Information about Middle East Oriente Medio

2. The Determinants of Long-Term Growth in the GCC Countries

Raphael Espinoza, Ghada Fayad, and Ananthakrishnan Prasad
Published Date:
November 2013
  • ShareShare
Information about Middle East Oriente Medio
Show Summary Details

2.1 Introduction

The member countries of the GCC have changed considerably over the last thirty years. The fast development of the region has spurred the creation of new cities, the development of infrastructure, and the expansion of new industries that have attracted capital and a new labor force from around the world. The growth of these economies has been considerably higher than that of advanced economies or other oil exporters as the size of the GCC economies has more than doubled since 1986 (see Table 2.1). However, economic development has been accompanied by very large inflows of foreign workers and the population has increased by more than 80 percent in the GCC (with the exception of Kuwait). As a result, real Gross Domestic Product (GDP) per worker, a measure used to assess the improvements in worker productivity, has declined in Bahrain, Kuwait, and the UAE and improved at very low rates in Saudi Arabia, Oman, and Qatar (last column of Table 2.1).

Table 2.1.Nominal GDP and annual growth rate of real GDP and real GDP per worker(1990 to 2009)
CountrySourceNominal GDP, in

billion US dollars,

Annual growth

rate, real GDP

Annual growth

rate, GDP per

Bahrain (BHR)IMF13.55.3−1.3
IMF (non-oil GDP)10.15.9−0.7
PWT/IMF (non-oil GDP)17.16.2−0.4
Kuwait (KWT)IMF80.84.4−3.0
IMF (non-oil GDP)35.67.1−0.3
PWT/IMF (non-oil GDP)65.84.6−2.8
Oman (OMN)IMF30.94.50.5
IMF (non-oil GDP)
PWT/IMF (non-oil GDP)
Qatar (QAT)IMF44.59.61.0
IMF (non-oil GDP)
PWT/IMF (non-oil GDP)
Saudi ArabiaIMF315.82.9−0.1
IMF (non-oil GDP)
PWT/IMF (non-oil GDP)
United ArabIMF180.65.4−3.4
Emirates (UAE)PWT195.06.6−2.2
IMF (non-oil GDP)118.78.5−0.2
PWT/IMF (non-oil GDP)
Oil exportersPWT3.73.7
developing c.
OECD (median)PWT2.82.8
Note: The oil exporters (excluding the GCC) used in the chapter are Algeria, Angola, Azerbaijan, Brunei, Chad, Ecuador, Equatorial Guinea, Gabon, Iran, Iraq, Kazakhstan, Libya, Nigeria, Rep. of Congo, Russia, Sudan, Timor-Leste, Trinidad and Tobago, Turkmenistan, Venezuela, and Yemen.Source: Penn World Tables 7, IMF, and authors’ calculations
Note: The oil exporters (excluding the GCC) used in the chapter are Algeria, Angola, Azerbaijan, Brunei, Chad, Ecuador, Equatorial Guinea, Gabon, Iran, Iraq, Kazakhstan, Libya, Nigeria, Rep. of Congo, Russia, Sudan, Timor-Leste, Trinidad and Tobago, Turkmenistan, Venezuela, and Yemen.Source: Penn World Tables 7, IMF, and authors’ calculations

The disappointing growth in GDP per worker has been a worry mostly in Bahrain, Oman, and Saudi Arabia where a large portion of the national population has relatively low incomes and where job prospects, especially for the growing young population, are scarce. Poor economic performance and youth unemployment have been one of the triggers of the political transitions taking place in the broader region. In the GCC countries, economic problems have not been as acute but the region is lagging in several development areas. The Human Development Index (HDI) compiled by the United Nations (UN), which takes into account quality of education and life expectancy, consistently ranks GCC countries at levels below what would be predicted by their GDP per capita.

Governments in the GCC have had an explicit objective of diversification away from oil, with policies of high investment financed by oil revenues and undertaken with the help of migrant workers. Understanding the factors of growth, the role of investment, and the role of skills and the labor force, and refining our measure of success on this enterprise therefore has a value even for the smaller and richer countries of the Gulf.

The modern analysis of long-term growth and productivity started with Solow (1957). His growth-accounting model assumes that production (output), usually measured by real GDP, is obtained by combining two inputs: capital and some measure of labor (e.g. the hours of work in a year or the number of workers). Solow (1957) showed however that for the United States, the two factors of production did not explain output well, and he interpreted the remaining unexplained part as technical change or Total Factor Productivity (TFP), a measure of efficiency in the use of factors of production. The literature on TFP flourished as economists debated on the sources of growth in East Asia (factor accumulation versus a growth miracle), but surveys showed that results on TFP were fairly sensitive to assumptions (Felipe 1997). Nonetheless, a growth-accounting exercise remains a useful start to interpret data of factor accumulation and to discuss the sources of growth.

The aim of this chapter is to go beyond the raw numbers presented in Table 2.1 to explain the drivers of GDP growth and of productivity growth. We find that the GCC countries have swiftly accumulated large stocks of physical capital but the population increase and the shift away from oil mean that capital intensity has actually decreased or remained roughly constant. On the other hand, the efforts that have been made to improve human capital have had positive effects on growth, though educational attainment remains below what is achieved by countries with similar levels of income.

Finally, the growth-accounting exercise suggests that the development of Bahrain and Saudi Arabia is hampered by declining TFP. In addition, although Qatar and the UAE have grown fast in the last twenty years, their TFP growth has been low. One potential explanation is that the kind of capital that has been accumulated in the region (aircraft, computer equipment, electrical equipment) is not fully productive because the labor force is not educated enough. We also find that the poor quality of institutions and the large size of government consumption, both of which are possible symptoms of a resource curse, could explain the disappointing TFP growth.

The chapter first describes the economic data in the region. Section 2.3 discusses the process of diversification and shows how factors of production were accumulated. In section 2.4, we apply the growth-accounting frame-work to the GCC. Section 2.5 concludes by considering the vast cross-country literature that has attempted to explain growth by a variety of institutions. We apply the econometric estimates typically found in the literature to throw light on the determinants of TFP.

2.2 Economic Data

The dataset most commonly used in the literature to assess long-term growth is the Penn Word Tables (PWT, Heston, Summers, and Aten 2011) because this dataset corrects for the differences in purchasing power that the dollar has in different countries. However, other data sources exist that provide statistics for GDP, population, investment, etc. In particular, the World Bank, the United Nations, and the IMF provide statistics that are comparable across counties, while national statistical agencies supply some detailed data. We describe in this section the basic series needed for assessing the determinants of long-term growth and we explain which source was chosen based on our assessment of data quality.

