Energy Import Dependency and Trade

The analysis on physical import dependency solely focused on the physical amount of energy required to fill the gap between domestic commercial energy use and production. No reference was made to the economic costs associated with the import of foreign energy sources. One possibility to measure the economic costs of import dependency is to compare the fuel import expenditures with the total merchandise export revenues of a country. As previously mentioned, current publications –including the World Development Reports– fail to provide up-to-date data on this cost measure. To compensate for this omission, I calculated the ratio of the value of fuel imports to the value of total exports by making use of recent UN Comtrade data.

This indicator is similar to the widely applied terms-of-trade indicators, which describe a country’s export to import ratio. The net barter terms-of-trade index is one of the most frequently encountered ratios. It measures the price of a country’s export basket relative to the price of its import basket. The index can also be regarded as the ratio of the unit values of exports and imports (Sell 1993: 242-247, Wagner 1997: 48). The terms-of-trade concept is frequently being applied to determine the impact of trade conditions on the economies of developing countries. Developing countries are especially exposed to terms-of-trade fluctuations because they cannot exert substantial influence on their export and import prices.

Especially the prices of basic commodities, which usually account for the largest share of developing countries´ export basket (e.g. cotton, timber, coffee, fuels), are first and foremost dictated by global market forces and the demand on the sales markets in industrialized countries. According to an UNDP study, there is a close link between commodity export dependence and indebtedness. Almost all of the heavily indebted poor countries (HIPCs) have a commodity export dependence greater than 50 per cent. For example, in 1997, nuts and fishery accounted for 95.8 percent of Guinea-Bissau’s exports, cotton and tea for 95.7 percent of Burundi’s exports and cotton and gold for 83.9 percent of Mali’s exports (UNDP 2001: 4).

Other studies found out that terms-of-trade fluctuations are twice as large in developing countries as in industrialized countries (Baxter et al. 2000) and can explain up to 50 percent of

Chapter 3: Quantitative Analysis of Energy Import Dependency developing countries´ output volatility (Mendoza 1995, Kose 2002). What holds true for export prices of basic commodities also applies to the import costs of foreign energy sources.

Energy importing countries normally have to accept the fact that fuel prices are outside their control and therefore have to treat them as exogenous. For these countries, a substantial increase in fuel prices –all other things equal– leads to a deterioration of their terms-of-trade.

Hence, the ratio of export prices to import prices becomes smaller. Applied to the indicator of fuel imports to total exports used in my analysis, the same price development leads to a higher indicator value. In the following, I do not attempt to examine terms-of-trade fluctuations over time. I would rather like to highlight the structural differences in energy import costs between the country income groups with regard to their export earnings.

The mean ratio of fuel import costs to total export earnings was calculated for all energy importers and exporters with available data for the years 1998 to 2002 (see Appendix B, Table B-3). The range of possible ratio values is positive and stretches from 0 percent to over 100 percent. A value of zero means that a country does not import fuels at all. This was not the case for any of the 166 countries in the main sample. As mentioned earlier, even major energy exporting countries import small amounts of foreign energies. In the observed period, the average ratio of fuel import costs to total export earnings was 0.1 percent for Saudi Arabia and 0.4 percent for Kuwait. However, the other extreme, a situation where a country’s fuel import costs surpass the total value of its export revenues can be found in the sample data. I will detail on these extreme cases shortly. As in the previous analysis on physical dependency, I wanted to restrict my statistical analysis only to inter-group differences between energy importing countries. The ratio of import costs to export revenues itself was not helpful in separating importers from exporters, because all countries have positive values and even some major energy exporting countries have considerable high ratio values. Bahrain is a good example. For the observed time period, the costs of fuel imports accounted for almost one third of the country’s total export earnings (30.8 percent). This is due to the fact that Bahrain, unlike other countries in the Gulf region, exports refined petroleum products rather than crude.

In 2001, Bahrain had a refinery output of 257 thousand barrels per day but produced only 37 thousand barrels per day on its own territory. The difference was met by crude oil imports from Saudi Arabia (EIA 2001: Bahrain – Country Energy Balance). These physical values are also reflected in Bahrain’s trade balance. In 2002, Bahrain imported fuels worth $1.7 billion but at the same time earned almost $4 billion from fuel exports (UN Comtrade data).

