Our data show that under this model, as long as cohort life expectancies are lower than 94 years of age for Norwegian men and lower than 92 years of age for Norwegian women, their cohort life expectancies will be below period life expectancies a number of years later, where the time interval equals the cohort’s life expectancy. When cohort life expectancies are higher, they will exceed the period life expectancies at this particular lag. The gap between cohort and period life expectancies will grow by roughly half a year for every oneyear increase in cohort life expectancy – or by about one year of age for every period of 67 years. The lag λ defined above widens rapidly for Norwegian men and women, by approximately three to four years for every oneyear increase of the cohort life expectancy. For women we find that compression of morbidity, as judged by the standard deviation of the AADD above age 30, went more than twice as fast in reality (i.e. in birth cohorts) than what we see by inspecting period data only.

Keilman, Nico (2017). Population Projection in Kosovo, 20172061: Methodology and key findings.

Keilman, Nico; Pham, Quang Dinh & Syse, Astri (2017). Mortality shifts and mortality compression in Norway 19002100: periods and cohorts.
Show summary
BACKGROUND: Life expectancies and compression indicators from period life tables give a distorted picture of average life times and concentration of time of death. Cohort life tables reflect reality better. OBJECTIVE: To derive expressions for life expectancy and variance of age at death in period and cohort life tables METHODS: Focus on d_xcolumns of period and cohort life tables. When radix l_0 = 1, d_x is the Age at Death Distribution (AADD). RESULTS: We derived translation formulae for period life expectancy and variation in period age at death as a function of the moments of the cohort AADD.

Bernhoft, Aksel; Albihn, Ann; Hessen, Dag Olav; HolmboeOttesen, Gerd; Keilman, Nico; Børresen, Trond; Haug, Ruth; Nicolaysen, Anna Marie & Andersen, Regine (2016). Hvordan skal vi sikre et landbruk som kan fø oss i framtiden?.
Show summary
Menneskenes endring av naturen og jordas ressurser truer framtidens landbruk. Så hvordan skal den sårbare matproduksjonen best foregå på lang sikt når jordas befolkning øker?

Keilman, Nico (2016). A twosex model for first marriage.

Keilman, Nico (2016). Barnefamilier, fruktbarhetsnivå og samfunnsplanlegging.

Keilman, Nico (2016). Befolkning: Statistisk sentralbyrå bør endre praksis og publisere sannsynlighetsprognoser. Samfunnsøkonomen.
ISSN 18905250.
(3), s 59 66
Show summary
Med jevne mellomrom beregner Statistisk sentralbyrå (SSB) prognoser for befolkningen fram til 2100: alderssammensetningen for hvert år, årlige antall fødsler og dødsfall, størrelsen på inn og utvandringsstrømmene, antall innvandrere med ulik landbakgrunn, osv. Publisering av en ny prognose er planlagt for den 21. juni i år. I denne kommentaren bruker jeg resultater fra SSBs forrige befolkningsprognose fra 2014 for å gjøre rede for et viktig poeng: den framtidige befolkningsutviklingen er (som alle prognoser) usikker, men noen utviklinger er mer sannsynlige enn andre. Derfor bør SSB publisere sannsynlighetsprognoser, og ikke bare befolkningstall for framtidige år uten at brukeren vet hvor stor sjanse SSB knytter til den ene eller den andre befolkningsutviklingen.

Keilman, Nico (2016). Fødsler og fruktbarhet i Norge.
Show summary
I dette vedlegget gjøres det rede for at fruktbarhetsnivået i Norge er høyere enn det Samlet fruktbarhetstall (SFT) har vist de siste årene. Samtidig er det meget sannsynlig med en økning i årlig antall fødsler fra rundt 60 000 nå til i underkant av 70 000 om ti år. En tilfredsstillende forklaring for dagens fruktbarhetsnivå er vanskelig å gi. Derfor er predikeringer av framtidens fruktbarhet usikre. En viktig konsekvens er at Statistisk sentralbyrå bør publisere sine befolkningsprognoser i form av sannsynlighetsprognoser. Hvis vi ønsker å løse de problemene en aldrende befolkning medfører med hensyn til vårt velferdssystem, er ikke fruktbarhetsnivået av stor betydning. I denne sammenhengen er det viktig at politikere og samfunnsplanleggere vurderer tiltak innenfor arbeidsmarkedet samt skatte, helse, og pensjonssystemene.