The region hosted 40 million people in 2009, a number that tripled in thirty years, with a similar growth pattern across country (see Table 2.2). Revisions on population data are frequent in countries that attract a large population of migrant workers, and the data from the Penn World Tables (PWT) and the United Nations report lower population estimates. An increasingly large fraction of this population is of foreign origin (Chapter 3 covers the data on migrant workers in more detail). Population growth was matched by a strong increase in total employment, from around 7 million workers in the GCC in 1990 to 16 million workers in 2009.

Table 2.2.Employment and population in the GCC, in millions
CountryBahrainKuwaitOmanQatarSaudi ArabiaUAEGCC
IMF (1990)0.120.530.520.25a4.650.726.8
IMF (2009)0.412.061.111.18b8.153.5416.4
IMF (1980)0.351.371.200.229.321.0113.5
IMF (1990)0.482.131.630.4215.191.8421.7
IMF (2000)0.672.222.400.6220.473.0029.4
IMF (2009)1.043.542.881.6425.524.9139.5
Other sources
World Bank (2009)0.792.792.851.4125.394.6037.8
United Nations (2009)0.721.581.581.2014.904.8224.8
PWT (2009)1.142.492.910.8325.334.8037.5

Source is World Bank, extrapolated from 1991 data

Source is World Bank

Source is World Bank, extrapolated from 1991 data

Source is World Bank

To assess the evolution of real income and of productivity, we use the series of real GDP from the IMF and from the PWT. Data quality for this series is variable. For instance, for many years, real GDP was not compiled in the region, except in Saudi Arabia. To obtain a measure of real GDP, statisticians must remove the effect of inflation on the nominal value of economic activity (“nominal GDP”),1 using information on prices that are specific for each industry. As these statistics were not available in the Gulf countries, economists in international organizations estimated these prices using other prices—for instance, consumer price indices or price indices for imported products. The Purchasing Power Parity (PPP) index of the Penn World Tables also corrects for the purchasing power of dollars in different countries. Table 2.1 shows different measures of GDP that take into account inflation (all converted in 2005 US dollars). The first measure is real GDP from the IMF, which in most cases yields the smallest estimate of growth (column 4). The second measure is GDP growth in US$ PPP from the PWT, which tends to be high because inflation in the costs of production in the GCC is lower than in the US.

Although real GDP is a commonly used measure of economic activity, its meaning in the region, as well as for other major oil producers, is questionable. In countries without a significant natural resources sector, real GDP is a useful measure of the amount of production in the economy, which affects crucially many markets and sectors of the economy. Higher volumes of output create jobs, yield taxes to the government, attract investors, etc.

But in countries where oil production is very large, real GDP (which includes the volume of production in the oil sector) is not a good measure of economic activity, especially when the focus is on the private sector. On one hand, the effect of oil prices is taken away and therefore increases in oil prices—which improve profitability in the energy sector, yield revenues to the government, and boost asset prices—have no effect on the measure of real GDP. On the other hand, increases in the number of barrels of oil extracted (even if sold at half the price) do improve real GDP, although they do not translate into more jobs or higher private-sector profits.

This line of reasoning suggests two measures of economic activity for the region, and the validity of each of them will depend on the intended use of the indicator. One measure is GDP in constant US dollars (i.e., after taking into account the loss of purchasing power of the US dollar). This measure will be of interest when assessing the development of wealth and income in the region, or when reporting indicators of assets (or liabilities) in proportion of GDP. A second measure is a measure of real economic activity excluding production of oil and gas—what is loosely called non-oil real GDP (and more specifically, real value added in the non-hydrocarbon sector). This measure will be used when discussing growth, either in the context of short-term fluctuations or in the study of long-term development and diversification.

Series for non-oil real GDP are provided by the IMF since 19802 and the UN since 1970,3 but the noise in the data is heightened by the focus on a smaller subset of the economy and the need to estimate a price index for the non-oil economy. As a result, the series coming from the two data sources are not consistent, especially for Saudi Arabia and the UAE, signaling an issue with the estimation procedures for prices and therefore for data quality. The collection of statistics in the UAE is also complicated by the federal structure of the country and the lack of homogeneity between statistics provided by the different statistical agencies. We also constructed a new series for non-oil GDP using the Penn World Tables value of total GDP, and subtracting from it the IMF series for oil GDP, after deflating it using oil prices. This latter series may provide a better source for cross-country comparison because it is based on the PWT prices.

Employment dynamics has been a source of economic growth but the standards of living in the GCC have also increased, with annual income per capita exceeding 20,000 US dollars in all GCC countries in 2009. Qatar, the UAE, and Kuwait are among the ten richest nations in the world. Bahrain, Saudi Arabia, and Oman are not as rich, but living standards there are equivalent to those of Portugal and the Slovak Republic, and the region stands out among its poor Middle-Eastern neighbors (see Figure 2.1). Income is however not evenly distributed. Data on income inequality is sparse in the region, but income inequality should be higher than that of Algeria, Egypt, Israel, or other developing economies (where Gini coefficients are around 40 percent; see Ali 2003).

Figure 2.1.World rankings of income per capita (2009)

Note: Income per capita is US$ PPP 81,000 in Qatar (using IMF statistics for population), US$ PPP 52,932 in the UAE, US$ 32,826 in Kuwait (using IMF statistics for population), US$ PPP 23,539 in Bahrain, US$ PPP 21,579 in Saudi Arabia, and US$ PPP 20,505 in Oman.

Source: Heston, Summers, and Aten (2011). Map source is

The fortune of the region depends on its energy exports. Indeed, average income has followed closely the evolution of hydrocarbon revenues (Figure 2.2). As oil prices fell to record lows after the first Gulf War, income declined to a bottom in 1995. Since then, oil revenues have soared thanks to the tripling in oil prices and the growth of hydrocarbon production.

Figure 2.2.Oil production and per capita income in the region (1990–2005)

Source: US Energy Information Agency, IMF, and authors’ calculations

Production increased by more than 50 percent in Kuwait and the UAE and by more than 100 percent in Oman and Qatar. Production was stable in Saudi Arabia and started declining in Bahrain (the country’s reserves were drained), but this did not prevent income per capita from more than doubling in the last fifteen years.

The hydrocarbon exports have supported the countries’ ambitious development agendas. Governments financed infrastructure, developed cities, improved educational attainment, and managed to build a more diversified and modern economy. The strategy has been relatively successful and although the GCC countries’ rankings in the UN HDI are still lagging behind their rankings in income per capita, Qatar, Bahrain, Saudi Arabia, and the UAE are now among the 20 to 30 percent most advanced countries according to the Human Development Index, a significant improvement since 1980.

