Chapter

Chapter 3. Determinants of Growth in Cambodia and Other Low-Income Countries in Asia: Evidence from Country Panel Data

Author(s):
Sumio Ishikawa, Sibel Beadle, Damien Eastman, Srobona Mitra, Alejandro Lopez Mejia, Wafa Abdelati, Koji Nakamura, Il Lee, Sònia Muñoz, Robert Hagemann, David Coe, and Nadia Rendak
Published Date:
February 2006
Share
  • ShareShare
Show Summary Details
Author(s)
Wafa Fahmi Abdelati

This chapter explores the determinants of long-term sustainable growth in Cambodia from a cross-country analysis of the sources of growth. Section A describes Cambodia’s recent economic growth performance and compares it with other low-income countries (LICs) for the period from 1970 to 2003. Section B gives an overview of the differences between Cambodia and some country groupings with respect to a number of growth determinants identified in the literature. Section C presents estimation results based on seven five-year-period averages for the 144 countries. Using these results, we consider the implications for steady-state growth in Cambodia by comparing its performance relative to that of countries in the Association of Southeast Asian Nations (ASEAN).

A. Cambodia’s Growth Experience and Prospects

In the last five years, Cambodia’s growth performance has been among the best of LICs. Cambodia’s GDP growth rate has averaged 6–7 percent during 1999–2003, reflecting both external factors and good macroeconomic policies. Per capita GDP growth was much lower, however, at an average 3.8 percent (Table 3.1). This growth performance is significantly higher than the average for all developing countries and for LICs.5

Table 3.1.Real GDP Per Capita Growth1(Annual average, in percent)
1970–20031999–2003
All (144)1.51.6
PRGF (70)1.01.5
LIC-nonfuel (68)1.01.0
Asia (23)2.82.1
Asia excluding China (22)2.61.8
Asia excluding islands (18)3.12.8
Asia-LIC (15)2.42.0
Transition (29)2.14.7
Transition LIC (13)1.54.3
ASEAN (9)3.43.2
ASEAN LIC (4)3.14.4
Cambodia3.53.8
Lao P.D.R.2.93.4
Vietnam3.73.9
Source: IMF, World Economic Outlook database.Notes: ASEAN=Association of Southeast Asian Nations; LIC=low-income countries; PRGF=Poverty Reduction and Growth Facility.

Excludes advanced economies. Asia and ASEAN exclude Brunei Darussalam.

Source: IMF, World Economic Outlook database.Notes: ASEAN=Association of Southeast Asian Nations; LIC=low-income countries; PRGF=Poverty Reduction and Growth Facility.

Excludes advanced economies. Asia and ASEAN exclude Brunei Darussalam.

Cambodia’s pace of growth is similar to that of other ASEAN low-income countries (Myanmar, Vietnam, and Lao P.D.R.) and transition countries.6

However, prospects for weaker growth in the period ahead call for a deeper exploration of the factors that can contribute to sustained growth. As will be disclosed in Chapters 4 and 5, Cambodia has benefited from preferential access to the United States since 1996 under the Multifiber Agreement. With the elimination of the quota system in January 2005, growth is expected to slow down as its garment industry will be exposed to direct competition with neighboring countries. Cambodia’s low labor productivity, inadequate and expensive infrastructure, and a cumbersome regulatory environment—as confirmed by recent World Bank value chain studies and investment climate assessment—do not bode well for future sustainable growth (World Bank, 2004). Identification of key impediments to growth has become an urgent agenda.

B. Overview of Growth Determinants

We begin the analysis by assessing Cambodia’s performance for the period 1970–2001 against a number of factors that have been positively associated with growth. These include initial conditions, macroeconomic polices, improvements in human and physical capital, institutional factors, and other exogenous factors.

  • Cambodia’s initial conditions in 1970 are among the weakest of LICs. In 1970, it had one of the lowest per capita GDP in purchasing power parity (PPP) terms, about one-third that of other LICs (Figure 3.1). Life expectancy, which is one indicator of human capital conditions, was about 41 years compared to the Asian average of 54 years.