Keilman, Nico (2016). Probabilistic demographic forecasts.

Keilman, Nico (2016). Probabilistic household forecasts for Denmark, Finland, and the Netherlands 20112041: Combining the Brass relational method with a Random Walk model.
Show summary
Probabilistic household forecasts to 2041 are presented for Denmark, Finland, and the Netherlands. Future trends in fertility, mortality and international migration are taken from official population forecasts. Time series of shares of the population in six different household positions are modelled as Random Walks with a Drift. Brass’ relational model preserves the age patterns of the household shares. Probabilistic forecasts for households are computed by combining predictive distributions for the household shares with predictive distributions of the populations, specific for age and sex. If current trends in the three countries continue, we will witness a development towards more and smaller households, often driven by increasing numbers of persons who live alone. We can be quite certain that by 2041, there will be between two and four times as many persons aged 80 and over who live alone when compared with the situation in 2011.

Keilman, Nico (2016). Tar SSB høyde for den usikre demografiske utviklingen i framtiden?.

Bijak, Jakub; Alberts, Isabel; Alho, Juha; Bryant, John; Buettner, Thomas; Falkingham, Jane; Forster, Jonathan; Gerland, Patrick; King, Thomas; Onorante, Luca; Keilman, Nico; O'Hagan, Anthony; Owens, Darragh; Raftery, Adrian E.; Ševcikova, Hana & Smith, Peter (2015). Letter to the Editor: Probabilistic Population Forecasts for Informed Decision Making. Journal of Official Statistics.
ISSN 0282423X.
31(4), s 537 544 . doi:
10.1515/JOS20150033
Show summary
Demographic forecasts are inherently uncertain. Nevertheless, an appropriate description of this uncertainty is a key underpinning of informed decision making. In recent decades, various methods have been developed to describe the uncertainty of future populations and their structures, but the uptake of such tools amongst the practitioners of official population statistics has been lagging behind. In this letter we revisit the arguments for the practical uses of uncertainty assessments in official population forecasts, and address their implications for decision making. We discuss essential challenges, both for the forecasters and forecast users, and make recommendations for the official statistics community.

Keilman, Nico (2015). Dimension reduction by Brass' Relational Model: Household Dynamics in Five European Countries.
Show summary
We use techniques of data dimension reduction to model and predict age patterns of household dynamics in a multicountry context. Probabilistic household forecasts are computed to 2041 for Denmark, Finland, Germany, the Netherlands and Norway. Time series of shares of the population by household position are modelled as Random Walk with Drift. Brass’ relational model preserves the age patterns of the household shares. Predictive distributions for the household shares are combined with predictive distributions of the populations. The predicted continuation of current trends towards more and smaller households is driven by increasing numbers of persons who live alone. Relative prediction uncertainty in more numerous household types is smaller than that in less numerous types.

Keilman, Nico (2015). Jordens befolkning, "antall munner å mette".

Keilman, Nico (2015). Probabilistic household forecasts for five countries in Europe. Full text in Research Archive.
Show summary
We show how techniques of data dimension reduction can be used to predict patterns of household dynamics in a multicountry context. Probabilistic household forecasts are presented for Denmark, Finland, Germany, the Netherlands, and Norway, spanning the period 20112041. Starting point is the population of each country broken down by age, sex, and household position as reported in the census round of 2011. Future trends in fertility, mortality and international migration are taken from official population forecasts. For changes in household structure we rely on time series of household data. Long series of household data, in which the population is broken down by household position, age, and sex, are available for Denmark (19812007) and Finland (19882009) from the population registers in these countries. For the Netherlands the series are rather short (19952011). Annual shares of the population by household position, age, and sex for the three time series countries are modeled using an approach that builds on Brass’ relational model originally developed to model the age pattern of mortality. We find that the household shares can be modelled as Random Walks with Drifts (RWD), independent of country. The Brass approach preserves the age patterns of the household shares. Future household shares are found by extrapolating the RWD processes. This results in household share forecasts, as well as standard errors of the forecasts. Correlations across ages and between men and women are estimated from model residuals. No time series data are available for Germany or Norway. For Germany, we use household transition rates borrowed from Denmark and Finland, but adjusted to cohabitation and marriage levels from the German Generation and Gender Survey. For Norway, we have household transition rates for the year 2010 from the population register. Future household patterns for these two countries are computed by using the multistate household model LIPRO, in which the household transition rates are applied to the household pattern from the census. Uncertainty parameters are borrowed from the time series analyses for Denmark, Finland and the Netherlands. The results show a continuation of current trends towards more and smaller households, often driven by increasing numbers of persons who live alone. The number of households increases faster than population size, which leads to falling average household size. A very consistent finding is that larger households are easier to predict than smaller households, at least when uncertainty is considered in a relative sense.

