This area of research remains controversial; but the hypothesis has anecdotal evidence to back up what economists and development experts have been alluding to for decades, which is that hot and humid climatic conditions are contributing factors to underdevelopment in the tropics. In a recent research entitled ” Climate and Economic Development: Further Evidence in Support of “The Tropical Effect,” Mariam Khawar provides more evidence to this effect. A short version of her research is presented here.

Mariam Khawar

Abstract

Economists have historically ignored the relationship between geographical factors and economic growth and development. However researchers in other fields, historians and biologists, have provided detailed and plausible explanations of the connection between geography/climate and economic progress. Recently, economists have also begun to examine the existence of this relationship by studying the effects of climate on agricultural and labour productivity, for example. Using both cross-sectional and panel data sets, studies have been conducted on the specific aspects of climate and weather that may influence economic outcomes. This paper adds to that literature by focusing in particular on the effects of climate as it pertains to temperature and rainfall, using ground station data from the Global Historical Climatology Network over a period of 30 years. The study finds empirical evidence suggesting that higher temperatures are negatively associated with the level of GDP per capita of a country. In addition, countries that have larger ranges of temperature extremes also have higher incomes. The relationship between temperature and GDP per capita growth rates turns out to be more complex but again the evidence indicates that temperature matters. Lastly, the paper discusses evidence that points to the importance of rainfall and stresses the need for further verification to pinpoint the relationship.

Introduction

The relationship between geographical factors and economic development has historically been ignored by economists. However, views about the correlation between temperature and climate have been expressed in works dating as far back as Montesquieu (1748) and Huntington (1915).

This paper examines specific features of climate, namely temperature and rainfall, as possible factors that might influence productivity and hence income per capita across countries. Building on previous studies by economists and ecologists who have studied the impact of climate on agricultural productivity and disease burden, this investigation seeks to pinpoint the characteristics of climate that are important in establishing that link.

Earlier works on this topic include Kamarck (1973) and later, economic historian Landes (1998). Researchers in other fields, Crosby (1986) and Diamond (1997), a historian and a biologist respectively, have provided detailed and plausible explanations of the connection between geographical, climatic and economic factors. Their studies have a historical focus extending over several centuries (Crosby (1986) and several millennia (Diamond (1997). Since then economists, beginning with Gallup et al. (1999) have conducted comprehensive cross-country studies to investigate this relationship. They have demonstrated that being a tropical country is negatively related to output per capita both in levels and growth rates. However as Sachs (2001) eloquently stated: “The most notable feature of global economic development – the continuing impoverishment of the tropics – remains to be explained.”

Gallup and Sachs (2000a), controlling for differences in technology, find that total factor productivity is less in tropical than in temperate climate zones. Most explanations of the geographical limitations of agriculture in the tropics focus on problematic soils in humid tropics, and rainfall variability and limited irrigation potential in the arid tropics. Some features of climate that can affect agricultural productivity have been studied along with effects on disease burdens (Gallup et al. (2000b)). The lack of freezing temperatures in the tropics causes a much greater number of agricultural pests. Masters and McMillan (2001) present convincing findings, which point to the presence or absence of frost as a significant factor influencing economic development. Human tropical diseases such as malaria reduce agricultural labor productivity. Additional factors explaining lower agricultural potential in the tropics are pest and disease loads, and net photosynthetic potential differences. Although the tropics are generally warmer and sunnier throughout the year than temperate zones, the climate has disadvantages for photosynthesis.

The humid tropics are often cloudy, blocking sunlight, and the high nighttime temperatures cause high respiration that slows plant growth. While discussing the thermal physiology of organisms, Lafferty (2009) explains how warm temperatures speed up biochemical reactions which require higher food consumption rates. These in turn can decrease survivorship rates. Thus the relationship between an organism’s productivity and temperature should follow a convex function. Panel estimates by Schlenker and Lobell (2010) find that higher temperatures reduce agricultural yields.Using panel data on rice firms in Asia, Welch et al. (2010) find that higher minimum temperature reduces yields but higher maximum temperature increases yields. While studying land invasions in Brazil, Hidalgo et al. (2010) estimate that rainfall deviations lower agricultural incomes. Haile (2005) finds that the rainfall pattern in Sub-Saharan Africa is influenced by large-scale intra-seasonal and inter-annual climate variability. Dell at al. (2014) provide an extensive summary of the literature on studies using panel data to estimate the effect of temperature and precipitation on industrial output. They note that the studies consistently estimate a 2 percent loss of output per 1°C. These studies are consistent with micro-level studies of labour productivity as well (Niemila et al. (2002)).

