The analysis here is based on a dataset which consists of per capita GDP levels, a measure of institutions, measures of geography, a measure of openness, measures of religion and a measure of human capital in (up to) 180 countries. Since I am combining data from diff erent sources, I have to deal with diff erent numbers of observations for diff erent variables. The data typically also come from diff erent years. In case of institutional meas- ures, I use averages in order to capture the long- run eff ect. The defi nitions and sources of all the variables are summarized in the Data Appendix.
Table 4.1 presents the summary statistics of these measures.
I divide the dataset into six major parts depending upon the variable that they are measuring. They are as follows: measures of economic devel- opment, measure of institutions, measures of geography, measures of religion, measure of openness and trade and measure of human capital.
Following is a brief outline of each of them. We also plot some of these variables to look at the correlation.
Economic development is measured by the level of per capita GDP in
Table 4.1 Summary statistics Variables Number
of obs
Mean Standard deviation
Minimum Maximum Measure of institutions
Rule of law index 171 0.0015 0.9441 − 2.19 2.37 Measures of geography
Distance 178 19.6293 17.0873 0 63.89
Malaria risk 160 0.3678 0.4390 0 1
Land area within tropics
146 0.4991 0.4779 0 1
Soil suitability 154 13.3965 9.8512 0 55.07 Land area within
100km of ocean or ocean- navigable river
146 46.0343 37.604 0 100
Measures of religion
Catholicism 174 31.3805 36.1243 0 97.3
Islam 174 22.0817 34.749 0 99.8
Measure of openness and trade
Log of trade share 146 4.0914 0.6384 2.58 5.76 Measure of human capital
Enrolment ratio in 1900
88 32.34 26.37 0.1 95
Measures of economic development Log initial income
(1820)
50 6.58 0.3893 5.98 7.52
Log initial income (1870)
60 6.88 0.5668 5.98 8.09
Log initial income (1900)
39 7.46 0.6096 6.30 8.41
Log initial income (1950)
136 7.29 0.9446 5.67 10.32
Log initial income (1960)
115 6.3153 0.8647 4.72 8.23
Log per capita GDP in 2000
147 8.588 1.1177 6.19 10.799
Note: For a detailed discussion of the defi nition and source of these variables, see Data Appendix.
2000. I also use the level of GDP per capita in 1960 as the initial level of development.
The average rule of law index is a measure of institutional quality. In particular, it measures the overall institutional quality and the quality of governance in a particular country. The range of this index varies from
−2.5 to 2.5 with higher values implying better institutional quality. I use an average value of this index from 1998 through to 2000 for each of 171 countries as the measure. The data show that Somalia has the weakest institutions in the world and Singapore has the strongest institutions.
In case of geography, I use fi ve diff erent measures: distance, malaria risk, land area within tropics, soil suitability and land area within 100 km of the ocean or an ocean- navigable river.
Distance is a measure of the distance from the equator. Following the argument of Sachs (2003a), I use this as a measure of climate. The greater the distance from the equator, the further the country is from the tropics and more temperate or cold it is. I exploit data from 178 countries.
Malaria risk measures the share of population at risk from malaria in 160 countries in the year 1997. I use this as a measure of disease burden. A higher value indicates greater risk for the population. Typically, tropical and subtropical countries register higher risk of malaria and the risk of malaria declines in the temperate climates.
Land area within the geographical tropics is used as a measure of the eff ects of geography on agricultural productivity. This is calculated as a proportion, a higher value implying that a higher proportion of the country’s land is tropical in nature and, therefore, the country’s agricul- ture is expected to have tropical characteristics marked by slow and low plant growth. I use soil suitability as an additional measure of agricul- ture. Ukraine registers as the country with highest proportion of suitable soil.
Finally, I use land area within 100 km of the ocean or an ocean- navigable river as a measure of market proximity. The higher the proportion of land within 100 km of the ocean or an ocean- navigable river, the greater is the chance for an economy to participate in maritime trade and have better access to larger markets. The bulk of the sub- Saharan economies is observed to be landlocked.
In measuring religion, I use the proportion of population following Catholicism and Islam in the year 1980 as a measure of the same, respec- tively. The data shows Spain, Ireland, Portugal and the most of Latin America to be predominantly Catholic and the Middle East, Indonesia, Pakistan and some of the ex- Soviet republics to be predominantly Islamic.
I measure openness to trade by using the log value of the actual trade
share from Frankel and Romer (1999). They calculate the actual trade share by taking the percentage of imports plus exports to GDP in 1985 from 146 countries. The higher the trade share, the more open is the economy. Singapore is observed to be the most open economy in the world and Myanmar is the most closed.
The human capital measure is from Benavot and Riddle (1988). They record data on the primary school enrolment ratio in 1900. I use this as a proxy measure of historical human capital attainment. The data shows that in the United States, 95 per cent of the relevant school going popula- tion were enrolled in primary school in 1900 whereas in Equatorial Guinea the same number was as low as 0.1 per cent.
Pairwise correlations are reported in Table 4.2. It shows that better institutions, less disease, better agriculture and soil conditions, more trade and higher initial income are all positively associated with development.
This is also revealed in the following scatter plots (Figure 4.1a–i).
The data shows that countries that were wealthier in 1960 were also likely to be wealthier in 2000. The data also shows that institutions measured by rule of law, trade and 1900 school enrolment rate are posi- tively correlated to the current level of development. This is in line with the prediction made by many theorists of root causes of development.
In contrast, malaria risk and land area within the tropics are negatively correlated to the current levels of development. The scatter plots also show that countries further from the tropics, countries with proximity to oceans and ocean- navigable waterways, and countries with Catholicism as the dominant religion are relatively prosperous. The obvious ques- tion is, does this imply causality? I will discuss this issue in detail in what follows.