INTERNATIONAL JOURNAL OF POPULATION GEOGRAPHY Int. J. Popul. Geogr. 8, 319-344 (2002) Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/ijpg.261 Populations of the Industrial World — A Convergent Demographic Community? D. A. Coleman* Department of Social Policy and Social Work, University of Oxford, Bamett House, Wellington Square, Oxford 0X1 2ER, UK ABSTRACT Europe's populations have emerged from a long period of demographic transition. The old demographic anciens regimes no longer exist. The end-point of demographic transition is often assumed to be convergence to a new stable post-transitional regime. Convergence in demographic characteristics among all European countries might be expected to follow their acquisition of common economic and social structures. Remaining differences would then merely reflect different positions on the same trajectory to a common statistical destination. However, socio-economic convergence remains incomplete and cannot be assumed to bring uniform demographic responses in its wake. This paper measures demographic convergence in industrial societies to see if they are acquiring a common pattern. Despite some clear upper limits to demographic variables, for example fertility, in proportional terms international demographic variation remains high and is declining only slowly, if at all. Either the expectation of convergence is wrong, or we are looking for it too soon. Copyright © 2002 John Wiley & Sons, Ltd. Received 25 September 2001; revised 4 April 2002; accepted 17 April 2002 * Correspondence to: D. A. Coleman, Department of Social Policy and Social Work, University of Oxford, Barnett House, Wellington Square, Oxford OX1 2ER, UK. E-mail: david.coleman@socres.ox.ac.uk Keywords: demographic convergence; demographic transition; international comparisons; fertility trends; mortality trends INTRODUCTION A 11 industrial societies have now emerged /\ from the turmoil of the first demo--Z. A. graphic transition. Many achieved replacement fertility over 60 years ago; all share a basic pattern of low birth and death rates, ageing population structures and low population growth. Has the developed world thus become 'globalised', sharing a new common form of demographic behaviour functionally appropriate to late industrial or post-industrial society (Chasteland and Chesnais, 1997)? RESEARCH QUESTIONS This paper asks whether tendencies towards convergence are evident in the industrial world today. That modest empirical goal is, in fact, quite hard to achieve. The Assumption of 'Convergence' The assumption of convergence follows from a number of considerations: firstly that socioeconomic characteristics and ways of life will become similar across the industrial world; secondly, that demographic characteristics depend upon such factors (e.g. Nerlove, 1973). It is expected that the end-product of transition would be a stabilisation around convergent low rates of birth and death which would generate zero or negative rates of natural increase (Coward, 1986), supported, Copyright © 2002 John Wiley & Sons, Ltd. 320 D. A. Coleman according to van de Kaa (1999) by high levels of immigration. On that view, current patterns of diversity can be accounted for by different rates at which various societies are moving (Roussel, 1992) towards the same demographic 'point omega'. If this is so, then the end of demographic history may be in sight. Socio-economic Convergence Outside demography, several large-scale theories predict socio-economic convergence. Marxist analyses hold convergence to be more or less historically inevitable, as nations develop along defined tracks only temporarily diverted by the contradictions of capitalism from the uniform certainties of the future workers' state. In so far as events may not necessarily have turned out to the advantage of such views, the eventual ascendancy of liberal democracy is held instead to bring history to a different but equally final and uniform conclusion (Fukuyama, 1992). Neo-classical economic theory holds that free markets bring economic convergence. Economic rationality and the competitiveness of the industrial mode of production in an increasingly globalised economy will force all industrial populations on to the same track of development (e.g. Abramovitz, 1986; UNECE, 2000). In addition, convergence may be a deliberate aim of economic or welfare policy. Regional, national and EU policies aim to equalise income levels, labour costs and welfare entitlements at different geographical scales. International systems impose both informal and formal, overt convergent pressure on national economies and welfare systems, and therefore on their demography. Convergence was also an explicit aim of the former Communist countries of Europe, involving the elimination of regional and social differences, accelerated industrialisation, regional planning and controlled population distribution, pronatalist policies and the 'abolition of the contradiction between town and country' (Macura, 1974). Finally, the experience of trade, migration (national and international) and the international flow of information through the media and information technology will lead to the diffusion of common preferences and aspirations: so-called 'globalisation'. Economic convergence has been widely believed to go hand in hand with convergence in social structure and politics, in labour relations and incomes, in work habits (Kerr et at, 1962; Kerr, 1983) and chances of social mobility (Erikson and Goldthorpe, 1993). In open societies, urbanisation and the forms of urban life are expected to develop similarly in different countries (Hawley, 1950). Some empirical studies show that institutional convergence has indeed occurred: for example, in EU social welfare polices (Greve, 1994, Kosonen, 1995) and, at least in the earlier post-war period, social security, education and unemployment (Williamson and Fleming, 1996). While convergence is probably the predominant view, other studies contradict it, in respect of the economy and other structures of industrial societies (e.g. employment, trade unions and labour relations: Goldthorpe, 1984; social mobility: Marshall et at, 1998). The economies of poorer countries often do not grow faster than rich ones; instead they tend to settle on different growth paths (UNECE, 2000). Persistent diversity is seen in national healthcare systems, reflecting specific and apparently durable differences in national political culture (Field, 1998). While overall women's workforce participation has been rising everywhere, great disparities remain between countries in the hours worked by women and in other conditions (OECD, 1994). Demographic Convergence However, given convergent socio-economic trends, demographic convergence is assumed to follow. Economic determinism in demography is challenged by the claim that in modern societies, demographic behaviour is influenced more by ideational factors than by material ones (van de Kaa, 1987, 1994; Lesthaeghe, 1995). But the expected convergent outcome remains similar. As developed economies satisfy basic needs, 'post-material' values emphasising self-realisation and autonomy gain ground. Accordingly, inhibitions on personal and sexual behaviour are weakened, and traditional norms relating to marriage and children are eclipsed. Globalisation of these attitudes is mediated by universal literacy, global trade, media and telecommunications, Copyright © 2002 John Wiley & Sons, Ltd. Int. ]. Popul Geogr. 8, 319-344 (2002) Convergent Demographic Communities 321 and the spread of common languages, notably English; in short, by many of the same processes by which economic convergence is also expected to arise (Inglehart, 1990). We are all post-materialists now, or soon will be. All these considerations suggest that uniform demographic patterns will develop in modern societies (Jones, 1993, Mellens, 1999), in particular in fertility (Roussel, 1992) and family type (Goode, 1963). Population projections used to assume a diminution of current fluctuations and diversity and the convergence at least of fertility on to similar and sometimes identical values. Standard Eurostat projections were of this type, although some recent projections made with Eurostat help have explored 'diversity' scenarios (de Beer and van Wissen, 1999). Remaining demographic contrasts would only persist for as long as it took for countries at different stages on the same sequence to catch up. On that view, rich and 'progressive' countries such as the Netherlands, Sweden and Denmark set the pace which the rest of Europe will follow. Convergence, Targets and Policy Human volition and public policy also affect demographic convergence. In the past some goals were beyond reach (longer life) or beyond imagination (control of fertility). Today, both are partially under control. All wish to avoid premature death. All modern states, explicitly or otherwise, have policies aimed at minimising mortality, in the motivation of which international comparisons figure prominently. There is no such obvious common target, however, with respect to fertility. Government Expenditure and Policy While most governments of industrial countries lack explicit policies intended to influence fertility (pronatalist policies in Eastern Europe have been muted or abandoned since 1990), a