Keilman, Nico (2015). Stochastic household forecasts for Denmark, Finland, Germany, Netherlands, Norway.

Keilman, Nico (2015). Stochastic household forecasts for five countries in Europe.

Keilman, Nico (2014). Dimension reduction of household parameter time series by the Brass logit model.

Keilman, Nico (2014). Norway’s new public pension system: Is it robust against unexpected life expectancy developments?.
Show summary
Norway introduced a new system for public old age pensions in January 2011. The new system leads to lower pension expenditures than the old system, because annual pension benefits under the new system are inversely proportional to the remaining life expectancy of those who retire. We can expect public pension expenditures equal to 170 billion Norwegian crowns (NOK) in 2030 and 288 billion NOK in 2050. But expenditures will be larger if retirees live longer than expected. We cannot be certain about the pace of mortality decline in the future. Therefore we have computed a probabilistic population forecast for Norway to 2050 and analysed the consequences of population growth for public old age pension expenditures. A new insight is that the new system is much less robust against unexpected longevity shocks than what was assumed earlier, in spite of the longevity adjustment. The reason is that annual pension benefits are determined when a person retires. After retirement, a retiree’s annual benefits remain the same, even when mortality changes.

Keilman, Nico (2014). Probabilistic population and household forecasts.

Keilman, Nico (2013). Pensjonsreformen kan bli mye dyrere enn mange tror.
Show summary
Det nye pensjonssystemet medfører lavere offentlige utgifter enn det gamle systemet. Statistisk sentralbyrå forventer en innsparing på 31 mrd. kroner i 2030 og 50 mrd. kroner i 2050. Hvis pensjonistenes levealder øker fortere enn forventet blir innsparingene mindre. Det er usikkert hvor høy denne levealderen kommer til å bli i fremtiden. Derfor har vi laget en sannsynlighetsprognose for befolkningsutviklingen fram til 2050 og beregnet konsekvensene for pensjonsutgifter. Vi finner at de forventede pensjonsutgifter i 2050 er langt fra sikre.

Keilman, Nico (2013). Probabilistic demographic projections.

Keilman, Nico (2013). Uncertainty in population projections  with special reference to the UK.

Keilman, Nico (2013). "What is happening in modern demography: life course analysis, policy evaluation, dimension reduction, prediction intervals in demographic forecasting".

Christiansen, Solveig Tobie Glestad & Keilman, Nico (2012). Probabilistic household forecasts based on register data: the case of Denmark and Finland.

Keilman, Nico (2012). Challenges for statistics on households and families.

Keilman, Nico (2012). College huishoudensprognoses.

Keilman, Nico (2012). Integrated modelling of European migration flows: Methodology and main results.

Keilman, Nico (2012). Missing girls in China will lead to fewer future births than previously thought.

Keilman, Nico (2012). Missing girls in China will lead to fewer future births than previously thought.

Keilman, Nico (2012). Onzekerheid in demografische prognoses.

Keilman, Nico & Van Duin, Coen (2012). Stochastic household forecasts by coherent random shares predictions.

Wisniowski, Arkadiusz; Keilman, Nico; Bijak, Jakub; Christiansen, Solveig; Forster, Jonathan; Smith, Peter & Raymer, James (2011). Augmenting migration statistics with expert knowledge.

