The contribution of reduced population growth rate to demographic - - PDF document

1 27th International Population Conference, Cape Town 30 October – 3 November 2017

The contribution of reduced population growth rate to demographic dividend

Jane N. O’Sullivan

Honorary Senior Research Fellow, School of Agriculture and Food Sciences, University of Queensland, St Lucia 4072 Australia. j.osullivan@uq.edu.au

Theme 14. Population and Development

Theme Convener: Cassio M. Turra, Cedeplar, Universidade Federal de Minas Gerais (UFMG)

Abstract

The economic stimulus experienced in countries following a fall in fertility has been widely attributed to the improvement of dependency ratio, as the proportion of children falls. But age structure can account for less than a third of the stimulus observed in East Asian countries and the correlation has been reported to be inconsistent elsewhere. Population growth rate itself has economic impacts, which have lately been overlooked in economic analysis of the fertility transition. This paper describes measurement of the macroeconomic cost of providing physical capital for additional people. In Australia and the UK, this cost was estimated at 6.5- 7% of GDP per one per cent population growth rate. In rapidly growing, high fertility countries, this implies a debilitating burden. The relationship between population growth rate and economic advance was found to be remarkably consistent when comparing countries with slow and fast fertility decline, demonstrating the strength of the constraint which population growth rate places on economic advance. While growth in GDP per capita was found to be highly dependent on fertility rate, the decline in fertility was not found to be dependent on wealth. The paper concludes that voluntary family planning can be highly effective development stimulus.

Introduction

It has been observed that countries experience an economic boost after fertility declines. The main explanation to date has been the effect of reducing the numbers of children leading to a higher proportion of working age people, referred to as the ‘demographic dividend’ (Bloom and Williamson 1998, Canning et al. 2015). However, even if the increase in working age proportion fully translates into increased workforce, it accounts for between a fifth and a third

• f the economic stimulus observed (Wang and Mason 2007; Bloom and Canning 2008).

Garenne (2016) found the relationship between economic growth and dependency ratio was inconsistent in longitudinal analyses of African countries. The ‘infrastructure dividend’ is less well appreciated. It is directly related to the population growth rate, and arises from alleviating the need to acquire additional physical capital, including infrastructure, equipment and training of professional service providers, to extend the existing quality of life and employment opportunities to additional people. It is likely to

2 have a greater and more sustained impact on development stimulus than the demographic dividend – indeed, it operates equally in ‘ageing’ countries where age dependency is rising, and even benefits declining populations. The burden of equipping a rapidly-growing population is not a novel idea. It is occasionally alluded to by those who advocate that family planning should be a priority for development, not merely for health and rights. Lee (2009) suggests “[it would seem] so obvious: Larger, more rapidly growing populations have fewer natural resources per person, less physical capital per worker, more dependents, and greater needs for new social infrastructure. Of course they must be economically worse off.” Bongaarts (2016a) defines ‘demographic dividend’ broadly as “the boost to economic growth that follows a decline in fertility”, without reference to mechanistic channels. He explicitly incorporates the infrastructure burden alongside age dependency (Bongaarts 2016b), explaining, “A better-educated workforce has fewer dependents and more resources, and this ‘demographic dividend’ spurs economic growth. Governments are better able to keep pace with – and even get ahead of – meeting infrastructure needs of their citizens.” Nevertheless, without explicit quantification, it is omitted from metrics and models from which policy advice is derived. Demographic dividend potential is measured only in terms of age dependency ratios, and economic models generate misleading conclusions by ignoring the cost of expanding durable assets. The purpose of this article is to outline the measurement and the scale of this neglected impost, which in many contexts overwhelms other impacts of demographic dynamics. Sauvy (1958) first attempted to calculate what he termed the ‘demographic investment’ required to provide physical capital for additional people. This represents a substantial call on the limited saving capacity of rapidly growing developing countries. It is inevitably the first call on these funds, at considerable opportunity cost, preventing expenditure which would increase the capital/labour ratio and the quality of services delivered. Robinson (1974) applied Sauvy’s concept to the budget for Bangladesh’s first five-year plan. He concluded that the cost of ‘standing still’ at the prevailing 3% per annum population growth represented around 75% of all the investment. With the planned level of investment, incomes might be raised by 30% over 20 years, but if population growth were at the European level (0.45% p.a. at that time) an income increase of 150% would be expected. More recent discourse has referred to ‘demographic investment’ as ‘capital widening’ in contrast to ‘capital deepening’ of improving the provision per person. However, since these early works, there has been little attempt to quantify this impost and its impact on economic

• development. Sauvy’s work is less remembered than that of Solow (1956) who framed

physical capital more narrowly as a production factor. Subsequent models have tended to express it as the “capital shallowing effect” of additional labour (eg. Ashraf et al. 2013), and consequently to apply the marginal productivity of capital. This runs the risk of underestimating the productivity of the total industrial capital stock, and ignoring the public and domestic capital stock which are equally prone to dilution and, apart from affecting productivity indirectly, affect the relationship between income per capita and the quality of life it can achieve. Rather than extrapolating the dilution of capital to the point of mass unemployment, modellers tend to assume that capital will adjust, with the help of the expanded economy and diminished labour costs (Rowthorn 2015; Productivity Commission 2016), without quantifying the ongoing cost of such adjustment. Thus they have not adequately dealt with the impact of expenditure diversion on other consumption, nor the complex system failures resulting from chronic failure to keep pace with population growth.

3 This paper initially reviews the conceptual framework for quantifying the ‘infrastructure dividend’, reports the scale of the growth rate burden in those countries where it has been quantified, and discusses its potential application in high-fertility developing countries. The multiple channels of economic effects of population growth are discussed to place the infrastructure burden in perspective. In the second part of the paper, the significance of this burden is tested by examining data relating population growth rate to economic performance

• f countries. The controversial findings, that high population growth rate has indeed been a

strong impediment to development, and that voluntary family planning efforts, rather than economic or educational advance, have been the dominant driver of population growth reduction, point to a development opportunity which is currently underappreciated.