2.3 Diversification and the Drivers of Long-Term Growth

2.3.1 Diversification

With the exception of Saudi Arabia, the GCC region’s non-hydrocarbon growth performance has been above that of other oil producers or of advanced economies for the period 1980–2009. Growth in Kuwait, Qatar, and the UAE was in fact at par with that of India and China. The sectors that contributed most to non-hydrocarbon growth (and that therefore increased their share in real GDP) were the manufacturing sector in Bahrain, Oman, and Saudi Arabia (driven by petrochemicals), the construction sector in Oman and Qatar, and the transportation sector in Kuwait, Oman, Qatar, and the UAE. The financial sector also grew strongly in Qatar and the UAE (see Table 2.3).

Table 2.3.Nominal value added by sectors, in percent of nominal non-oil GDP
BahrainKuwaitOmanQatarSaudi ArabiaUAE
Trade, hotels, and restaurants14111510211911139102922
Transport and communication9117169134877913
Finance, insurance, and real estate353826281818172719141926
Government services2417191825163838273087
Community and social services6825221315446444
Memorandum: oil GDP/total GDP193048654858445832534039
Note; Remaining component of non-oil GDP includes imputed bank service charges, which is usually negative.Source: Country authorities and authors’ calculations
Note; Remaining component of non-oil GDP includes imputed bank service charges, which is usually negative.Source: Country authorities and authors’ calculations

The push for diversification was successful in the UAE where the share of oil to GDP decreased by more than 20 percent in real terms, although in nominal terms oil GDP grew faster than non-oil GDP. The share of hydrocarbon production in real GDP also decreased in Bahrain and Oman but this is explained as much by the limitations of oil resources as by growth in the non-hydrocarbon sector. In Qatar, gas production increased dramatically, driving the reduction in the share of the non-hydrocarbon sector in the economy. On average, the diversification efforts have paid off as non-oil growth reached high levels and was relatively symmetric across sectors, but oil production still accounts for more than 50 percent of GDP in Kuwait, Oman, Qatar, and Saudi Arabia.

Diversification can also be assessed by looking at the structure of exports. For the GCC, as for other resource-rich countries, non-hydrocarbon export competitiveness may be undermined through the Dutch disease effect. Despite the push for diversification across the GCC and their advantageous ability to alleviate bottlenecks through access to a perfectly elastic supply of foreign workers, hydrocarbon exports still overwhelm non-hydrocarbon exports, the bulk of which are exports of energy-intensive and subsidized manufactures. (See Chapters 3 and 4 for more discussion of the structure of non-hydrocarbon exports, subsidies, and how the GCC countries have dealt with the risk of Dutch disease).

2.3.2 The Stock of Capital

What are the main drivers of growth? The economics literature has emphasized the role of investment and of the stock of capital because capital is used in the production process.

Over the period 1980–2009, investment has not been significantly higher in oil exporters or in the GCC than in other countries. Countries invest typically around 22 percent of their production and the GCC is no exception. However, the ratio that matters to assess the extent of capital formation in the economy is investment in proportion to non-oil GDP, and this has been very high for oil-exporting countries.4 Investment to non-oil GDP was around 33 percent over the period 1980–2000 in the GCC, and the ratio increased to 40–50 percent after 2000.

Given a series for investment, it is possible to compute the capital stock K, using the perpetual inventory method:

where δ is the physical rate of depreciation of capital, assumed to be 6 percent.5

Data for investment and in particular for investment in real terms (i.e., deflating for the change in the price of investment) is of poor quality. Because national sources do not publish a deflator for investment, except in Saudi Arabia, two methods are possible. The first one is to use the deflator used in Saudi Arabia and apply it to the IMF series for investment in current prices (see Table 2.4, columns (a), (b), and (c)). This method corrects for inflation in the GCC and therefore corrects for differences in the price of investment across time. The second method uses the series provided by the Penn World Tables that also take into account the differences in costs of production across countries. As a result, investment in PWT is constantly higher than in the IMF database because production costs are low in the GCC. However, the PWT series only start in 1986 in the GCC and this is too late to compute a stock of capital in 1990. We therefore need to extrapolate the PWT investment series back to 1965 using the IMF series. The results for the capital stock are presented in columns (d) and (e) in Table 2.4.

Table 2.4.Investment and growth in the stock of capital
Capital stock, cumulative growth rate, in percentCapital stock per workerCapital stock per workerCapital stock per workerCapital stock per worker
1990–2009in 2005 USS,in 2005 USS,in 2005 PPPin 2005 PPP
19902009USS, 1990USS, 2009
Saudi Arabia18280,40983,340151,694220,616
Other oil producers57,45745,024
Other developing c.19,78622,990
OECD (median)120,110172,024
Source: IMF, PWT, and authors’ calculations. See text for details.
Source: IMF, PWT, and authors’ calculations. See text for details.

The second method probably provides a better picture of the cross-sectional differences in investment, in particular when one wants to compare capital intensity with other developing or advanced economies. Column (e) in Table 2.4 shows that capital intensity remains high in the GCC, at par or slightly superior to capital intensity in advanced countries. However, the first method provides a better proxy of the evolution across time of capital, because the capital stock in 1990 was constructed using actual IMF series starting from 1965 (as opposed to extrapolated PWT series). This is why we will use the statistics in columns (a), (b), and (c) of Table 2.4 for the growth-accounting exercise (i.e., to decompose growth through time), although we will also use the statistics from column (d) and (e) in our attempt to compare productivity across countries.

Between 1990 and 2009, the capital stock would have increased by around 200 percent in Bahrain, Kuwait, Saudi Arabia, and the UAE. Growth in the stock of capital would have reached 30 percent in Oman and 600 percent in Qatar. Although the GCC countries have invested massively in the last twenty years, the stock of capital per worker has been declining in Bahrain, Kuwait, and the UAE because the population increase has outpaced the rate of investment. These countries also had sizeable stocks of capital in 1990 because the oil sector was already developed, and since 1990, diversification strategies have led to the development of less capital-intensive sectors attracting a large migrant population, in particular in real estate and services. In Qatar, gas production was not yet developed in 1990 and this sector has required massive investments, yielding higher stocks of capital per capita. In Saudi Arabia and in Oman, where population growth was slower and where the energy sectors have been mature for many years, the dynamics of capital intensity have been more moderate.

2.3.3 Human Capital

A second factor that can explain growth is the level of qualification of the labor force (“human capital”). Based on the empirical literature on the returns to education, in particular Psacharopoulos (1994), Hall and Jones (1999) have suggested modeling human capital as h = eφ(s), where s is the average years of schooling of population aged 15 and over, and φ is a piecewise linear function capturing the findings that returns to education are decreasing. In Sub-Saharan Africa this return is about 13 percent, but returns to education decrease with higher levels of education, to about 10 percent on average in the world, and to about 7 percent in OECD countries. This functional form has become standard in the growth-accounting literature and the value s is obtained from the dataset of Barro and Lee (2010).