  • Physical and human capital development has been anemic. Physical infrastructure, as proxied by the number of telephones per thousand inhabitants, has remained very low and only began to increase in the last decade (Figure 3.2). Illiteracy rates have remained high, particularly compared to the fast-growing economies of ASEAN and transition countries.

  • Macroeconomic policy indicators, on the other hand, have been better than the average for ASEAN and for transition economies. Accordingly, inflation has remained relatively subdued, and budget balances within the range for LICs (Figure 3.3).

  • Favorable external conditions, including foreign aid flows and trade agreements, have helped propel recent growth. Cambodia’s per capita aid has amounted to 10 percent of per capita GDP in the period 1970–2001 and has increased to 20 percent in the period 1995–2001 (Figure 3.4). Moreover, its terms of trade have remained relatively favorable and stable (Figure 3.5).

  • As with other transition economies, Cambodia lags in institutional capacity and its markets are underdeveloped. Financial markets remain shallow, with bank credit to the private sector at around 7 percent of GDP, and the ratio of broad money to GDP under 20 percent (Figure 3.6).

Figure 3.1.Initial Conditions, 1970

Source: IMF, World Economic Outlook database.

Notes: ASEAN=Association of Southeast Asian Nations; LIC=low-income countries; PRGF=Poverty Reduction and Growth Facility; PPP=purchasing power parity.

Figure 3.2.Human and Physical Capital

(Average 1970–2001)

Sources: IMF, World Economic Outlook; and World Bank, World Development Indicators.

Notes: ASEAN=Association of Southeast Asian Nations; LIC=low-income countries; PRGF=Poverty Reduction and Growth Facility; PPP=purchasing power parity.

Figure 3.3.Macroeconomic Policies

(In percent, average 1970–2001)

Sources: IMF, World Economic Outlook; and World Bank, World Development Indicators.

Notes: ASEAN=Association of Southeast Asian Nations; LIC=low-income countries; PRGF=Poverty Reduction and Growth Facility; PPP=purchasing power parity.

Figure 3.4.Aid Per Capita

(Aid flows as percent of GDP, average 1970–2001)

Sources: IMF, World Economic Outlook; and World Bank, World Development Indicators.

Notes: ASEAN=Association of Southeast Asian Nations; LIC=low-income countries; PRGF=Poverty Reduction and Growth Facility; PPP=purchasing power parity.

Figure 3.5.Terms-of-Trade Volatility

(Average 1970–2003)

Sources: IMF, World Economic Outlook; and World Bank, World Development Indicators.

Notes: ASEAN=Association of Southeast Asian Nations; LIC=low-income countries; PRGF=Poverty Reduction and Growth Facility; PPP=purchasing power parity.

Figure 3.6.Financial Development Indicators

(In percent, average 1970–2001)

Sources: IMF, World Economic Outlook; and World Bank, World Development Indicators.

Notes: ASEAN=Association of Southeast Asian Nations; LIC=low-income countries; PRGF=Poverty Reduction and Growth Facility; PPP=purchasing power parity.

Weak governance has become the Achilles’ heel for growth as transition economies, including Cambodia, increasingly depend on private sector development. An earlier World Bank cross-country study shows Cambodia as weaker than most developing countries on a number of different governance indicators (Table 3.2). The average index shows that, overall, Cambodia scores lower than the average for Asian LICs and transition LICs, and well below the ASEAN average in each of the six different indicators.