Chapter 3: Quantitative Analysis of Energy Import Dependency I therefore had to rely on other selection criteria. Firstly, I considered only those countries as being energy import dependent, which had a positive ratio of net fuel import expenditures to GDP in the period from 1998 to 2002⁴⁰. In other words, I included only countries where fuel imports constitute a cost factor and lead to a loss of GDP. For energy exporting countries net fuel import expenditures are negative indicating that the energy exports actually add to a country’s GDP. In the case of Bahrain, the country was excluded due to a ratio value of -28.8 percent. In other words, roughly one third of Bahrain’s GDP was earned through fuel sales. A detailed overview of the fuel import-GDP indicator, its calculation and the country results will be given in the next section. Secondly, I checked whether the countries identified as energy import dependent in the first step were consistent with those countries included in the analysis of physical import dependency in section 3.2. Normally, net importers of commercial energy sources should also be expected to have positive net fuel import expenditures, whereas net exporters of commercial energy sources should have negative net fuel import expenditures (positive net fuel export revenues). This reasonable assumption holds true for almost all countries. Only two countries, South Africa and Paraguay, do not fit into this pattern. The indicator “energy imports, net (as % of commercial energy use)” identifies the two countries as net energy exporters. In 1999, South Africa had an indicator value of -31.7 percent and Paraguay of -62.8 percent. However, in the five-year period from 1998 to 2002, both countries paid more for fuels imports than they earned from fuel exports. The costs of net fuels imports as a percentage of GDP amounted to 1 percent for South Africa and to 4 percent for Paraguay. How can this apparent inconsistency be explained? Paraguay is the fourth largest producer of electricity in South America, behind Brazil, Venezuela and Argentina. Of the 48.4 billion kWh produced in 2002, 95 percent was exported, mainly to Brazil. This made Paraguay the second largest net electricity exporting country in the world, behind only France.

On the other hand, Paraguay has no crude oil reserves and relies entirely on the import of petroleum products. In 2002, the country imported 22 thousand barrels of oil per day (EIA 2004: Paraguay – Country Analysis and Data Report). Obviously, the bill of oil imports exceeded the revenues from exporting electricity. However, when the amount of generated electricity is measured in tons of oil equivalent, the physical quantities of fuel exports by far outweigh the physical quantities of oil imports. Paradoxically, Paraguay can be described as a fuel importing country in economic terms but at same time as a fuel exporting country in physical terms. The same explanation is likely to be true for South Africa. The country currently belongs to the world’s top coal producing and exporting countries. Similar to

40 Net fuel import expenditures are calculated by subtracting fuel export revenues from the costs associated with fuel imports. Hence, a positive value means that a country is a net importer.

Chapter 3: Quantitative Analysis of Energy Import Dependency Paraguay, South Africa has to meet its crude oil demand almost entirely through imports.

Based on the data given above, the expenditures on oil imports seem to surpass the foreign exchange earnings from coal exports, even though the physical amount of energy imports is apparently lower than the physical amount of energy exports (EIA 2003b: South Africa – Country Analysis). To achieve maximum consistency between the indicators of physical and economic dependency, I excluded Paraguay and South Africa from the country sample examined in this section.

The application of the two selection criteria finally led to a sample of 115 countries (27 HICs, 21 UMCs, 30 LMCs, and 37 LICs). The statistical results for this sample are presented in Table 3.3.2-1 below. The box plots show that the value distributions for all four country groups are asymmetrical and contain various outliers. The outliers with ratio values above 100 percent are the Netherlands Antilles in the HIC group, Lebanon and the Seychelles in the UMC group, the Dominican Republic in the LMC group, and Sierra Leone and Sao Tome and Table 3.3.2-1: Value of Energy imports (% of total export earnings) 1998-2002

Overview of Selected Countries

High Income Countries Upper Middle Income Countries Highest Ranking Countries

Greece 31.24 Grenada 37.75 Belize 45.12 Burkina Faso 65.28 Countries cllose to Median

Iceland 8.44 Brazil 13.31 Bulgaria 23.78 Mali 34.38

Source: Own Calculations based UN Comtrade Data.

Boxplot: Data Distribution (Only Net Energy Importer)

Statistical Data

HIC UMC LMC LIC N (= 115) 27 21 30 37

Mean 17.45 28.42 29.06 47.96

Standard Deviation 20.91 30.66 27.51 54.34 Coefficient of Variance 1.20 1.08 0.95 1.13

Chapter 3: Quantitative Analysis of Energy Import Dependency Principe in the LIC group⁴¹. Because all outliers are located outside the upper whiskers of the boxes, the distributions are positively skewed with mean values being greater than the respective group medians. For example, the HIC income group has a mean value of 17.5 percent, which is almost double the value of its median value (8.8 percent). To avoid any distortions caused by the asymmetrical nature of the value distributions, it seems appropriate to compare only the median values of the four income groups. A typical energy import dependent country close to the group’s median pays around 9 percent of its export revenues for fuel imports in the high income group (e.g. Netherlands, New Zealand), 16 percent in the upper middle income group (e.g. Uruguay, Slovak Republic), 24 percent in the lower middle income group (e.g. Guyana, Guatemala) and 35 percent in the low income group (e.g. India, Burundi). These inter-group differences are graphically reflected in the step-function-like increase of the box plots´ median lines from the left to the right. The results show that energy import dependency is especially damaging to the trade performance of low income countries.

