Tuesday, September 02, 2008

Eurostat on European populations in 2060

I wanted to call attention to the recent results of Eurostat's metadata projection of the projected population changes of the European Union's twenty-seven member-states as well as Norway and Switzerland between now and 2060. (My thanks go to the The Financial Times's Tony Barber's post.) The Eurostat report Ageing characterises the demographic perspectives of the European societies" provides a detailed exploration of these changes. Table 1, describing some of the most notable of these changes, is below.

There have been all kinds of reactions to this news in individual nations. Let's start with the smaller member-states. Observers in the Czech Republic pointed out that the Eurostat data presuming a decline from 10 to 9 million underestimate immigration and births, with some arguing that the population could instead rise to 13 million by 2060. People in the Republic of Ireland are reacting to the news that, with an estimated 2060 population of 6.7 million, the island of Ireland would have regained its pre-Famine population of eight million. News that the population of Estonia might decline by one-sixth to 1.1 million have been greeted with concern, along with the news that Bulgaria's population is projected to fall by 29%, as have news that Romania will certainly see rapid and perhaps economically unsustainable population aging as the population falls by 4.5 million.

The changes among the largest European Union states are perhaps especially noteworthy for their influence on the balances of economic and perhaps political power, Britain's projected growth to 77 million people, giving it the largest national population in Europe, is fitting into national concern over "uncontrolled" immigration, while metropolitan France's expected growth to nearly 72 million--not, it should be noted, out of line with 2005 projections charting a French population of 75 million by 2050--coexists with a Gemran population projected to fall to less than 71 million and a Spain projected to grow to just short of 52 million people. Italy's population is projected to remain stable at 59 million, but quite frankly the numbers look cooked--is a natural decrease of 12.0 million really going to be almost entirely balanced out by an immigration of 11.8 million? Who knows, perhaps it is the recent rivalry with Spain at work. Poland, at present the sixth EU member-state by population at 38 million is projected to see a fall to 31 million. Barber is quite right to note that all these changes will of necessity influence the development of Europe.

It is hard to believe that such massive changes, which Eurostat says will take place in spite of immigration into the EU, would not have a big impact on the distribution of power in the EU. For example, the Lisbon treaty - which, of course, may never come into force - recognises Germany’s present pre-eminence by allocating Germany more European Parliament members than any other country. But that arrangement surely could not last if Eurostat’s forecast were to prove accurate.

As for Poland, its leaders cited its population size last year as an argument for more weight in the EU’s institutions. But if its population were to shrink as much as Eurostat predicts, it would be difficult to make the case that Poland deserves the same influence as, say, Spain.

Meanwhile, the UK would find itself in the remarkable position of being the largest country in an organisation that it has never seemed entirely sure it wants to be part of. Of course, the secession of Scotland (with just over 5m people now) from the UK would make a difference.

One glaring omission from the Eurostat report is Turkey, an official candidate for EU membership. Hostility to Turkey’s bid in countries such as Austria and France stems partly from the objection that Turkey is already so big (more than 70m people) that its admission would fundamentally change the EU’s nature. But I see that, according to a recent United Nations Population Fund forecast, Turkey will keep on growing and have over 100m people by 2050.

Everyone, everywhere, is concerned about population aging, much more rapid in some countries--Bulgaria and Romania stand out particularly--than in others.

It should be noted that as we've seen with Spain, these projections can change radically if you account for the possibility of large-scale migration described in the report's Table 12.

Maybe the populations of high-income places like the Czech Republic, Slovenia or Estonia will grow as Ukrainians, Vietnamese, and others move to pleasantly high-income societies. Maybe France will open its dors to la francophonie while Spain will shut its doors. Maybe the populations of Romania and Bulgaria will decline even more quickly than projected as their economies get caught in downward spirals. Who knows for certain how fine details will evolve over the next 52 years? All that I'm willing to say is that for now, thi projection provides a useful starting point for discussions about population trends in their national and European contexts.