Graziani, Rebecca & Keilman, Nico (2010). The sensitivity of the Scaled Model of Error with respect to the choice of the correlation parameters: A Simulation Study. Memorandum from Department of Economics, University of Oslo. 22.
Show summary
Abstract: The Scaled Model of Error has gained considerable popularity during the past ten years as a device for computing probabilistic population forecasts of the cohortcomponent type. In this report we investigate how sensitive probabilistic population forecasts produced by means of the Scaled Model of Error are for small changes in the correlation parameters. We consider changes in the correlation of the agespecific fertility forecast error increments across time and age, and changes in the correlation of the agespecific mortality forecast error increments across time, age and sex. Next we analyse the impact of such changes on the forecasts of the Total Fertility Rate and of the Male and Female Life Expectancies respectively. For age specific fertility we find that the correlation across ages has only limited impact on the uncertainty in the Total Fertility Rate. As a consequence, annual numbers of births will be little affected. The autocorrelation in error increments is an important parameter, in particular in the long run. Also, the autocorrelation in error increments for age specific mortality is important. It has a large effect on long run uncertainty in life expectancy values, and hence on the uncertainty around the elderly population in the future. In empirical applications of the Scaled Model of Error, one should give due attention to a correct estimation of these two parameters. Key words: Scaled model of error, Stochastic population forecast, Probabilistic cohort component model, Sensitivity, Correlation JEL classifications: C15, C49, C63, J41

Keilman, Nico (2010). On future households.
Show summary
We develop a method for computing probabilistic household forecasts which quantifies uncertainty in the future number of households of various types in a country. A probabilistic household forecast helps policy makers, planners and other forecast users in the fields of housing, energy, social security etc. in taking appropriate decisions, because some household variables are more uncertain than others. Deterministic forecasts traditionally do not quantify uncertainty. We apply the method to data from Norway.We find that predictions of future numbers of married couples, cohabiting couples and oneperson households are more certain than those of lone parents and other private households. Our method builds on an existing method for computing probabilistic population forecasts, combining such a forecast with a random breakdown of the population according to household position (single, cohabiting, living with a spouse, living alone etc.). In this application, uncertainty in the total numbers of households of different types derives primarily from random shares, rather than uncertain future population size. A similar method could be applied to obtain probabilistic forecasts for other divisions of the population, such as household size, health or disability status, region of residence and labour market status.

Keilman, Nico (2009). Failure and success: Insample and outofsample demographic forecasts.
Show summary
There is a growing literature on the quality of population forecasts. All of these forecasts relate to the future, and their accuracy can be established ex post facto. However, a large literature is concerned with explanation and prediction of current, not future behaviour. This paper explores the predictive success of those insample predictions. Likelihood statistics for 96 life course analyses published in key demographic journals are investigated. To what extent does predictive success depend on subject matter, on sample size, on type of dependent variable? We find that variables for fertility/reproduction and pair formation/dissolution are difficult to explain, in particular events. For mortality and health variables the situation is somewhat better.

Keilman, Nico (2008). Concern in the European Union about Low Birth Rates. European View.
ISSN 17816858.
7, s 333 340 . doi:
10.1007/s1229000800555
Show summary
While the European Union recognizes the importance of ageing and falling populations in a number of official documents, the findings require focus. In this article, the author first identifies the documents’ failure to name a target birth rate. Second, he stresses that the often mentioned fertility level of an average of 1.5 children per woman underestimates the real figure. Thirdly, he points out the futility of fertility policies when ageing processes will continue regardless of birth rate changes. In response, policies must be coordinated in a number of areas including gender policy, employment policy, immigration policy, housing policy, family policy and economic policy.

Keilman, Nico (2008). Emerging family and household types in Europe: Issues, definitions and classifications.
Show summary
Earlier work in connection with the preparation of the 2010 round of Population and Housing Censuses for member countries of the United Nations Economic Commission of Europe revealed that in the 2000 round countries experienced difficulties in collecting data for reconstituted families and samesex couples. Other problematic issues were the distinction between private and institutional households, and the notions of de facto and legal marital status. In addition to problems encountered in population censuses, some emerging family and household types pose problems when mapped by means of sample surveys. The UNECE Task Force on Families and Households has identified a number of additional types for which problems exist. In this paper I shall briefly discuss the problems mentioned above, and refer to possible definitions and classifications for reconstituted families, samesex couples and couples in a LATrelationship (Living Apart Together).

Keilman, Nico (2008). Scenarios of demographic change: the role of probabilistic approaches in population projection.
Show summary
This brief note first summarizes the four main demographic trends that are expected for the coming decades, and discusses two techniques that demographers use for mapping future population trends, namely scenarios and probabilistic projections.

Keilman, Nico (2008). Uncertain population forecasts.

Keilman, Nico (2008). Using deterministic and probabilistic population forecasts. Interdisciplinary Communications.
ISSN 08098735.