For these countries the loss of export earnings due to fuel imports is on average three times higher than for high income countries. When mean values are compared, the ratio is even 4 to 1. These ratios also imply that fuel price fluctuations have larger impacts on the terms-of-trade in developing countries than in industrialized countries. All other things equal, a 10 percent increase in energy import prices would reduce monetary gains from trade by 3.4 percent in developing countries but only by 0.9 percent in high income countries. The other way round, a global slump in energy prices can be assumed to have a greater positive effect on the trade balances of less developed countries.

41 To make sure that these high figures are reliable, I checked if the countries ran into trade deficits during the period in question. This was actually the case for all 6 countries (Netherlands Antilles: -900 thousand US$

(1998), Lebanon: -6 billion US$ p.a. (1998-2001), Seychelles: - 220 thousand US$ p.a. (2001-2002), Sierra Leone: -310 thousand US$ (2002), Sao Tome & Principe: -25 thousand US$ p.a (1999-2002)). However, the data basis for the Netherlands Antilles, Sierra Leone and the Seychelles was only limited to one or two observation years. In a second step, I examined whether the countries´ severe import dependency can also be observed from the EIA country energy fact sheets. For all countries, except the Netherlands Antilles, the physical quantities of energy imports by far outweighed those of energy exports. The Netherlands Antilles proved to be an important oil refining country in the Caribbean, similar to Bahrain in the Persian Gulf. In 2001, the Netherlands Antilles imported 242 thousand barrels of crude oil and at the same time exported 230 thousand barrels of refined oil products per day (e.g. gasoline, jet fuel). Domestic daily oil consumption was only around 72 thousand barrels. The ratio calculated for the Netherlands Antilles (103.2%) is therefore unlikely to give a real picture of the country’s energy import dependency. Either the ratio is based on wrong UN Comtrade data, or the country’s energy situation has dramatically changed in the period from the latest UN Comtrade observation year 1998 to the latest 2001 EIA data. The data quality for Sierra Leone can also be questioned. It is unlikely that the war-torn country is capable to effectively and comprehensively report on its foreign trade. Due to smuggling and activities in the shadow economy, import and export figures can be easily over- or underestimated.

Chapter 3: Quantitative Analysis of Energy Import Dependency From Figure 3.3.2-1 one can quickly observe in which geographical regions energy import dependency is particularly detrimental to a country’s trade balance. In the yellow coloured countries, energy imports reduce export revenues by 15 to 30 percent, in the orange coloured countries by 30 to 60 percent and in the red coloured countries by 60 percent and more. Most severely affected are countries in Central America and the Caribbean, in West Africa and most parts of East Africa stretching from Egypt in the North to Zimbabwe in the South, in South-East Asia including India, in various countries located at the periphery of the former Soviet Union (e.g. Mongolia, Kyrgyz Republic, Georgia) and in South-East and Eastern Europe. Countries where the share of export revenues lost to the purchase of fuel imports is less pronounced are coloured in dark green (5-15%) and light green (less than 5%). This particularly applies to the net energy exporting countries, which were excluded from the statistical analysis but are now also colour-coded to make the graphical representation more expressive. Net energy exporters like Canada, Mexico and Saudi Arabia are all coloured in light green, indicating that these countries have ratio values well below 5 percent.

Furthermore, the world map clearly illustrates the privileged position of net energy importing high income countries. The majority of these countries pays less than 15 percent of their export revenues for energy imports and is, thus, also coloured in green (mostly dark green).

