Demographics and long-run growth

Thomas Cooley, Espen Henriksen 11 June 2018

a

A

The recovery from the Great Recession in both the US and Europe was remarkably slow until just recently. There have been two popular accounts for why current and expected future growth rates are low. One view, articulated by Gordon (2016), holds that productivity growth may be impaired because the opportunities for technological innovations in the future are less than they have been in the past. An alternative view, associated with former Treasury Secretary Larry Summers, holds that slow growth is a consequence of the fall in the market-clearing real interest rate and the inability of monetary policy authorities to cut nominal interest rates to ‘stimulate’ investment demand (e.g. Summers 2014, 2016). 

The common element in these accounts is that the growth slowdown will be persistent. We do have other examples of persistent stagnation in previously fast-growing economies, the most notable being Japan. Japan has been stagnant since the early 1990s in spite of aggressive monetary and fiscal stimulus. Similarly, in both Europe and the US growth rates were below the trend in the period leading up to the financial crisis. After the crisis, the growth rates have been below that in previous recoveries from recessions, despite aggressive monetary and fiscal stimulus.

The challenge is to identify factors that can account for the persistence of the slowdown. One low-frequency factor is the demographic structure of an economy. It has been suggested that demographic changes may have implications for economic growth, but the channels through which these changes work have been less well understood. Previous contributions, such as Favero and Galasso (2015), have shed important light on the potential growth consequences of changing demographics. In a recent paper, we make them more precise in the context of a life cycle model with rich demographics (Cooley and Henriksen 2018). 

There is increasing recognition that demographic changes are an important driver of persistent current account imbalances, capital flows, and decline in real interest rates. Backus et al. (2014) showed the effect of demographic changes on international capital flows and interest rates in the US and Japan. Recently, Carvalho et al. (2016), and Ikeda and Saito (2014) have shown the impact of demographic changes on the real interest rate and investment in the US and Japan. All these papers, and others, have focused on how demographic change affects the supply of and demand for capital. 

Growth accounting

Standard growth accounting shows that aggregate growth rates may be decomposed into: 

  • total factor productivity (TFP); 
  • increases in capital supply relative to labour supply;
  • labour supply on the extensive margin, that is, whether people are in the labour force;
  • labour supply on the intensive margin, that is, how many hours those who are in the labour force on average work; and
  • population growth.

All of these elements may partly be accounted for by changing demographics. Population growth is, trivially, part of demographics. Changes in labour supply, both on the intensive and extensive margin, can be due in part to changing demographics. Across countries, labour supply, both on the intensive and extensive margin, shows a hump-shaped pattern with respect to age. Individual decisions on whether to stay in the labour force and how many hours to work may also depend on remaining life expectancy. Savings, and changes in the capital-to-labour ratio through it, may be effected by individuals life-cycle savings – as explored in the aforementioned papers. And lastly, and maybe most subtly, demographic change may have a sizable impact on measured productivity growth. 

Growth accounting across the G7 countries shows that when comparing those countries that have aged fastest (such as Japan) with those that have aged slower (such as the US), the fastest-ageing tend to have been slower growing, to have a positive growth contribution from higher capital accumulation, and to have negative growth contributions from TFP and from labour supply on the intensive and extensive margins.

Policy analysis and making projections: Functional forms and elasticities

To account for secular trends, such as long-run growth rates, we ultimately rely on models. Models are also useful for making predictions of future growth rates based on demographers’ projections for fertility, migration, and mortality rates. Finally, in order to evaluate alternative policies, which may mitigate the effects of demographic trends, we need models. 

A key element in such a model are individuals who, conditional on their life expectancies, make savings-consumption choices and labour-leisure choices both on the extensive and intensive margin. One way to account for the fact that people tend to gradually leave the workforce is that the cost or disutility of working increases with age. In order to distinguish between labour supply on the extensive margin and labour supply on the intensive margin, the model needs to incorporate idiosyncratic productivity shocks. 

A model’s ability to account for historical growth rates and to predict future growth rates, and its suitability for policy analysis will, ultimately, rely on the calibration of functional forms and elasticity parameters. Key functional-form questions are how individuals trade off consumption and leisure over time, and whether disutility of labour is a function of age or expected remaining years of life. The latter is particularly important given projected gains to life expectancy. Key elasticity parameters are those governing intertemporal consumption choices, labour-leisure choices, and the disutility of labour as an individual ages.

Cohort distributions, life expectancies, and general equilibrium

Demographic change has many facets. The discipline of general-equilibrium models provides a framework to identify those aspects of demographic change that affect economic growth. Those are changes in life expectancy and changes in the age-cohort distribution of the population. Changes in life expectancy impact individuals’ decisions, including their life-cycle savings decisions and their labour-supply decisions. Changes in cohort distributions affect the aggregation of these individual decisions. We distinguish analytically between these two channels of demographic change.

The evolution of cohort distributions is a function of three variables: past and present fertility, immigration, and conditional mortality rates. Making demographic projections, such as expected future cohort distributions, is best left to demographers. In our research, we study how projected future demographics affect individual decisions and how these decisions are aggregated. 