Between 1990 and 2010, all GCC countries pushed forward plans to increase schooling. The increase in the average number of years of schooling of the population was impressive in Bahrain (from 6.5 to 9.5 years), in Saudi Arabia (from 5.9 to 8.5 years), and in the UAE (from 6.1 to 9.2 years). In Kuwait (from 5.9 to 6.3 years) and in Qatar (from 5.6 to 7.5 years), the increase was more moderate.6 The estimated growth in the stock of human capital may however be overstated because the quality of education in the region has been disappointing, as noted for instance in the OECD’s assessment of educational systems (the “PISA” study).

2.4 A Growth-Accounting Exercise

A formal growth-accounting exercise, as applied by Artadi and Sala-i-Martina (2002), is the step usually taken to investigate in more depth the role of the different factors of production. We follow Caselli’s (2005) description of the model as

where Y is output (real GDP or non-oil real GDP), K is the stock of capital in the economy, L is the number of workers, and h is the measure of human capital. A, which is a residual in equation (1) capturing the unexplained component of GDP, is called Total Factor Productivity (TFP) and is considered to be a measure of efficiency in the use of factors of production. Because there are no available series on investment in the oil sector versus investment in the non-oil sector, we are not able to subtract the capital stock in the oil sector when trying to explain GDP in the non-oil sector using capital in the non-oil sector. Similarly, we do not distinguish between employment in the whole economy and employment in the non-oil sector because employment in the oil sector is small.

The parameter α is an important parameter capturing the elasticity of growth to the stock of capital. Under the additional assumption that factors are paid their marginal product, the wage rate is w = ∂Y/∂L = (1 - α)Y/L and therefore α can be estimated from the share of factor payments in GDP, i.e. α = 1- (wL)/Y. Barro and Sala-i-Martin (2004: chapter 10) report α estimated from the national accounts data on factor payments for several OECD and developing countries. Their α ranges from 0.26 (Taiwan) to 0.69 (Mexico), but its value is thought to be around 0.3 to 0.5 for most countries. α can also be estimated by regressing GDP on the factors of production (i.e., estimating equation 1), but such estimations are fraught with difficulties. Simple regressions are incorrect and overestimate α because of the common issue of reverse causality: higher GDP (which results in higher profits) finances investment and therefore a higher capital stock. Disentangling the effect of capital stock on GDP from the reverse effect is difficult. Senhadji (2000) attempted such an estimation using long-term cointegration relationships and a correction for endogenenity. His results do not point to a specific range for oil exporters. Senhadji (2000) found that α was as high as 0.7 in Algeria, 0.89 in Norway, and 0.64 in Venezuela, but estimates for Ecuador, Iran, and Nigeria were all below 0.4. When no specific estimate for α is available, the literature has tended to use α = 1/3, which is what we apply for our growth-accounting exercise.

Dividing equation (1) by the number of workers L, we define y = Y/L and k=K/L, which leads to

What does the model say about TFP? Our results are presented in Figure 2.3, where TFP in 2008 in oil-exporting countries is shown as a ratio to TFP in the US. Chart (a) shows TFP computed using GDP (vertical axis) and non-oil GDP (horizontal axis) and the stock of capital deduced from the IMF series, whereas Chart (b) shows TFP computed using the PWT series. The model suggests that TFP is higher in the GCC than in most other oil exporters. In addition, TFP is higher in the smaller countries of the region. In fact, Qatar and the UAE have productivities roughly at par with that of the US, even when TFP is calculated on non-oil GDP (horizontal axis) and when the relatively large estimate of the stock of capital from PWT is used (Chart (b)). TFP for Saudi Arabia, Oman, and Bahrain is found to be lower, around 50–60 percent of that of the US when computed on non-oil GDP.

Figure 2.3.Total factor productivity relative to the US (total GDP and non-oil GDP)

(logarithmic scale, 2008)

Source: IMF, PWT, and authors’ calculations

The model also allows us to decompose the contributions to growth coming from capital, human capital, and TFP (the unknown factor). The contribution from the stock of capital can be computed using the two different series described earlier, but the choice of the series is innocuous because both series show similar changes in the stock of capital per worker between 1990 and 2009 (although the levels are different). We choose the series computed using IMF data and the Saudi Arabia price deflator for investment.

Table 2.5 and Figure 2.4 show that the efforts made to improve skills were important drivers of GDP per worker in all countries but in Kuwait. However, in Bahrain, Kuwait, and to some extent the UAE, the decrease in capital intensity has been strong enough to drive a decline in worker productivity. Overall, capital and skills cannot explain the long-term performance of the GCC. In particular, the model suggests a decline in TFP in Bahrain and in Saudi Arabia even when looking at non-oil GDP.

Table 2.5.Growth accounting of GDP per capita, (contributions, in percentage points, 1990–2009)
Model with Total GDPModel with non-oil GDP
Δy (total GDP, IMF)αΔk(1-α)ΔhΔTFP (total GDP)Δy (non-oil GDP, IMF)ΔTFP (non-oil GDP)
Saudi Arabia−−1.00.5−0.4
Source: IMF, PWT, and authors’ calculations
Source: IMF, PWT, and authors’ calculations

Figure 2.4.Contributions to the annual percentage change in GDP per worker (1990–2009)

Note: DZA stands for Algeria; GAB for Gabon; IRN for Iran; LBY for Libya; SDN for Sudan; VEN for Venezuela; ALB for Albania; CHN for China; POL for Poland; ROM for Romania; LKA for Sri Lanka; and GBR for Great Britain.

Source: Authors’ calculations; model using total GDP for the GCC

Positive TFP growth is important because it indicates that factors of production are used efficiently. Indeed, most growth successes in the past twenty years can be attributed to positive developments in TFP (see the factors of growth for Albania, China, and the other non-oil-exporting emerging countries shown in Figure 2.6 that are offered as examples of growth success stories). Even slowly growing advanced economies such as the UK have benefitted from positive TFP. In line with the findings of a resource curse in earlier periods (Sachs and Warner 1995, 2001), many other oil producers (Algeria, Gabon, Libya, Sudan, Venezuela) also suffered from negative TFP in the period 1990–2008, despite high oil prices in the years 2002–8. The objective of the remainder of the chapter is to investigate what could have driven TFP down in the GCC region.

Figure 2.5.Imports of capital, by type (darker color for increasingly R&D-intensive capital)

Source: Feenstra (2000) and authors’ calculations following classification in Caselli and Wilson (2004)

Figure 2.6.Contributions to TFP (1991–2009), in difference from median non-oil-exporting country

Note: the median non-OECD country in the sample has a TFP over the period 1990–2008 very close to 0.

