Academic journal article Economic Inquiry

An Estimate of the Age Distribution's Effect on Carbon Dioxide Emissions

Academic journal article Economic Inquiry

An Estimate of the Age Distribution's Effect on Carbon Dioxide Emissions

Article excerpt


We exploit temporal and geographic demographic variation to estimate the relationship between a population's age distribution and carbon dioxide (C[O.sub.2]) emissions in a panel of 46 countries from 1990 to 2006. To reduce potential omitted variable bias, we instrument for the age distribution with lagged birth rates and control for total population and total gross domestic product (GDP). A higher share of prime working-age (35-49) individuals within a country leads to higher C[O.sub.2] emissions, while younger and older populations have lower C[O.sub.2] emissions according to our results.

Emissions of C[O.sub.2] account for about half of the radiative forcing from anthropogenic sources that are considered the primary contributors to global warming. The Intergovernmental Panel on Climate Change (2007) predicts increases in average surface temperatures in the range of 1.1 -6.4[degrees]C (2-11,5[degrees]F) by the end of the century. Climate change has created the threat of substantial environmental damage, with the possibility of catastrophic consequences for many throughout the world. As a result, both academics and policy-makers are interested in C[O.sub.2] emissions.

The literature aimed at explaining C[O.sub.2] emissions usually focuses on the size and affluence of the population. This approach can be traced back to the seminal work by Ehrlich and Holdren (1971), the biologist and physicist who argued that population size has a disproportionate impact on the environment. The impact is generally assumed to depend upon technology, leading to the well-known IPAT identity (or Kaya identity for C[O.sub.2] emissions), I = P x A x T, where I is environmental impact, P is population size, A is affluence (GDP per capita), and T is a technology index. More recent literature has focused on models of the technology index and how the index changes over time (e.g., due to improvements in abatement technology). The widely cited Intergovernmental Panel on Climate Change studies use several variations of the IPAT model to produce regional emissions forecasts, which are then aggregated to the global level. Structural studies by economists (e.g., Nordhaus 2009) have also attempted to forecast global C[O.sub.2] emissions using estimates of GDP from production functions and world population size. Although these studies have frequently used geographic density of the population as well as its size as explanatory variables, and have allowed for nonlinearity in size, most have ignored the age distribution within the population. (1)

A few papers have begun to address this potential oversight by explicitly including the age distribution in studies of C[O.sub.2] emissions. Dalton et al. (2008) build a structural overlapping generations model to forecast U.S. C[O.sub.2] emissions. (2) In their model simulations, population aging has a big effect on future emissions. Our empirical findings support Dalton et al. (2008), but our approach is most closely related to Cole and Neumayer (2004). (3) Cole and Neumayer (2004) also use a cross-country panel to estimate the effect of demographic change on C[O.sub.2] emissions in an IPAT type regression model. While Cole and Neumayer (2004) include age group shares as controls, the groups (younger than 15, 15-64, and 65+) are too wide to capture the effect we posit, and they do not instrument for the age distribution as we do. Cole and Neumayer (2004) mainly focus on changes in total population, whereas we are interested in the age distribution effect for a given population size.

Our ordinary least square (OLS) estimates indicate that a country's level of C[O.sub.2] emissions depends on its age distribution. In addition to population size, the regressions include total GDP and a full set of time and country dummies as controls. We use the fraction of the population aged 35-49, or working share, as the main explanatory variable. Changes in the working share across time, not common to all countries, and independent of GDP and total population provide the variation used to estimate the age distribution's effect. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.