Another climate related factor potentially affecting productivity is humidity. As temperature and humidity increase, malaria transmission can increase from zero to epidemic rates (Lafferty (2009)).The diversity of infectious diseases of humans is higher in countries near the equator than in countries at higher latitude (Guernier et al. 2004). The diversity of all disease categories increases with the maximum range of precipitation, and most disease categories increase with monthly temperature range. Wolfe et al. (2007) found that infectious diseases of humans were equally likely to have originated in tropical or temperate regions. The early humans that migrated out of Africa and into temperate latitudes initially left several infectious diseases behind: only one of the 10 major tropical diseases, cholera, followed into temperate latitudes. However 11,000 years ago, several infectious diseases of newly domesticated temperate animals jumped to humans and most of these novel infectious diseases subsequently spread into the tropics (Wolfe et al. 2007).

The high diversity of infectious diseases in the tropics could result from a high diversity of vectors, perhaps due to differences in vector diversity. The inability of human tropical diseases to spread from the tropics to temperate regions may be due to the higher fraction of tropical diseases that have a specific vector (80% tropical vs. 13% temperate) and/or a wild animal reservoir (80% tropical vs. 20% temperate; Wolfe et al. 2007). A seminal paper by Acemoglu, Johnson and Robinson (2001) takes a different route and suggests that the quality of institutions plays a more prominent role in comparative development outcomes. They further pinpoint the type of colonization that a country was subjected to as being responsible for institutional quality. In the end though, even this finding traces back to climate because Acemoglu et al. (2001) conclude that the type of colonization was determined by the mortality rates of the colonizers in the conquered countries, which in turn were determined by the disease ecology of those lands.

An important caveat to keep in mind with studies that control for the effects of institutions is one suggested by Dell et al. (2014). The authors point out that if hot climates caused low-quality institutions which in turn lead to low income, then controlling for institutions can have the effect of partially eliminating the explanatory power of climate, even if climate is the underlying cause. Thus claims by researchers as to the supremacy of institutions as the primary determinant of income may be subject to this critique (Rodrik et al. 2004).

Studies measuring aggregate economic activity and climate have also found a link between the two. Nordhaus (2006) introduces data on global economic activity, the G-Econ database, which measures economic activity for all large countries, measured at a 1° latitude by 1° longitude scale. Amongst other results, he finds that the relationship between temperature and output is negative when measured on a per capita basis. Dell et al. (2009), using data for 12 countries in the Americas find a statistically significant negative relationship between income and average temperatures but little or no impact of average precipitation levels. Newer studies using panel data (Dell et al. 2012, Hsiang (2010)) report a negative link between temperature and per capita income but again no effects of precipitation. Barrios et al. (2010) demonstrate that higher rainfall is associated with faster growth in sub-Saharan Africa but not elsewhere. Thus the summary evidence on climate and average income, demonstrates a definite link for temperature but a weaker one for precipitation.

Following that thread, this paper focuses its attention on climate specific variables to try to weed out the effect that ‘the tropical effect’ could have on levels and growth rates of GDP per capita. Instead of a broad category that represents the percentage of land that lies in the tropics as in Gallup et al. (1999), the contribution of this paper is to use data on temperature and rainfall to examine whether or not those particular variables contribute to the ‘tropical effect’ or whether it is other features of the geographical tropics that are more significant.

The next section describes the data used in the study. This is followed by an analysis and interpretation of the results. The last section concludes with some comments on the direction of future research on this topic.