Keilman, Nico & Veløy, Chris (2008, 17. september). AFP en ulykke for landet. [Internett].
http://nrk.no/programmer/tv/schrodingers_katt/1.6224469.

Keilman, Nico & Veløy, Chris (2008, 17. september). Europa kan halveres på to generasjoner. [Internett].
http://nrk.no/programmer/tv/schrodingers_katt/1.6224278.

Alho, Juha; Cruijsen, Harri & Keilman, Nico (2006). Empiricallybased specification of forecast uncertainty.

Keilman, Nico (2006). European demographic forecasts have not become more accurate during the past 25 years.
Show summary
Nowadays, demographers, population statisticians, and population forecasters have richer data, more refined theories of demographic behaviour, and more sophisticated methods of analysis than they had two or three decades ago. This scientific progress should have made it easier to predict demographic behaviour. But my analyses of the errors in old forecasts show that demographic forecasts published by statistical agencies in 14 European countries have not become more accurate over the past 25 years. My findings demonstrate that scientific progress in population studies during the previous two to three decades might have been too slow to keep up with less predictable demographic behaviour of populations in European countries. There is no reason to be more optimistic about US Census Bureau forecasts. I argue that population forecasts are intrinsically uncertain, and hence should be couched in probabilistic terms.

Keilman, Nico (2006). Livslängd och invandring spräckte prognoserna. Välfärd.
ISSN 16516710.
(3), s 6 7
Show summary
En analys av befolkningsprognosernas träffsäkerhet i 18 europeiska länder har visat att demograferna har varit för försiktiga i sina antaganden. I Europa har både medellivslängden och invandringen ökat kraftigare under de senaste decennierna än vad demograferna räknade med.

Keilman, Nico; Cruijsen, Harri & Alho, Juha (2006). Diverging views of future demographic trends.

Alders, Maarten; Keilman, Nico & Cruijsen, Harri (2005). Assumptions for national stochastic population forecasts in 18 European countries.
Show summary
In order to develop and implement stochastic population forecasting for European countries, the European Commission has commissioned the project �Changing population of Europe: uncertain future� to researchers from Joensuu University and the statistical offices of Finland, the Netherlands and Norway. The objective was to develop a general methodology for assessing predictive distributions for fertility, mortality and migration. Generally speaking the assumptions underlying stochastic population forecasts can be assessed by means of analyses of errors of past forecasts, modelbased estimates of forecast errors, and expert judgement. All these approaches have been thoroughly investigated in the project. The paper summarises and discusses the results of the different approaches. It demonstrates how the, sometimes conflicting, results can be synthesised into a consistent set of assumptions. Finally, the paper presents the main assumptions about the total fertility rate, life expectancy at birth of men and women, and net migration for 18 European countries.

Alho, Juha; Alders, Maarten; Cruijsen, Harri; Keilman, Nico; Nikander, Timo & Pham, Quang Dinh (2005). Population decline postponed in Europe. Statistical Magazine.
Show summary
The United Nations and Eurostat recently predicted that the population in 18 European countries will fall in the future. However, the results of a new forecast show that there is reason to expect higher immigration and lower mortality than that predicted by the UN and Eurostat. Hence, the population decline will happen later, and perhaps as late as 2050. For Norway, a population of 5.75 million is anticipated by 2050, slightly higher than the official forecast by Statistics Norway in 2002.

Keilman, Nico (2004). Long term public old age pension obligations in a stochastic demographic setting in Norway.

Keilman, Nico Willem (2004). Conséquences démographiques et sociales d'une faible fécondité pour les structures familiales en Europe.

Keilman, Nico & Kinsella, Kevin (2004). Families and households in the 2000 round of censuses in ECE member countries.

Keilman, Nico & Vogt, Yngve (2004). Utrygg pensjonsreform: Samfunnet kan tape skyhøye beløp. Aftenposten (morgenutg. : trykt utg.).
ISSN 08043116.
Show summary
Datasimuleringer viser at pensjonskommisjonen og politikerne er naive. De har ikke tatt hensyn til usikkerhetsanalyser i befolkningsutviklingen under planleggingen av det nye pensjonssystemet. Det kan koste samfunnet dyrt.