The cost of ‘capital widening’

The quality of life that a nation may provide for its citizens depends greatly on its stock of durable man-made assets, in addition to its endowment of natural assets. Man-made assets include all forms of infrastructure, from private housing, industrial and commercial structures to hospitals, utilities, transport and public amenity. In addition, they include all forms of equipment, from domestic appliances to vehicles and major industrial installations. Further, the supply of professional and trade services implies a prior investment in training, which also creates a durable asset. Each of these durable assets has a limited useful lifespan, and hence a proportion of total economic activity each year must be used for durable asset acquisition, to maintain the stock (O’Sullivan, 2012). In a stable, stationary population, the annual investment would be inversely proportional to the lifespan of the asset class: 100 divided by the lifespan in years equals the annual percentage turnover. Thus, if power stations last for 50 years, on average 2% of them would need to be replaced each year to maintain a stable stock. If municipal buses are in service for 10 years, 10 per cent of the fleet would need to be purchased per year. It is important to recognise that the total value of all durable man-made assets is typically several times greater than total annual GDP. In any one year, a society can only afford to provide a fraction of the stock. Durability allows many years’ worth of acquired assets to be enjoyed at any one time. Quality of life therefore depends greatly on the durability of the things we create. Population growth requires that the stock of all durable assets is expanded at the same growth rate, in order to maintain the current level of productivity, amenity and service provision that the population has already attained. The effect of population growth on durable asset expenditure is disproportionately higher than the rate of growth itself. For instance, a cost-weighted average lifespan of all infrastructure is in the order of 50 years, implying a replacement need of 2% of the total stock per annum. A population growth rate of 1% per year implies the need to expand the stock by 1% in that year, in order to keep pace with population growth. Consequently, society’s burden of annual infrastructure acquisition is raised from 2% to 3% of the existing stock, a 50 per cent increase. Similar calculations can be made for other categories of assets. Illustrative examples are given in Table 1. If trained professionals on average spend 33 years in the workforce after graduation, a stable, stationary population would need to graduate 3% of the workforce in

4 that profession annually to replace retirees. If that population were to grow at 1% per annum, it would need to graduate 4% of the workforce: 3% to replace retirees plus 1% to expand the

• workforce. This is a 33% increase over the burden carried by a stable population. The result

is that the percentage increase in annual acquisition needed for each 1% growth is equal to the working lifespan in years. Table 1. Illustrative examples of the burden of expansion to cater for a population growing at 1% per annum, relative to that of maintaining a constant stock to serve a stable population.

Asset class Working lifespan (illustrative estimation only) Annual acquisition burden to maintain a constant stock Annual acquisition for maintenance and expansion by 1% Increase in burden per 1% population growth Power stations 50 years 2% 3% 50% Buses 10 years 10% 11% 10% Nurses 33 years 3% 4% 33%

In proof-of-concept analyses based on Australian and UK data (O’Sullivan 2012, O’Sullivan 2013a), national accounts of Gross Fixed Capital Formation (GFCF), Household Final Consumption Expenditure, and tertiary education were used to collate national spending on asset classes of different estimated lifespan. Historical population data allowed the estimation

• f the proportion of stock in each lifespan group needing to be replaced in any year,

accommodating the effect of growth over the lifespan of an asset class reducing the turnover rate by diluting the oldest cohort due for replacement. Population growth rate in any year dictated the proportion of current stock needing to be acquired for capacity expansion. Thus the actual spending on durable assets could be apportioned to maintenance and expansion. Using this methodology, it was found that the replacement value of capital stock in the UK averaged 6.9 times GDP over the period 1968 to 2007 (O’Sullivan 2013a), and 6.5 times GDP between 1964 and 2004 in Australia (O’Sullivan 2014a). Thus, over that period, capital widening cost 6.9% of GDP per 1% population growth rate in the UK, and 6.5% of GDP per 1% population growth rate in Australia. While the UK averaged only 0.25% per annum population growth, the burden of capital widening was only 1.76% of GDP, but it has since climbed to nearer 5% with elevated population growth. For Australia, growing at an average

• f 1.44% per annum, the burden was 9.3% of GDP over the 40-year reference period.

The latter study found evidence for an escalation in cost per added person after population growth rate accelerated since 2004, suggesting that diseconomies of density and growth rate

• utweigh economies of scale. Diseconomies of density include requiring more costly

structures to cope with congestion, such as road tunnels and high-rise buildings, or substituting environmental services (such as gravity-fed water supply) with technological alternatives (such as regionally pumped water or desalination). Diseconomies of growth rate include shortening the lifespan of assets (decommissioning installations before they are worn

• ut) through the need to replace them with higher-volume versions, or to reallocate scarce

space to higher priority uses. Governments often neglect to factor infrastructure costs into budgets on the grounds that they are investments, which will ultimately fund their own financing. It should be stressed that ‘capital widening’ is a recurrent cost (consumption) of a growing population, not an investment (Figure 1A). By Sauvy’s (1958) definition of ‘demographic investment’, an

5 investment implied prior saving, but in modern monetary systems it implies borrowing from the future, justified by the net improvement it yields for the future users. Capital widening yields no betterment, it is required merely to stand still, in terms of economic and social

• standards. Like other recurrent costs, such as purchasing groceries, debt-financing tends to

spiral into insolvency. It merely adds the cost of servicing this ‘investment’ to the cost of further capacity expansion in subsequent years. This applies equally to other deferred payment arrangements, including privatisation of public utilities and public-private partnerships, where the cost is recouped by higher user charges. Thus, failure to pay for expansion with current income decreases future ability to keep pace with growth, either by adding debt repayments incurred to expand infrastructure, or by crowding the infrastructure which was not sufficiently expanded. As illustrated in Figure 1B, the debt quickly

• compounds. This may be manifested by government austerity reducing services and income

support for the poor, crowded schools and hospitals lowering education and health standards, reduced productivity affecting competitiveness of industries, and/or mounting unemployment as businesses fail to expand. A. B. Figure 1. Conceptual illustration of: A. the components of durable asset acquisition as recurrent cost or investment, and B. the escalation of deficit if capacity expansion fails to keep pace with population

• growth. From O’Sullivan (2012).