Source: Sala-i-Martin et al. (2004) database, IMF, and authors’ calculations. Data for inflation and terms of trade was not available for Iraq. Data for trade openness was not available for Albania, Kazakhstan, and Libya.

2.5 Total Factor Productivity and Country Characteristics

In the absence of a “standard model” of economic growth, the unexplained component of production, Total Factor Productivity, has been linked to various factors in the economic literature, either explicitly in analyses of TFP or indirectly in broader analyses of growth. There are a number of country characteristics that could matter, and formal arguments have been made to explain why the type of capital, geography, history, but also macroeconomic policy, trade openness, political and legal institutions, etc., might be important for the growth process.

2.5.1 Type of Capital

Caselli and Wilson (2004) have argued that not all investments are alike and that heterogeneity in the technology content of different capital goods can explain part of the unexplained factor of growth. In other words, countries that use more productive capital (for instance, computers if the labor force is skilled) will produce more for a given value of the stock of capital. To investigate the type of capital used in the GCC countries, we use Feenstra (2000) data on capital goods imports, which proxy for capital stocks. The use of this proxy is justified because for most countries (including for the GCC), capital goods cannot be produced domestically and are imported. The distribution of capital imports by type of capital is shown in Figure 2.5. The data shows that the GCC countries have invested large amounts in high-tech equipments, especially aircraft (Bahrain, Saudi Arabia, Qatar, and the UAE), communication equipment (Kuwait, the UAE, Saudi Arabia). In contrast, Oman has invested in relatively low-tech capital (motor vehicles).

In Figure 2.5, countries were sorted by their TFP growth between 1991 and 2008 and capital goods were sorted by their R&D content, as estimated by Caselli and Wilson (2004). This ordering shows that there is no clear pattern between the share of high-tech capital and the TFP growth. As a result, it is problematic to interpret, for instance, the disappointing growth performance of Oman in light of the lower technological content of its capital. Indeed, it is not so much the R&D content of capital that matters as it is its complementarity with country characteristics, such as human capital or geographical remoteness. In particular, Caselli and Wilson (2004) documented that for the average country, aircraft, computing, communication and electrical equipment are relatively inefficient given the average country’s low level of human capital. As capital imports in the GCC are relatively high-tech whereas skill levels are average, one argument could therefore be that the capital stock accumulated by the GCC is not well exploited by the labor force, yielding the decreasing TFP estimates.

Caselli and Wilson (2004) however showed that computers, motor vehicles, and communication equipment are relatively efficient for countries remote from the world’s largest economies. Several countries in the GCC have successfully developed their trade and transportation sector, basing this success on the geographical location between Asia and Europe. Therefore, the high-tech nature of capital in the GCC can be justified given its location.

2.5.2 Institutions and the Empirical Growth Literature

Barro (1991) estimated a reduced-form econometric model where income per capita is regressed on a vector of candidate variables that could explain growth. The results, however, were found by subsequent research to be sensitive to the inclusion of different variables. Although specifying the correct model remains elusive, the body of empirical work is now large enough that a survey of the literature should be able to identify the factors that have been found to be statistically and economically significant. This section discusses different surveys (or meta-analyses) that have been written in the recent years and uses coefficient estimates that reflect our view of the literature to assess the drivers of TFP growth in the GCC. The exercise is similar to that of Hakura (2004) who explains the low performance of the GCC (and of the rest of MENA) by using an econometric model of long-term growth, but instead of relying on a small set of regressions, we prefer to base our results on the existing literature with the objective of relying on robust relationships between growth and its determinants.

We start from the results of Sala-i-Martin et al. (2004) as a benchmark measurement of the relevance of each of the six factors. Sala-i-Martin et al. use the sixty-seven candidate regressors chosen by Sala-i-Martin (1997) to estimate all possible combinations of a growth regression with seven explanatory variables and they use Bayesian updating methods to compute the posterior probability that the coefficient of a particular variable is non-zero. Based on this methodology, they rank regressors by their potential significance, and we focus on those factors with high significance that have been most studied by economists and political scientists: initial income per capita, size of the government, macroeconomic stability, terms of trade, participation in international trade, democracy, and institutional quality.7 Although the interpretation of some of these variables remains open to debate,8 there is a growing consensus that the above categories, in one form or another, are all important determinants of economic growth.

Initial Income Per Capita and Convergence

The negative relation between initial income per capita and economic growth is among the most robust in the empirical literature. The relation—known as beta-convergence after the customary Greek letter used as a regression coefficient—is grounded in the neoclassical growth theory of Solow (1956) and Swan (1956). The model yields predictions for the path of output but it is important to note that convergence depends on the rates of population growth, capital depreciation, and technological progress, as well as the elasticities of output to the various factors of production. The Solow–Swan model is said indeed to predict conditional convergence. Mankiw, Romer, and Weil (1992) estimated a regression on OECD countries that yielded a speed of convergence of 2 percent. Abreu, de Groot, and Florax (2005) conducted a meta-analysis of beta-convergence and examined the results of 619 different growth regressions from forty-eight different published papers. They found that taking stock of country differences matters for the estimated rate of convergence, which is exactly what the Solow growth model predicts. Therefore, it is difficult to attribute a large fraction of growth to convergence for very heterogeneous economies. We thus apply the commonly estimated 2 percent convergence rate only within the GCC. We apply a similar rate of convergence within non-GCC oil exporters, within non-OECD countries and within OECD countries. Figure 2.6 shows our decomposition of TFP by factors. Qatar and the UAE were the richest countries in the GCC and therefore conditional convergence would have slowed down their growth. On the contrary, countries like starting from lower levels of income, such as Sudan or China, would have benefitted from conditional convergence.

Size of the Government

The size of the public sector is potentially an important factor in growth performance, and Barro (1991) had already noted that the coefficient on government consumption was negative in growth regressions. Sachs and Warner (1995) also argued that one of the possible reasons why resource-endowed countries have tended to grow slower than resource-poor countries has to do with outsized governments. Artadi and Sala-i-Martin (2002) applied this argument to MENA and claimed that the large income that these governments receive—and subsequently spend—from oil revenues creates rent-seeking behavior. Rather than concentrate their efforts in productive activities, the incentives are for agents to concentrate on securing as big a share as possible from the oil revenues. For the GCC, the government is clearly a driving force in the economy, in particular for factor accumulation: investment, immigration, and educational improvements are to a large extent financed by the government. The argument is however that from the supply side, TFP (in the long run) is hurt by rent-seeking activities, and unproductive government consumption is a good proxy for this effect.

