Kuwait: Selected Issues and Statistical Appendix
Author: International Monetary Fund
Publication Date:
18 Jun 2012
eISBN:
9781475504590
ISBN:
9781475504590
Language:
English
Keywords:
ISCR; CR; Kuwait; investment company; asset; gasoline price; income elasticity; interaction terms; price difference; energy price; Income
Download PDF (805.4 KB)

This Selected Issues and Statistical Appendix paper on Kuwait focuses on recent development in investment companies (ICs) and the business environment in the country. ICs continue to be vulnerable to swings in financial and real estate markets. They continue to have large exposures to domestic, regional, and international equity and real estate markets. Local banks’ lending to ICs declined in 2010–11 owing to banks’ write-offs of ICs loans. The financial situation of many ICs remains precarious, and there are 15 listed investment companies in a dire situation.

Abstract

This Selected Issues and Statistical Appendix paper on Kuwait focuses on recent development in investment companies (ICs) and the business environment in the country. ICs continue to be vulnerable to swings in financial and real estate markets. They continue to have large exposures to domestic, regional, and international equity and real estate markets. Local banks’ lending to ICs declined in 2010–11 owing to banks’ write-offs of ICs loans. The financial situation of many ICs remains precarious, and there are 15 listed investment companies in a dire situation.

III. Fuel Subsidies and Energy Consumption—A Cross-Country Analysis1

A. Introduction

1. GCC countries have among the highest levels of per capita energy consumption per capita in the world (Figure III.1). Data produced by British Petroleum’s Statistical Review of World Energy indicate that Kuwait, Qatar, Saudi Arabia, and the United Arab Emirates (no data are reported for Bahrain and Oman) were among the top ten per capita consumers of energy in 2010.

Figure III.1.
Figure III.1.

Energy Consumption per Capita, 2010

(In barrels of oil equivalent)

Citation: IMF Staff Country Reports 2012, 151; 10.5089/9781475504590.002.A003

Sources: British Petroleum Statistical Review of World Energy and IMF staff calculations.

2. GCC countries’ high energy consumption is in part related to their high percapita income levels, but is also linked to low domestic energy prices. For example, gasoline prices in the GCC are highly subsidized (compared to their opportunity cost) and are among the lowest in the world (Figure III.2). A highly subsidized gasoline price is a feature of several other fuel exporters (e.g., Algeria, Ecuador, Iran, Turkmenistan, and Venezuela).

3. The high level of energy subsidies raises the question of whether resources are being spent in the most efficient manner. In particular, untargeted energy subsidies tend to benefit low-income households less than high-income households, because the latter consume more energy. In addition, energy subsidies tend to encourage inefficient excess consumption of energy.

Figure III.2.
Figure III.2.

Price of Gasoline, end-2010

(US$ per liter)

Citation: IMF Staff Country Reports 2012, 151; 10.5089/9781475504590.002.A003

Sources: IMF, FAD database on cross-country gasoline prices and GTZ.

4. In this chapter, we explore the relationship between energy prices and consumption, and discuss its implications for subsidy reform—for Kuwait in particular. In Section B, we report the findings from the empirical literature on this topic. In Section C, we estimate the sensitivity of energy consumption to energy prices using a panel dataset of 66 countries. Finally, we explore the implications of our estimates for subsidy reform.

B. Responsiveness of Energy Consumption to Energy Prices—Findings of the Empirical Literature

5. While it is usually argued that high energy subsidies encourage inefficient consumption, quantifying this link is generally difficult. This is largely linked to the difficulty in estimating the price elasticity of demand. Two issues are particularly important: First, short-term and long-term demand elasticities are likely to differ, since consumers may need time to adapt to a different price environment (e.g., they may need time to change from a high energy-intensive technology to a less energy-intensive one in response to higher prices). Second, empirical estimations face identification problems; the data observed usually reflect the interaction of both supply and demand factors, making it difficult to isolate demand-specific elasticities.