Data

To facilitate a direct comparison between the results in this study and those of Gallup et al. (1999), their original dataset for the economic, social, policy and geographical variables was used. All but the geographical variables are from established, widely available sources. The physical geography and malaria index variables are contributions of Gallup et al. (1999). The variables relating to temperature and rainfall are direct contributions of this paper. The variables were calculated based on a data set compiled by the National Climatic Data Center (NCDC) (1997). The data set contains information on worldwide temperatures and precipitation for at least one location in each country throughout the world, whenever possible. For large countries the stations are selected to provide comprehensive geographical coverage. The data are presented as an annual average calculated over a record length ranging from 3-105 years, averaging about 30 years for most countries. The temperature data consists of values of average daily temperature in January, April, July and October, as well as extreme maximum and extreme minimum temperatures, all in Fahrenheit. The precipitation data consists of average precipitation in each month as well as an annual total, all in inches. Dell et al. (2014) summarize the various data sets commonly used in the climate and economic analyses. Out of the two methods suggested for aggregating the data, spatially or using population weights, this study uses the former. From the raw data set, country averages for total annual rainfall, mean temperature and the difference between the extreme maximum and minimum temperatures were computed.

The main result that this study highlights is that mean temperature is a significant determinant of GDP per capita and is negatively related to output, indicating that warmer temperatures have detrimental effects on output. This is not the same as stating that tropical countries have lower per capita income (the main finding of the Gallup et al (1999) study) since being tropical includes a variety of features pertaining to climate, vegetation, soil etc. At the very least it singles out temperature as an important determinant of the ‘tropical’ characteristic. Moreover, this finding holds while correcting for the effect of institutions, something that has not been found in earlier studies. The reasons for this phenomenon and the channels through which heat can affect economic activity have been discussed earlier in this paper. They include the impact of infectious diseases and hence life expectancy which may influence labor productivity in manufacturing and services. The same factors could also affect crop yields and labor productivity in agriculture, which together could influence agricultural output and productivity.

Also the infrastructure of developing countries is susceptible to many factors and rainfall could easily disrupt many basic utilities such as energy, water and transportation (due to the conditions of roads). However at sufficiently high levels of precipitation the benefits accruing to agriculture may outweigh these factors and result in an overall positive impact on GDP per capita.

This interpretation is certainly open to debate. A better measure of precipitation would probably be the variability in rainfall. As Kamarck (1973)7 points out “Rainfall in the tropics is usually too much or too little. Average annual rainfall means little when one year may receive three times as much rain as the next, or when it does not rain evenly throughout a given season of the year but falls in torrents within brief periods”.

Hence a measure of the variation in rainfall might be a better indicator to test whether or not precipitation is a factor in affecting output.

Conclusion

This study has provided additional evidence to suggest that climate, as defined specifically by temperature and rainfall, may have an important role in determining both the levels of output per capita and how fast a country grows. As suggested by Dell et al. (2014), since climatic and geographic variables are (largely) exogenously determined, reverse causation is unlikely to be of concern with these results. The puzzling nature of the link between climate and economic growth warrants further investigation and should serve as a springboard for more studies on this topic. It is possible that the measurement of some variables such as political and institutional factors might influence the results based on the construction of those variables. To avoid the possible effects of such differences, an alternative approach might be to study income differentials within a country, such as the United States, and test to see whether climate has played a role in regional economic growth.

The increased availability of data on global weather has led to an increase in the use of GIS (geographical information systems) software and data sets in investigating weather-related phenomena as evidenced by recent research in the area. GIS has the advantage of being potentially more accurate since it corresponds to the particular surface area being analyzed instead of a countrywide average. This would be particularly helpful if one were to study differences in output and climate within a certain region or country, for instance Brazil.

As a final note, it should be stated that the purpose of studies such as this are not to suggest that geography alone is responsible for determining the economic outcome of a country, a concept that has come to be known as ‘geographical determinism’. Instead, the intent is to draw attention to the fact that geography and climate do matter and how they matter is an area worthy of further investigation.

If a particular technology or policy prescription works in a certain environment because of the right conditions, then adapting it to work in a different one where conditions are substantially altered would certainly require a commitment to R&D that may be beyond the scope of poorer countries but could be pursued in the developed world. In addition, policy or development might increasingly be tailored to regional conditions, even in more developed countries.