number express concern about the low level of their birth rate (Macura et at, 1995). Governments tend to regard replacement fertility as acceptable. Were effective policies to become general, they would therefore have a convergent effect. In practice, pronatalist policies differ little from those adopted to support families and working mothers ostensibly on welfare grounds. However, it has been difficult to demonstrate unequivocal effects of family and welfare policies upon demographic behaviour on an international basis (Gauthier, 1996). Strong effects on fertility have been claimed in specific cases, however, notably in (former) Eastern Europe and in Sweden (Hoem and Hoem, 1997; Olah, 1998). The EU Commission believes that EU demographic and other social characteristics are already converging (European Commission, 1995) and will need similar policy responses (Hantrais, 1997). Demographic convergence is already regarded as one indicator of a desirable harmonisation of European social conditions. Although so far no Directive relating to standard European birth rates, death rates or household structure has been proposed, the possibility of harmonised EU demographic optima has already been raised (Gesano, 1999). Demographers' Views on Convergence Demographers' views on this, like the evidence (Blayo, 1987), are mixed. For some the long view reveals a long-standing social unity in Europe, including similar family models (Kaelble, 1989). Roussel (1994, 1995) insisted that current variety in family and fertility merely reflects the unequal progress of societies heading for the same destination. However, a test of convergence in family type up to the 1970s, using 15 variables including demographic patterns, found enough ambiguity in industrial societies to make the convergence hypothesis 'difficult to assess' (Inkeles, 1981). Data from ten countries in 1990-92 from the 'Family Life Forms and Everyday Family Life in European Comparison' suggest that our 'era of affluence' may truly have become an 'era of choice' (Kuijsten, 1996: 140-141), a view supported by wider samples (Rothenbacher, 1995). Marked demographic divergence between Western and Eastern European countries is well known (e.g. Tomka, 2002). Demographers have also noted major divergences in demographic variables within Western Europe. For example: 'because of variation in timing and intensity Copyright © 2002 John Wiley & Sons, Ltd. Int. ]. Popul Geogr. 8, 319-344 (2002) 322 D. A. Coleman Table 1. Variation between developed countries for selected demographic variables, 1995 (statistics are unweighted). Region Mean Coeff of SD variation Region n Mean Coeff of SD variation Total fertility rate Northern Europe 7 1.84 Western Europe 7 1.52 Southern Europe 4 1.28 Central/Eastern 7 1.34 Europe Balkans 5 1.88 Former Soviet Union 7 1.42 Neo-Europes 4 1.87 Asia 5 1.55 Total Europe 37 1.55 Total 46 1.58 Age-specific fertility rate for ages 15-19 Northern Europe 7 15.3 Western Europe 7 8.9 Southern Europe 4 12.1 Central/Eastern 7 30.3 Europe Balkans 4 27.2 Former Soviet Union 7 43.7 Neo-Europes 4 34.0 Asia 5 7.3 Total Europe 36 23.5 Total 45 22.6 Percentage births to mothers over age 30 Northern Europe 7 43.5 Western Europe 7 44.3 Southern Europe 4 41.0 Central/Eastern 7 18.4 Europe Balkans 4 25.2 Former Soviet Union 7 19.1 Neo-Europes 4 40.5 Asia 4 26.7 Total Europe 36 31.7 Total 44 32.1 Illegitimacy ratio per 1000 births Northern Europe 7 424.9 Western Europe 7 183.1 Southern Europe 4 102.9 Central/Eastern 7 208.2 Europe Balkans 4 154.6 Former Soviet Union 7 210.4 Neo-Europes 4 318.2 Asia 3 31.4 Total Europe 36 228.2 Total 43 222.9 Life expectation at birth, males 0.12 6.6 Northern Europe 7 74.27 1.62 2.2 0.13 8.7 Western Europe 7 74.01 0.69 0.9 0.10 8.1 Southern Europe 4 73.85 1.59 2.2 0.27 19.8 Central/Eastern 7 67.78 2.00 2.9 Europe 0.53 28.1 Balkans 3 69.93 0.35 0.5 0.17 11.8 Former Soviet Union 7 62.08 2.08 3.4 0.16 8.6 Neo-Europes 4 74.33 1.33 1.8 0.26 16.7 Asia 4 74.12 3.20 4.3 0.33 20.9 Total Europe 35 70.06 5.00 7.1 0.32 20.1 Total 43 70.84 4.88 6.9 Infant mortality rate 7.77 50.9 Northern Europe 7 5.1 1.12 21.8 4.64 52.4 Western Europe 7 5.3 0.46 8.6 6.24 51.7 Southern Europe 4 6.6 0.85 12.9 14.31 47.2 Central/Eastern 7 12.1 5.15 42.7 Europe 13.70 50.3 Balkans 5 16.8 9.96 59.4 11.10 25.4 Former Soviet Union 7 16.3 3.29 20.3 16.25 47.9 Neo-Europes 4 6.5 0.82 12.5 5.58 76.1 Asia 4 5.0 1.24 24.7 15.93 67.9 Total Europe 37 10.3 6.51 63.0 16.17 71.6 Total 45 9.5 6.17 64.8 Crude marriage rate 4.61 10.6 Northern Europe 7 4.9 0.91 18.5 7.78 17.6 Western Europe 7 5.2 0.47 9.0 8.24 20.1 Southern Europe 4 5.7 0.80 14.0 5.11 27.8 Central/Eastern 7 5.1 1.01 19.7 Europe 4.33 17.2 Balkans 4 5.8 1.66 28.9 2.41 12.6 Former Soviet Union 7 6.5 1.53 23.4 3.94 9.7 Neo-Europes 4 6.6 1.54 23.2 10.47 39.3 Asia 5 7.5 1.17 15.6 12.97 40.9 Total Europe 36 5.5 1.18 21.4 12.45 38.8 Total 45 5.8 1.36 23.3 Aged dependency ratio 132.31 31.1 Northern Europe 7 22.25 3.68 16.5 105.15 57.4 Western Europe 7 21.95 1.56 7.1 64.43 62.6 Southern Europe 4 22.46 1.15 5.1 107.05 51.4 Central/Eastern 7 19.21 2.36 12.3 Europe 103.87 67.2 Balkans 4 16.46 2.52 15.3 120.17 57.1 Former Soviet Union 7 18.44 2.23 12.1 64.02 20.1 Neo-Europes 4 18.2 0.91 5.0 21.44 68.2 Asia 5 12.47 5.26 42.2 146.00 64.0 Total Europe 36 20.24 3.11 15.3 146.92 65.9 Total 45 19.19 4.05 21.1 Sources: Council of Europe, UN Demographic Yearbook, National Demographic Yearbooks. SD, Standard deviation. Copyright © 2002 John Wiley & Sons, Ltd. Int. ]. Popul. Geogr. 8, 319-344 (2002) Convergent Demographic Communities 323 ..., one cannot simply conclude that countries follow the same trajectory in sequential fertility trends with certain time lags. When looking ... at the tempo effects, one sees that the between country variation is large. ... the results of this study lead me to assert that a divergence in fertility trends has occurred.' (Bosveld, 1996: 253-4) Recuperation of births in recent cohorts show further diversity (Lesthaeghe and Moors, 2000; Frejka and Calot, 2001a). Event history analysis of Italy and Germany shows 'divergent postponement' such that 'a convergence in patterns of union formation and first births may not occur and Europe is likely to be characterised by distinct national and regional patterns in the near and immediate future' (Biliar! and Köhler, 2002: 10). By the early 1990s, EU demographic projections assumed the continuation of differences in tempo and quantum of cohort fertility, and life expectation in all their variants (Eurostat, 1996). More recent attempts have discussed the prospects of permanent diversity more explicitly (de Beer and van Wissen, 1999). DATA AND METHODS How do we measure trends towards demographic convergence and establish criteria for its attainment? Economists recognise two forms of convergence: to a common absolute level, and a parallelism of trends. The former seems a more suitable aim in this case. A statistically significant sustained downward trend in demographic variability establishes a trend towards convergence. Depending on the variables concerned, the appropriate measure could be the variance, the coefficient of variation, multivariate distance measures arising from correlation matrices, or the magnitude of coefficients derived from scaling and clustering exercises. All of these will be used. Specific indices of convergence (or divergence) do not seem to have been developed. The indices of dissimilarity and related statistics such as indices of polarisation favoured by geographers are more suited to one-by-one comparisons of specific groups or of regional distinctiveness within countries (Hamm and Litsch, 1987). The 1930s would be the best choice as a starting point, but data are sparse for that time, and in most European countries trends were disrupted by the Second World War. Data for many countries remain patchy until around 1950, and some series can scarcely be used comparatively before 1960. Some countries, and the rural provinces of others now regarded as modern, retained transitional characteristics up to the 1950s, and in the Balkans up to the present day. While demographic behaviour from about 1970 may be more appropriate, that benchmark gives us less than 30 years of data. Results will be reported from several time-periods. It is more difficult to choose a yardstick to tell us when convergence has finally been attained. Its choice must be somewhat arbitrary. At the aggregate level it seems clear that convergence has not yet arrived in fertility, when at the end of the 1990s total fertility rates (TFRs) in some countries of Europe were 60% higher than in others, and the proportion of births outside marriage varied tenfold (Table 1). A difference of 5% might be accounted trivial, clearly indicating convergence, while one of 25% would be more interesting. For health variables a narrower, functional criterion might be suitable: that level of similarity at which governments cease to be concerned about the relative international position of their populations - probably just a few per cent. Here, however, rank order also matters. So far it has been assumed that the nation-state, rather than its constituent regions, is the appropriate unit for large-scale comparisons. Space does not allow this assumption to be justified here (see Coleman, 1996). The demographic utility of country-level analysis follows from the (usual but not universal) coincidence of culture and language with national boundaries, and also with the common influences on demographic behaviour of uniform national systems of taxation and of family welfare arrangements. National boundaries appear to have become more coincident with boundaries of demographic distinctiveness (Decroly and Vanlaer, 1991; Watkins, 1991). The coefficient of variation in fertility within countries seldom goes below 5. That seems also to be a reasonable minimum for differences between countries also, since zero variation is not to be expected. Copyright © 2002 John Wiley & Sons, Ltd. Int. ]. Popul. Geogr. 