The application of ‘infrastructure dividend’ in high-fertility underdeveloped countries

It would be useful for African policy-makers to know how much it costs to equip an extra citizen with the infrastructure and service capacity to be no poorer than current citizens. At

Year 2

depreciation = investment (getting ahead) maintenance expansion improvement

Year 1

recurrent cost (treading water)

}

Replacement Cost of All Durable Assets $

demand growth due to population growth current inventory

Year 2

depreciation = investment (getting ahead) maintenance expansion improvement

Year 1

recurrent cost (treading water)

}

Replacement Cost of All Durable Assets $

demand growth due to population growth current inventory

Year 1 Year 2 Year 3

demand growth d e p r e c i a t i

• n

acquired within budget infrastructure deficit

• r budget deficit

Replacement Cost of All Durable Assets $ Year 1 Year 2 Year 3

demand growth d e p r e c i a t i

• n

acquired within budget infrastructure deficit

• r budget deficit

Replacement Cost of All Durable Assets $

6 the national scale, what proportion of GDP is being diverted to this purpose? How much economic capacity would be freed up by a reduction in population growth rate of, for example, 0.5%? In whose hands would this extra spending power lie, and how could it be spent? This section does not attempt to resolve these issues, but only to reflect on their scope and implications. When population growth is reduced through a change in birth rate rather than a change in net migration, there are differences in timing of specific impacts. The latter immediately impacts household formation and employment opportunities. The former may immediately impact women’s engagement in the workforce (a demographic dividend) and demand for health services, and will soon affect education systems, but the need for housing and jobs will not be affected for a couple of decades. However, in both scenarios, a near-term increase in household saving capacity is expected, either through reduced expenditure on children or a reduced proportion of underemployed people and of highly-indebted new mortgagees among households. There are also differences in what is being measured, and how capital widening is borne by the community. Most developing countries still have large subsistence sectors, in which not

• nly food but materials for construction and equipment are derived from the natural

environment directly. They also have significant informal sectors, where monetary or in-kind trade occurs without being measured. Official GDP statistics may estimate the scale of such activities differently from one country to another (Jerven 2013). Where economic data are of variable coverage, there may be opportunities to measure specific channels of the impact, such as the proportion of imports dedicated to capital widening, and their impact on balance

• f trade.

Although a majority of people may be currently employed in the subsistence and informal economy, capital widening may be mostly restricted to the formal economy. Increasingly, additional people are moving to cities, and even those in rural settings are opting for a higher proportion of purchased materials for construction, exchanging traditional tools for manufactured equipment and accessing the services of professional personnel. If GDP mainly represents the economy of the urban and industrial areas, rather than the nation as a whole, then it may be that the relevant population growth rate is that of the urban and industrial areas also – often double that of the nation as a whole. Let us take an example, in which the total stock of infrastructure and equipment in a city has a replacement value of seven times the GDP generated in that city annually. While the national population growth rate may be 2.5% per annum, that of the city is 5% per annum. This implies that 5 x 7 = 30% of total economic activity must be directed to building or acquiring new infrastructure, equipment and professional expertise, merely to prevent the population getting poorer on average. The rural areas, with higher birth rates but high out- migration, may not be growing as fast. But the government’s capacity to deliver services to them will be constrained by the demands of capital widening in the city. This dynamic may go part of the way to explaining the apparent inequity of resource distribution, where rural regions appear relatively neglected. Despite being poorer, they may attract a smaller fraction of government spending, as their situation may appear more stable while that in the city threatens to worsen rapidly if not sufficiently addressed. Hence it may be the growth rate, rather than the development deficit, which dominates the allocation of

• resources. Only a reduction in growth rate allows increased attention to development deficit.

7

Contribution of the infrastructure dividend to overall impacts of population growth

Population growth acts on economic dynamics through several channels, including density, age structure, and growth rate. Density: The ratio of people to the natural resource base is most often discussed in terms of regional carrying capacity. It is manifested through food and water insecurity, or environmental degradation and pollution. Raising the productivity of the natural resource base, through technological or institutional innovations, can increase carrying capacity, but as Thomas Robert Malthus (1798) famously observed, this is an incremental process which can’t indefinitely outpace the exponential tendency of unchecked population growth. Around the time Malthus wrote, global carrying capacity began a sudden and unprecedented expansion with the recruitment of fossil fuels and the globalisation of agricultural commodities. Having evolved in this time of plenty, modern economic theory is inclined to disregard natural resources as a limiting factor, relying on any form of income generation to provide access to all necessary resources through global markets. However, this strategy exposes the population to increasing risk of external shocks, and forces activity to be export-oriented in an increasingly competitive market. The niches for densely populated, trade-dependent countries are already crowded. The strategy has worked well enough while overpopulation is localised and other countries have surplus products of natural resources that they are willing to trade. Indeed, trade globalisation has eased the population pressure in many countries in recent decades. During this time, many commentators argued that scarcity would always generate a work-around through innovation (Simon, 1981). But as humanity increasingly tests the planetary limits of resource capacity, including fresh water availability and the capacity to buffer changes caused by greenhouse gas emissions and other pollutants, carrying capacity is likely to become the main issue of concern once more. Age structure: Profound changes in age structure are an inevitable consequence of the demographic transition, from high to low rates of both deaths and births. The initial impact of reduced infant mortality was an enormous increase in the proportion of children, creating a burdensome ‘youth dependency ratio’. As fertility falls, this burden is eased, as has already been discussed in terms of the ‘demographic dividend’. Subsequently, as low-fertility populations stabilise or gradually decline, the proportion of older people increases. This ageing trend has been the subject of much misinformed concern in recent years. Contrary to popular belief, ageing will not continue inexorably but would stabilise with a majority of people still being of working age. The presumption, for both the demographic dividend in mid-transition countries and the ‘senescent’ ageing countries, is that a shortage of labour constrains economic advance. However, there is little evidence that this is the case. In countries where adult labour is oversupplied and underutilised through lack of physical and human capital, the benefits of a large working age proportion may be weak or even negative. Furthermore, despite dire predictions, ‘ageing’ societies have not yet seen any shrinkage of