Standard regression estimates did not fully support the findings of Barro (1991) (Levine and Renelt 1992; Nijkamp and Poot 2004) but the formal Bayesian approach of Sala-i-Martin et al. (2004) found nonetheless that the effect of government share of consumption is significant and economically important: a ten percentage points increase in the share of government consumption over GDP would reduce annual growth by 0.4 percent. We apply this coefficient in our estimates, using the ratio of government consumption to non-oil GDP for oil producers, and find that for Qatar, Libya, and to some extent Kuwait, government consumption was large enough that it may have been a drag on TFP (Figure 2.6). The impact is relatively small for the other GCC countries.

Inflation and Macroeconomic Stability

High inflation, volatile export revenues, or variable economic growth add to the risks faced by agents and thereby worsen the prospects of economic activities that pay off in the future. These uncertain environments are therefore detrimental to investment, to skills learning, and in general to the development of businesses. Empirical studies have indeed found a significant statistical relationship between inflation and growth, even after controlling for fiscal performance, wars, droughts, population growth, openness, human and physical capital, and after allowing for simultaneity bias. Based on a cross-country regression of 101 countries over 1960–89, Fischer (1993) found that high inflation reduces output growth by reducing investment and productivity growth. Using annual data for eighty-seven countries over 1970–90, Sarel (1996) found evidence of a structural break at an 8 percent inflation rate. Inflation and growth are positively correlated below 8 percent but negatively correlated above that, suggesting that ignoring this nonlinearity would significantly underestimate the impact of inflation on growth. Ghosh and Phillips (1998) confirmed the presence of nonlinearities and found that increasing inflation from the optimal level (2–3 percent) to 5 percent reduces annual growth by 0.3 percent. Khan and Senhadji (2001) and Espinoza, Leon, and Prasad (2012) reexamined this result and found that for developing countries and for oil producers, inflation becomes costly only when it exceeds 10 percent. We choose the nonlinear specification adopted by both these papers in our application. For the GCC, inflation was lower than 10 percent during the period under study, and therefore it is unlikely that poor macroeconomic management should be a cause of low TFP in the region, contrary to the experience of Sudan and Venezuela (Figure 2.6).

Volatility and Growth

Output variability and exports or terms-of-trade volatility are the other commonly used measures of macroeconomic stability, and since Ramey and Ramey (1995) the literature has confirmed the negative relationship between growth and volatility. Hnatkovska and Loayza (2005) estimated that a one point increase in the volatility of GDP per capita decreases annual growth by around 0.3. Hnatkovska and Loayza (2005) also found that the negative link is stronger for poor countries, for countries with underdeveloped institutions, for countries with intermediate levels of financial development, and for countries that do not implement countercyclical fiscal policies. Similarly, Kose et al. (2006) suggested that the relationship between growth and output volatility is positive at high levels of income but negative at low levels of income, and depends on trade and financial liberalization. Loayza et al. (2007) provide a survey of some recent findings. Van der Ploeg and Poelhekke (2009) also found that a one point increase in volatility reduces growth by around 0.3 percentage points. They reinterpreted these findings for resource-rich countries and argued that volatility is indeed the key channel for the resource curse, dwarfing the Dutch disease effect that had been at the center of the literature. We apply a coefficient of 0.3 in our analysis and find that macroeconomic volatility could be a significant drag on growth for all oil producers. The standard deviation of growth per capita in Kuwait,9 Qatar, and Saudi Arabia exceeded 5 percent since 1992.

Openness to Trade

Barro (1991) argued that price distortions significantly affected growth, and a growing literature has confirmed that trade openness is positive for growth (Sachs and Warner 1995; Edwards 1998), despite some critics (Rodriguez and Rodrik 2001). The recent literature has suggested that the effect of trade openness on growth is positive albeit small (for instance, Lee, Ricci, and Rigobon 2004; Billmeier and Nannicini 2009). Mookerjee’s (2006) meta-analysis harvested the results of ninety-five regressions from seventy-six different papers on trade and growth and confirmed the near unanimous consensus among economists that exports orientation is positive for growth.10 However, Makdisi et al. (2006) suggest that the link between trade and growth is weaker in MENA than it is on average. They use the same measure of openness as Sala-i-Martin et al. (2004) but find that the marginal effect of openness in MENA is only one fourth of what it is in other countries. However, it is not clear this result holds for the GCC, which had a very different tariff regime than the rest of MENA in most of the period covered. We therefore follow Sala-i-Martin et al. (2004) who find that for every four more years a country remained open in the period 1950–94, GDP growth was higher by 0.12 percent per year. The average growth of terms of trade has also been clearly associated with growth. Mendoza (1997) found that an increase of one percentage point in the terms of trade increases consumption growth by 0.05 percent, a result we take into account in our calculations. Terms of trade have been moving favorably for oil producers in the last twenty years, but the results shown in Figure 2.6 suggest that the effect was not strong enough to improve TFP in any GCC country.


As standard growth models failed at explaining the differences in growth across countries, the literature investigated whether non-economic factors, in particular institutions, could explain growth performance. Institutional quality is notoriously difficult to measure. Different freedom indices—political freedom, economic freedom, democracy—tend to be highly correlated to each other and insignificantly correlated to growth. Sala-i-Martin et al. (2004), for example, find that an index of political rights is not significantly correlated to growth. In a meta-analysis of the growth–democracy nexus, Doucouliagos and Ulubasoglu (2008) find that three quarters of the 470 regressions they examine do not have a positive, robust correlation between democracy and economic growth. In fact, their meta-analysis does not find any direct effects from democracy to growth.

Even though political regime variables were not found to correlate with growth, narrower measures of institutional quality were found to consistently correlate with growth. Economists and political scientists have accumulated theoretical arguments and empirical results showing that bureaucracy, corruption, or the strength of property rights matter for the growth process. However, it is important to use institutional measures parsimoniously in order to avoid multicollinearity problems. Acemoglu and Johnson (2005), for example, found that a variable capturing “contracting institutions” is significant only when the variable for “property rights institutions” is not included. Once both institutional measures are included as regressors, “contracting institutions” loses explanatory power. If the initial institutional measure is chosen appropriately, the marginal information provided by additional institutional regressors is unlikely to be significant. This is why we focused on a single variable measuring the quality of institutions, and opted for the measure of corruption of the International Country Risk Guide (ICRG), which is available for a wide range of countries. The claim is not that corruption is the only relevant feature for growth, but rather that high levels of corruption are surely symptomatic of institutional breakdown. Indeed, in the ICRG database, the correlation between the corruption and bureaucracy indices is 80 percent, and the correlation between corruption and the ICRG’s composite rating is 75 percent.11

Ugur and Nandini (2011) conducted a meta-analysis from seventy-two different studies on corruption and economic growth. They found that corruption retards growth both directly as well as through a decline in human capital and a worsening of public finance. More importantly, they also found that corruption has a stronger negative effect in middle- and high-income countries than it does in low-income ones. Using the full sample of countries, they find that a one-point drop in the corruption index corresponds to an increase in the annual growth rate of 0.86 percent.12 We keep this coefficient for our estimates. In several countries in the GCC (Qatar, Saudi Arabia, and the UAE), the ICRG graded corruption to be worse than in the median country in our sample. As a result, the analysis suggests that poor institutions in these countries could have accounted for around −0.1 to −0.6 percentage points, per year, to the disappointing TFP of the period 1990–2008.