Table 3.2.Governance Indicators
AllPRGFAsiaAsia-LICTransitionTransition LICASEANLao P.D.R.VietnamCambodia
Combined index1–0.2–0.4–0.2–0.4–0.2–0.5–0.1–0.6–0.4–0.7
Voice and accountability–0.2–0.4–0.3–0.4–0.2–0.6–0.6–1.0–1.2–0.7
Political stability–0.1–0.4–0.1–0.30.1–0.20.21.00.4–1.1
Government effectiveness–0.3–0.6–0.1–0.3–0.3–0.50.2–0.1–0.2–0.7
Lack of regulatory burden–0.2–0.4–0.1–0.4–0.2–0.60.1–1.1–0.5–0.3
Rule of law–0.3–0.6–0.2–0.6–0.3–0.6–0.1–1.3–0.5–0.9
Control of corruption–0.3–0.6–0.3–0.6–0.3–0.6–0.2–0.9–0.6–0.9
Source: Kaufman, Kraay, and Zoido-Lobatón (1999).Notes: ASEAN=Association of Southeast Asian Nations; LIC=low-income countries; PRGF=Poverty Reduction and Growth Facility.

Each of the six governance indicators is measured in units ranging from –2.5 to 2.5, with higher values corresponding to better governance outcomes.

Source: Kaufman, Kraay, and Zoido-Lobatón (1999).Notes: ASEAN=Association of Southeast Asian Nations; LIC=low-income countries; PRGF=Poverty Reduction and Growth Facility.

Each of the six governance indicators is measured in units ranging from –2.5 to 2.5, with higher values corresponding to better governance outcomes.

C. Results from Econometric Analysis and Implications for Cambodia

A growing literature has focused on the theoretical and empirical investigation of the impact of policies and conditioning factors on the steady-state rate of growth. Empirical investigation has generally taken the form of either comparative regression analysis, growth accounting, or, more recently, a combination of the two.7 The growth accounting approach estimates the contribution of capital accumulation and improvements in total factor productivity, but does not capture the influence of economic policies and external factors (such as changes in the terms of trade). The more eclectic cross-country approach, inspired by the theory of endogenous growth, attempts to explain differences in growth rates by a wide range of macroeconomic, structural, and external factors. Attempts have been made to combine both approaches by adding factor contributions and conditioning factors to estimate their contributions to growth in the same equation, or to estimate the influence of policies and conditioning factors on the rate of human and physical capital accumulation and, thus, on growth.8

For developing countries, two cross-country studies by IMF staff have focused on drawing lessons from the impact of macroeconomic and structural policies on differences in growth rates across countries. A 1999 study investigated the impact of macroeconomic and structural policies on growth in 84 low- and middle-income nontransition countries, subdividing them into PRGF-countries and non-PRGF countries (IMF, 1999). The study found that the gap between the growth rates of PRGF and non-PRGF countries has narrowed, and confirmed the positive role that good policies—single-digit inflation, low budget deficits, outward-oriented policies, and streamlined governments—can play in improving growth. A 2003 study covering 94 countries, including 69 low-income countries, found that institutional quality has a more significant impact on growth and performs better than macroeconomic policy variables (with the exception of trade openness) in explaining the differences in the level of income, in growth rates, and in the volatility of growth.9 Another 2003 regional study that combined the growth accounting approach and institutional quality for a cross-country sample of 74 countries, including 53 low- and middle-income countries, found the lower growth rates of Middle Eastern countries can be explained by the larger size of government, poor quality of institutions, misalignment of the real exchange rate, terms of trade volatility, and barriers to trade.10

Cambodia and other transition economies have been typically excluded from cross-country studies of long-term growth. One reason might have been that structural rigidities and weak influence of market forces were thought to make it difficult to distinguish the role of macroeconomic policies in promoting capital accumulation and productivity growth. More practically, however, data shortcomings have precluded the inclusion of these countries, either because many have become independent states only since the early 1990s, or because earlier data collection methods were deemed unreliable. Accordingly, while some improvements have been made in data quality, the results of the following analysis still reflect such weaknesses. Nevertheless, inclusion of these countries would produce more relevant results for assessing Cambodia’s medium-term growth prospects.