• workforce. So far, they have maintained similar proportions of people actually employed,

through greater workforce participation rates (O’Sullivan 2014b, Betts 2014). The responsiveness of participation rates to labour demand suggests that labour is oversupplied even in ageing countries. Having fewer people chasing the available jobs has more

8 advantages than disadvantages. The increased burden of old-age pensions would be at least partly off-set by lower unemployment and income support payments for working age people, and probably further off-set by lower law-enforcement and incarceration rates of disenfranchised youth. Growth rate: The rate of change in population, as a factor with unique impacts, has been relatively neglected in recent population-development discourse. Conceptualisation has been a major barrier. Economic training in cost-benefit analysis has tended to encourage thinking in terms of a one-off addition of a worker or child, balancing their future productivity against the cost of provisioning them. This is a problematic approach, which tends to assume the short-term impacts are short-lived (Rowthorn, 2015), forgetting that each cohort is followed by another. It further ignores that provisioning costs for people who are ‘additional’ differ from those who are ‘replacement’, while their future productivity, in general, does not. In contrast, the concept of capital widening draws on the logic of calculus, in which an ongoing rate of change confers measurably different characteristics at every point in time, compared with a higher or lower rate. One way to conceptualise the burden of capital widening in relation to demographic dependence, is to imagine this burden as attributable to a sub-group of dependent people, the ‘not-yet-added’. The not-yet-added pay no taxes and do no work, so the costs of providing for them are born by the current workforce. (It may seem that we are always widening capacity for people who are already present, but only to the extent that we are always playing catch-

• up. If they stopped being added, we would soon catch up and then the need for that activity

would disappear.) Figure 2 illustrates the impact on dependency ratio, when this burden of population growth rate is included. The circles represent the economic ‘pie’ of Gross National Income. Dependency ratios assume that the wealth generated by working age people must be distributed to the whole population, so the larger the proportion of working age, the easier it is to provide well for everyone. Accepting this oversimplification for the moment, Figure 2A shows Australia representing the ‘demographic dividend’, being near its high-point in proportion of working-age people. By comparison, high-fertility Uganda carries a heavy burden of youth dependency, while Europe and Japan illustrate the relatively modest increase in dependency due to ageing. When we add in the proportion of GDP required for capital widening to accommodate the not-yet-added (Figure 2 B), the challenge for rapidly growing countries is dramatically increased, while the disbenefit of an ageing population is largely annulled.

9 Figure 2. Economic dependency ratios, A: the pie of gross national income divided among demographic categories, on a notional ‘per capita’ basis (i.e. according to the percentage of population under 15, 15- 64 and over 65), and B: the distribution of GNI when the cost of capacity expansion is included. The population growth rate is given under each country name. Capacity expansion (the cost attributed to the ‘not yet added’) is provisionally assumed to have a cost of 6.7% of GNI per percentage of annual population growth rate. Population growth rates and percentage under 15 and over 65 are from Population Reference Bureau (2011). The burden of capital widening (or, conversely, the infrastructure dividend) is not the only impact of growth rate, but all tend to interact in a reinforcing way. It is widely accepted that

• versupply of labour puts downward pressure on wages and shifts the distribution of the

gains of economic activity from labour to capital, widening inequality of income. Furthermore, the inflation of land values is driven largely by population growth rate. While competition for land may be seen to be related to population density, it is the expected

• ngoing increase in demand which attracts speculative investment in real estate. This drains

investment away from productive projects, while increasing the cost of housing (Bezemer and Hudson, 2016). The wealth-circulating powers of a market economy are truncated as an ever-greater proportion of wage income is diverted to housing, through mortgages or rents, and less to consumption which generates employment. In developed countries, the constraint

• n consumption demand has been mitigated by expanding access to credit, mortgaged against

inflated land values. Capital widening simultaneously strains fiscal budgets, promoting austerity measures and raising service charges, which further reduce poorer households’ purchasing power and access to services. Thus population growth rate drives widening inequality and deepening debt. Capital widening is also resource-intensive, meaning less of this expenditure is on labour income and more on non-renewable resources, environmental impact and its mitigation. Consequently, population growth also increases environmental impact per capita, independent of the behaviours of the incumbent population, simply because the impacts of the ‘not-yet-added’ are being attributed to them.

10 Most of these changes are poorly monitored by governments. Aggregate GDP statistics are generally stimulated by population growth, particularly through the creation of new debt. Many economists are inattentive to the disparity between headline GDP growth and per capita net national income, let alone trends in inequality and personal debt. Exceptions to these dynamics may prevail when population growth is outpaced by expanding access to natural resources (particularly land and energy), such as prevailed in 19th century colonial economies, and the oil era between WWII and the OPEC crisis of 1973. They are

• ften cited as proof that population growth is economically benign, although it is difficult to

demonstrate that these one-off resource booms would not have been even more enriching without population growth. It is more convincing to examine the patterns we see among countries as they vary in population dynamics, as the next section attempts.