Keilman, Nico Willem & Pham, Quang Dinh (2004). Empirical errors and predicted errors in fertility, mortality and migration forecasts in the European Economic Area. Discussion papers. No. 386.
Show summary
We analyse empirical errors observed in historical population forecasts produced by statistical agencies in 14 European countries since 1950. The focus is on forecasts for three demographic variables: fertility (Total Fertility Rate  TFR), mortality (life expectancy at birth), and migration (net migration). We inspect forecast bias and forecast accuracy in the historical forecasts, as well as the distribution of the errors. Finally, we analyse for each of the three variables correlation patterns in forecast errors across countries and, for mortality, the correlation between errors for men and women. In the second part of the report we use time series model to construct prediction intervals to 2050 for the TFR, the life expectancy for men and women, and net migration in 18 European countries. GARCH models are used for fertility and mortality, while net migration is modelled as an autoregressive process. Keywords: stochastic population forecast, empirical forecast errors, prediction intervals, GARCHmodels, TFR, life expectancy, net migration, EEA. JEL classification: C22, J11.

Keilman, Nico Willem (2003). Demographic and social implications of low fertility for family structures in Europe.
Show summary
The study sets out to investigate the relationship between low fertility and new patterns in the family and nonfamily sectors. It examines the social implications of childlessness, single child families, and other family sizes, with an emphasis on questions of social cohesion. Firstly a theoretical perspective on childlessness is given. This is followed by an analysis of the impact of changes in birth orderspecific fertility on family size using the results from a simulation study which analyses how family sizes change when the level and timing of age and birth orderspecific fertility change. The final section discusses possible consequences for social cohesion and social exclusion of the trends identified in the previous sections with a focus on poverty.

Keilman, Nico Willem (2003). Demographic implications of low fertility for family structures in Europe.
Show summary
This text reports the findings from the study on �Demographic and social implications of new patterns in the family and nonfamily sectors� undertaken for the Council of Europe. The study investigates the relationship between low fertility and new patterns in the family and non family sectors. The study discusses also the social implications of childlessness, single child families, and other family sizes, thereby focusing on social cohesion. The study is limited to birth orderspecific fertility, and couples, thereby not distinguishing between married and cohabiting couples. First, a theoretical perspective on childlessness is given. Next, the impact of changes in birth orderspecific fertility on family size is analysed, by using the results from a simulation study, in which I analyse how family sizes change when the quantum and tempo of age and birth orderspecific fertility change. The model applied here is a multidimensional life table. The final section discusses possible consequences for social cohesion and social exclusion of the trends identified in the previous section, with a focus on poverty.

Keilman, Nico Willem (2003). Do Cause of Death analyses improve the accuracy of mortality forecasts?.

Keilman, Nico Willem (2003). Model based errors and empirical errors in fertility and mortality forecasts for the Nordic countries.

Keilman, Nico Willem (2002). Book review of S. K. Smith, J. Tayman, and D. Swanson ¿ State and Local Population Projections: Methodology and Analysis¿. Population Studies.
ISSN 00324728.
56(3), s 340 341

Keilman, Nico Willem (2002). DEMOGRAPHIC AND SOCIAL IMPLICATIONS OF LOW FERTILITY FOR FAMILY STRUCTURES IN EUROPE.
Show summary
This text is a progress report for the study on ¿Demographic and social implications of new patterns in the family and nonfamily sectors¿ undertaken for the Council of Europe. The study investigates the relationship between low fertility and new patterns in the family and nonfamily sectors. The nonfamily sector is to be understood as consisting of oneperson households and childless couples (i.e. couples who have not, or not yet, given birth to any children). The study discusses also the social implications of childlessness, single child families, and other family sizes, thereby focusing on social cohesion. The study is limited to birth orderspecific fertility, and couples, thereby not distinguishing between married and cohabiting couples. First, a theoretical perspective on childlessness is given. Next, the impact of changes in birth orderspecific fertility on family size is analysed, by using the results from a simulation study, in which I analyse how family sizes change when the quantum and tempo of age and birth orderspecific fertility change. The model applied here is a multidimensional life table. The final section discusses possible consequences for social cohesion and social exclusion of the trends identified in the previous section.

Keilman, Nico Willem & Pham, Quang Dinh (2002). Time series based errors and empirical errors in fertility forecasts in the Nordic countries.
Show summary
We use time series models to derive model based prediction intervals for the Total Fertility Rate (TFR) in Norway, Sweden, Finland, and Denmark up to 2050. For the short term (510 yrs), expected TFRerrors are compared with empirical forecast errors observed in historical population forecasts prepared by the statistical agencies in these countries since 1969. Moreover, we analyse mediumterm and longterm (up to 50 years) error patterns based on socalled naïve forecasts, i.e. forecasts that assume that recently observed TFRlevels also apply for the future.

