Energy and Economic Growth: The Stylized Facts.

• Author: Csereklyei, Zsuzsanna

1. INTRODUCTION

   Kaldor (1957, 1961) highlighted six "stylized'' facts that summarized the patterns that economists had discovered in national income accounts with a view to shaping the growth models being developed to explain them. Recently, Jones and Romer (2010) introduced a set of "new Kaldor facts" for growth economics. In this paper, we attempt to summarize what we know about energy and economic growth in a similar set of stylized facts with the intention of informing the development of models of energy and economic growth. Though we examine the previous literature, we do not take on faith the facts laid out there. Rather, we carry out a systematic analysis of a global dataset covering the 1971-2010 period and also look at the longer run historical data that are available. This reveals a new set of stylized facts that is sometimes at odds with the received wisdom.

   Stylized facts are empirical regularities that can be seen clearly without using sophisticated econometric techniques (Summers, 1991). Stylized facts are not relations that are true in all countries in all periods of time but are statistical tendencies. Such regularities are not necessarily structural relationships. Rather, they may be the outcomes of complex processes. The stylized facts discussed in this paper should, therefore, not be seen as necessarily describing functional relationshipsbetween the variables in question. Rather, they are historical characteristics of the data that models of energy and economic growth should be able to reproduce.

   Several previous authors have attempted to characterize the stylized facts of energy and growth. Smulders and de Nooij (2003) sought to develop a model of the role of energy in economic growth that was consistent with the main stylized facts concerning energy use and growth. They list four such stylized facts, which they took from Jones (2002). For U.S. data over the period 1950-1998:

1. Energy intensity--the ratio of energy use to GDP--in the U.S. declined at an annual rate of 1.4% on average;

2. Per capita energy use increased at an average annual rate of about 1%;

3. The ratio of energy costs to GDP declined at an average of about 1% per annum, though in the 1970s the energy cost share rose temporarily;

4. The relative price of energy to labor declined. This fact is based on Nordhaus (1992), who shows that the relative price followed a negative trend since at least 1870.

   Smulders and de Nooij (2003) showed that there were similar trends for the first two variables in Japan, France, West Germany, and the UK for the period from 1960 to 1990.

   Kander et al. (2013) list several stylized facts for a set of today's developed countries over the past two centuries, though not all of these actually relate to energy. The energy-related facts are that over time:

1. The energy/capital ratio falls;

2. The energy cost share falls;

3. The real price of energy falls;

4. The quality of the energy mix increases; and that

5. In the 20th century energy intensity fell and converged across countries; and

6. There was a clear trend break in the energy services to GDP ratio in the 1970s.

   Kander et al. (2013) and Smulders and de Nooij (2003), therefore, concur on some of the key features of the data for individual countries, but with the exception of Kander et al.'s comment on the convergence of energy intensity across countries, there is nothing in these stylized facts about how the relationship between energy and income varies across countries and no discussion of energy use in developing countries. Several studies have, however, examined these relationships.

   Zilberfarb and Adams (1981) examined cross-sections of 47 developing countries in 1970, 1974, and 1976; finding that the elasticity of per capita energy with respect to purchasing power parity (PPP) adjusted income per capita was greater than unity. This implies that energy intensity increases with income. However, this data did not include traditional biomass use. (1) Ang (1987) found that, for a cross-section of 100 countries in 1975, energy intensity (including non-commercial energy) rose with PPP adjusted income. The effect was stronger when he excluded non-commercial energy and when non-PPP income was used there was a decline in energy intensity at high income levels. Medlock and Soligo (2001) examined the patterns of the development of energy use by end-use sector--transportation, industry, residential etc.--for a panel of data from 28 countries. (2) They only included commercial energy, omitting traditional fuels (see also Galli, 1998). They concluded that energy intensity follows an inverted-U shaped curve with increasing income. They found that the share of industry in commercial energy use declines over time, that of transportation increases, and the share of residential and commercial use rises and then levels out. Judson et al. (1999), who examined a much larger sample of countries, found that the household sector's share of aggregate energy consumption tends to fall with income, the share of transportation tends to rise, and the share of industry follows an inverse-U pattern. Schafer (2005), who, unlike Judson et al. (1999), included traditional biomass use, found similar results for the residential and industrial sectors and that the share of services in energy use also rises monotonically.

   In a recent study, Lescaroux (2011) looks at commercial energy only and uses market exchange rates and finds that energy intensity declines monotonically with income per capita. Jakob et al. (2012) examine a panel of 30 developing and 21 developed countries. They investigate the effect of income growth (market exchange rates) on total primary energy use (including biomass) as well as individual fuels and end-use categories for developed and developing countries separately. They find that the elasticity of total primary energy use with respect to income is 0.631 for developing countries and -0.181 (but statistically insignificant) for developed countries. (3)

   A mixed picture emerges from these studies of the cross-sectional relationship. Some researchers find that energy intensity increases with income; some find it decreases; and others find that it follows an inverted U. This depends on the way the data are measured, the sample of countries used, and possibly the period of time considered too. There do not appear to be recent cross-sectional studies that both include traditional energy and use PPP adjusted income. Furthermore, studies appear to either investigate the time series behavior of energy and income or the cross-sectional behavior, but do not relate the two and, therefore, the linkages between the cross-section and time series behavior have not been explored.

   The purpose of our paper is to determine what robust global patterns exist between energy use and economic growth in an up-to-date data set in both the cross-section and over time, by linking the cross section results to the time series dimension in a panel that is as large as possible in both the cross-section and time-series dimensions. Our data includes non-commercial energy and reports income in PPP-adjusted terms. Our main analysis uses an annual panel data set for 99 countries over the period from 1971 to 2010. Rather than carry out a standard panel regression analysis, we look separately at the time series and cross-sectional dimensions of the panel and the relationship between them. We also examine some longer-run time series and cross-sections for the U.S., Canada, and several European and Latin American countries to determine whether these relationships appear to hold in the previous century and a half too. We also look at the issue of the cost share of energy for which we only have long-run data for two countries--the United Kingdom and Sweden.

   Based on the literature we have reviewed above, we select the following variables and relationships for investigation: energy use per capita, energy intensity, the energy/capital ratio, energy mix, and the energy cost share. Consistent with previous research, we measure energy use in joules of primary energy use. While it would be interesting to also examine quality-adjusted energy use, construction of such series requires detailed price data and the IEA Database only provides limited coverage of OECD countries. We include non-commercial energy despite greater uncertainty in estimates, because otherwise we would exaggerate the increase in energy use that occurs in the early stages of economic development and give the false impression that the poorest countries have very low energy intensity. We use purchasing power parity adjusted income because we aim to characterize the relationship between energy use and economic activity in each country, which is best compared across countries on a purchasing power parity basis. (4)

   Our main conclusions are, first, that we find a stable relationship between energy use per capita and income per capita over the last four decades. The elasticity of energy with respect to income is less than unity. This implies that energy intensity is negatively correlated with income and that decreases in energy intensity are related to economic growth. Energy intensity does not decline and may even increase in the absence of growth. Thus, energy intensity has declined globally as the world economy has grown. Second, there is unconditional convergence in energy intensity over time, both in the recent period and over the last two centuries. This means that energy intensity tends to increase in countries that have relatively low energy intensity for their income level and that there was much greater variation in energy intensity at each given income level in the 19th Century than today. Third, though we have limited evidence, we find that the cost share of energy declines over time. We also show that the energy capital ratio behaves similarly to energy intensity and that energy quality increases with income and over time.

   The...