6. It is therefore not surprising that there is significant variability in the price elasticities of energy demand found in the empirical literature. As illustrated in Table III.1, the findings of the empirical literature suggest a large variability for both short-term and long-term price elasticities. For instance, the work summarized in Hamilton (2009) puts short-term price elasticities for gasoline and oil consumption in a range from nearly zero to -0.25, and long-term price elasticities in a range from -0.21 to -0.86. More recently, IMF (2011) estimated much lower elasticities for oil consumption: a short-term price elasticity of nearly zero, and a long term price elasticity of -0.08.

Table III.1.

Income and Price Elasticities of Energy Demand: Literature Review

article image
Source: IMF staff literature review.

7. There is more consensus, however, that income elasticities are close to one, but recent evidence may suggest lower values. The work summarized by Hamilton (2009) indicates a range for the income elasticity of oil and gasoline consumption between 0.88 and 1.32 (Table III.1)—i.e., close to unitary income elasticity. IMF (2011), however, suggests lower long-term income elasticities for oil consumption, arguing that this may reflect substitution away from oil and toward other energy sources.

C. Re-examining the (Long-term) Price Elasticity of Energy Consumption

8. Given the wide range of empirical estimations in the literature, we reexamine the issue by comparing cross-country energy consumption and its determinants. The use of cross-country data provides key sources of heterogeneity to estimate income and price elasticities of energy demand, as countries differ widely in per capita income per capita and also in their energy prices. The price differences are mainly due to different taxation regimes, as some countries tend to tax energy heavily, while other countries apply lower taxes or even subsidize energy consumption. Cross-country data also provide other sources of heterogeneity that can affect energy consumption, such as weather conditions and the level of development (e.g., more developed countries typically have stricter environmental regulations that can affect energy consumption).

Analytical framework

9. Our starting point for estimating the determinants of cross-country energy demand is the following standard demand equation:

qi,t=αγtAiyi,tδpi,tβ(1)

where qit denotes energy demand of country i at time t, a denotes a constant, yt denotes a time-varying parameter (e.g., technology), At denotes country-specific factors (e.g., weather or environmental regulations), yi,t denotes the (real) income of country i at time t (i.e., demand for all goods and services), pi,t denotes the real price of energy, δ is the income elasticity of energy demand—which is expected to be positive, and β is the price elasticity of energy demand—which is expected to be negative.

Taking natural logs on both sides of equation (1) we obtain:

ln⁡(qi,t)=ln(α)+ln(γt)+ln(Ai)+δln(yi,t)+βln(pi,t)(2)

Data

10. The data are collected from a variety of sources and are available for periods of different length:

  1. Data on energy consumption (measured in million tons of oil equivalent) come from British Petroleum’s 2011 Statistical Review of World Energy. These data are available for the period 1965–2010 for a group of 66 countries, which defines the country dimension of the dataset.

  2. US CPI data—used to deflate variables denominated in U.S. dollars—and GDP per capita statistics (in U.S. dollars)—used as a proxy for countries’ income—come from the IMF’s World Economic Outlook database. Data on population come from the United Nations. These data are available for the period 1965–2010.

  3. Gasoline prices—used as a proxy for overall energy prices—for 2002–09 come from the IMF’s Fiscal Affairs Department data prepared for the paper by Coady and others (2010), while 2010 gasoline prices come from GTZ. These data are only available for the period 2002–10, and therefore restrict the time dimension of the dataset.

  4. Raw daily weather data come from the U.S. National Oceanic and Atmospheric Administration (NOAA).2 These data were processed to compute a cold weather index (average temperature of the three coldest months) and a hot weather index (average temperature of the three hottest months). The time coverage of these data varies from country to country. For our empirical exercise (below) we have collapsed these two variables into two time-invariant country dummies by taking the average of the specific variable during the observation period available for each country.3

  5. An advanced countries dummy follows the IMF World Economic Outlook country classification. This classification implies that our sample is composed of 29 advanced economies and 37 non-advanced economies.

In total, we have a highly balanced panel (for some countries gasoline prices were missing in some years) of 66 countries covering the period 2002–10.