Estimation Approach

A standard growth model was estimated for 144 low- and middle-income countries. Transition economies are included where data permit. The countries (denoted by i) include 71 LICs, of which 15 are Asian LICs. Variables are averaged for seven five-year periods (denoted by t), with the seventh period ranging from two to four years, depending on data availability. Real per capita GDP growth is the key dependent variable, and growth in labor productivity is used as an alternative dependent variable.11 The explanatory variables and their means are described in Appendix Table A3.1 and the countries are listed in Appendix Table A3.2 in the appendix to this chapter. The basic regression takes the following form:

A number of estimation methodologies are used to test the robustness of the coefficients. Ordinary least squares (OLS) has been typically used with either annual pooled data or period averages. Using the panel data organized in seven five-year period averages, we use a random effects application while assuming that the independent variables are independent of the unobserved individual country effects (μit) and the true disturbance term (υit) for all i and t.12 A key concern is the endogeneity of macroeconomic and institutional factors and their possible correlation with the unobserved omitted factors. This could be addressed by using two-stage least squares with appropriate instrumental variables for the endogenous explanatory variables, but it is typically difficult to obtain good instruments for these variables.13 In the absence of readily available instruments, we used three approaches in addition to OLS. The first is a generalized least squares (GLS) estimation that allows for heteroscedastic effects between the country panels. The second is the Hausman-Taylor estimation method whereby some variables are designated as exogenous and used to instrument for variables suspected to be endogenous. The third alternative uses the lagged dependent variable along with first differences of the independent variables and applies the Arellano-Bond estimator.

Results

The results shown in Table 3.3 are consistent with other studies in the literature. The empirical analysis confirms that higher real per capita growth is associated with lower initial income levels, better macroeconomic performance, faster human and physical capital accumulation, smaller government, and stronger institutions and governance. Variations in trade openness and trade restrictiveness did not yield significant coefficients for explaining growth performance. Labor force growth had the wrong sign, possibly due to widespread unemployment and underemployment.14 Similar results are obtained when labor productivity growth (change in output per worker) is used as the dependent variable. While there is room to further improve variable measurement and estimation methods, overall, the results are useful for illustrating the implications of the determinants of sustained growth for Cambodia.