8, 319-344 (2002) 324 D. A. Coleman 30 4- 26 ■24 -C 3.00 2.80 + 2.60 es 2-40 H C 2.20 2.00 1.80 1.60 1.40 00O O -M c RESULTS Fertility Space permits the analysis of only a few variables. Fertility can only be analysed here according to one simple variable, the total fertility rate (TFR). International comparisons of components of fertility are presented by Lutz (1989), Sardon (1990), Grasland (1990), Beets (1995), Bosveld (1996), Hobcraft and Kiernan (1995), Lesthaeghe (2001), Frejka and Calot (2001b) and others. Since 1945 the general trend in mean TFR, taking 46 developed countries together, is of almost linear decline, falling by 1 every 26 years (Fig. 1). The lack of time series from republics of the former Soviet Union or FSU (except the Russian Federation and Ukraine) restricts the number of years for which comparisons can be made. More recent decades exhibit divergent trends. A more restricted set excludes East Germany whose extreme decline dates only from its non-existence as a country in 1990 (Conrad et ah, 1996), and four of the five Far Eastern countries, which were not 'developed7 before the 1960s. The ratio of maximum to minimum has changed rather little over time; 120% in 1958 and almost 110% as late as 1987. During the 1990s the excess fell to about 77%, partly due to radical declines in some less-developed Balkan countries and Moldova. Groups of countries show a more complex picture. The 12 Northern and Western European countries (in the old UN classification, and excluding Iceland and Ireland) show modest variability and no downward trend since the mid-1970s - if anything the reverse. Germany and its neighbours speaking Germanic languages show a consistent and tightly grouped near-zero trend since 1975. Despite cliches about 'Europe's declining fertility', the TFR increased in most Scandinavian countries from the early 1980s to the early 1990s, and remained stable in most of Western Europe and in the countries of European origin overseas (the 'Neo-Europes'). Sweden's striking increase to over 2.0 in 1992 was followed by a sharp decline to 1.5 in 1999; elsewhere, increases mostly levelled off. The last quarter of the twentieth century was the most stable period in the birth rate in these countries since the onset of the fertility transition. The experience of the Northern (Scandinavia plus UK and the Irish Republic) and Western European countries is paralleled by that of the 'Neo-Europes' overseas: Australia, New Zealand, Canada and the US. These two pairs of countries share a similar trend in TFR: a generally higher trend of fertility than in Western Europe, a more or less continuous baby boom from the end of the Second World War to the 1970s, and the avoidance of really low fertility. They share with Western Europe the same timing of the baby boom. They share Copyright © 2002 John Wiley & Sons, Ltd. Int. J. Popul. Geogr. 8, 319-344 (2002) Convergent Demographic Communities 325 0 H LO ON On I-H OJ 1 LO <* ON 3 O, 0 W rd OJ -t-J o 13 O OJ bC rd rd O" 3 13 bo >-, bo OJ (N OJ rd H OJ O, O W Ó OJ §3 g Ph rd m "ČO 6 OJ £1 rH -M OJ 0 fi cfl Í-H OJ rd 3 U W W I 3 3 O W (XI a, 0 bC 6 OJ -M OJ OJ a, 0 3 13 OJ 5 W -t-J u JĽ "Ôj cfl 6 OJ 0 OJ a, 0 w OKHtnMM^ninoooioooiinooMri^aM^iHooo (^ I-H o tv en tN <* í) li (N ^ N N inruomiíiMic^incoMOifinotriN HlOiOlNHMtOn r( H rl (N H N tí tí »S (S (N ** rH (N CONN^MIO^^HOS fi fi CO LO O (N o os mn H H N CO (N C5 CO (N ^ <*LOcom<*ooc\|en<* >>>>>>>"^ **tv**VO'*LOCNrH'* <* O 1 "2 O c« UT3 W c/5 Copyright © 2002 John Wiley & Sons, Ltd. Int. ]. Popul. Geogr. 8, 319-344 (2002) 326 D. A, Coleman ■& ■& CO CO On On c JN On o\ o\ o\ On On On On ■& ^O CO CO On On On On On On On On On On On Scandinavia NeoEuropes mean ■ Scandinavia NeoEuropes CoV (d) 35 > C o ü C 1 ^ OJ CO O ^ OJ CO c Jn On On On On On On On On ■& ■& CO CO On On C )N On On On On On On On On ■FSU(exM) CEE Balkans ■FSU(exM)- CEE Balkans 11111111 M 1111111 M 1111111 M 1111111 M 111111 M 11111 ■ Southern mean ■ Southern CoV Figure 2. (a) TFR trends, Scandinavia and Neo-Europes, 1950-99. (b) Coefficient of variation of TFRs for Scandinavia and Neo-Europes, 1950-99. (c) TFR trends, former communist countries, 1950-99. (d) Coefficient of variation of TFRs, former communist countries, 1950-99. (e) TFR trends, for Southern European and Asian countries, 1950-99. (f) Coefficient of variation of TFRs, for Southern European and Asian countries, 1950-99. Note: 'ex M' = excluding Moldova. Copyright © 2002 John Wiley & Sons, Ltd. Int. ]. Popul. Geogr. 8, 319-344 (2002) Convergent Demographic Communities 327 with Northern Europe a rise in the birth rate since the mid-1980s, which took TFR in the US and in New Zealand back to replacement levels by the early 1990s. Fertility in New Zealand did not fall below 1.96 during the 1990s and rose slightly to 2.01 in 2000. Fertility in Australia and Canada is falling somewhat and thereby diverging from a previous pattern of relatively high fertility in the English-speaking countries (Chandola et al., in press). But in the US the TFR has remained, uniquely, at around replacement level and indeed exceeded it in 2001 (TFR = 2.135) Southern and Eastern Europe now have much lower levels of fertility, although since the early 1990s the populations of Southern Europe have halted the free fall in fertility which began around 1980 (earlier in Italy; Michaeli, 1996). An altogether more rapid decline was precipitated in Eastern and Central Europe from 1989, a sharp discontinuity from the gradual trends which preceded it. This has taken the birth rate down to Southern European levels in just a few years. Whether this should be regarded as 'crisis' or adaptation is beyond this paper (Lechner, 2001; Philipov and Köhler, 2001; Philipov, 2001). By the end of the 1990s the decline at national level appears to have stopped at about the same level as Southern Europe - around 1.2, perhaps a new empirical lowest level. The European Republics of the former Soviet Union, amongst which fertility levels had converged very closely by the 1980s, have experienced almost identical and severe fertility declines to a similar level, not excluding the Baltic States. Does this add up to convergence? Different regional groupings of countries give different answers. The coefficient of variation in the Western European group of six countries in the EU core, and of the four Scandinavian countries (excluding Iceland), is down to a third of its level in 1950. The four countries of Southern Europe and the six countries of Central and Eastern Europe (CEE) have also become more similar. The European Republics of the former USSR converged to near-identity by the 1970s and diverged somewhat from the 1980s onwards. These were formerly 'provinces' of the same country where it was explicit policy to eliminate regional diversity. In the 1950s fertility in the Eastern group was exceptionally heterogeneous. Some countries were predominantly rural and only semi-transitional. All became subject to uniform systems of pervasive state control and forced urbanisation and industrialisation, punctuated by erratic bursts of transiently effective pronatalist policy, now withdrawn (Macura, 2000). The trends in the proportion of women childless in the two halves of Germany summarise East-West divergence up to the end of the 1980s. From a common position of 8% childless among the birth cohorts of 1935, childlessness among women in West Germany rose steadily to 20% by the birth cohorts of 1955, by which time only 6% of their sisters in the East remained without children. Most of the Balkan statistical outliers, the republics of former Yugoslavia with high fertility (Macedonia, parts of Serbia such as Kosovo), had moved in a similar direction by that date, although still not convergent. But in the free markets of the post-transitional West, birth rates can go up as well as down. The unprecedented economic boom of the 1950s and 1960s passed by the communist East. Almost all shared a decline in the birth rate from around 1990, and small increases in the late 1990s. The variation of fertility in the Neo-Europes remains modest throughout the whole 45-year period (Table 2, Fig. 2) although the trend in Canada is looking divergent. The sharp convergence of the Far Eastern countries cannot be considered in the same light as the others; it arises from the rapid economic development and demographic transition in four of the countries which joined the developed world, including Japan, in the 1960s (Leete, 1994; Leete and Alam, 1993). Between groupings of countries, and in the industrial world from 1960 to 1980, there was a modest trend towards statistical convergence. This trend ceased after the 1970s, and in the last decade a more divergent pattern dominates. Trends towards similarity were mostly completed by the mid- to late 1970s. From then on, relatively little change is seen, although the general level of variation remains modest. More importantly, while most groups of countries may have become more homogeneous by the end of the baby-boom period, Copyright © 2002 John Wiley & Sons, Ltd. Int. ]. Popul Geogr. 