Information about Asia and the Pacific Asia y el Pacífico
Journal Issue
Share
Article

Cambodia: Selected Issues

Author(s):
International Monetary Fund
Published Date:
October 2004
Share
  • ShareShare
Information about Asia and the Pacific Asia y el Pacífico
Show Summary Details

I. Growth and Poverty

Chapter 1. Determinants of Growth in Cambodia and Other LICs in Asia: Evidence from Country Panel Data1

1. This chapter explores the implications for long-term sustainable growth in Cambodia from 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 overviews the difference 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 ASEAN countries.

A. Cambodia’s Growth Experience and Prospects

2. In the last five years, Cambodia’s growth performance has been amongst the best. 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 2½ percent (Table 1). This growth performance is significantly higher than the average for all developing countries and for LICs.2 Cambodia’s pace of growth is more similar to that of other ASEAN low income countries (Myanmar, Vietnam and Laos) and of transition countries.3

Table 1.Real GDP Per Capita Growth1(Annual average, in percent)
1970-20031999-2003
All (144)1.52.2
PRGF (70)1.01.9
LIC-Non-Fuel (68)1.01.9
Asia (23)2.81.9
Asia excluding China (22)2.61.8
Asia excluding islands (18)3.12.3
ASIA-LIC (15)2.41.9
Transition (29)2.13.7
Transition LIC (13)1.53.6
ASEAN (9)3.42.9
ASEAN LIC (4)3.13.3
Cambodia3.52.6
Laos, PDR2.95.4
Vietnam3.72.2
Source: WEO database.

Excludes advanced economies. Asia and ASEAN exclude Brunei Darussalam.

Source: WEO database.

Excludes advanced economies. Asia and ASEAN exclude Brunei Darussalam.

3. However, prospects of weaker growth in the period ahead call for a deeper exploration of the factors that can contribute to sustained growth. As elaborated in Chapter 2, Cambodia has benefited from economic rents from textile quota since 1996 under the Multi Fiber 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 the recent World Bank value chain studies and investment climate assessment—do not bode well for future sustainable growth.4 Identification of key impediments to growth has become an urgent agendum.

B. Overview of Growth Determinants

4. Before estimating the growth equations, Cambodia’s performance is assessed 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. In 1970, it had one of the lowest per capita GDP in PPP terms, about one third that of other LICs (Figure 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 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).
  • 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 4). Moreover, its terms of trade have remained relatively favorable and stable (Figure 5).
  • As with other transition economies, Cambodia lags in institutional and market development. Financial markets remain shallow, with a bank credit to the private sector at around 7 percent of GDP, and the ratio of broad money to GDP under 20 percent (Figure 6).

Figure 1.Initial conditions, 1970

Figure 2.Human and Physical Capital

(Average 1970–2001)

Figure 3.Macroeocomic Policies


(In percent, average 1970–2001)

Figure 4.Aid Per Capita


(Per capita aid flows as percent of per capita GDP, average 1970–2001)

Figure 5.Volatility of Terms of Trade


(Average 1970–2003)

Figure 6.Financial Development Indicators


(In percent, average 1970–2001)

5. weak governance has become the Achilles’ heal 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 shown in Table 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 2.Governance Indicators
AllPRGFAsiaASIA-LICTRNTRN LICASEANLao PDRVietnamCambodia
Combined Index 1−0.2−0.4−0.2−0.4−0.2−0.5−0.1−0.6−0.4−0.7
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: World Bank WP 2195, “Aggregating Governance Indicators”, Daniel Kaufman, Aart Kray and Pablo Zoido-Lobaton, 1999.

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

Source: World Bank WP 2195, “Aggregating Governance Indicators”, Daniel Kaufman, Aart Kray and Pablo Zoido-Lobaton, 1999.

Each of the six governance indicators are 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

6. Cambodia and other transition economies have been typically excluded from cross-country studies of long-term growth. One reason might have been the structural rigidities and weaker influence of market forces that would make it more 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 became independent states only since the early nineties, 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 needs.

7. A standard growth model, described in Annex I, was estimated for 144 low- and middle-income countries. The sample includes 70 LICs, of which 15 are Asian LICs. Real per capita GDP growth and labor productivity growth are used as alternative dependent variables.5 The explanatory variables included in the estimation exercise and their group means are described in Annex Table 1. The regression equation takes the following form:

8. The results shown in Table 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. Similar results are obtained when labor productivity (change in output per worker) is used as the dependent variable.

Table 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−0.974−1.048−0.743−0.826−0.849−1.011D1−0.990−1.052
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, lagged0.0320.0330.0210.0170.0270.027D10.0510.040
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,0.0300.0310.0410.0510.0850.064D10.0690.018
as percent of GDP per capita0.0200.0210.012***0.015***0.027***0.026**0.041*0.041
Telephones per '0000.0020.0020.0020.0010.0010.002D10.0090.004
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 enrollment0.0190.0000.010−0.0030.0450.023D1(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***
No. of Observations640640640640640640463462
R-Squared0.2990.367
rho0.6160.629
ST40.65 (14)39.54 (14)
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 over-identifying restrictions.

Time dummy variables were used in equations (3) to (6).