Estimation strategy

11. We use OLS (with robust standard errors and adjusting for within-cluster correlation) for estimating equation (2). The choice of the estimation method is not trivial. An alternative to our estimation method would be to use fixed-effects (Within) estimation, but this procedure may be problematic, given the short time series of our dataset and given that this method will likely capture short-term elasticities. On this issue, Baltagi and Griffin (1984) have shown with Monte Carlo simulations that OLS estimation is likely to be superior to fixed-effects (and random effects) estimation if the objective is to estimate long-term elasticities, particularly for panels with a short time dimension.

12. Before proceeding with the estimation, it is illustrative to view the data assuming unit income elasticity.4 In this case equation (2) can be rearranged to obtain:

ln⁡(qi,tyi,t)=ln(α)+ln(γt)+ln(Ai)+βln(pi,t)(3)

where the left-hand side is the share of energy consumption in total (real) GDP.5 Figure III.3 shows the share of energy consumption in total demand against the domestic gasoline price (which, as mentioned above, is taken as a proxy for the overall domestic cost of energy). Despite some outliers that show quite low prices and medium levels of energy consumption, the chart suggests, as expected, an inverse relation between energy consumption share and energy prices.

Figure III.3.
Figure III.3.

Share of Energy Consumption (per-capita) to GDP (per-capita) and Energy Prices in a Sample of 66 Countries, 2002-10

Citation: IMF Staff Country Reports 2012, 151; 10.5089/9781475504590.002.A003

Source: Authors’ calculations using data from: British Petroleum Statistical Review of World Energy, FAD and GTZ databases on gasoline prices, and World Economic Outlook database.

D. Estimation Results6

13. The estimation of equation (2) suggests a relatively good fit and a statistically significant link between energy consumption and the explanatory variables. The explanatory power of the estimated equation is quite high; R-square is in the order of 0.75–0.81. In terms of significance of the regressors, looking at column (3) of Table III.2 we obtain that:

  1. Real per capita GDP has the expected (positive) sign, is statistically significant, and implies an income elasticity of around 0.75. This elasticity is somewhat lower than the range reported by the empirical literature mentioned in section B.

  2. The gasoline price also has the expected (negative) sign, is statistically significant, and implies a (long-term) price elasticity of around -0.3. This elasticity is in the lower end of the range reported by the empirical literature mentioned in section B.

  3. The weather variables are both statistically significant and have the expected sign. They imply that having a winter (summer) period that is on average 10 degrees Celsius colder (hotter) will increase energy consumption by 2.9 (2.7) percent.

  4. The advanced countries dummy was not significant.

Table III.2.

Estimating Energy Demand

article image
Note: Standard errors in parentheses. Standard errors are robust and adjust for within-cluster correlation. Time dummies included but not reported. Three, two, and one asterisk denote significance levels of 1, 5, and 10 percent, respectively. Estimation is done using OLS for the period 2002–10.

14. Testing for different coefficients across country groups (i.e., advanced versus nonadvanced) and for the impact of outliers suggests a lower income elasticity in advanced countries and a higher price elasticity (but equal) across countries.

  1. Column (1) of Table III.3 shows the results of a regression with interaction variables that allow the coefficients to differ between advanced and nonadvanced countries. This column shows that the interaction terms are only statistically significant for the income elasticity, whose interaction term is negative. This implies an income elasticity for advanced countries in the order of 0.44. Another feature of this regression is that the advanced countries dummy becomes positive and significant, which suggests that advanced countries have a higher energy consumption per capita that is not fully explained by their higher income.

  2. Column (2) of Table III.3 shows the results dropping the observations of two countries whose gasoline prices were at the bottom of the distribution, which based on Figure III.3 are outliers in terms of energy consumption. These results show that the price elasticity increases to around -0.5 (in the middle range of the empirical literature), but remains equal for advanced and nonadvanced countries. The R-square increases to 0.84.

Table III.3.

Estimating Energy Demand

article image
Note: Standard errors in parentheses. Standard errors are robust and adjust for within-cluster correlation. Time dummies included but not reported. Three, two, and one asterisk denote significance levels of 1, 5, and 10 percent, respectively. Estimation is done using OLS for the period 2002–10. The second column excludes observations for which the real price is below $0.14 cents per liter.