8, 319-344 (2002) 328 D. A, Coleman 0^0 í» e o 190 o o o « ^H 'C h 170 JS 0) > o. Sw 0> O M C3 150 Ö 'C 130 0> 0) o rS u S O 110 S 70 50 I I 11 I 11 I 11 I 11 I 11 I 11 I 11 I 11 I 11 I 11 I 11 I 11 I 11 I 11 I 30 80 75 70 63 e o 4-1 55 O a 0> 50 ifl 0> o 45 U 40 35 -B— Mean ■ Coeff var ■ Mean ■ Coeff var (c) 250 -, r i" 230- M f ~ V3 £ •a 210- ^\ / - í» o 190~ t 170-e, : /♦ A " |° 150 -•e e 130- r/ V outside o J ^ S 90-■-S 70- ^ SESŕŕ^ 50- I 1 1 1 1 1 II 1 1 1 1 1 II 1 1 1 1 II 1 1 1 1 1 II 1 1 1 1 II 1 I - 95 85 80 75 70 65 60 55 50 ON CO t-~ \r-, \o \o o> o> o> o> o> o> o> ■ Mean ■ Coeff Var (d) 500 450 a o o o o a ^H 'C a 0) > 0) O .s e C3 g o 0) 2 u *3S s o 400 350 300 M 250 200 S 150 ■S íoo 50 Australia Canada New Zealand USA Southern Europe Mean Scandinavia mean 0 1111 m 11111111 m 111111111 m 11111111 m 11111111 m 111111111 Figure 4. Illegitimacy ratios, unweighted means and coefficients of variation, (a) 18 'Western' European countries, 1956-1996. (b) 10 'Eastern' European countries, 1955-1997. (c) 28 European countries, 1959-1998. (d) Illegitimacy ratio trends, Neo-Europes compared with Southern Europe and Scandinavia, 1945-1999. Copyright © 2002 John Wiley & Sons, Ltd. Int. ]. Popul. Geogr. 8, 319-344 (2002) 330 D. A. Coleman O N ^ IT) lO N Os CC HCOiriO^ "^ CC OS iO^ mo ro co > lO Os (N (N «ON ■=* CO so cO (N CO (N O SO CO T-H CC Os LO Os Os co CO T-H (N N t-h cO coOSCTitNlrHiONNNOrOOCN^rOlNCONcOHcoNiOOCTirO^^MH os so so c^rooioc^co^^co^r^rooco^ioio^cocriCNlrncococTíCNroiocoN rococo lOiOiOroc^rororo^cor^roroocorONNr^r^^'í^rHioiOrotNCN^ rslOlOC^C^HC^HO^H^^^C^^CO^HN^OrOrsICO^NCOCOHrntNOlOrOtN N CN o ^ HNNrONHcO^moNN^NrorNHOcoiOOScocoo^rO^rOrOH'í^ocorOH o so ** t-h ^coNcoiosoiococc^cocorNlccrNlcocoosNcorNl^rNlcoso^soassoi^ ^ŕ t-h ro cc^rNioc^t>>iOT-HrNic^ioc^rNic^^t>>t>>t>>io^ccT-H^c^ as os n cn Nco^^soiosoorNlNccasccN^rNlcosoasasNOsoas^rNlcoNT-H^No o^řroco coi>>coio^iococo^rocorNll>>rNlrococoi>>rorNll>>rNlro^^^co^io ^ŕ t-h (N OCOC^lOO^lOC^OC^rONlOlOHrOrOHCOHC^H^OHCO^OrNCO^OrH^OC^CO T-H-rŕ ONOs^LOioos^sococcccasN^LorNiasasNcc^asrNlsocorNlT-Hcco^ m ro ^co^coioioiocot>>^cocorNlt>>rNlcocot>>^corNlt>>rNlco^^^ N co I>> O (N (N ro ^ H 10 ro T-H lO CO H ro ro N ro I>> (N ro T-H MO ^ CO O ^H CO O (N CO "^ so "& lO lO (N T-H ^O o (N so lO I>> O so On (N ^ On ^h ^h 'CirOC^COOSlOlOrOHCOlNlOlNOSrN^ fN^COO^lOHrOHcOrNllO^^HvO roi>>rN|^roi>.I>.^ř^coroco^^^co IT) HfN^ (N O SO CO CO CO OS O ro CN CO lO T-H (N (N CO CO (N (N CO ^P T-H (N IN UO H IO lO CO CO CO O. CO CO T-H OS T-H T-H O. co (N os co co Tř io (N 10 co cO so cO ** ^ ^ ^ cO cO so lO lO Os OS T-H Tř Os SO O CN T-H so lO so (N lO O (N O co co co ^ ■^ T-H (N "* fN "* iH Oi "* fN CN CO t-> in ■* ■* T-H Tř O Os so ^ CN O T-H CN CO SO OS CN CO lO ro in ^n ^n ^n O CO H N O CO ^ CN CO so T-H O-so so -3 so CO (N OS CO On T-H (N CO so CO lO lO CO CO lO so lO ■^ CO CO ^ C/} ^ so so so so lO Os so so ^ SO T-H so Tř T-H Os CO cO O. so (N mo co so o os \o m co 10 10 "* CN OS so M ON ^ ^ lO COON CO SO CO ^ŕ co ^ "S Q OS T-H SO CO ^h T-H I U CO CO T-H CO O, CO -Q 3 a OJ ^ 13 •íS CO s .rt rä ^ ctí ia T3 S £ CO 3 3 3 'bo _bp 3 ľroati ľzech )enm; C o CO § 3 bo ta "O "O bo C ~ C - ÜJ QJ > g TJ n Ä ■§ OJ -'S Pi —i ■r oj oj ^J >* L^ ^ co.3 ^r^r-=LÍ5 ?< ^ tu ro.^ ^j^h^^^ u u ^ -j_2_2 C^ ž ž í^ „ 3 'S -° S S 6 ŕ 3 u §3-i£šl ss-assl« T3 «TS bO C ö ^ älľ c -a TJ C OJ i -s ia ° o o q T3 ' bol •S Ĺ O) " • S ^ i ■X C -Sf>Ä Q O »2 ,Q ^ P ej-, ž en Copyright © 2002 John Wiley & Sons, Ltd. Int. ]. Popul. Geogr. 8, 319-344 (2002) Convergent Demographic Communities 80 75 70 65 60 331 55 50 05 CT> CT> CT> CT> CT> O O O CT> O O O O CT> O O O O —♦—France —"—Greece —*—Hungary —x— Japan —*— Russia Figure 5. Expectation of life at birth for males of selected countries, 1945-1999. unweighted), its variability was already high (coefficient of variation about 100, Fig. 4a-d). This reflects persistent traditional differences in reproductive habits dating from much earlier times (Wall, 1983; Ajus and Henye 1994; Viazzo, 1986). In 1950, for example, the illegitimacy ratio in Iceland was 25%, in Austria nearly 20%, in Portugal 12% and in Sweden 10%. Elsewhere the figure was well under 10%, and for the first two post-war decades it fell rather than rose. Since the mid-1960s, although the mean and range of the illegitimacy ratio in 'Western' Europe has increased five-fold, the coefficient of variation actually fell over the period from about 100 to about 70 (Fig. 4a). What was once highly variable but little evident has become salient and obvious, and of course of much greater social and economic significance. In this respect any ratio measure such as the coefficient of variation, which corrects for differences in mean, removes the main feature of interest. That one country has 50% of its births outside marriage and another 5% is much more interesting, from the point of view of population diversity, and defining statistical norms of behaviour, than if one has 5% outside marriage and the other 0.5%: the same ratio. Measures of dispersal that take into account the absolute magnitude of differences, such as the interquartile range, may be more appro- priate. Thus taking 42 countries together, the standard deviation and interquartile range in 1950 were 53 and 46 respectively (per 1000 live births), and in 1995 had trebled or more to 149 and 198 respectively. There are no regular data on births outside marriage at all for the individual republics of the former Soviet Union except Russia, or for Romania before 1970 or for Albania except 1951-1964. Elsewhere in the East around 1950, births outside marriage were somewhat higher than in the West, nearer 10% than the Western average of 5% (East Germany: 12%). In the FSU as a whole, where policy had intermittently tended to discourage traditional family structures, the proportion reached 18% in 1948. Elsewhere, there was much less variation over space and time than in Western Europe. That situation remained little changed up to the end of the 1970s. However, average levels rapidly increased through the 1980s, substantially in East Germany, the Czech Republic and Hungary, much less so in Poland. In this respect Eastern Europe ended the half-century more varied statistically than the West (Fig. 4b). Overall, the statistical effect in Europe as a whole is of convergence of the coefficient of variation (Fig. 4c), but marked divergence using measures based on absolute values (Table 3). The populations of the Neo-Europes have moved together in this respect - all have Copyright © 2002 John Wiley & Sons, Ltd. Int. ]. Popul. Geogr. 8, 319-344 (2002) 332 D. A. Coleman (a) 76 74 -f 0> Ä £ 72 Ä ■^ Sh pfi 70 +^ Ä 68 «M O ä 66 o +^ CS 64 0> Ä X W 62 7.0 (b) 68 o.u 0> 13 67 ä S 5.0 o ä +^ +^ « '- '- -o 66 4.0 & > «M O +^ ^5 3.0 «M O 65 la ä o ?.o 0> o U u 64 1.0 tí 60 llllllllll llllllllllllllllllllllllllllllllllll 0.0 \ i 63 lllllllllllllllllllllllllllllllllllllllllllll 1.0 6.0 5.0 4.0 3.0 2.0 ä o .2 'u la o U Mean Coeff. Var ■ Mean Coeff. Var (c) 73 72 13 IV S ä 70 ■^ '- pfi 69 -^ Ä 68 «M O 67 S O 66 -^ w O) & K 65 tí 64 63 lllllllllllllllllllllllllllllllllllllllllllll ä o o U (d) 72 71 70 í 69 «ô 68 4- .Sž o 68 3 67 + tí 66 2.5 A 65 lllllllllllllllllllllllllllllllllllllllllllll 2.5 5.5 5.0 ä o 4.0 3.5 3.0 o -^ ä o U Mean Coefficient of Variation Mean ■ Coeff. Var Figure 6. Expectation of life at birth for males, with means and coefficients of variation, (a) 18 Western European countries, 1950-96. (b) Selected Eastern European countries, 1955-96. (c) 33 developed countries, 1950-96. (d) 37 developed countries, 1955-90. shown substantial increases to about the same level since the early 1960s. The coefficient of variation is correspondingly modest. Its abrupt increase from 1960 to 1961 (Fig. 4d) reflects the full incorporation in 1961 of data from the New Zealand Maori population, among whom a high proportion of births are outside marriage. Canadian data are also incomplete and inconsistent, although not enough to upset the general conclusions. Data are only available for two or three of the five Asian countries (Singapore will provide none) and, are not shown here. Few births are outside marriage (usually less than 2%) and show little sign of increasing (in 1995, 0.5% in Hong Kong, 1.2% in Japan and 2.8% in Taiwan). Copyright © 2002 John Wiley & Sons, Ltd. Int. J. Popul. Geogr. 