Table 3.3.Summary Regression Results for Panel Data
(1)(2)(3)(4)(5)(6)(7)(8)
OLS (robust SE)GLS (Hetero panel)RE (Hausman-Taylor)Arellano-Bond Estimator
Dependent VariablerGr_PCGr_Y/LrGr_PCGr_Y/LrGr_PCGr_Y/LrGr_PCGr_Y/L
Initial period GDP (log)–0.755–0.807–0.555–0.424–1.957–1.989D1–6.174–5.016
0.252***0.248***0.190***0.179**0.595***0.589***1.000***1.005***
Labor force growth–0.340–0.669–0.243–0.622–0.250–0.634D1–0.076–0.334
0.119***0.141***0.098**0.094***0.133*0.131***0.1860.182*
Log of inflation–0.974–1.048–0.743–0.824–0.849–1.011D1–0.990–1.052
0.143***0.137***0.071***0.087***0.148***0.145***0.198***0.196***
Government consumption to GDP–0.112–0.094–0.128–0.118–0.110–0.110D1–0.151–0.141
0.023***0.025***0.010***0.015***0.033***0.032***0.055***0.054*
Terms of trade change, lagged
0.0320.0330.0210.0170.0270.027D10.0510.040
0.018*0.019*0.011**0.010*0.015*0.0150.018***0.018**
Terms of trade volatility0.0000.0000.0000.0000.0000.000D10.0000.000
0.0000.0000.0000.0000.0000.0000.0010.001
Weather: crop decline–3.318–3.871–1.924–2.682–3.578–3.880D1–4.362–4.826
0.799***0.734***0.411***0.420***0.711***0.700***0.983***0.976***
Broad money to GDP0.000–0.013–0.0010.0030.003–0.006D1–0.018–0.042
0.0290.0280.0170.0170.0290.0280.0460.046
Aid per capita, as percent of GDP per capita0.0300.0310.0410.0510.0850.064D10.0690.018
0.0200.0210.012***0.015***0.027***0.026**0.041*0.041
Telephones per thousand0.0020.0020.0020.0010.0010.002D10.0090.004
0.0030.0030.0020.0020.0030.0030.005*0.005
Trade restrictiveness index–0.0110.031–0.055–0.016–0.0200.025D1–1.273–1.574
0.0580.0620.0370.0370.1740.1760.9600.950
Gross capital formation to GDP0.1280.1750.1410.149(EN)0.1380.197D10.1940.289
0.023***0.040***0.014***0.015***0.025***0.025***0.038***0.038***
Trade to GDP–0.009–0.004–0.008–0.010(EN)–0.003–0.004D10.0080.011
0.0160.0160.0090.0090.0140.0140.0250.025
Secondary school enrollment
0.0190.0000.010–0.0030.0450.023D1(dropped)(dropped)
0.010**0.0100.006*0.0060.019**0.019
Dummy for fuel exporters–0.449–0.408–0.925–0.2960.9410.818D1(dropped)(dropped)
0.5860.5950.377**0.4411.4261.434
Government efficiency1.4501.5351.5441.436(EN)2.2742.101D1(dropped)(dropped)
0.338***0.309***0.202***0.208***1.358*1.358
Lagged dependent variableLD–0.0010.022
0.0500.049
Constant9.3489.9356.1416.21514.71815.8901.2061.035
1.9071.9481.3331.2714.5344.4920.226***0.229***
Number of observations640640640640640640463462
R-squared0.2990.367
rho0.6160.629
ST40.65(14)39.54(14)
Sources: Author’s estimates based on IMF, World Economic Outlook; and World Bank, World Development Indicators.Notes: Standard errors in italics. Significance of the coefficients at the 1, 5, and 10 percent level are designated by *, **, and ***, respectively. rho is the fraction of the variance due to u_i.(EN) = variables designated as endogenous variables in Hausman-Taylor estimation method.D1 = first differenced variables in the Arellano-Bond method, no lags were used for the independent variable.ST refers to chi -squared value of the Sargan test for overidentifying restrictions.Time dummy variables were used in equations (3) to (6).
Sources: Author’s estimates based on IMF, World Economic Outlook; and World Bank, World Development Indicators.Notes: Standard errors in italics. Significance of the coefficients at the 1, 5, and 10 percent level are designated by *, **, and ***, respectively. rho is the fraction of the variance due to u_i.(EN) = variables designated as endogenous variables in Hausman-Taylor estimation method.D1 = first differenced variables in the Arellano-Bond method, no lags were used for the independent variable.ST refers to chi -squared value of the Sargan test for overidentifying restrictions.Time dummy variables were used in equations (3) to (6).

Although most of the methods yielded similar coefficients, equation (5) in Table 3.3 was used to draw implications for Cambodia. When pooling the data, the OLS estimator in equations (1) and (2) does not take advantage of the benefits of panel data analysis that can capture the impact of country-specific effects stemming from unobserved, and hence, omitted variables. The Arellano-Bond estimator is deemed less suitable as it did not yield a significant coefficient for the lagged dependent variable, negating the usefulness of this estimator.15 The Hausman-Taylor estimator was preferred as it allows relaxation of the assumption of exogeneity of all regressors, and some of the exogenous variables could be used as instruments for governance indicators, trade openness, and capital formation to GDP, yielding similar coefficient estimates.

Lessons for Cambodia

Lessons for achieving more robust sustainable growth can be drawn by comparing Cambodia to strong performers. For illustration purposes, the results from equation (5) in Table 3.3 are used to estimate the contribution to per capita growth in Cambodia of the main growth determinants. The estimated contribution in the last column of Table 3.4 suggests that Cambodia has benefited from high per capita aid flows and stability in the terms of trade.16