E. Implications for Subsidy Reform

15. The price elasticity of energy demand estimated in the previous section suggests that subsidy reform could potentially have large implications on energy consumption in several countries in the long-term, including Kuwait. As shown in Table III.4, bringing energy prices to reflect the opportunity cost would constitute a large price adjustment for GCC countries and other countries in the MENA region.7,8 For the case of Kuwait, increasing the gasoline price to its opportunity cost would mean a 183 percent increase in the retail price of gasoline. If that price increase was also reflected in price increases of other forms of energy (e.g., electricity), Kuwait could see a reduction of its energy demand of around 27–41 percent in the long term.

Table III.4.

Change in per Capita Energy Consumption if Energy Priced at its Opportunity Cost

article image
Note: For Kuwait, energy-related subsidies/transfers are projections for FY 2011/12. Source: IMF staff calculations.

Change in gasoline price needed to reach its opportunity cost (defined as the tax/subsidy-free cost of gasoline). In order to calculate the tax-free cost of gasoline we take the U.S. price and subtract the average tax in the U.S. (11 cents per liter). For end-2010 this would have implied a gasoline price of 65 cents per liter.

Changes are with respect to the end-2010 price. Notice that this exercise corresponds to changes vis-à-vis end-2010. Iran has since then adjusted its gasoline prices as it has embarked on a reform of its energy subsidies.

The opportunity cost is defined as the tax/subsidy-free cost of gasoline. In order to calculate the opportunity cost of gasoline we took the U.S. price in our sample for end-2010 and subtracted the average tax in the U.S. (11 cents per liter). For end-2010 this would have implied an opportunity cost (in U.S. dollars) of 65 cents per liter.

16. The findings suggest that countries with high energy subsidies could benefit from subsidy reform accompanied by safety nets to protect the lower income segments of the population. Let’s take Figure III.4 to illustrate this point: Ss is the energy supply in a country that is selling energy at the subsidized price Ps, which is lower than the international market price Pm (which is also the opportunity cost). Given the country’s demand curve, the country ends up consuming the amount of energy Qs, which is higher than the energy it would consume if the price of energy were equal to its opportunity cost (Qm). The rectangle formed by the areas A, B, and dw constitutes the total subsidy that the government is giving to consumers. Notice that dw is a deadweight loss from the subsidy—that is, the area in which willingness to pay by consumers (given by the height of the demand curve) is below the opportunity cost.

Figure III.4.
Figure III.4.

Illustration of Impact of Subsidy Reform

Citation: IMF Staff Country Reports 2012, 151; 10.5089/9781475504590.002.A003

Source: IMF staff analysis.

17. Now, if the government removes the subsidy (i.e., price increases to Pm) it saves for itself the areas A, B, and dw, but the welfare of consumers declines because they are now consuming less of the good and paying more for each unit. The government can then decide on how to rebate part of the resources it has saved to compensate consumers. If the government wants to keep consumers completely indifferent, then it can rebate back areas A and B, which would still provide the government a net saving equal to dw—for the case of Kuwait this would mean a net permanent long-term saving between 0.9–1.4 percent of GDP (Table III.4). But other alternatives are also possible: for instance, the government could only rebate area A—i.e., only rebate the subsidy for the amount of energy that should have been consumed without the subsidy—in which case the net permanent savings would amount to areas B and dw—which for the case of Kuwait would be between 1.8 and 2.8 percent of GDP. Finally, the government could choose a more targeted approach, focusing only on low-income consumers, in which case the savings would depend on the scope of the government’s rebate program.

F. Final Considerations

18. The findings of this chapter suggest potentially large long-term benefits from reforming energy subsidies. Nevertheless, it is important to take into account short-term considerations. In particular, as the empirical literature mentioned in Table III.1 indicates, the short-term price elasticity is likely to be lower than the long-term elasticity, which suggests that the loss of consumer welfare in the short term is likely to be higher than the long-term loss.9 The higher impact that changes in energy prices would have on consumers’ welfare in the short term calls for a gradual approach to subsidy reform or, alternatively, more generous safety nets in the short term.