8, 319-344 (2002) Convergent Demographic Communities 333 (a) 60 55 I: | 35 •3 30 | 25 U 20 10 50.00 45.00 I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I - coeff. of variation (b) 70 |60 .2 50 40 30 20 10 120 I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I - coeff. of variation (c) 80 70 i: | 40 1 30 u 20 10 I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I 80 70 60 50 40 30 20 10 - coeff. of variation ■ Figure 7. Male infant mortality rates, unweighted means and coefficients of variation, (a) 18 Western European countries, 1950-98. (b) 13 CEE countries, 1950-97. (c) 40 industrialised countries, 1950-97. Sources: Council of Europe, UN Demographic Yearbook, National Demographic Yearbooks. Mortality Only mortality in males will be considered, as in post-war Europe, males have experienced the more volatile, diverse and damaging trends. Mortality, of all demographic variables, would be expected to have converged, as all individuals and all policies desire its reduction. Indeed, if there is a natural absolute limit to low mortality, which all populations Copyright © 2002 John Wiley & Sons, Ltd. Int. ]. Popul. Geogr. 8, 319-344 (2002) 334 D. A. Coleman are approaching at their own pace, convergence can eventually be expected merely on technical grounds. However, that day, if it is coming, appears to have been postponed. Instead, trends in male expectation of life at birth in a variety of industrial countries presents a grossly divergent and uneven picture (Figs 5 and 6). While the overall life expectancy (of 37 industrial countries) has increased unsteadily over the last half-century, there has been no general trend towards convergence. The coefficient of variation, although never high, is much as it was 50 years ago. The main reason is obvious: the failure since the 1960s of the Eastern Bloc countries (with some important Balkan exceptions: Meslé et at, 1996, Meslé and Hertrich, 1997; Gjonca and Bobak, 1997) to follow Western improvement in survival, exacerbated by the additional mortality crisis of the 1990s from which only some have now clearly emerged. Outside the Eastern Bloc the developed world shows a more or less linear increase in life expectancy (19 Western European countries) and an equally linear decline in coefficient of variation to very low levels (less than 2), albeit with little change since the mid-1980s. In the Neo-Europes, trends in life expectancy have been nearly identical and the coefficient of variation accordingly very low (under 2), although increasing over time (not shown). By comparison with their mean value, divergences between the Western countries are relatively small and surely pass most statistical requirements for convergence. For obvious reasons, however, persistent international differences in indicators of health and mortality, and the rank-order of particular countries, provoke controversy in the less favoured nations. The expectation of life, however, is a summary measure which, because of the inevitability of very high death rates at very old age, cannot respond to even quite substantial proportional differences in death rates at younger ages. These can remain statistically impressive, and salient in terms of welfare and politics. For example, in visual terms the graph of the infant mortality rate (IMR) seems to provide clear evidence of near-complete convergence: an (unweighted) average of 66 per 1000 live Table 4. Infant mortality rates for 19 CEE and FSU countries, 1970-1993, selected statistics. Year Mean SD Coeff. var. 1970 36.8 24.2 65.9 1975 30.6 15.8 51.8 1980 24.7 11.7 47.5 1984 21.6 10.6 49.1 1985 20.4 8.8 42.9 1986 20.3 11.6 57.3 1987 18.7 8.4 44.8 1988 17.4 8.0 45.9 1989 17.2 8.2 47.4 1990 16.1 6.8 42.5 1991 16.3 6.5 39.6 1992 16.6 6.9 41.4 1993 15.6 6.4 40.7 Sources: Council of Europe, National Demographic yearbooks. births in 37 countries in 1950 had fallen to 6 in 1993 (Fig. 7). In the context of health policy, this must count as convergence, even though in relative terms the coefficient of variation of the IMR has ended the period almost unchanged at over 55. Eighteen Western European countries have clearly converged since the 1960s to a relatively low coefficient of variation by 1993 (about 20). The four Neo-Europes have converged even more strongly, but the Asian industrial countries rather less so, because of indifferent progress in Korea and Taiwan. Taking the developed world as a whole, the main impediment to statistical convergence in IMR comes from Central and Eastern Europe (Fig. 7c), with no clear trend towards convergence since the early 1970s. Although some CEE countries ended the century with very low IMRs, in others infant mortality is still a real health problem. This is a more variable picture than in the 1950s, when all IMRs were high. The CEE countries are a source of divergence, not convergence, in the industrial world. The former Soviet Union is hard to incorporate into the analysis because of the limitations of data for individual republics from before 1980 and the need to correct the reported levels (Ksenofontova, 1994). Contrasts in mortality rates are most marked in later adult life. Age-standardised mortality rates (ASMRs) for males aged 0-64 (standard- Copyright © 2002 John Wiley & Sons, Ltd. Int. ]. Popul. Geogr. 8, 319-344 (2002) Convergent Demographic Communities 335 (a) 4.0 -. 3.5 H 1.5 (b) 2.0-1.9-1.8- «u 1.7 - -^ es u £ 1.6- 56 1.5 1.4 H 1.3 1.2 1.1 56 POL V SLK V IRE ROM POR 0> 3.0- + YSM V £> RUS < HUN ■*^ > FRVA ^pCZR «M CRO 3 < LIT + V r© 2.5- > BEL H 2.0- GRRGRE AUT ^+ LUXO FRG O O LAT > NL O NOR SWI UK' O SWE 58 —T" 60 —T" 62 —T" 64 66 68 70 Geographical area > FORMER SOVIET < BALKANS V CENTRAL AND EASTERN EUROPE + SOUTHERN EUROPE O WESTERN EUROPE • NORTHERN EUROPE MAC < YSM < MOL HUN POL v RUS > LIT > UKR BLR tfc> SLK v 72 IRE NOR ■ ■ DENFIN FRAUK LUXc* O BEL XTT O NL O, EST > ROM v POR + LAT > CZR SLO AUT O FRG SWI O GRE BUL ESP + ITA + 58 60 62 64 66 70 72 74 76 Expectation of life at birth (m) Figure 8. Total fertility rates (TFR) and male life expectancy for European countries, (a) 1955. (b) 1995. ised on the Standard European Population) for a rather limited selection of 18 European countries show much more variation than does expectation of life at birth. In 1970 the maximum ASMR was 83% higher than the lowest; by 1992 it was over double. The coefficient of variation correspondingly increased from 16 to 40. The data in Table 4 understate the growing contrast, as suitable ASMRs were not available for the FSU countries (where mortality is highest) until the 1980s. A Broader Picture The European demographic space is shown more effectively by taking two or more of these Copyright © 2002 John Wiley & Sons, Ltd. Int. J. Popul. Geogr. 8, 319-344 (2002) 336 D. A. Coleman variables together. Two key parameters are fertility (TFR) and male expectation of life at birth (e0m). There is almost no correlation between the two variables in modern, post-transitional countries, but in combination they mark out a variety of demographic options. In 1955, towards the height of the baby boom, the industrial world was concentrated around e0m of 65 and a TFR of 2.5. Diverse countries from all parts of Europe shared apparently similar fertility and mortality characteristics. Outliers included all the Neo-Europes with higher birth rates, some Scandinavian countries with clearly superior life expectancy, and some Eastern European countries where the reverse was true. By 1970, at the end of the baby boom, the same features of overlap - convergence -still predominated, with only Scandinavia standing out with its particularly favourable mortality. By 1980, however, the poor mortality record of the communist countries was beginning to segregate them from the rest of the industrial world. By 1990 that separation was absolute - even more by 1994 - and furthermore the collapse of the birth rate in southern Europe and its buoyancy in the North (and in the Neo-Europes) had produced a pleasingly cartographic segregation of countries: a rough mirror-image of the European map (Fig. 8b). Multidimensional Approaches Several demographic variables can be combined to position countries in a multidimensional demographic space. Their spatial relationship then reflects their relative similarity or distance, and the trend in the overall size of the space needed to accommodate all the counties would measure their 'convergence' or otherwise. A number of techniques can do this; multidimensional scaling (Kruskal and Wish, 1978) is presented here (Figure 9a). This technique computes matrices of proximities or dissimilarity between objects (here up to 41 countries), usually from standardised variables (here up to 15 demographic variables), and then attempts to find structure in the data by positioning close together in n-dimensional Euclidean space those countries that are in overall terms most similar according to the statistical distances in the matrix derived from the data. All variables must be present for all years analysed, so only a moderate number of basic variables with a modest number of populations can be used; even then it is difficult to make comparisons before 1960. Variables include TFR, percentage of births over age 30, rate of natural increase, median age, percentage of population over age 65, illegitimacy ratio, crude marriage and divorce rates, IMR and expectation of life at birth. Recognisable structures for 30 countries can be identified for 1960 and 1995. Some of the Balkans and the former Soviet republics have to be omitted. In 1960 the picture is much less coherent, although the Western and Southern European populations are distinguishable from the others (with Japan's honorary membership of Southern Europe already apparent). The results are shown (for a restricted group of countries) in Fig. 9. A Western European group emerges clearly, except the Netherlands, out on a limb doubtless because of their then very high fertility. The Central and Eastern European countries (then with only 15 years experience of communism) are distinct from the rest but scattered. Southern and Northern Europe are closer to each other than they subsequently become. By 1995, more coherent relationships are evident (Fig. 9). CEE countries occupy the bottom right-hand quadrant, and Scandinavia and the English-speaking countries the opposite quadrant, with Western and Southern Europe distinct but close. On this scale, Western European countries are rather close together. Slovenia is close to Southern and Western Europe (and was so in 1960), with the Czech Republic and Hungary relatively close by. Poland and the Slovak Republic occupy the same space. Russia and Bulgaria remain outsiders. Japan remains in Southern Europe. Such techniques do not directly answer the question 'are populations converging?' They do, however, indicate that certain geographical groupings of countries have preserved some coherent demographic similarities over half a century, although more clearly at the end of the period than in the middle, during the baby boom period. In the 1990s the groupings into which they fall are in some ways more distinct Copyright © 2002 John Wiley & Sons, Ltd. Int. ]. Popul. Geogr. 