Monday, September 01, 2008

"Murderous identities and population paranoia"

Over at Himal South Asian, Mohan Rao's article "Murderous identities and population paranoia" explores the uses and misuses of popular demography in the wider world and particularly in India, host to a very large and diverse population. Indian Muslims, a population marked by period fertility rates higher than the average for Hindus and consequently a higher rate of population growth--over the 1991-2001 period, the Muslim population grew by 30% while the Hindu population grew by 20%--are the subject opf theories not wildly different from Eurabia claiming that Muslims will outreproduce Hindus and take over the country.

Although the population argument is seen to lend itself to a variety of prejudiced political projects across the world, the 20th century has also witnessed ‘communalisation of demography’ in the Subcontinent. In India, this has helped propel the discursive power of the communal myths (and outright lies) that surround the population question – or what some have termed ‘saffron demography’, an idea that carried particular weight in the election in Gujarat soon after the 2002 anti-Muslim carnage. Ethnic nationalism, combined with various fundamentalisms, gives saffron demography a particularly vicious punch. Of course, the communalisation of the population question is not unique to India. Present within Southasia (and many parts of the rest of the world), it seeks to legitimise offensive policies against minorities in order to build political consensus across newly created political majorities, fixed around ascribed identities that are all too often manufactured. Such a proposition is useful not just in manufacturing these majorities, but also in instilling fear, which is seen to bear political dividends.

There is an offensive slogan currently in use in North India, which is a play on a popular population control slogan of yesteryear – Hum do hamare do; woh paanch, unke pachees, which crudely translates as ‘We [Hindus] are two and have two children; they [Muslims] are five and have 25 children’. The suggestion is simple and beguilingly appealing, but also deeply flawed. The reference is, of course, to the fact that Hindus are not allowed by civil law to have more than one wife, while Muslims in India can have four. What this does not reveal is that data clearly shows that unlawful bigamous or polygamous marriages are more prevalent among Hindus than among Muslims. For example, as per the available data, the percentage incidence of what are called polygynous marriages (in which a man has more than one wife) is 5.8 among Hindus, while it is 5.73 among Muslims. What this also overlooks is that, assuming a situation of relatively equal males and females, a Muslim man with four wives would actually contribute less to population growth than if each of the wives were to be married to different men. Another oversight is that Muslims, like Hindus, are not a monolithic, homogenous community. Muslims in Kerala or Tamil Nadu, indeed in South India in general, typically have smaller families than Hindus in states such as Uttar Pradesh and Bihar in North India. Clearly, then, religion is not the main issue at play.

In response to a recent series of articles in the Indian press attempting to debunk myths about high Muslim population growth rates, this writer received a barrage of vehemently critical letters. Some suggested a change of name and conversion to Islam; others argued that he was an enemy both of Hindus and of India in general, and that he should go to Pakistan. Some others offered some bizarre facts to ‘refute’ what was being said. There was a long letter, enclosing two papers presented at international conferences for non-resident Indians, arguing that Muslims seek, through population growth, to overrun Europe. The correspondent was a retired inspector-general of police, a member of what seems to be a front organisation of the Rashtriya Swayamsevak Sangh (RSS) named the Patriotic Front. What is remarkable among these people is that facts are relatively unimportant; they are useful only if they further prejudice.

Although I'd quibble with some of his arguments on the role on demographics in determining wider political and cultural trends, the article's quite worth reading.

Economic Growth and the Demographic Dividend in Chile

By Claus Vistesen

There are many perspectives through which to look at economic development and growth. Geography, institutions or perhaps just plain good old physical capital accumulation are all important parameters. This small piece suggests a further metric and attempts to frame the argument with Chile as a case study.

Specfifically, this note explains the process known as the demographic dividend and conceptualizes it in a Chilean context. The analysis shows how Chile during the last two decades has benefited from the dividend proxied by the increasingly favorable trend in overall age structure of the society. By some measures Chile’s demographic dividend is thus ending during these very years. Yet, by adapting a slightly broader definition of the optimal working age and subsequent productivity profile, it appears that Chile still finds itself in the proverbial sweet spot and will continue to do so for the next decade. Coupled with the favorable windfall from copper exports and the subsequent transformation of this into an unprecedented net wealth position of Chile’s public accounts, the economy looks on a very solid footing to face whatever travails that might come next.