Table 3.4.Difference Between ASEAN Average and Cambodia on Growth Determinants1
Regression CoefficientsASEAN Mean Value (1970–2001)Cambodia Mean Value (1970–2001)Impact on Per Capita GDP Growth
Secondary school enrollment0.04541.017.5-6.0
Gross capital formation to GDP0.13824.614.0–10.5
Government consumption to GDP–0.11010.510.40.1
Trade to GDP–0.00392.658.00.2
Lagged improvement in terms of trade0.0271.22.31.3
Terms of trade volatility–0.00112.28.70.0
Weather (years of low crop yield)–3.5780.10.10.0
Broad money to GDP0.00347.123.6–0.3
Government effectiveness22.2742.71.8–113.7
Telephones per thousand0.00155.11.2–4.4
Aid per capita as percent of per capita GDP0.0854.310.45.0
Sources: IMF, World Economic Outlook; World Bank, World Development Indicators; and Kaufman, Kraay and Zoido-Labatón (1999).

Regression results from equation (5) in Table 3.3.

Scale adjusted from between –2.5 and +0.25 to between 0 and 5. Similar coefficients obtained for the other governance indicators.

Sources: IMF, World Economic Outlook; World Bank, World Development Indicators; and Kaufman, Kraay and Zoido-Labatón (1999).

Regression results from equation (5) in Table 3.3.

Scale adjusted from between –2.5 and +0.25 to between 0 and 5. Similar coefficients obtained for the other governance indicators.

In contrast, Cambodia’s growth performance has been constrained by a number of factors. These include lower levels of education and capital formation (infrastructure development as proxied by the number of telephones). Accordingly, improvements in those areas could potentially yield significant improvements in long-term growth. Above all, improved government effectiveness could be an important contributor to boosting growth. The same result was obtained by substituting government effectiveness with each of the other governance indicators shown in Table 3.2.

D. Conclusions

Cambodia has experienced more rapid growth than other LICs since the Asian crisis. The higher growth rates are partly consistent with the experience of other LICs and transition countries, which are starting from a lower base. Cambodia has also benefited from large aid inflows that have boosted economic activity. Relative macroeconomic stability, compared to other LICs, has also helped support higher growth rates.

The crucial question for Cambodia is how to sustain high growth rates in the presence of a number of adverse developments that are likely to lead to slower growth. Compared to the fast-growing Asian economies, Cambodia and other LICs have weaker human and physical capital base and institutional infrastructure. Sustaining such high growth rates in the future would require Cambodia to catch up with other countries in labor skills, market institutions, infrastructure, and strengthened governance. At the same time, continuing with the macroeconomic stability and a relatively open trade system will remain crucial to supporting private sector activity.