8, 319-344 (2002) Convergent Demographic Communities 337 (a) MD-SCAL 30 countries 1960 2.0 1.5 1.0 | 0.0 -1.0 -1.5 -2.0 ICE D RUS POR POL n CAN USA NZL D SLK IRE AUS D BUL D SLO D ESP GRE NL rJTA D HUN AUT DEN □CZR GDR D FRG SWE D JAP D FRA d UKü DSWI£ELNOR ° _ n n -2.50 i—i—i—i—i—i—i—r~ -2.00 -1.50 i—i—i—i—i—i—i—r -1.00 -.50 i—i—i—i—i—i—i—i— 0.00 .50 DIM2 n—i—i—i—i—i—i—r 1.00 1.50 n—i—i—i—i—i—i 2.00 2.50 MD-SCAL 41 countries 1995 2.0 - SWE D 1.0- ICE D IRE D ™„DEN MT cwT JAP NOR FRANL SW1 nESP d FIN d □ D □ AUS *" L™ BEL GRE " CAN UK TT SLOo ITA D HK GDR D D D POR n D AUS D o.o- NZL D D CZR HUNn -1.0- USA D POL D SLK ^ LIT EST BUL D LAT -2.0- MAC D MOL D BLRROMD □ D UKR RUt? D D -3.0- -3.00 -2.00 -1.00 0.00 1.00 2.00 DIM2 Figure 9. (a) Multidimensional scaling, industrial countries 1960 and 1995, 11 demographic variables, (b) Multidimensional scaling of industrial countries, 1960 and 1994. Demographic variables include: median age difference between sexes, percent of births to mothers aged 30+, crude marriage rate, crude divorce rate, dependency ratio of youth and aged, life expectation at birth for men and women, illegitimacy ratio, rate of natural increase, net immigration rate, percent of population aged 15-64, total fertility rate. Copyright © 2002 John Wiley & Sons, Ltd. Int. J. Popul. Geogr. 8, 319-344 (2002) 338 D. A. Coleman (b) 3 ■J pOR NZL CAN ICE * * * * USA * * AUS FIN ESP* * * GRE NL ITA * * IRE * FRG * AUT * LUX SWI* * J NOR SWE USA ■ * NZL IRE B POR " ICE AUS -----------1------------ GRE B ■attV^H^ UK CAN ■AUT DEN PIN FRAnOR FRGH *^*SWI. --NL ■ ITA ■ ■ ■■ ■ SWE ■ --------------------------------------1--------------------------------------------------1------------------------------------------ YEAR ■ 1994 * 1960 0 1 DIM2 Figure 9. Continued. and are more closely related to geographical contiguity and to boundaries of culture and language. Can these techniques answer the initial question on convergence? They can measure the size of multidimensional demographic space occupied by our populations and show if it is expanding or diminishing. A graphical impression is given in Fig. 9b, which shows the relationships between a more restricted set of 22 countries using the same variables on the same scale. It is apparent that a smaller demographic space is needed in 1994, although this is partly due to the migration of outliers such as the Irish Republic. A numerical answer can be given by determining the absolute average and standard deviation (and thence coefficient of variation) of all the distances between each pair of countries at successive points in time, derived from the matrix of proximities computed from the correlation matrix. However, the means (equivalent to the average radial distance from the centroid) at different time-periods cannot be compared directly. The variables are automatically transformed to z-scores with mean zero and standard deviation 1, so the mean based on these standardised variables is therefore made constant over time. The variation about the mean in the distance matrix, however, will show whether countries are converging or diverging. We can get round this problem by standardising each variable across all time periods at once, and then selecting the appropriate subset of data pertaining to each time period being analysed. The values of the average distances will then not be equal, and neither will their standard deviations (Fig. 10). In the Western European countries by themselves, the mean distance between countries for the set of 14 variables fell by about 30% over the 40 years from 1955 to 1995. The Copyright © 2002 John Wiley & Sons, Ltd. Int. J. Popul. Geogr. 8, 319-344 (2002) Convergent Demographic Communities 339 5.00 45.00 4 40.00 2 s o 3.20 0.00 1960 1970 1980 1990 1995 Figure 10. Mean and coefficient of variation of absolute Euclidean distance between 25 European countries, 1960-1995. standard deviation also fell, so the coefficient of variation was therefore more constant at about 25 - much the same as that for fertility and other key variables. That must be considered to indicate some overall convergence, although some variables (including those influenced by past trends, e.g. median age) showed marked convergence, while others (mortality, household size) did not. When put together with Eastern Europe, however, the overall European picture shows less net convergence. The interpretation of the absolute value of that global distance between countries, and a coefficient of variation of 30, is more difficult. It may depend on the comparison with variability during previous times, during and before the transition, or during the 1930s. DISCUSSION AND CONCLUSIONS Convergence is expected on various plausible grounds, but the data from 1950 to 1995 lead us to mixed conclusions. A contraction of the demographic space occupied by the 'developed7 countries is indeed apparent from the figures, mostly as a result of the completion of the first demographic transition in a relatively small number of 'outlier7 populations such as Iceland, the Irish Republic, Albania and Macedonia. Geographically peripheral, and initially more rural and poorer than average, they have become more 'developed7 over the post-war period. Some had never achieved independent statehood before the twentieth century; some were 'provinces7 until ten years ago. Against that are divergent developments which are wholly new. Most important among these are the adoption of very low fertility by the Southern European countries, along with the accentuation of family living by the young, and the popularity of other, new living arrangements in NW Europe and overseas (not discussed in this paper). The latter may be embraced by the 'second demographic transition7 model; the former certainly cannot be. Copyright © 2002 John Wiley & Sons, Ltd. Int. J. Popul. Geogr. 8, 319-344 (2002) 340 D. A. Coleman Other major departures are the chronically severe mortality patterns in the CEE and FSU countries, and more recent radical changes in family and fertility. The former inheritance of communism may be self-healing; the latter may, paradoxically, represent the beginning of the end of the old Hajnal distinctions and eventually herald a greater homogeneity between East and West. In looking to the future there are two problem areas: (a) the supposed determinants of demographic characteristics, economic and ideational, are difficult to forecast and are not necessarily converging themselves; (b) they may not be demographic determinants at all. Once (and if) the under-developed outliers have become completely 'modern', further convergence may not be expected. Some engines of demographic change (for example, women's workforce participation) may no longer have the demographic effects that seemed obvious in the 1970s. While welfare systems may emancipate union formation and childbearing from free-market constraints, the Swedish case shows that they remain vulnerable to economic forces. The trend towards 'post-material' values and attitudes is presented by the votaries of the 'second demographic transition' as an historically inevitable universal development of irresistible force. Identification of 'leader countries' which others follow has proved difficult; there does not seem to be one single trajectory. Liberating forces need not lead to convergence, unless all agree to be liberated in the same direction. The expected demographic consequences are in any case incoherent. Populations most enthusiastic for non-traditional living arrangements (NW Europe and the Neo-Europes) tend to have the highest fertility, while 'traditional' attitudes persist in rich countries with very low fertility 0apan, Italy: Bettio and Villa, 1996). In addition, high levels of divorce and lone parenthood, which transfer some of the costs of the consumption of women and the production of children to the general taxpayer, may not be affordable in the long run. The age of entitlement may have only temporarily insulated people from the consequences of their reproductive