Evidence for the macroeconomic impact of high fertility and its reduction

Rapid population growth is a product of the delay between reducing mortality rates (particularly of children) and reducing fertility. Although child mortality remains unacceptably high in many countries, it is far lower than pre-transition levels. Fertility rate is consequently the main determinant of population growth rate in most underdeveloped countries. Kohler (2012) reviewed literature on drivers of fertility decline, and on the efficacy of family planning in reducing fertility. Pritchett (1994) and others have claimed that family planning programs have little effect because desired fertility, not contraception access, dominate family size outcomes. This underestimates the role of family planning programs in reducing desired family size, particularly where this has been an explicit goal for the program, as was generally the case in Asia and much less so in Latin America. While respecting individual freedom to choose, demand generation may be direct through marketing benefits of delayed, spaced and limited child-bearing, providing role models in popular media, and recruiting community leaders, including religious leaders, in promoting acceptability and advisability of family planning. In addition, a ‘social multiplier effect’ occurs by stimulating exchange between neighbours (Kohler, 2012), a powerful means of overcoming distrust of contraception and stigma of low fertility. Kohler (2012) further attempts to evaluate the cost-benefit ratio of family planning, including such impacts as maternal and infant mortality, women’s labour force participation and household incomes. He concludes the return on investment might lie in the range of 90:1 to 150:1, excluding the benefits of reduced inequality, environmental sustainability and political stability, which are acknowledged to be significant but not easily quantified. Most socioeconomic projections continue to assume that economic and educational advances drive fertility decline, without reference to the presence or absence of family planning efforts (eg. KC and Lutz, 2014). An analysis of country level data found that fertility decline was not dependent on levels of wealth or education, but was very responsive to voluntary family planning programs intended to reduce fertility (O’Sullivan 2013b). Economic advance was found to gather pace only after fertility had fallen substantially. Garenne (2017) has similarly documented the relationship between family planning effort and fertility decline in African

11 countries, and the lack of support in longitudinal data for a dominant effect of income or education levels. He notes, “economic development is not the main driver of success in this field: political will matters more.” On the basis of a few studies in which family planning interventions could be analysed as a quasi-controlled experiment (most notably the cases of Matlab in Bangladesh and Navrongo in Ghana), it is often claimed that family planning causes a quantum of fertility decline, of 1 – 1.5 units (Canning et al. 2015). An alternative, more robust experiment is presented by the diverse population programs and fertility transitions experienced by the community of nations over the past sixty years. From the national-level data, it is evident that family planning interventions act on the rate rather than the quantum of fertility decline. The quantum thus depends on how long the program initiatives persist. Family planning programs have driven rapid decline to below replacement rate in countries such as South Korea, Thailand and Iran. However, where programs were neglected before reaching replacement level, such as in Indonesia, Bangladesh and Algeria, fertility has stalled or rebounded. To reap the economic advantages of population stabilisation, the importance of maintaining programs to well below replacement rate should be stressed. Figure 3 provides an example contrasting two countries in the same region, which implemented very different population policies. The timing of change in GDP per capita is compared with that of total fertility rate (TFR, the average number of children born to each woman over her lifetime). Similar patterns were found by comparing countries with high and low provision of family planning, in various world regions. The initiation of rapid fertility decline in Thailand (Figure 3A) corresponds with the start of its voluntary family planning programs, not with an economic or educational trigger. Only some two decades later did its economic performance diverge from that of the Philippines (Figure 3B). The relationship between TFR and GDP per capita (Figure 3C) showed a steeply concave curve. This pattern is repeated for most countries, with TFR falling to between 3 and 2 before enrichment

• accelerated. Indeed, rapid transition and slow transition countries followed similar paths: it is

rare to see sustained economic advance while fertility remains high. This suggests that the pace of economic advance has largely depended on the pace and extent of fertility decline. Figure 3. Fertility and wealth time courses for Thailand, a strong family planning adopter, and the Philippines, a weak adopter. Left: the change in total fertility rate (TFR, average children per woman)

• ver 5-year intervals from 1950 to 2015; middle: the average GDP per capita over 5-year intervals from

1960 to 2015 (adjusted to constant year 2005 US$); and right: the relationship between TFR and GDP per capita. From O’Sullivan (2013b).

12 To test how representative such examples are, data from all countries were aggregated, grouped according to their maximum rate of fertility decline over any twenty-year period. The results are summarised in Figure 4. Group 1 contained countries which had TFR above 5 in 1950, and where TFR fell after a particular date, with a peak two-decade rate of decline exceeding 1.5 units per decade. Each of these countries adopted voluntary family planning measures around the time that the birth rate began to fall, although not all maintained them until reaching replacement level fertility. Group 2 also showed considerable decline in TFR after a given date, but at a slower rate, between 0.5 and 1.5 units per decade. Group 3 showed no distinct start date for fertility decline, and most still have fertility rates above 3. Group 4 (not plotted) were considered ‘advanced transition’ countries, which had fertility rates below 5 and falling at the start of the data series in 1950-55. Countries where either immigration or emigration contributed more than 15% to population change are excluded from plots. Before averaging each group’s data, they have been synchronised with respect to the timing of their fertility transition by designating Year 0 to be the start of the transition, or 1970 where no distinct change in fertility path is observed (Group 3), and both national population and inflation-adjusted GDP per capita are expressed as per cent of that in Year 0. Averages are not population-weighted. The sixteen countries included in Group 1 were Algeria, Bangladesh, Bhutan, Cambodia, Chile, China, Costa Rica, Iran, South Korea, Libya, Maldives, Mongolia, Oman, Thailand, Tunisia, Viet Nam. Group 1 countries rejected due to high migration were Hong Kong, Macao, Singapore, Aruba, Kuwait and Saudi Arabia (high immigration), and Mauritius and Guyana (high emigration). Group 2 contained 39 included countries and 48 high-migration countries, group 3 contained 26 included countries and 19 high-emigration countries. Only Group 1 have achieved a tapering of their population growth (Figure 4B). Most of these will achieve a peak population around 2 – 2.5 times the population when they started addressing family planning. Group 2 countries have lessened population growth, but have not reduced family size as fast as the number of families has increased. Hence most are still adding more people each year than ever before. Group 3 have seen population triple in the same time. Due to their high proportion of young people yet to start families, they have another doubling in store, if they choose to embrace family planning now (and more if they do not). The impact of fertility decline on wealth can be seen dramatically in Figure 4C. Rapid fertility decline has been associated with dramatic economic improvement. Slow-transition countries have seen virtually none. Figure 4D plots the fertility rate as a function of wealth. It contains two features which are at odds with the popular belief that development drives fertility decline:

• 1. The relationship between fertility and wealth is steeply concave, as fertility fell first

before economic development accelerated.