A Good Run

As can be observed below, Chile did indeed lose a substantial amount of output surrounding the Latin American debt crisis in the 1980s as well as the Asian currency crisis in 1997. Yet, and although Chile’s economy did not emerge unscathed from the past three decades of emerging market crises, the economy still managed to recover in terms of output. [1]

Chile's growth performance depicted by the chart is interesting in so far as it shows us the period that some scholars have dubbed Chile's Golden Age (Gallego and Loayza, 2002) due to the extended period of high growth rates. Between 1984 and 1998 Chile's growth rate in output per capita averaged 5.15% a year with a volatility of 2.64% p.a. This compares with an average growth rate in output per capita between 1998 and 2008(f) of 2.61% and a subsequent volatility of 1.73%. The 1985-1998 figures are remarkable and thus deserve some explanation.

According to Gallego and Loayza (2002) Chile's impressive growth performance primarily comes down to improvements in total factor productivity induced by increased investment in human capital and the development of a sound and coherent institutional setup. As such, and not unlike other growth accounting exercises the authors initially find that TFP accounts for the biggest share of output growth alongside the usual suspects of capital accumulation and growth in the labour force, the latter which is (in)famously coined as synonomous with population growth in the neo-classical growth model

The empirical approach is rather straight forward in terms of methodology, and is closely related to the tenets of endogenous growth theory as well as of course Mankiw, Romer and Weil's (1992) seminal findings that investment in human capital be considered an important part of capital accumulation. Formally, the authors first estimate a cross-section regression framework (GMM) based on a, more or less, standard neo-classical growth model augmented with human capital (schooling rates and life expectancy). The authors also include; government consumption to GDP, financial market development, terms of trade shocks, trade openess, and a black market premium. They find that this model account for 43% of the growth observed in Chile.

Unsatisfied with this result, the authors imbue the model with a number of variables whose origin in the growth theory framework are inspired by the tenets of endogenous growth theory. These variables include proxies for the political system, governance, public services and infrastructure, and with these, the new model moves reaches a coefficient of determination of 73%.

In line with endogenous growth theory the authors consequently find that this initial "residual" best be explained by improvements in the institutional edifice of Chile's economy. As a result and although the notorious convergence effect will tend to lead to lower overall growth rates in period t0 than in period t-1, the authors suggest that Chile focus further on institutional improvements to foster growth in the future.

Far be it from me to take issue with these results. However, in the following I propose another way to look at the past and future growth performance of Chile. It is important to understand that the two approaches are not mutually exclusive but ultimately directs the attention to a different set of governing mechanisms when it comes to economic growth.

A Demographic Dividend?

In one of their many papers on the subject David E. Bloom and David Canning (see Demographic Challenges, Fiscal Sustainability and Economic Growth, PGDA Working Paper no. 8) provide a useful historical sweep of the different approaches to demographic changes and their significance on the economic edifice. From the Malthusian epoch to a more optimist view on the benefits of vibrant population dynamics (see e.g. Simon Kuznets, Julian Simon, and Ester Boserup) and on to what Bloom and Canning coin as the “neutralists” [3] , the perspective on the importance of demographics has certainly changed a lot.

One crucial lesson to draw from the historical prism of demographic discourses is that the demographic transition is a far more complicated process than a mere transition in population growth rates as well as one of sectoral shifts in the economy. Lee (2003) consequently shows how the demographic transition also fundamentally changes the age structure of society whereas others such as Malmberg and Sommerstad (2000) and Hugh (2006) have suggested that the demographic transition be re-thought all together. Common for these contributions is the shifts in age structure, the complex mechanisms which govern these changes, and their subsequent effect and operationalization on the macroeconomic edifice.