actions, and thereby only transiently permitted a wider spectrum of behaviour. If symmetrical gender equity is an important determinant of fertility levels (McDonald, 2000) then convergence may indeed be slow, as its acquisition would require a major cultural transformation in the countries with very low fertility. Finally, modern countries maintain longstanding differences quite separate from the realm of 'materialism'. These include political systems and electoral preferences, national characteristics of various kinds (Inkeles and Masamichi, 1996), and attitudes to risk-taking. One striking example of non-convergence is the persistence of large differentials in standardised mortality rates from accidents and violence in 31 industrialised countries, in particular traffic accidents, despite near identity in the cars driven, the standards of road and signs, the use of seat belts, laws against drink-driving, and so on (Adams, 1985). It was reasonable to expect convergence within the CEE group of populations, and their divergence from the West, until 1990. Until then all were run on a centrally planned basis with little operation of market forces, but with erratic and sometimes effective attempts to promote fertility. In union formation and fertility, intended and unintended effects of public policy not only maintained but reinforced previous patterns of universal and early marriage and childbearing, with much reduced variance of family size. The divergence in mortality between East and West is the most obvious difference, and in some respects the least interesting, because it is easiest to account for. Its first component is the failure of improvement in survival in Eastern and Central Europe and Russia since the 1960s, a matter too complex to be dealt with here (see Lutz et at, 1994; Meslé et at, 1996). A unique political system has put these populations under unnatural constraints for decades. These constraints, summarised by Elwood Carlson as the 'State Socialist Mortality Syndrome', have driven them off the trajectory of improvement in survival which would have been expected earlier in the century. On top of that is the recent increase in the death rate in many (not all) of the countries in transition. Secondly, there is diversity within the former communist bloc. Some Central European countries, especially those formerly aligned to 'Western' Europe and forcibly Copyright © 2002 John Wiley & Sons, Ltd. Int. ]. Popul. Geogr. 8, 319-344 (2002) Convergent Demographic Communities 341 'Easterniseď after 1945 (e.g. the Czech Republic), have recovered faster in economic and health terms. The Balkans, backward rural areas in the late 1940s with high mortality, then had the lowest expectation of life in Europe. But they did not share the worst of the Eastern Bloc mortality pattern. Thirdly, even within Western Europe there is still enough lack of convergence to excite the epidemiologists. An early group of low-mortality achievers (the Netherlands, Denmark and Norway, for example) reached a high expectation of life in the 1950s and 1960s (about 70) but then failed to make much further progress. The statistical mediocrity of survival in very successful economies such as that of Germany, with high expenditure on medical services, suggests that further convergence cannot be expected simply on economic grounds (Valiin et at, 2002). Models of economic inequality (Wilkinson, 1996) compete with dietary hypotheses. The superior survival of the populations of the 'olive oil belt' of poorer Southern Europe compared with the richer 'sausage, beer and chips' belt of Northern Europe confounds economic explanation of mortality differentials, but does connect persuasively with parallel advantages shown surprisingly in such places as Albania (Gjonca and Bobak, 1997) and Macedonia. It may be that the twentieth century saw so many upheavals that its outcome is beyond the reach of theory, and the turbulence created has unhelpfully muddied the demographic waters. The expectation for any future regime^) must be one of constrained variety. Past trends may be some guide, in that demographic characteristics do seem to be geographically located and some salient characteristics today connect with the particularities of the past, for example the familism and household patterns of Southern Europe (Reher, 1998), and the Scandinavian history of births outside marriage, now shared by Estonia. The ancient division of Europe by Hajnal's line, which outlasted the Ottoman and Russian Empires and was if anything reinforced by the Soviet one, looks as though it will not survive the end of the Iron Curtain. But what shape it will assume, and to which of the diverse patterns of Western Europe its components will converge, remains beyond prediction. ACKNOWLEDGEMENTS This analysis was produced with the assistance of Dr Tarani Chandola. Previous help was given by Miss Sara Beale and Mr David Thomas. It formed part of a project supported by grants from the Economic and Social Research Council, London (Award Number L315253006) and by the World Society Foundation, Zurich. REFERENCES Abramovitz M. 1986. Catching up, forging ahead, falling behind. Journal of Economic History XL VI: 2. Adams JGU. 1985. Risk and Freedom: The Record of Road Safety Regulation. Transport Publishing Projects: London. Ajus F, Henye I. 1994. Illegitimacy in Hungary 1880-1920. Journal of Family History 19(4): 369-388. Beets G. 1995. Does the increasing age at first birth leads to increases in involuntary childlessness? In Evolution or Revolution in European Population, Vol. 2, Contributed Papers Session I-IV. Franco Angeli: Milan; 15-30. Bettio F, Villa P. 1996 A Mediterranean Perspective on the Breakdown of the Relationship between Participation and Fertility. Universita degli Studi di Trento Dipartimento di Economia, Trento. Discussion Paper No 5. Billari FC, Kohler H-P. 2002. The impact of union formation dynamics on first births in West Germany and Italy: are there signs of convergence? In Fertility and Partnership in Europe: Findings and Lessons from Comparative Research, Klijzing E, Corijn M (eds). United Nations: New York. Blayo C. 1987. La fecondité en Europe depuis 1960: Convergence ou Divergence? Paper presented at the European Population Conference, Jyväskylä, 1987. Bosveld W. 1996. The Ageing of Fertility in Europe: A Comparative Demographic - Analytic Study. PDOD Publications, Thesis Publishers: Amsterdam. Chandola T, Coleman DA, Hiorns RW. In press. Distinctive features of age-specific fertility profiles in the English-speaking world: common patterns in Australia, Canada, New Zealand and the United States. Population Studies. Chasteland J-C, Chesnais J-C. 1997. Quelle 'Mon-dialisation'? In La Population du Monde: enjeux et problémes, Chasteland JC, Chesnais JC (eds). INED / Presses Universitäres de Paris: Paris; 5-8. Coleman DA. 1996. New patterns and trends in European fertility: international and sub-national Copyright © 2002 John Wiley & Sons, Ltd. Int. J. Popul Geogr. 8, 319-344 (2002) 342 D. A. Coleman comparisons. In Europe's Population in the 1990s, Coleman DA (ed.). Oxford University Press: Oxford; 1-61. Conrad C, Lechner M, Werner W. 1996. East German fertility after unification: crisis or adaptation? Population and Development Review 22: 331-359. Coward J. 1986. The analysis of regional fertility patterns. In Population Structures and Models, Woods RI, Rees P (eds). Allen and Unwin: London; 45-67. de Beer J, van Wissen L. (eds). 1999. Europe: One Continent. Different Worlds. Population Scenarios for the 21st Century. Kluwer: Dordrecht. Decroly J-M, Vanlaer J. 1991. Atlas de la Population Européenne. Editions de l'Université de Bruxelles: Brussels. Erikson R, Goldthorpe JH. 1993. The Constant Flux -a Study of Class Mobility in Industrial Societies. Clarendon Press: Oxford. European Commission 1995. The Demographic Situation in the European Union. 1994 Report. Office for Official Publications of the European Communities: Luxembourg. Eurostat 1996. Demographic Statistics 1996. Office for Official Publications of the European Communities: Luxemburg. Field MG. 1998. Comparative health systems and the convergence hypothesis: the dialects of universalism and particularism. In Health Care Reform: International Perspectives, Powell F, Wessen A (eds). Sage: New York. Frejka T, Calot G. 2001a. Cohort Childbearing Age Patterns in Low-Fertility Countries in the 20th Century: Is the Postponement of Birth an Inherent Element? IUSSP seminar on International Perspectives on Low Fertility: Trends, Theories and Policies, Tokyo, March 2001: ISSUP: Paris http://demography.anu.edu/VirtualLibrary/ ConferencePapers/IUSSP2001/Program.html. Frejka T, Calot G. 2001b. Cohort reproductive patterns in low-fertility countries. Population and Development Review 27: 103-132. Fukuyama F. 1992. The End of History and The Last Man. Hamish Hamilton: London. Gauthier AH. 1996. The State and the Family: A Comparative Analysis of Family Policies in Industrialized Countries. Oxford University Press: Oxford. Gesano G. 1999. Maastricht requirements for European populations? Demotrends - quadrimes-trale sulla realta demografica italiana 1999(1): 1. Gjonga A, Bobak M. 1997. Albanian paradox: another example of protective effect of Mediterranean lifestyle? Lancet 350(9094): 1815-1817. Goldthorpe J. 