• 2. All three groups have followed the same path. Those which accelerated fertility

decline progressed more rapidly to economic development. With rare exceptions, “development first” has not been an achievable option. The change shown on the graph is relative to the wealth at the start of programs, so doesn’t depict initial differences in wealth. At time zero, Group 1 was approximately 20% poorer, on average, than Group 2, but twice as rich as Group 3. Therefore, there is a possibility that poverty presented a barrier to family planning achievement in Group 3 countries. However, both Group 1 and Group 2 contained several countries below the average wealth of Group 3

13 at the start of their programs. The wealthier Group 3 countries have shown no more propensity for economic development than the poorest. Figure 4. The averaged time-course for (A) fertility, (B) population and (C) GDP per capita (inflation-adjusted US$), and (D) the relationship between TFR and per capita GDP, for developing countries grouped according to the rate of their fertility transition. Each of the rapid transition countries (Group 1) deployed successful family planning programs. Many countries in Group 2 had programs that were weaker or not sustained. Group 3 countries generally did not have widespread family planning efforts. Year 0 is the start of the fertility transition in each country, or 1970 for weak adopters (Group 3). Population and fertility data from UNDESA (2013), economic data from World Bank economic database. The direction of causation was further investigated, by comparing the level of wealth per capita with the change in TFR over the subsequent five years, and conversely the change in GDP per capita as a function of the initial level of TFR, using data from all countries, for all five year intervals covered by the World Bank’s databank, which span from 1960 to 2010 (Figure 5). Although it is commonly asserted that fertility decline follows and depends on economic development, no such relationship was found (Figure 5A). Very poor countries were as likely as middle-income countries to achieve rapid fertility decline, if they chose to prioritise it. In contrast, the impact of high fertility on economic advance was profound. While fertility remains above four children per woman, the chance of sustained economic improvement has proven to be extremely low. It improves steadily as fertility falls to below the ‘replacement rate’ of 2.1 children per woman (Figure 5B). Only a few oil-rich states showed brief periods of rapid enrichment while fertility remained high. While low fertility has not guaranteed enrichment in any five year period, over 20 year intervals all low fertility

14 countries made considerable gains in wealth, including those with shrinking populations. Fertility decline appears to be a necessary, if not sufficient, precondition for economic development. Figure 5. A. The relationship between the wealth of nations and their rate of change in total fertility rate (TFR, births per woman), and B. the relationship between the level of TFR and the change in GDP per capita over the subsequent 5-year period. Each country is represented by multiple data points, one for each 5-year interval with available data between 1960 and 2010. Boxes span the 25th, median and 75th percentile; whiskers extend to 10th and 90th percentile. TFR data are from the United Nations World Population Prospects (2015 Revision); real GDP per capita data, expressed in constant-value 2005$US, are from the World Bank databank. These results do not suggest that ending poverty does not influence family size, but that reducing poverty as a population control strategy simply has not proven possible for most high-fertility countries. Family planning programs, on the other hand, have proven achievable even in the poorest settings, and the economic benefit to families has translated into societal enrichment. It must be stressed that the successful voluntary family planning programs of the past did not rely solely on ensuring access to contraception. Large desired family size remains the main determinant of high fertility (Population Media Center, 2017). Even among those women who do not want to become pregnant, social and spousal pressure, and misconceptions about side-effects, are more commonly cited reasons for not using modern contraception than lack

• f access and affordability (Sedgh et al., 2016). The most successful programs promoted the

benefits of fewer, more widely spaced children, employed culturally appropriate means to change social norms around family size and women’s roles, and addressed the many barriers to achieving fertility regulation (Campbell et al., 2009).

15

Reconciling empirical data with simulation modelling

The data presented above are simple presentations of widely available data sets, and are deliberately as inclusive as possible. Yet they reveal a strength and consistency of relationship between population growth and economic development that has been elusive to the past three decades of development literature. This disconnect requires explanation. Detailed treatment is beyond the scope of this paper. Here some observations are offered on simulation models. Without adequately including the cost of capital widening, economic models have failed to anticipate the drag caused by population growth. The study by Ashraf et al. (2013) is among the few arguing a clear benefit from fertility reduction, but their claim is likely to be overly

• modest. They simulated outcomes for Nigeria under the UN’s medium and low fertility

projections (UNDESA 2011) by combining multiple channels of economic impact from population growth. These two projections have fertility falling in parallel, only 0.5 units

• apart. This is an improbably small impact of population policies. The more common

experience has been a sustained divergence of TFR based on differing rates of fertility decline, as Figures 3 and 4 portray. Ashraf et al. (2013) further present the impact of lower fertility as a variation from the baseline ‘medium fertility’ scenario, not the change from present conditions. This fails to consider that the baseline scenario may have no per capita economic growth, or even negative growth. The data presented in Figure 5 show that this has been the rule, rather than the exception, for all countries as long as fertility remains high. More broadly, multifactorial models assume independence of factors and consistency of causal functions across the range modelled, often calibrating from sparse correlations of limited relevance and range. Economists might learn much from ecology, where relationships are recognised to be conditional and range-sensitive. A better basis for simulation might be derived from the ‘law of the minimum’ popularised by Liebig (1840) and originally applied to the mineral nutrition of plants, which states that productivity will be limited by the most limiting factor, regardless of the supply of other factors. There are, of course, various interactions between factors, which may be partially substitutable, complimentary or antagonistic, and an adaptive system may redistribute effort to equalise the limitation by several factors (Tilman, 1980). But the principle remains: no matter the marginal productivity