Keilman, Nico Willem; Pham, Dinh Quang & Hetland, Arve (2001). Norway's uncertain demographic future.
Show summary
The demographic future of any country is uncertain. There is not just one possible future, but many. Some of these are more probable than others. Therefore, an exploration of a country's demographic future should include two elements: first, a range of possible outcomes, and second, a probability attached to that range. Together, these two constitute a predictive interval for the variable concerned. The current paper presents predictive intervals for the future size and agesex composition of Norway's population up to the year 2050. These intervals have been obtained by stochastic simulation, on the basis of repeated runs of the cohortcomponent model. We describe the method briefly. We also compare our results with those of the official population forecast computed by Statistics Norway, and explain why the latter forecast does not give an appropriate indication of forecast uncertainty.

Keilman, Nico Willem (2001). Age structural transitions in historical perspective: What can we learn from the past?.
Show summary
The paper reviews what can we learn from age structures, and changes therein, when using stable population theory, inverse/back projections and generalizations thereof, variable growth rate method, and old population forecasts. Following the discussion of the accuracy of age structure forecasts for the elderly, I also speculate how we can improve the accuracy of mortality forecasts. The focus is on industrialized countries.

Keilman, Nico Willem (2001). Children and time: The Norwegian model. NIDI Hofstee Lectures Series. 6.
Show summary
Norway's family policy has received widespread attention in the Netherlands in recent years. The extremely generous parental leave arrangements and reasonable level of childcare facilities are designed to make it easier for Norwegian women to combine a family and a professional career. In this lecture I shall look into the impact of this family policy on the labour force participation and fertility of women in Norway. The more unintentional effects, such as the time pressures this places on many families with young children, will also be dealt with. Should the Netherlands follow Norway's example?

Keilman, Nico Willem (2001). Population Forecasting.
Show summary
Cohort component method, ratio method, empirical forecast accuracy, stochastic forecasts, use of population forecasts

Keilman, Nico Willem (2001). Types of models for projecting mortality.
Show summary
The short paper presents a systematic overview of various types of models for projecting mortality. The framework is that of population forecasts of the cohortcomponent type for the elderly at the national level.I distinguish between demographic and nondemographic models. Demographic models are rather mechanistic, although they include various degrees of judgement. Demographic models can be divided further into extrapolation models and interpolation models. Nondemographic models build on epidemiology and biomedical sciences, and thus are much more of a causal nature than the demographic ones. I distinguish between macro models and micro models.

Keilman, Nico Willem; Pham, Quang Dinh & Hetland, Arve (2001). Norway's Uncertain Demographic Future. Sosiale og økonomiske studier. 105.
Show summary
This report presents prediction intervals for the future population of Norway broken down by age and sex to the horizon 2050. We combined Monte Carlo simulation with a cohortcomponent population projection model to obtain 5,000 simulations of the population. The odds are four to one (80 per cent chance) that Norway's population, now 4.5 million, will number between 4.3 and 5.4 million in the year 2025, and 3.76.4 million in 2050. There is a clear tradeoff between greater accuracy (higher odds) and higher precision (narrower intervals). Odds of 19 to one (95 per cent chance) result in a wider interval: 4.15.7 million in 2025, and 3.27.3 million in 2050. The probabilistic population forecasts of the youngest and the oldest age groups show largest uncertainty, because fertility and mortality are hard to predict. As a result, prediction intervals in 2030 for the population younger than 20 years are so wide, that the forecast is not very informative. International migration shows large prediction intervals around expected levels, but its impact on the age structure is modest. In 2050, uncertainty has cumulated so strongly, that intervals are very large for virtually all age groups, in particular when the intervals are judged in a relative sense (compared to the median forecast). According to our statistical model, the expected accuracy of the total population size forecast published by Statistics Norway is somewhat below twothirds on the long run, and a little above that level on the short run.