We conclude by summarizing our findings presented in Figure 2.4 and in Figure 2.6. Total factor productivity growth has been disappointing in the GCC in the last twenty years, as it has in many other oil exporters—although with positive growth numbers in GDP it is difficult to point at a resource curse. The emerging consensus in the growth literature is that high initial income per capita, oversized governments, instable macroeconomic environments, restrictions on trade, declining terms of trade, and poor institutions can explain declining TFP.

The GCC has had stable inflation and benefitted from terms-of-trade and trade policies in the region that were favorable to exports. However, relatively poor institutions (especially in Saudi Arabia and the UAE), oversized governments (in Kuwait and the UAE), and volatile growth in the region would have contributed negatively to TFP. Although we attempted to take into account the major lessons learnt from the growth literature, a significant part of declining TFP remains unexplained, especially in Bahrain, Oman, and the UAE. In Qatar, on the contrary, TFP was stable despite several factors that could have harmed growth (in particular the share of public consumption in GDP and the relatively poor quality of institutions).

References GrootH. L. F. and FloraxR. J. G. M. (2005). “A Meta-Analysis of ß-Convergence: The Legendary 2%Journal of Economic Surveys19 (3): 389420.

    • Search Google Scholar
    • Export Citation

    AcemogluD. and JohnsonS. (2005). “Unbundling institutionsJournal of Political Economy113 (5): 94995.

    AliA. A. G. (2003). “Globalization and inequality in the Arab region.Kuwait: Arab Planning Institute.

    BarroR. (1991). “Economic growth in a cross section of countriesQuarterly Journal of Economics106 (2): 40743.

    BarroR. and LeeJ. W. (2001). “International data on educational attainment: Updates and implications” Oxford Economic Papers53 (3): 54163.

    • Search Google Scholar
    • Export Citation

    BarroR. and LeeJ. W. (2010). “A new data set of educational attainment in the world, 1950–2010.” NBER Working Paper No. 15902.

    • Search Google Scholar
    • Export Citation

    BarroR. and Sala-i-MartinX. (2004). Economic Growth2nd edn. Cambridge, Mass.: MIT Press.

    BillmeierA. and NanniciniT. (2009). “Trade openness and growth: Pursuing empirical glasnostIMF Staff Papers56: 44775.

    CaselliF. (2005). “Accounting for income differences across countries” in P.Aghion and S.Durlauf (eds) Handbook of Economic Growth Volume 1A. New York: North-Holland679741.

    • Search Google Scholar
    • Export Citation

    CaselliF. and WilsonD. J. (2004). “Importing technologyJournal of Monetary Economics51: 132.

    DoucouliagosH. and UlubasogluM. H. (2008). “Democracy and economic growth: A meta-analysisAmerican Journal of Political Science52: 6183.

    • Search Google Scholar
    • Export Citation

    EdwardsS. (1998). “Openness, productivity, and growth: What do we really know?Economic Journal108 (2): 38398.

    EspinozaR.LeonH. and PrasadA. (2012). “When should we worry about inflation?World Bank Economic Review26 (1): 10027.

    • Search Google Scholar
    • Export Citation

    FeenstraR. (2000). “World trade flows, 1980–1997.Mimeo. Center for International DataUniversity of California Davis.

    FelipeJ. (1997). “Total factor productivity growth in East Asia: A critical survey.ADB EDRC Report Series No. 65.

    FischerS. (1993). “The role of macroeconomic factors in economic growthJournal of Monetary Economics32: 485512.

    GhoshA. and PhillipsS. (1998). “Warning: Inflation may be harmful to your growthIMF Staff Papers45: 672710.

    HakuraD. (2004). “Growth in the Middle East and North Africa.IMF Working Paper 04/56.

    HallR. E. and JonesC. I. (1999). “Why do some countries produce so much more output per worker than others?Quarterly Journal of Economics114: 83116.

    • Search Google Scholar
    • Export Citation

    HestonA.SummersH. A. R. and AtenB. (2011). “Penn World Table Version 7.0.” Center for International Comparisons of Production, Income and PricesUniversity of Pennsylvania.

    • Search Google Scholar
    • Export Citation

    HnatkovskaV. and LoayzaN. (2005). “Volatility and growth,” in L.Aizenman and B.Pinto (eds) Managing Economic Volatility and Crises: A Practitioner’s Guide. Cambridge: Cambridge University Press65100.

    • Search Google Scholar
    • Export Citation

    KhanM. S. and SenhadjiA. (2001). “Threshold effects in the relationship between inflation and growthIMF Staff Papers48: 121.

    • Search Google Scholar
    • Export Citation

    KoseM. A.PrasadE. S.TerronesM. E. (2006). “How do trade and financial integration affect the relationship between growth and volatility?Journal of International Economics69: 176202.

    • Search Google Scholar
    • Export Citation

    LeeH. YRicciL. A. and RigobonR. (2004). “Once again, is openness good for growth?Journal of Development Economics75: 45172.

    • Search Google Scholar
    • Export Citation

    LevineR. and ReneltD. (1992). “A sensitivity analysis of cross-country growth regressionsAmerican Economic Review82: 94263.

    • Search Google Scholar
    • Export Citation

    LoayzaN.RancièreR.ServénL. and VenturaJ. (2007). “Macroeconomic volatility and welfare in developing countries: An introductionWorld Bank Economic Review21 (3): 34357.

    • Search Google Scholar
    • Export Citation

    MakdisiS.FattahZ. and LimamI. (2006). “The determinants of economic growth in the MENA region,” in J.Nugent and M. H.Pesaran (eds) Explaining Growth in the Middle East Contributions to Economic Analysis 278. Amsterdam: Elsevier3160.

    • Search Google Scholar
    • Export Citation

    MankiwN. G.RomerD. and WeilD. N. (1992). “A contribution to the empirics of economic growth,Quarterly Journal of Economics107: 40737.

    • Search Google Scholar
    • Export Citation

    MendozaE. G. (1997). “Terms-of-trade uncertainty and economic growthJournal of Development Economics54: 32356.

    MookerjeeR. (2006). “A meta-analysis of the export growth hypothesisEconomics Letters91: 395401.

    NijkampP. and PootJ. (2004). “Meta-analysis of the effect of fiscal policies on long-run growthEuropean Journal of Political Economy20: 91124.