• f a unit of labour or an investment in training when it is limiting, it will have none when in
• surplus. Nor will human needs and wants translate into effective demand for services and

manufactures, if the natural resource base offers no surplus above subsistence, or where such a surplus is absorbed in rents and debt-servicing. The failure of neoclassical economics to differentiate adequately between natural resources and financial assets, or between the activities of production/consumption and the rentier economy of finance, insurance and real estate (FIRE) sectors (Bezemer and Hudson, 2016), corrupts both its metrics and its models. These models do not accommodate unemployment, debt or monopoly rent-taking. Simulations may be set up for failure by flawed assumptions. Examples include that capital investment always generates productive capacity rather than speculative acquisitions bidding up the price of existing assets, and that investment is determined by a national savings rate which is regarded as a constant feature of the economy regardless of the relative need for capital widening under different scenarios (Akraf et al., 2013). That ‘land’ (fixed factors of production) has a constant contribution to total factor productivity (Akraf et al., 2013) or that it can be omitted entirely from production functions (Docquier et al., 2010), such that its crowding will not constrain opportunities for labour or

16 capital, defies every geographer’s sense of the wealth of nations. Assuming that labour will always be fully employed at its level of skill, and that productivity gains will translate into higher wages rather than profit-taking, lead to predictably rosy prognoses for immigration- driven population growth (Docquier et al., 2010) that don’t tally with common realities of engineers driving taxis and graduates’ lifetime earnings and consumption trimmed by years

• f unpaid internships trying to get entry into their oversupplied sector. While the principle of

integrating micro-level responses into macro-level simulations has merit, it requires a greater understanding of feedbacks and boundary conditions than is typically demonstrated in economic simulation models.

Conclusion

That nations struggle to fund the infrastructure needed to advance their economies is a truism. However, the size of this burden, in relation to population growth rate, has remained inadequately measured and largely ignored in the vast literature speculating and modelling economic impacts of population growth. One means for its measurement is described in this paper, which reveals population growth to be a formidable economic millstone. Voluntary family planning programs were regarded as vital components of poverty eradication in the 1970s. Bongaarts (2016a) attributed their loss of priority to a shift in views among economists, who argued either that population growth was not relevant to economic development, or that family planning interventions were ineffective in influencing population growth, or both. Consequently, family planning was reframed only as an issue of women’s health and rights, competing among many pressing needs in health agenda. Political and financial support faded, so that the shift has ironically been to the great detriment of women’s health and rights. As Bongaarts (2016a) notes, almost no recent modelling of future socioeconomic and environmental outcomes examine alternative population trajectories, reflecting the prevailing view that future population growth is not amenable to policy interventions. Hence the wide- reaching benefits of a lower population path have remained relatively unstudied. In the world of socio-economic policy, messaging speaks louder than data. The message of ‘social capital’, compounded by increasing population density, was a seductive argument for the cornucopian optimism championed by Simon (1981) and sustained during the 1980s-90s. The message of ‘sexual and reproductive health and rights’ adopted after the 1994 International Conference on Population and Development, was supposed to incentivise universal access to contraception without the need for economic or demographic motives but has failed to sustain political will for family planning services. Oblique references to “how population dynamics affect the major development challenges of the 21st century” (UNFPA et al. 2013) dog-whistle to those who harbour concern about population growth that it is still

• n the agenda, but remain obscure to those whose political will is being sought.

The message of the demographic dividend is that reducing fertility may offer a window of improved economic performance, if certain conditions are met. The impression given is that this benefit is optional, conditional and ephemeral. The message of the infrastructure dividend is that lowering fertility is an absolute prerequisite for sustained economic advance. To delay fertility decline is to delay development.

17 It is concluded that the most rapid fertility transitions were not driven by economic advance but by voluntary family planning programs, and that these transitions generated enormous economic benefits for adopting countries. The ‘infrastructure dividend’ explains this economic stimulus to an extent that changes in age structure do not. The narrative of demographic dividend has been useful in motivating family planning effort and investment in youth, but its misrepresentation as purely a function of age dependency ratios risks distracting from the necessity to bring fertility to below replacement level (O’Sullivan and Martin 2016). Greater appreciation of the economic impost of population growth rate will allow better targeted interventions.

References

Ashraf Q.H., Weil D.N. and Wilde J. (2013) The effect of fertility reduction on economic growth. Population and Development Review 39(1): 97–130. Betts K. (2014) The ageing of the Australian population: triumph or disaster? A report prepared for the Monash Centre for Population and Urban Research. https://theaustralianpopulationresearchinstitute.files.wordpress.com/2015/01/betts-ageing-of-the- austn-popn-2014.pdf Bezemer D. and Hudson M. (2016) Finance is not the economy: Reviving the conceptual distinction. Journal of Economic Issues 1(3): 745–768. Bloom D.E. and Canning D. (2008) Global Demographic Change: dimensions and economic

• significance. Population and Development Review, Vol. 34, Population Aging, Human Capital

Accumulation, and Productivity Growth (2008), pp. 17–51. http://www.jstor.org/stable/25434758 Bloom D.E. and Williamson J.G. (1998) Demographic transition and economic miracles in emerging

• Asia. World Bank Economic Review 12(3): 419–455.