Appendix
Table A3.1.Description of Data and Group Means for 1970–20031
All (144)PRGF (70)LIC-Nonfuel (68)AsiaAsia Excluding ChinaAsia Excluding Islands (18)ASIA-LIC (16)Transition (29)Transition LIC (13)ASEAN (9)ASEAN LIC (4)Lao P.D.R.VietnamCambodiaNumber of Observations2
Dependent variables
Real GDP per capita1.40.90.82.52.33.02.12.01.43.43.12.63.53.41,008
Growth of ouput per labor
1.10.70.72.62.43.02.41.70.83.23.33.23.23.5921
Initial conditions
Log of GDP (1970)7.77.17.17.27.27.16.97.97.37.36.56.56.66.51,007
1970 GDP in U.S. dollars, PPP127656057147448841937411157195122172072721791,008
Labor growth
Labor force growth2.42.32.32.42.52.42.31.12.02.52.02.02.21.9921
Population growth2.12.22.22.12.22.22.10.91.62.22.22.52.02.21,008
Human capital
Life expectancy (in years)60.254.554.760.059.659.957.666.061.459.552.047.862.145.71,000
Log of life expectancy4.14.04.04.14.14.14.04.24.14.13.93.94.13.81,000
Illiteracy rate31.644.143.430.130.331.836.411.223.422.629.946.411.040.6847
Primary school enrollment91.083.483.596.495.596.293.199.498.899.698.296.0107.593.9855
Secondary school enrollment
45.130.530.638.738.138.533.075.261.840.927.320.946.817.5847
Tertiary school enrollment11.56.06.17.27.47.63.622.715.310.32.91.44.21.3807
Physical capital
Number of telephones per thousand69.528.529.034.334.334.614.3114.157.855.14.43.011.71.2941
Capital formation to GDP22.821.521.524.423.824.621.825.623.424.616.818.923.114.0823
Macroeconomic policies
Inflation55.273.174.611.812.213.014.294.1120.917.830.234.748.521.01,006
Log of inflation2.22.32.31.92.01.92.12.12.22.02.53.02.81.7974
Government balance–4.2–5.7–5.6–4.0–4.2–3.8–5.2–3.4–6.4–2.7–5.2–8.7–5.2–4.3945
Government consumption to GDP15.715.115.213.213.311.813.215.614.310.59.08.37.610.4813
Openness
Trade to GDP85.782.282.579.682.573.065.4108.4135.792.643.751.376.758.0906
Trade Restrictiveness Index
4.64.64.64.24.24.23.84.84.84.84.32.56.04.51,008
External factors
Terms of trade (ToT) change
2.22.52.30.70.70.70.93.97.70.92.12.02.61.7998
Terms of trade change, lagged2.22.52.30.80.80.80.94.08.01.22.50.93.62.3997
Variance of ToT change17.721.020.915.816.415.619.520.838.412.218.59.536.78.71,001
Fraction of years of low crop yield0.20.10.20.10.10.10.10.20.20.10.10.00.10.1924
Aid per capita in U.S. dollars
45.454.155.348.350.228.262.625.831.011.216.031.59.218.4900
Aid per capita to GDP per capita8.713.013.011.712.07.615.96.19.94.38.115.94.110.4899
Financial sector development
Credit to GDP28.218.518.732.530.235.318.222.010.644.88.46.418.05.4826
Broad money to GDP47.047.547.844.245.838.839.455.872.547.125.431.347.623.6914
Institutional factors
Composite index–0.2–0.4–0.4–0.2–0.2–0.2–0.4–0.2–0.5–0.1–0.7–0.6–0.4–0.71,001
Voice and accountability–0.2–0.4–0.3–0.3–0.3–0.5–0.4–0.2–0.6–0.6–1.1–1.0–1.2–0.71,001
Political stability–0.1–0.4–0.4–0.1–0.1–0.1–0.30.1–0.20.2–0.21.00.4–1.1875
Government effectiveness–0.3–0.6–0.5–0.1–0.10.0–0.3–0.3–0.50.2–0.5–0.1–0.2–0.7987
Lack of regulatory burden–0.2–0.4–0.4–0.1–0.10.0–0.4–0.2–0.60.1–0.7–1.1–0.5–0.3994
Rule of law–0.3–0.6–0.6–0.2–0.2–0.2–0.6–0.3–0.6–0.1–1.0–1.3–0.5–0.9889
Control of corruption–0.3–0.6–0.6–0.3–0.3–0.3–0.6–0.3–0.6–0.9–0.1–0.9–0.6–0.9798
Sources: IMF, World Economic Outlook; World Bank, World Development Indicators; and Kaufman, Kraay and Zoido-Labatón (1999).

For many variables, data cover 1970–2001.

Number of observations refers to number of five-year period averages per variable.

Sources: IMF, World Economic Outlook; World Bank, World Development Indicators; and Kaufman, Kraay and Zoido-Labatón (1999).

For many variables, data cover 1970–2001.

Number of observations refers to number of five-year period averages per variable.