1984. The end of convergence: corporatist and dualist tendencies in modern western societies. In Order and Conflict in Contemporary Capitalism: Studies in the Political Economy of Western European Nations, Goldthorpe J (ed.). Clarendon Press: Oxford; 315-343. Goode WJ. 1963. World Revolution and Family Patterns. Free Press, Collier-Macmillan: London. Grasland C. 1990. Systěmes demographiques et systěmes supranationaux: la fécondité européenne de 1952 ä 1982. European Journal of Population 6: 163-191. Greve B. 1994. Development in the EC and its impact on the Welfare State in Europe — trends towards convergence in the last ten years. History of European Ideas 19: 147-152. Hamm B, Litsch M. 1987. Sunbelt versus Frostbelt: a case for convergence theory? International Social Science Journal 39(2): 199-214. Hantrais L. 1997. Exploring relationships between social policy and changing family forms within the European Union. European Journal of Population 13: 339-379. Hawley AH. 1950. Human Ecology: A Theory of Community Structure. Ronald: New York. Hobcraft J, Kiernan K. 1995. Becoming a parent in Europe. In Evolution or Revolution in Human Population - European Population Conference, Milan, 1995. European Association for Population Studies. FrancoAngeli 1., Milan. Plenary Sessions: 27-66. Hoem B, Hoem JM. 1997. Sweden's Family Policies and Roller-coaster Fertility. Stockholm Research Reports in Demography 115, Demography Unit, Stockholm University. Inglehart R. 1990. Culture Shift in Advanced Industrial Society. Princeton University Press: Princeton, NJ. Inkeles A, Masamichi M (eds). 1996. Comparing Nations and Cultures: Readings in a Cross-disciplinary Perspective. Prentice Hall: Englewood Cliffs, NJ. Iwasawa M. 2001. Partnership transition in contemporary Japan: prevalence of childless non-cohabiting couples. ISSUP seminar on International Perspectives on Low Fertility: Trends, Theories and Policies, Tokyo, March 2001. IUSSP Working Paper Series. IUSSP: Paris. http://demography. anu.edu/VirtualLibrary/ConferencePapers/ ISSUP2001/Program.html. Jones GW. 1993. Is demographic uniformity inevitable? Journal of the Australian Population Association 10: 1-16. Kaelble H. 1989. ĽEmergence ďune société européenne. Futuribles 138: 19-27. Kerr C, Dunlop JT, Harbison FH, Meyers CA. 1962. Industrialism and Industrial Man. Heinemann: London. Kerr C. 1983. The Future of Industrial Societies: Copyright © 2002 John Wiley & Sons, Ltd. Int. ]. Popul Geogr. 8, 319-344 (2002) Convergent Demographic Communities 343 Convergence or Continuing Diversity? Harvard University Press: Cambridge, MA. Kosonen P. 1995. European Welfare State models: converging trends. International Journal of Sociology 25: 81-110. Kruskal JB, Wish M. 1978. Multidimensional Scaling. Sage: Beverly Hills. Ksenofontova N. 1994. Trends in infant mortality in the USSR. In Demographic Trends and Patterns in the Soviet Union before 1991, Lutz W, Scherbov S, Volkov A (eds). Routledge: London; 359-378. Kuijsten AC. 1996. Changing family patterns in Europe: a case of divergence? European Journal of Population 12: 115-143. Lechner M. 2001. The empirical analysis of East German fertility after unification: an update. European Journal of Population 17: 61-74. Leete R. 1994. The flight from marriage and parenthood among the overseas Chinese in Asia. Population and Development Review 20(4): 811-829. Leete R, Alam I. (eds) 1993. The Revolution in Asian Fertility: Dimensions, Causes and Implications. Oxford University Press: Oxford. Lesthaeghe R. 1995. The second demographic transition in Western countries: an interpretation. In Gender and Family Change in Industrialized Countries, Mason KO, Jensen A-M (eds). Clarendon Press: Oxford; 17-62. Lesthaeghe R. 2001. Postponement and Recuperation: Recent Fertility Trends and Forecasts in Six Western European Countries. IUSSP seminar on International Perspectives on Low Fertility: Trends, Theories and Policies, Tokyo, March 2001. ISSUP: Paris. http://demography.anu.edu/ VirtualLibrary/ConferencePapers/ISSUP2001/ Program.html. Lesthaeghe R, Moors R. 2000. Recent Trends in Fertility and Household Formation in the Industrialized World. Interuniversity Papers in Demography IDP-WP 2000-2. Interface Demography, Free University of Brussels. Lutz W. 1989. Distributional Aspects of Fertility. Academic Press: London. Lutz W, Scherbov S, Volkov A. (eds). 1994. Demographic Trends and Patterns in the Soviet Union before 1991. Routledge: London. McDonald P. 2000. Gender equity in theories of fertility transition. Population and Development Review 26(3): 427-440. Macura M. 1974. Population policies in socialist countries of Europe. Population Studies 28(3): 369-379. Macura M. 2000. Fertility decline in the transition economies 1989-1998: political, economic and social factors. In Economic Survey of Europe. United Nations: Geneva; 189-207. Macura M, Eggers M, Frejka T. 1995. Demographic change and public policy in Europe. In Population, Family and Welfare: A Comparative Survey of European Attitudes, Moors H, Palomba R (eds). Clarendon Press: Oxford; 8-44. Marshall G, Swift A, Roberts S. 1998. Against the Odds? Social Class and Social Justice in Industrial Societies. Clarendon Press: Oxford. Mellens M. 1999. Uniformity and diversity defined. In Europe: One Continent, Different Worlds. Population Scenarios for the 21st Century, de Beer J, van Wissen L (eds). Kluwer: Dordrecht; 33-44. Meslé F, Shkolnikov VM, Hertrich V, Valiin J. 1996. Tendances récentes de la mortalite par cause en Russie 1965-1994. INED: Paris. Meslé F, Hertrich V. 1997. Evolution de la mortalite en Europe: la divergence s'accentue entre l'Est et l'Ouest. In International Population Conference Beijing 1997, 3, Volume 2, IUSSP (ed.). International Union for the Scientific Study of Population: Liege; 479-508. Michaeli GA. 1996. New patterns of family formation in Italy. Which tools for which interpretations? Genus LII: 15-52. Nerlove M. 1973. Household and economy: towards a new theory of population and economic growth. Journal of Political Economy 82, 200-218. OECD. 1994. Women and Structural Change: New Perspectives. OECD: Paris. Olah LS. 1998. Do Public Policies Influence Fertility? Evidence from Sweden and Hungary from a Gender Perspective. Stockholm Research Reports in Demography 130. Stockholm University Demography Unit: Stockholm. Philipov D. 2001. Low fertility in Central and Eastern Europe: culture or economy? Paper presented at the IUSSP Symposium on International Perspectives on Low Fertility: Trends, Theories and Polices, Tokyo March 2001. International Union for the Scientific Study of Population Working Paper Series: Paris. Philipov D, Kohler HP. 2001. Tempo effects in the fertility decline in Eastern Europe: evidence from Bulgaria, the Czech Republic, Hungary, Poland and Russia. European Journal of Population 17: 37-60. Reher DS. 1998. Family ties in Western Europe: persistent contrasts. Population and Development Review 24(2): 203-234. Rothenbacher F. 1995. Household and family trends in Europe: from convergence to divergence. Eurodata Newsletter 1: 3-9. Roussel L. 1992. La famille en Europe occidentale: divergences et convergences. Population 47: 133-152. Roussel L. 1994. Fertility and family. Proceedings of Copyright © 2002 John Wiley & Sons, Ltd. Int. J. Popul. Geogr. 8, 319-344 (2002) 344 D. A. Coleman the European Population Conference Geneva 1993 1: 35-110. Roussel L. 1995. Vers une Europe des families? Futuribles 200: 47-62. Sardon J-P. 1990. Le remplacement des generations en Europe depuis le debut du siěcle. Population 45(6): 947-968. Tomka B. 2002. Demographic diversity and convergence in Europe 1918-1990: the Hungarian case. Demographic Research 6: 2. http://www. demographic-research.org UNECE 2000. Catching up and falling behind: economic convergence in Europe. In Economic Survey of Europe 2000 United Nations: Geneva; 155-187. Vallin J, Meslé F, Valkonen T. 2002. Trends in Mortality and Differential Mortality. Population Studies no. 36. Council of Europe: Strasburg. van de Kaa DJ. 1987. Europe's second demographic transition. Population Bulletin 42: 1. van de Kaa DJ. 1994. The second demographic transition revisited: theories and expectations. In Population and Family Life in the Low Countries 1993: Late Fertility and Other Current Issues, Beets GCN, et al. (eds). Swets and Zeitlinger: Lisse; 81-126. van de Kaa DJ. 1999. Without maps or compasses-population scenarios for the 21st century. European Journal of Population 15 (4): 309-316. Viazzo PP. 1986. Illegitimacy and the European marriage pattern: comparative evidence from the Alpine area. In The World we have Gained: Histories of Population and Social Structure, Bonfield L, Smith R, Wrightson K (eds). Blackwell: Oxford; 100-121. Wall R. (ed.) 1983. Family Forms in Historic Europe. Cambridge University Press: Cambridge. Watkins SC. 1991. From Provinces into Nations: Demographic Integration in Western Europe. 1870-1960. Princeton University Press: Princeton. Wilkinson R. 1996. Unhealthy Societies. Routledge: London. Williamson JB, Fleming JJ. 1996. Convergence theory and the social welfare sector: a cross-national analysis. In Comparing Nations and Cultures, Inkeles A, Sasaki M (eds). Prentice Hall: Englewood Cliffs, NJ; 348-355. Copyright © 2002 John Wiley & Sons, Ltd. Int. ]. Popul Geogr. 8, 319-344 (2002)