Figure 3.3.2-1: Fuel Imports (% of Total Merchandise Exports) 1998-2002

Source: Own Illustration based on own Calculations.

less than 5 % 5 – 15 % 15 – 30 % 30 – 60 % > 60 % no data

Chapter 3: Quantitative Analysis of Energy Import Dependency 3.3.3 Energy Import Dependency and Economic Prosperity

The costs of energy import dependency can further be measured in relation to the economic power of a country. For this purpose, I calculated the average ratio of net energy import expenditures to a country’s GDP for the time period from 1998 to 2002⁴². This indicator was already shortly introduced as a selection criterion for energy importing countries in the last section. Trade data was again taken from the UN Comtrade database. GDP data was obtained from the World Development Indicators 2002 and 2004. As before, the statistical analysis is restricted to net energy importing countries. Therefore, I only included countries with positive ratio values indicating that the costs of energy imports place a burden on the nation’s gross domestic product. This was the case for 113 countries (26 HICs, 22 UMCs, 29 LMCs, 26 LICs). The countries are listed in Table B-4 in Appendix B.

The calculations revealed that the industrialized countries are once again in an economic favourable position. A characteristic high income country pays roughly 2 percent of its GDP for energy imports. Even high income countries depending almost completely on the provision of foreign energy sources like Singapore and Luxembourg have energy import bills of less than 3.5 percent of their GDP. In contrast, typical middle and low income countries with ratios close to their respective group medians have expenditure shares between 3.5 and 4 percent of their GDP. If mean values are compared, the average ratio of import costs to GDP for LMC and LIC rises to 4.6 and 5 percent respectively. Inter-group differences are further to be found in the data range and variance of the four country groups. The cost ratios for HICs range from 0.78 (Netherlands) to 3.76 (New Caledonia), for UMCs from 0.86 (South Africa) to 6.0 (Slovakia), for LMCs from 0.69 (China) to 16.73 (Guyana) and for LICs from 0.37 (Cote d'Ivoire) to 17.64 (Moldova). In addition, the value distributions in the LMC and LIC groups are far less homogenous than in the HIC and UMC groups. This can statistically be proved by calculating the coefficients of variance for each country group⁴³. The coefficients for the HIC and UMC groups are slightly above 0.4, whereas the coefficients for the LMC

42 Net energy import expenditures are calculated by subtracting energy export revenues from the gross energy import expenditures of a country. For many energy importing countries, the distinction between gross and net imports is minor. Nevertheless, in some cases, a sole focus on gross import expenditures would overstate the economic costs of energy import dependency. In the case of Bahrain, gross energy import expenditures amounted to 22 percent of GDP in 2002. However, if energy export revenues are taken into account, the net ratio changes to -29 percent. In other words, Bahrain did not lose 22 percent but actually gained 29 percent of its GDP through energy trade.

43 The coefficient of variance is a measure of how much variation exists in relation to the mean. To calculate the

coefficient, the standard deviation of each group is divided by the corresponding mean value:

Chapter 3: Quantitative Analysis of Energy Import Dependency and LIC groups are around 0.7 (see Table 3.3.3-1). The following comparison illustrates the broad spectrum of import dependency ratios in the LIC group: Countries in the group’s first quartile are in a situation quite similar to their counterparts in the HIC group. For most of these countries, the ratio of net fuel import expenditures to GDP is around 2 percent of GDP or below (e.g. Rwanda: 2.15 percent, Uganda 2.14 percent, Bangladesh: 1.27 percent).

However, for countries in the last quartile of the LIC group, the economic costs of fuel import dependency are already four times higher and lead to a loss of GDP in the order of 7 to 8 percent (Ghana: 7.8 percent, Armenia: 7.7 percent, Nicaragua: 8.0 percent) or more (Mongolia: 10.3 percent, Ukraine: 13.8 percent, Moldova: 17.6 percent, Sierra Leone 17.9 percent⁴⁴).

Table 3.3.3-1: Costs of Net Energy Imports (% of GDP) 1998-2002

Overview of Selected Countries

High Income Countries Upper Middle Income Countries Countries close to Median

Finland 1.68 Czech Rep. 3.10 Sri Lanka 3.30 Nepal 3.51

Source: Own Calculations based on UN Comtrade Data.

The numerical differences in the ratio of net fuel import costs to GDP are graphically portrayed in Figure 3.3.3-1. In the period from 1998 to 2002, only the four countries Moldova,

44 The trade data quality for Sierra Leone can be questioned. For details see footnote 41.

Boxplot: Data Distribution (Only Net Energy Importer)

Chapter 3: Quantitative Analysis of Energy Import Dependency Mongolia, Guyana and Ukraine paid more than 10 percent of their GDP for the import of foreign energy sources. These exceptional cases are coloured in red. Countries with lower but

Chapter 3: Quantitative Analysis of Energy Import Dependency Mongolia, Guyana and Ukraine paid more than 10 percent of their GDP for the import of foreign energy sources. These exceptional cases are coloured in red. Countries with lower but