Bloom, Canning and their fellow scholars on the PGDA at Harvard, [4] have furthermore showed how age structure makes a much more solid demographic yard stick, for gauging economic trends, than merely looking at population growth and absolute size of the population. This, I think, is the ultimate lesson to derive from decades worth of thinking on demographic processes. I would essentially divide the lesson into two irrefutable points. One is that age structure matters much more than population growth and that a simple metric such as median age can give us a tremendous amount of information on an economy's given and future growth path. The second points is simply that the demographic transition is not, by a long shot, over. In fact, nobody knows when it will end.

It is within this framework that the process known as the demographic dividend enters, and not surprisingly, it is all about age structure and how economies who go through the demographic transition at some point will find themselves with above average conditions for growth as the working age as well as productive share of the population is maximized. In terms of median age and as a crude benchmark, we can say that those economies with median ages between 25-35 are situated in or close to the optimal age structure for economic growth. Nothing comes for free however, and it is crucial to point out that the demographic dividend provides an opportunity rather than a sure benefit. For example, it seems that Eastern Europe and Russia, by and large, have gone through their demographic dividends without experiencing the corresponding win-win situation in which favorable growth conditions coincides with advances in terms of institutional quality and political stability.

The demographic dividend operates through two interconnected mechanisms in the form of falling fertility and declining infant mortality. In most countries, falling mortality as the economy moves through the demographic transition has been accompanied, with a lag, by falling fertility Bloom and Canning (2006). If we add a steady increase in life expectancy to proxy the general improvement in the health of the population these interconnected processes endow an economy with a period of, let us say, 15-20 years in which the young and working cohorts of the society are relatively big compared to the dependent cohorts. The former are often defined as the cohorts aged <20-25 [5] years and for the latter's part >65. As for quantitative importance, Bloom and Canning (2004) have shown this to have a positive effect on per capita output as well as they have famously shown how one third of the East Asian Tiger economies’ impressive growth spurt in the latter part of the 20th century can be explained by the demographic dividend.

More generally Bloom & Canning et al. (2007) have also demonstrated, through cross sectional regression data, how age structure can significantly improve the forecast of economies' growth rate relative to world GDP.

Chile’s Demographic Dividend

If large parts of East Asia have already had their demographic dividend what about Chile then. Is Chile about to receive, or more aptly; is she in the middle of her demographic dividend?

As can readily be seen, Chile almost displays a textbook case of economic development. In this way, infant mortality has fallen back sharply since the middle of the 1970s as well as life expectancy has increased. Outside the immediate realms of economics, biologists and health economists speak of the process known as the epidemiological transition to explain the progression of the change in (and drivers of) variables such as mortality, life expectancy, and other public health metrics.

The reduction of, and subsequently the current level of, infant mortality in Chile rivals that of many developed economies. According to Albala and Vio (1995) Chile managed to reduce infant mortality by 82% between 1970 and 1992 and Jimenez and Romero (2007) further shows how provisions of services to counter perinatal risks and acute respirator distress have helped Chile to reach an impressive infant mortality rate of 8.9 infants per 1000 thousands in 2000.

With respect to life expectancy Albala and Vio (1995) describe how the mortality rate of people aged 65 and more decreased 73% between 1970 and 1992 . Especially, a reduction in the mortality from cardiovascular causes is highlighted. In a more recent paper Albala, Vio et al. (2002) also latch on to increasing risk posed by a transition from a prevalence of infectious diseases to on in which chronic diseases ascend in importance. The usual suspects here would be an increase in obesity as a result of malnutrition through the consumption of high-fat/high-carbohydrate energy-dense foods and a decrease in physical activity. Chronic diseases which spring from such developments would then be e.g. type 2 diabetes and cardiovascular diseases. Evidence of this development appears in the context of school children; from 1987 to 2000 the prevalance of obesity among first grade school children rose more than 100% for both boys and girls.

Much debate has and will be devoted to the extent that such adverse developments from economic development could, at some point, break the curve in terms of life expectancy. At this point however, it seems as if advances in healt care services and the subsequent improvements in old age life expectancy are enough to keep the curve ticking upwars.