Bongaarts J. (2016a) Slow down population growth. Nature 530: 409–412. Bongaarts J. (2016b) Family Planning Is an Investment in Development. New Deeply 14 July 2016 https://www.newsdeeply.com/womenandgirls/family-planning-is-an-investment-in-development/ Campbell M, Sahin-Hodoglugil NN and Potts M (2009) Barriers to Fertility Regulation: A Review of the Literature. Studies in Family Planning 37(2):87–98. Canning D., Raja S. and Yazbeck A.S., eds. (2015) Africa’s Demographic Transition: Dividend or Disaster? Africa Development Forum series. Washington, DC: World Bank. doi:10.1596/978-1- 4648-0489-2. Docquier F., Özden C. and Peri G. (2010) The wage effects of immigration and emigration. National Bureau of Economic Research, Cambridge MA, 44 pp. http://www.nber.org/papers/w16646 Garenne M (2016) Demographic dividend in Africa: macro- and microeconomic effects. N-IUSSP December 5, 2016. http://www.niussp.org/2016/12/05/demographic-dividend-africa-macro- micro-economic-effectsdividende-demographique-en-afrique-effets-economiques-macro-et- micro/ Jerven M. (2013) Poor Numbers: How We Are Misled by African Development Statistics and What to Do about It. Cornell University Press, Ithaca and London. KC S and Lutz W (2014) The human core of the shared socioeconomic pathways: Population scenarios by age, sex and level of education for all countries to 2100. Global Environmental Change doi:10.1016/j.gloenvcha.2014.06.004 Kohler H-P. (2012) Copenhagen Consensus 2012: Challenge Paper on “Population Growth”. PSC Working Paper Series. 34. http://repository.upenn.edu/psc_working_papers/34 Lee R.D. (2009) New perspectives on population growth and economic development. Working Paper, University of California at Berkeley, Center on the Economics and Demography of Aging, http://www.ceda.berkeley.edu/Publications/pdfs/rlee/UNFPANewPerspectives09.pdf Liebig J. (1840) Organic chemistry in its applications to agriculture and physiology. English translation by Lyon Playfair. London: Printed for Taylor and Walton. 414 pp. Available at https://archive.org/details/organicchemistry00liebrich

18

Malthus TR (1798) An essay on the principle of population, as it affects the future improvement of

• society. With remarks on the speculations of Mr. Godwin, M. Condorcet, and other writers.

London: J. Johnson. O’Sullivan J.N. (2012) The burden of durable asset acquisition in growing populations. Economic Affairs 32(1): 31-37. http://onlinelibrary.wiley.com/doi/10.1111/j.1468- 0270.2011.02125.x/abstract;jsessionid=9079E4E881757354969065CEA605CD52.d04t04 O’Sullivan J.N. (2013a) The cost of population growth in the UK. Population Matters, UK. https://www.populationmatters.org/documents/cost_population_growth.pdf O’Sullivan J.N. (2013b) Revisiting demographic transition: correlation and causation in the rate of development and fertility decline. 27th IUSSP International Population Conference, 26–31 August 2013, Busan Korea. http://www.iussp.org/en/event/17/programme/paper/4775 O’Sullivan J.N. (2014a) Submission to the Productivity Commission Inquiry into Infrastructure provision and funding in Australia. http://www.pc.gov.au/__data/assets/pdf_file/0004/135517/subdr156-infrastructure.pdf O’Sullivan J.N. (2014b) Ageing paranoia, its fictional basis and all too real costs. In Sustainable Futures: Linking Population, Resources and the Environment. (Eds J Goldie and K Betts). CSIRO Publishing, Melbourne. ISBN: 9781486301898 pp 47–60. O’Sullivan J. and Martin R. (2016) The risk of misrepresenting the demographic dividend. N-IUSSP April 18, 2016. http://www.niussp.org/2016/04/18/the-risk-of-misrepresenting-the-demographic- dividendle-risque-dune-interpretation-erronee-du-dividende-demographique/ Population Media Center (2017) Overcoming bias against contraception in the world’s developing countries: The case for social norm interventions. PMC, Washington DC. http://3lh07xhlls346a7ggo9psa1d-wpengine.netdna-ssl.com/wp-content/uploads/2017/08/PMC- Unmet-Need.pdf Population Reference Bureau (2011 – 2015) World Population Data Sheet. http://www.prb.org/Publications/Datasheets/ Pritchett L.H. (1994). Desired fertility and the impact of population policy. Population and Development Review 20(1): 1–55. Productivity Commission (2016) Migrant Intake into Australia, Inquiry Report. Commonwealth of

• Australia. 713 pp.

Robinson A. (1974) The economic development of Malthusia. Modern Asian Studies 8(4): 521-534. Rowthorn R. (2015) The Costs and Benefits of Large-scale Immigration: Exploring the economic and demographic consequences for the UK. Civitas, London. 100 pp. Sauvy A. (1958) De Malthus à Mao Tsé-Toung. Éditions Denoël, Paris. Accessed in English translation: “Fertility and survival: population problems from Malthus to Mao Tse-tung.” (Translation by Christine Brooke-Rose, 1961) Chatto & Windus, London. 252p. Sedgh G, Ashford LS and Hussain R (2016) Unmet Need for Contraception in Developing Countries: Examining Women’s Reasons for Not Using a Method. Guttmacher Institute, New York. 93pp. https://www.guttmacher.org/report/unmet-need-for-contraception-in-developing-countries Simon J.L. (1981) The Ultimate Resource. Princeton University Press. Solow R.M. (1956) A contribution to the theory of economic growth. Quarterly Journal of Economics 70(1): 65-94.

• D. Tilman, (1980), Resources: a graphical-mechanistic approach to competition and predation,

American Naturalist 116 (3), 362–393. UNDESA (2015) World Population Prospects, the 2015 revision. Population Division, United Nations Department of Economic and Social Affairs. http://esa.un.org/unpd/wpp/ UNFPA, UNDESA, UN-HABITAT, IOM (2013) Population and sustainable development in the Post-2015 agenda. Report of the Global Thematic Consultation on Population Dynamics. http://www.unfpa.org/publications/population-and-sustainable-development-post-2015-agenda Wang F. and Mason A. (2007) Population Aging in China: Challenges, Opportunities and Institutions. In: Zhao Z and Guo F. (Eds.), Transition and Challenge: China’s Population at the Beginning of the 21st Century. Oxford University Press, pp. 177–196.