To account for individual decisions, the general equilibrium effects of demographic change are also crucial. The wage rate and the real rate of interest enter into individuals’ budget constraints. These aggregate factor prices change as a result of changes in the relative supply and demand for labour and capital. The factor-price changes may magnify or dampen individual life-cycle decisions.

One crucial modelling question is whether individuals perfectly foresee the entire future factor-price path, and how it may be affected by changing demographics. A generally poorly understood nuance of modelling is that the difference between solving a demographic transition as a sequence of steady states or solving it as one transition, essentially amounts to whether individuals base their current choices on the reasonable belief that current factor prices will prevail, or whether they perfectly foresee the future path predicted by demographics. We find the former to be the most parsimonious assumption.

Reported TFP: Past demographic dividends

It is well known and understood that demographic change has given most OECD countries a ‘demographic fiscal dividend’ in the post-WWII era. During these decades, the ratio of working age individuals to individuals outside working age increased gradually, allowing a steady increase in government programs, benefiting both young and old. In most OECD countries, this ratio has either changed or is about to change.

A more subtle result from our structural analysis is that demographic change also may have given the same countries a reported TFP-growth dividend during the same time period. For each individual, the productivity profile over the life cycle follows a hump shape. Individual productivity increases steadily until the late 40s before it plateaus or gradually declines.

A consequence is that as the average age of the labour force has increased from the mid-30s to the mid-40s, measured average TFP has also increased. Part of the past measured TFP growth has probably simply been due to the fact that the average age has increased, and the average worker is more experienced. Going forward, there will probably be a demographic deficit with respect to measured TFP, just as it was a demographic dividend in the past.

Results from projections

We calibrate a general equilibrium model with the features described above for Japan and the US. We then simulate it between 1990 and 2007 with the evolving conditional life expectancy over this period, and aggregate individual decisions using the evolving cohort structure. The results show that changing demographics may account for about one sixth of the level of economic growth in this period.

The results from feeding projected demographic trends through the same model suggest that the economy will face headwind going forward. There will be no further gains to measured TFP and higher capital accumulation will not outweigh negative growth contributions from lower labour supply, both on the intensive and extensive margins.

Policy implications

Demographic factors cannot fully account for the slowdown of long-run growth, but they likely represent an important contributing factor. Economic policies that strive to mitigate the effects of ageing on long-run growth will have to target households’ incentives to supply capital and labour over their lifecycles; in particular late-working-life labour supply. Even more subtle policies may strive to affect households’ productivity as they age.

Our analysis highlights that effective policies are those that affect consumption-saving choices, labour-leisure choices, and human capital accumulation over the lifecycle. The welfare implications of different policies will rely crucially on how they affect individuals’ tradeoffs between labour and leisure at different stages of life, and, in particular, how they may change as life expectancy changes.

Concluding remarks

Demographic change is persistent and predictable. Across the most advanced economies, average age will increase and life expectancy will almost surely increase. This may have important consequences for long-run economic growth.

We provide a parsimonious analytical framework to quantitatively estimate the impact of changing demographics on measured long-run growth. The analysis shows that the ageing of populations may have large impacts on long-run growth. 

Our paper identifies key margins both for modelling and policy analysis. The quantitative magnitude depends on the functional forms and the calibration of key elasticity parameters, in particular with respect to old-age productivity and the evolution of disutility of labour as life expectancies increase. Incentives for late-life labour supply may stimulate economic growth, but should be carefully weighed against the disutility of working when old.

References

Aksoy, Y, H Basso, T Grasl and R Smith (2015), “Demographic structure and the macroeconomy”, VoxEU.org, 8 April. 

Backus, D K, T F Cooley and E Henriksen (2014), “Demography and low-frequency capital flows”, Journal of International Economics 92: 94–102.

Carvalho, C, A Ferrero and F Nechio (2016), “Demographics and real interest rates: Inspecting the mechanism”, European Economic Review 88: 208–226.

Cooley, T F and E Henriksen (2018), “The demographic deficit”, Journal of Monetary Economics 93: 45–62.

Favero, C A and V Galasso (2015), “Demographics and the secular stagnation in Europe”, CEPR Discussion Paper 10887. 

Gordon, R J (2016), The Rise and Fall of American Growth, Princeton University Press.

Ikeda, D and M Saito (2014), “The effect of demographic changes on the real interest rate in Japan”, Japan and The World Economy 32: 37–48.

Niepelt, N and M Gonzalez-Eiras (2011), “Does population ageing reduce productivity growth?”, VoxEU.org, 24 June. 

Summers, L H (2014), “U.S. economic prospects: Secular stagnation, hysteresis, and the zero lower bound”, Business Economics 49: 65–73.

Summers, L H (2016), “The age of secular stagnation”, Foreign Affairs 95(2): 2–9.

a

A

Topics:  Labour markets Macroeconomic policy

Tags:  demographics, population ageing, lifecycle, long-run growth rate, demographic change, labour, leisure, consumption, human capital

Professor of Economics, Stern School of Business and Faculty of Arts and Science, New York University

Associate Professor of Financial Economics, BI Norwegian Business School

Events

CEPR Policy Research