Returning to the question of demographic dividend in Chile, the trend of the decline in infant mortality exhibits the expected negative concave relationship as per function of the fact that the value cannot fall below 0. In order to build a simple model framework and by applying the logic expressed through theory above, we can construct a rudimentary econometric model to formalize the argumet.

Consequently, we let the lagged change (one year) [6] in the infant mortality rate predict the change, in year 0, of the fertility rate. Given the properties of the time series in question, and the theoretical framework above we would expect a positive but also a concave relationship since both variables are bound by the fact that they cannot fall below 0. In general terms, this model clearly assumes that the process of decline in fertility throughout the demographic transition is infinitely simpler than it really is. The crucial point here is that while the decline in infant mortality may be able to explain the decline in fertility on a certain part of the curve it cannot, and may in fact see its sign reverse, as we move further towards replacement level fertility and beyond. One could even with reasonable claim ask whether in fact the decline in fertility towards replacement levels is driven by infant mortality reductions alone. Nevertheless, the model estimated looks as follows where both variables are in changes.

Which leads to the following estimation:

The visual inspection of the model can furthermore be derived from the graph below.

In general, the model is far from solid but it manages to get the message across in the sense that it links the decline in fertility to the lagged decline in infant mortality [7] . The key thing to remember is the implicit and theoretical concave relationship cited above; a relationship also confirmed by the scatter plot.

The interesting thing about Chile here is that, according to standard demographic theory, the demographic transition should, by and large, end now as fertility trends towards replacement level. Not a lot of serious scholars would believe that however and we can thus expect fertility to decline below replacement level (see e.g. Wolfgang Lutz here). The extent to which it does not, Chile would clearly constitute something of a remarkable case. This is also why policy makers would be wise to consider implementing steps to avoid fertility dropping into lowest-low territory [8] , since what we know with almost certainty is that the demographic transition does not stop once infant mortality hits near rock bottom.

This point also highlights the idea that while the demographic dividend presents a window of opportunity so does the backdrop represent a penalty. This point is crucially related to the fact that only very few economies (e.g. the US and perhaps also France) have been able to stay at, or return to, replacement levels of fertility. In most other cases, fertility seems set bound to fall further and the only real metric to gauge is the speed by which this occurs. In an emerging market context the evidence is worrying to the extent that many economies have seen their fertility rates crash completely over the course of less than a decade. The next 5-10 years in Chilean, and indeed Latin American context, will be extremely interesting to watch in this regard.

Given the fact that Chile's fertility level is already approaching replacement level, the model cited above has, in all likelihood, run its course. What will likely cause Chile's fertility rate to fall below replacement level requires an entirely different set of explanatory variables and also theoretical edifice. Key trends would for example include an elaboration of the quantum and tempo effect of fertility in a context of rapid economic development and changing social norms.

To summarize the argument in a Chilean context, the ultimate data series to gauge, in the context of the demographic dividend would be age structure and the effect from the processes described above.

As per usual, beauty is in the eye of the beholder since depending on which definition you ascribe to the optimal age structure, Chile could be said to be in and out of the demographic dividend. The truth probably is that Chile is in the twilight hours of its demographics dividend. However, with a median age of about 30 years Chile still enjoys, and will continue to do so in the immediate future, the benefits of an age structure conductive to balanced economic growth.

One important point to note here is that the 25-44 bracket peaked sometime in the middle of the 1990s. Much evidence suggests though that it is a bit untimely to make the cut at the 44 year old age group, since many people are perhaps not far from their productive peak between 44 and 64. On the other, the peak of the 25-44 age bracket may still constitute an upper level of economic capacity if viewed as the ability and propensity to sustain housing booms, large negative external balances etc.


Chile still has ,and will continue to enjoy for the immediate future ,a favorable age structure for harboring economic growth and dynamism. Favorable is in this context defined through the spectrum of the demographic dividend and the subsequent increase in, and high proportion of, working age people to total population. Depending on fall in fertility, the demographic dividend is definitely tapering off at this point. If experience from East Asia is anything to go by Chile as well as its Latin American peers are now set to enter a new phase of the the demographic transition in which fertility steadily moves below and beyond replacement levels. The speed here is crucial. If it happens slowly, Chile can expect to posses a relatively balanced age structure in the decades to come but if the decline is swift and lingering the effect could be otherwise.