References

  • Al-Mutairi, N. and M. Eltony (1996), “Price and Income Elasticities of Energy Demand: Some Estimates for Kuwait Using Two Econometric Models,” The Journal of Energy and Development, Vol. 20, (2).

    • Search Google Scholar
    • Export Citation

  • Baltagi, B. and J. Griffin (1984), “Short and Long-Run Effects in Pooled Models,” International Economic Review, Vol. 5, (3).

  • Coady, D., R. Gillingham, R. Ossowski, J. Piotrowski, S. Tareq, and J. Tyson (2010), “Petroleum Product Subsidies: Costly, Inequitable, and Rising,” IMF Staff Position Note, SPN/1005 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

  • Hamilton, James (2009), “Understanding Crude Oil Prices,” Energy Journal, Vol. 30, (2).

  • International Monetary Fund (2011), “Oil Scarcity, Growth, and Global Imbalances,” World Economic Outlook (April, Chapter 3).

  • Narayan, P. and R. Smyth (2007), “A Panel Cointegration Analysis of the Demand for Oil in the Middle East,” Energy Policy, Vol. 35, pp. 6258–6265.

    • Search Google Scholar
    • Export Citation

  • Parry, I. and K. Small (2005), “Does Britain or the United States Have the Right Gasoline Tax?” The American Economic Review, Vol. 95, (4).

    • Search Google Scholar
    • Export Citation
1

Prepared by Pedro Rodriguez, Joshua Charap, and Arthur Ribeiro da Silva.

2

For a number of countries, more than one weather collection point was available. In such cases, we chose the collection point in the capital of the country, or closest to the capital of the country.

3

This was done to avoid losing observations due to missing weather data and under the view that weather patterns are in general similar over time (i.e., a country with a relatively cold winter is in general going to have cold winters, especially when compared with other countries).

4

The case of unit income elasticity is appealing as an illustrative tool on several grounds. First, the existing empirical data suggest that income elasticity should be close to one. Second, an income elasticity equal to one can be easily obtained from standard assumption on demand functions, such as when consumer preferences are assumed to be homothetic—in which case an increase in income will trigger an equally proportional increase in the demand for all the goods in consumers’ consumption basket.

5

As mentioned above, nominal variables in U.S. dollars have been deflated by the U.S. CPI to obtain the equivalent variable in real terms.

6

Given that the main information in our dataset comes from the cross-section dimension (we have 66 countries but only nine time periods), our price elasticity should be mainly interpreted as a long-term elasticity—as per the findings of Baltagi and Griffin (1984) mentioned above.

7

The opportunity cost is defined as the tax/subsidy-free cost of gasoline. In order to calculate the opportunity cost of gasoline we took the U.S. price in our sample for end-2010 and subtracted the average tax in the U.S. (11 cents per liter). For end-2010 this would have implied an opportunity cost (in U.S. dollars) of 65 cents per liter.

8

Notice that this exercise corresponds to changes vis-à-vis end-2010. Iran has since then adjusted its gasoline prices as it has embarked on a reform of energy subsidies.

9

Because consumers would not be less able to substitute away from energy in the short term.

Same Series

Other IMF Content

Other Publishers

Cover IMF Staff Country Reports
Kuwait: Selected Issues and Statistical Appendix
Author: International Monetary Fund
Volume/Issue:
Volume 2012: Issue 151
Publisher:
International Monetary Fund
ISBN:
9781475504590
ISSN:
1934-7685
Pages:
86
DOI:
https://doi.org/10.5089/9781475504590.002
  • View in gallery

    Energy Consumption per Capita, 2010

    (In barrels of oil equivalent)

  • View in gallery

    Price of Gasoline, end-2010

    (US$ per liter)

  • View in gallery

    Share of Energy Consumption (per-capita) to GDP (per-capita) and Energy Prices in a Sample of 66 Countries, 2002-10

  • View in gallery

    Illustration of Impact of Subsidy Reform

Energy Consumption per Capita, 2010

(In barrels of oil equivalent)

Price of Gasoline, end-2010

(US$ per liter)

Share of Energy Consumption (per-capita) to GDP (per-capita) and Energy Prices in a Sample of 66 Countries, 2002-10

Illustration of Impact of Subsidy Reform