    • Search Google Scholar
    • Export Citation

    PloegF.vander and PoelhekkeS. (2009). “Volatility and the natural resource curseOxford Economic Papers61: 72760.

    PsacharopoulosG. (1994). “Returns to investment in education: A global updateWorld Development22 (9): 132543.

    RameyV. and RameyG. (1995). “Cross country evidence on the link between volatility and growthAmerican Economic Review85 (5): 113859.

    • Search Google Scholar
    • Export Citation

    RodriguezF. and RodrikD. (2001). “Trade policy and economic growth: A skeptic’s guide to the cross-national evidence,” in B.Bernanke and K.Rogoff (eds) NBER Macroeconomics Annual 2000. Cambridge, Mass.: MIT Press261338.

    • Search Google Scholar
    • Export Citation

    SachsJ. D. and WarnerA. (1995). “Economic reform and the process of global integrationBrookings Papers in Economic Activity1: 1118.

    • Search Google Scholar
    • Export Citation

    SachsJ. D. and WarnerA. (2001). “The curse of natural resourcesEuropean Economic Review45 (4–6): 82738.

    Sala-i-MartinX. (1997). “I just ran two million regressionsAmerican Economic Review87 (2) Papers and Proceedings of the 104th Annual Meeting of the American Economic Association: 17883.

    • Search Google Scholar
    • Export Citation

    Sala-i-MartinX. and ArtadiE. (2002). “Economic growth and investment in the Arab world,” in P. K.Cornelius (ed.) The Arab World Competitiveness Report 2002–2003 World Economic Forum. New York: Oxford University Press2233.

    • Search Google Scholar
    • Export Citation

    Sala-i-MartinX.DoppelhoferG. and MillerR. I. (2004). “Determinants of long-term growth: A Bayesian averaging of classical estimates (BACE) approachAmerican Economic Review94 (4): 81335.

    • Search Google Scholar
    • Export Citation

    SarelM. (1996). “Nonlinear effects of inflation on economic growthIMF Staff Papers43: 199215.

    SenhadjiA. (2000). “Sources of economic growth: An extensive growth-accounting exerciseIMF Staff Papers47: 12957.

    SolowR. M. (1956). “A contribution to the theory of economic growthQuarterly Journal of Economics70 (1): 6594.

    SolowR. M. (1957). “Technical change and the aggregate production functionReview of Economics and Statistics39 (3): 31220.

    • Search Google Scholar
    • Export Citation

    SwanT. W. (1956). “Economic growth and capital accumulationEconomic Record32 (2): 33461. Country world map available online at <> accessed May312013.

    • Search Google Scholar
    • Export Citation

    UgurM. and NandiniD. (2011). “Corruption and economic growth: A meta-analysis of the evidence on low-income countries and beyond.MPRA Working Paper 31226.

    • Search Google Scholar
    • Export Citation
1Nominal GDP can itself be compiled in different ways. From the production side, nominal GDP is the sum of the value of output of all sectors (typically calculated by surveying companies’ sales and inventories), subtracting the cost of intermediate inputs (GDP is a measure of value added). From the expenditure side, nominal GDP is the sum of consumption, private-sector investment, public expenditure, and exports minus imports. However, the production method is more reliable in the region as surveys on spending are incomplete.
2Non-oil GDP series from the IMF World Economic Outlook and Regional Economic Outlook for the Middle East and Central Asia start in 1990. Data pre-1990 was collected manually using older individual IMF country reports.
3The UN reports data on value added at constant prices, excluding the mining and the utilities sectors. The utilities and non-oil mining sectors in the GCC are quite small though, so this measure is a good proxy of non-oil growth.
4One would normally want to remove the component of investment that is used in the oil sector, but this data is not available for a cross-section of countries. We therefore abstract from the difference between non-oil investment and total investment. For Saudi Arabia, oil investment represents around 10 percent of total investment only.
5An initial level of capital for 1960 and an assumption for δ are needed to compute the series. We use the steady-state relationship between capital and investment K0=I0/(g + δ), where g is the average growth rate in the data and I0 is investment in year 1960 (we use year 1965 for the GCC). For the non-GCC countries, if investment data in 1960 is not available, it is extrapolated from investment to GDP ratios using a linear time trend. Caselli (2005) argued the initial level of capital stock and that the specific value for δ did not affect dramatically the performance of the model in explaining the dispersion of income per capita. However, this does not imply that the assumptions are innocuous for the diagnostics of growth for individual countries.
6The data for Oman was not available. The World Development Indicators database of the World Bank provides numbers for the literacy rate in all the countries in the GCC, and a regression showed that the sensitivity of educational attainment to the literacy rate is 0.12 in the GCC. This coefficient was used to estimate educational attainment in Oman of 5.4 years in 1995, 6 years in 2000, 6.5 years in 2005, and 7.2 years in 2010.
7Since we focus on TFP, we do not discuss the role of physical capital or of education. These have been covered in section 2.4 using estimates in line with the growth-accounting literature. Among the 25 most significant variables in Sala-i-Martin et al. (2004), we do not discuss the variables that are geographic, religious, or ethnic dummies (East Asian dummy, African dummy, Latin American dummy, fraction of tropical area, Spanish colony, fraction Confucian, fraction Muslim, fraction Buddhist) as they are not fundamentally informative about the growth process (they merely capture omitted variables). Neither do we discuss initial life expectancy because it is highly correlated with initial income per capita.
8For example, institutional quality might be measured as the strength of property rights, the presence of democratic institutions, or the level of bureaucracy.
9We compute the volatility after the First Gulf War for this exercise.
10There are typically two ways to measure an economy’s “openness” or its disposition to trade: one is a measure of net exports (or exports plus imports) as a share of GDP, the other one is the number of years an economy is open, using Sachs and Warner’s (1995) binary index of openness. For this index an economy is considered closed—and the index is set to zero—if it satisfies at least one of the following five criteria: nontariff barriers covering 40 percent or more of trade, average tariff rates of 40 percent or more, a black market exchange rate that is depreciated by over 20 percent relative to the official rate, a socialist economic system, or a state monopoly over major exports. If an economy does not exhibit any of the above traits it is considered open and receives a score of one. Sala-i-Martin et al. (2004) noted that the numbers of years the economy is open is statistically more robust than the ratio of trade to GDP in growth regressions.
11We prefer the corruption index to the composite rating because the former is the one surveyed by the meta-analysis of Ugur and Nandini (2011). We use this meta-analysis to calibrate the impact of corruption on growth.
12This coefficient is practically unchanged when removing the effect via human capital and physical capital accumulation, which are arguably already taken into account in the growth-accounting exercise.

    Other Resources Citing This Publication