Table A3.2.List of Economies Included in the Analysis
Africa
Angola
Benin
Botswana
Burkina Faso
Burundi
Cameroon
Cape Verde
Central African Republic
Chad
Comoros
Congo, Dem. Rep. of
Congo, Rep. of
Côte d’Ivoire
Djibouti
Equatorial Guinea
Ethiopia
Gabon
Gambia, The
Ghana
Guinea
Guinea-Bissau
Kenya
Lesotho
Madagascar
Malawi
Mali
Mauritania
Mauritius
Morocco
Mozambique
Namibia
Niger
Nigeria
Rwanda
São Tomé and Príncipe
Senegal
Seychelles
Sierra Leone
South Africa
Sudan
Swaziland
Tanzania
Togo
Tunisia
Uganda
Zambia
Zimbabwe
Asia
Bangladesh
Bhutan
Cambodia
China
Fiji
India
Indonesia
Lao P.D.R.
Malaysia
Maldives
Myanmar
Nepal
Pakistan
Papua New Guinea
Philippines
Samoa
Singapore
Solomon Islands
Sri Lanka
Thailand
Tonga
Vanuatu
Vietnam
Europe
Albania
Bulgaria
Croatia
Cyprus
Czech Republic
Estonia
Hungary
Latvia
Lithuania
Macedonia, FYR
Poland
Romania
Slovak Republic
Slovenia
Former Soviet Union
Armenia
Azerbaijan
Belarus
Georgia
Kazakhstan
Kyrgyz Republic
Moldova
Mongolia
Russia
Tajikistan
Ukraine
Middle East
Algeria
Bahrain
Egypt
Iran, I.R. of
Jordan
Kuwait
Lebanon
Libya
Malta
Oman
Qatar
Saudi Arabia
Syrian Arab Republic
Turkey
United Arab Emirates
Yemen, Republic of
Western Hemisphere
Antigua and Barbuda
Argentina
Bahamas, The
Barbados
Belize
Bolivia
Brazil
Chile
Colombia
Costa Rica
Dominica
Dominican Republic
Ecuador
El Salvador
Grenada
Guatemala
Guyana
Haiti
Honduras
Jamaica
Mexico
Netherlands Antilles
Nicaragua
Panama
Paraguay
Peru
St. Kitts and Nevis
St. Lucia
St. Vincent and the Grenadines
Suriname
Trinidad and Tobago
Uruguay
Venezuela, R.B. de

LICs are defined as the group of Poverty Reduction and Growth Facility (PRGF)-eligible countries. ASEAN excludes Brunei Darussalam due to data limitations.

It should be noted, however, that Cambodia’s GDP per capita growth over the longer period is misleading because it reflects the sharp reduction in population in the 1970s.

Early papers that spurred research include Barro (1991) and Fischer (1993).

Bosworth and Collins (2003) review the recent literature and apply the combined approach to 84 high- and low-income countries, utilizing Barro and Lee’s (2000) data set of educational attainment and by extending the data set on initial capital stock contained in a 1993 World Bank study.

Chapter on “Growth and Institutions,” IMF (2003d).

Section on “How Can Economic Growth in the Middle East and North Africa Region Be Accelerated?” IMF (2003c).

Further work is needed to develop estimates of initial physical capital stock and human capital for many of the countries included in this study, thereby allowing application of the growth accounting approach to decompose the sources of growth.

This is a restrictive assumption that is arguably difficult to support. However, a fixed effects model, which does not require this assumption, is excluded because it ignores the time-invariant variables, such as the institutional factors that are of particular interest here. In principle, however, alternative time-varying measures of institutional indicators could be used, if readily available.

In the study on the impact of institutions in the September 2003 World Economic Outlook (WEO; IMF, 2003c), geographic latitude and ethno-linguistic diversity were used as instruments for institutions, but no instruments were used for macroeconomic variables. Data sets of instruments used in cross-country analysis, such as the percent of population speaking a foreign language or the origin of the legal system, have typically excluded transition economies.

This may be due to mismeasurement of labor inputs: very few countries report hours worked or overall employment figures, and employment was therefore measured by labor force growth or population growth. The 2003 WEO study used “economically active population growth differential,” measured as the rate of growth in the labor force minus the population growth rate, but using this measure yielded a wrong sign in our analysis as well, and the different result may stem from excluding advanced economies. The growth in the labor force is apparently not a good measure of labor input, possibly due to the prevalence of underemployment in many developing countries, particularly in the rural areas and in state-owned enterprises.

    Other Resources Citing This Publication