Keilman, Nico Willem (2000). Andrew Hinde, Demographic Methods. London: Arnold Publishers 1998. European Journal of Population.
ISSN 01686577.
16(2), s 187 188

Keilman, Nico Willem (2000). Demographic Translation: From Period To Cohort Perspective And Back. Memorandum fra Sosialøkonomisk institutt. 06/2000.
Show summary
[Internet http://www.sv.uio.no/sosoek/memo/memopdf/memo0600.pdf] When successive birth cohorts of women get their children at progressively lower ages, births that would have occurred during a certain period without changes in the timing, are now "squeezed" into a shorter period. This pushes period fertility up, and the period Total Fertility Rate (TFR) will be inflated, compared to the TFR that would have occurred without changes in the timing. In general, even when the number of children per woman is constant over successive birth cohorts, period fertility levels may vary  they are inflated in years in which childbearing is accelerated, and deflated when women postpone childbearing. Thus period fertility cannot be used as a reliable indicator for the level of cohort fertility: period fertility may be "distorted" in times of tempo changes in cohort fertility. At the same time cohort fertility cannot be fully understood without studying periods. These qualitative links between period and cohort fertility are straightforward. But the detailed interplay between period and cohort fertility, both its quantum (level) and tempo (timing) aspects, can be formalized mathematically. The resulting expressions constitute the core of what has become known as the theory of demographic translation, a term coined by Norman Ryder. This chapter gives a brief general overview of demographic translation theory. It integrates Ryder's findings from the 1960s that he applied to agespecific fertility, with more recent insights, which can be used for analysing other demographic processes, such as childlessness, first marriage, and divorce.

Imhoff, Evert Van & Keilman, Nico Willem (1999). On the quantum and tempo of fertility: Comment.
Show summary
[Internet: http://www.nidi.nl/public/nidi_wp_1999_2.pdf] The paper comments on the Bongaarts/Feeney adjustment method for period Total Fertility. First, the method is based on unsuitable fertility measures, which tend to exaggerate translational distortion caused by tempo shifts in the fertility of birth cohorts. Second, the BF assumption that periodbyperiod changes are independent of age and cohort is not supported by the data.

Keilman, Nico Willem & Hetland, Arve (1999). Simulated confidence intervals for future period and cohort fertility.
Show summary
[Internet www.unece.org/stats/documents/1999.05.projections.htm] Confidence intervals for future period and cohort fertility in Norway are computed. A multivariate ARIMA model is estimated for the TFR, the mean age at childbearing, and the variance in that age, all three in a period perspective. The data are derived from Norwegian vital statistics, 19451995. Predicted values for the three parameters and their confidence intervals up to 2050 are obtained by means of simulation. A Gamma curve is used to generate future agespecific fertility rates and their confidence intervals. These are rearranged to yield confidence intervals for the Completed Cohort Fertility for birth cohorts up to 2010. A comparison is made with assumptions on Completed Cohort Fertility used in Statistics Norway's most recent population forecast. Our results indicate that the predictability of fertility is limited.

Keilman, Nico Willem & Pham, Quang Dinh (1998). Predictive Intervals for AgeSpecific Fertility. Documents  Statistics Norway, Research Department. 98/19.
Show summary
A Vector Autoregression (VAR) model is combined with a Gamma curve to predict confidence intervals for agespecific birth rates by oneyear age groups. The method is applied to observed agespecific births in Norway between 1900 and 1995, and predictive intervals are computed for each year up to 2050. The predicted confidence intervals for Total Fertility (TF) agree well with TFerrors in old population forecasts made by Statistics Norway. The method gives useful predictions for agespecific fertility up to around 2030. For later years, the intervals become too wide. Methods which do not take account of estimation errors in the VARmodel coefficients underestimate the uncertainty for future TFvalues. The findings suggest that the margin between high and low fertility variants in official population forecasts for many Western countries are too narrow.

Tysse, Tone Ingrid & Keilman, Nico Willem (1998). Utvandring blant innvandrere. Rapporter / Statistisk sentralbyrå. 15.
Show summary
Innvandrernes utvandring i perioden 19751995 analyseres, med fokus på innvandringskullet som kom i 19861990. Fra dette kullet var 43 % utvandret før utgangen av 1995. Vestlige innvandrere utvandret hyppigst. Yrkesaktive, flyktninger, og familiegjenforenede blir i stor grad værende. Arbeidsledighet hemmer også utvandring. De som kom i ung alder har høy utvandringstilbøyelighet. Ugifte utvandrer hyppigere enn gifte.
View all works in Cristin
Published Sep. 23, 2010 10:40 AM
 Last modified May 26, 2021 2:46 PM