Abstract
Climate plays an important function in many aspects of human lives and, not surprisingly, receives considerable attention in several academic disciplines, including economics. The idea that climate is one of the fundamental determinants of economic development is not entirely new and has been vigorously debated for centuries. To this end, there is a large existing literature that is devoted to investigating the impacts of climate on the aggregate economy. The thesis aims to contribute to this literature by exploring the topic using three commonly used methods in economic analysis: cross-sectional analysis, panel analysis and time-series analysis. A shared characteristic in these three analyses is that they are based on reduced-form approaches; the main objective is to provide a neutral ‘big picture’ perspective of the relationship between climate and the aggregate economy without being dictated or constrained by specific (pre-determined) economic models or theories.
The cross-sectional analysis is primarily motivated by the decade-old geography versus institutions debate found in the deep determinants of development literature. The debate arises due to the disagreement between scholars regarding the roles of geography and institutions in influencing the course of long-run economic growth of countries. While some have argued for the direct effect of geography (including climate) on the economy, others have negated it and suggested that institutional factors, such as rule of law and political stability, are more likely to be the genuine drivers of economic progress. Using recently constructed historical and contemporaneous temperature data (1730-2000) for a cross-section of over 160 countries, the analysis employs an instrumental variables (IV) approach to examine whether the temperature impacts on cross-country income distribution reported in the literature are mediated by institutions. The regression estimates provide evidence of a positive effect of institutions on current income per capita, even after controlling for historical and contemporaneous temperatures. Nevertheless, depending on the choice of instrument(s), there are significant historical temperature impacts on income that cannot be explained by institutions. This implies that the effects of climate and institutions on economic growth are not mutually exclusive. The empirical results from the cross-sectional analysis provide, to a certain extent, support for the geography hypothesis.
The panel analysis aims to identify the short and long-run dynamics between climate and the aggregate economy by exploiting the year-to-year fluctuations of temperature and precipitation in a panel of over 120 countries for the period 1960-2005. To improve the reliability of estimates and to avoid spurious results, the analysis relaxes three empirical assumptions that are commonly applied in the panel climate-economy literature – (i) climate is a stationary process; (ii) climate parameters are homogeneous across panels; and (iii) there is cross-sectional independence between panels. Through a combination of qualitative, unit root and residual tests, all of these assumptions are determined to be inappropriate and require reconsideration. From estimates obtained using nonstationary panel estimation, there is evidence to support short-run negative temperature effects on economic growth in poor countries while short-run precipitation effects appear to be trivial. However, results from error-correction-based panel cointegration tests suggest that long-run equivalent climatic effects are tenuous. Evidence of cointegration between income per capita, temperature and precipitation is found in a small subset of countries only. Consequently, it seems that the climate-economy relationship is likely to be dominated by short- to medium-run effects and these effects may not persist in the long run owing to possible adaptations to climate change.
The time-series analysis focuses on a single country, Kenya, and examines whether long-term changes in climate could impact the state of food security. It is a slight departure from the other two analyses in that the outcome variable is a measure of agricultural production rather than overall output of the economy. However, because agriculture forms a substantial part of Kenya’s economy and food production is intimately related to climate, the analysis provides an alternative strategy to investigate the relationship between climate and the aggregate economy. Using a 50-year data set of annual observations of temperature, precipitation and food supply in Kenya, a vector error-correction model (VECM) is estimated to determine if there is a cointegrating relationship among the variables. The results suggest that, over time, deviations in climate are followed by changes in food supply and the variables appear to be converging towards a long-term equilibrium, albeit slowly. This implies that there is a causal element in the relationship and any increase in long-term average temperature or precipitation variability is likely to weaken food security and subsequently the economy in Kenya.