This small piece has also touched upon the way we conceptualize economic growth and development. I would not want to discount methods such as the one deployed in Gallego and Loayza (2002). However, I have suggested that a different perspective is a also considered. I would, in particular, emphasise this in the context of the future drivers of economic growth. Nobody can disagree with the impetus to move forward on strong institutional settings. Yet, economic development is not only accompanied by a demographic dividend but also, arguably, a demographic penalty which occurs as the effect of the dividend recedes and the decline in fertility continues. This would be where concepts such as the quantum and tempo effect of fertility comes in. it is also where policy makers would be wise to consider that a relentless strive to reach the apex of the value chain will also bring with it a deficit in terms of the proper quantity/quality mix of human capital.

List of References

Albala, Cecilia; Vio, Fernando; Kain, Juliana and Uauy, Ricardo (2002) - Nutrition transition in Chile: determinants and consequences, Institute of Nutrition and Food Technology (INTA), University of Chile

Albala, Cecilia and Vio, Fernando (1995) - Epidemiological transition in Latin America: The case of Chile, Institute of Nutrition and Food Technology (INTA), University of Chile

Bloom, D and Williamson, J (1998) Demographic transitions and economic miracles in emerging Asia. World Bank Economic Review. 12(3) 419-456.

Bloom DE et al. (2007) - Does Age Structure Forecast Economic Growth? PGDA Working Paper no. 20.

Bloom, DE & David Canning (2006) – Demographic Challenges, Fiscal Sustainability and Economic Growth, PGDA Working Paper no. 8.

Bloom, DE and Canning, D (2004) - Global demographic change: dimensions and economic significance, In Global demographic change: economic impacts and policy challenges (proceedings of a symposium, sponsored by the Federal Reserve Bank of Kansas City Jackson Hole)

Gallego, Francisco & Loayza, Norman (2002) - The Golden Period for Growth in Chile: Explanations and Forecasts, Working Paper, Central Bank of Chile no. 146

Hugh, Edward (2006) - Rethinking the Demographic Transition (can be downloaded by request)

Jiménez, Jorge and Inés Roméro, Maria (2007) - Reducing Infant Mortality In Chile: Success In Two Phases, Health Affairs, 26, no. 2 (2007): 458-465

Lee, Ronald (2003) - The demographic Transition: Three Centuries of Fundamental Change, Journal of Economic Perspectives, 17 (fall 2003), pp. 167-190

Malmberg, Bo & Lena Sommestad (2000) - Four Phases of the Demographic Transition, Implications for Economic and Social Development in Sweden, Working Paper 2000:6, Institutet for Framtidstudier

N. Gregory, Mankiw; Romer, David, and David N., Weil (1992) - A Contribution to the Empirics of Economic Growth, Quarterly Journal of Economics, vol. 107.

Kuznets, S (1967) Population and economic growth, in Proceedings of the American Philosophical Society, III (3).

Simon, J (1981) the ultimate resource. New Jersey: Princeton University Press.

[1] Although Chile did not recover from the Asian currency crisis to pre 1997 levels.

[2] Bloom & Canning (2006) – Demographic Challenges, Fiscal Sustainability and Economic Growth, PGDA Working Paper no. 8.

[3] Basically, this would be the modern institutional paradigm that has emerged within the economic growth/development discourse (see e.g. Daron Acemoglu, Dani Rodrik and Amartya Sen).

[4] See numerous contributions here: http://www.hsph.harvard.edu/pgda/working.htm

[5] I would argue that this is the right threshold (unlike the <15>

[6] The time series are in changes to correct for non- stationarity. As for the lag, the optimal number of lags could be more rigorously verified on the basis of theory and the statistical properties of the time series in question (VAR)

[7] Although, as can also be observed in the graphs, it cannot predict sudden reversals in fertility trends; i.e. these would essentially be treated as exogenous shocks to this model.

[8] A TFR of <1.5