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 over-identifying restrictions.

Time dummy variables were used in equations (3) to (6).

9. 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 are used to estimate the contribution to per capita growth in Cambodia of the main growth determinants.6 The estimated contribution in the last column of Table 4 suggests that Cambodia has benefited from high per capita aid flows and stability in the terms of trade.7

Table 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
Dweather (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
Telephone per’0000.00155.11.2−4.4
Aid per capita as per cent of per capita GDP0.0854.310.45.0

Regression results from equation 3 in Table 5.

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

Regression results from equation 3 in Table 5.

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

10. In contrast, Cambodia’s growth performance has been constrained by a number of factors. They 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 boost growth. The same result was obtained with substituting government effectiveness with each of the other governance indicators shown in Table 2.

D. Conclusions

11. 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, who are starting from a lower base. Cambodia has also benefited from large aid inflows which have boosted economic activity. Relative macroeconomic stability, compared to other LICs, has also helped support higher growth rates.

12. 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 relative open trade system will remain crucial to support private sector activity.

Annex: Investigating the Sources of Growth from Panel Data

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 of the theory of endogenous growth has generally taken the form of either comparative regression analysis or growth accounting or, more recently, a combination of the two.8 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 crosscountry approach, inspired by the theory of endogenous growth, attempts to explain differences in growth experience 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.9

Two cross-country studies by IMF staff have focused on drawing lessons from the impact of macroeconomic and structural policies on growth in developing countries. A 1999 IMF staff 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.10 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 the volatility of growth.11 Another 2003 study that combined the growth accounting approach and institutional quality for a cross-country 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.12

Estimation Approach

The approach used here is applied to 144 low- and middle-income countries, and excludes all advanced economies. Transition economies are included where data permits. The countries (denoted by i) include 71 LICs, of which 15 are Asian LICs. Variables are averaged for 7 five-year periods (denoted by t), with the seventh period ranging from 2-4 years, depending on data availability. Real per capita GDP growth is the key dependant variable, and growth in labor productivity is used as an alternative dependent variable. The explanatory variables and their means are described in Annex Table 1. The basic regression takes the following form:

A number of estimation methodologies are used to test the robustness of the coefficients. Ordinary least squares has been typically used with either annual pooled data or period averages. Using panel data with seven 5-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.13 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 endogeneous explanatory variables, but it is typically difficult to obtain good instruments for these variables.14 In the absence of readily available instruments, we used three approaches in addition to the OLS. The first is a GLS estimation that allows for heteroskedastic effects between the country panels.15 The second is the Hausman Taylor estimation method whereby some variables are designated as exogeneous and used to instrument for variables suspected to be endogenous. The third alternative is to use the lagged dependent variable along with first differences of the independent variables and apply the Arellano Bond estimator.

Although most of the methods yielded similar coefficients, equation (5) was used to draw implications for Cambodia. As for the OLS estimator, it does not take advantage of the benefit inherent in panel data analysis which 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. However, tests confirmed the absence of first order correlation in the residuals and the existence of second order correlation, and rejected the null hypothesis in the Sargan test for over-identifying restrictions. The Hausman-Taylor estimator was preferred as it allows relaxation of the exogeneity of all regressors, and some were used as instruments for governance indicators, trade openness and capital formation to GDP, yielding similar coefficient results.

Not all variables yielded significant results. Moreover, growth in labor inputs yielded a coefficient with a negative sign, suggesting that greater labor input results in a lower per capita growth or a lower labor productivity. 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.16 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. 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.

Annex Table 1.Description of Data and Group Means for 1970–20031/
All (144)PRGF (70)LIC Non-Fuel (68)AsiaAsia excl ChinaAsia Excl Islands (18)ASIA-LIC (16)TRN (29)TRN LIC (13)ASEAN (9)ASEAN LIC (4)Lao PDRVietnamCambodiaNo. Obs. 2/
Dependent Variables
Real GDP per capita1.40.90.82.52.33.02.12.01.43.43.12.63.53.41008
Growth of Ouput per Labor1.10.70.72.62.43.02.41.70.83.23.33.23.23.5921
Initial conditions
Log of GDP (1970)7.77.17.17.27.27.16.97.97.37.36.56.56.66.51007
1970 GDP in $, PPP127656057147448841937411157195122172072721791008
Labor growth
Labor force growth2.42.32.32.42.52.42.31.12.02.52.02.02.21.9921
Population growth2.12.22.22.12.22.22.10.91.62.22.22.52.02.21008
Human capital
Life expectancy (in years)60.254.554.760.059.659.957.666.061.459.552.047.862.145.71000
Log of life expectancy4.14.04.04.14.14.14.04.24.14.13.93.94.13.81000
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 enrollment45.130.530.638.738.138.533.075.261.840.927.320.946.817.5847
Tertiary school enrollment11.56.06.17.27.47.63.622.715.310.32.91.44.21.3807
Physical capital
Number of telephones per’00069.528.529.034.334.334.614.3114.157.855.14.43.011.71.2941
Capital formation to GDP22.821.521.524.423.824.621.825.623.424.616.818.923.114.0823
Macroeconomic policies
Inflation55.273.174.611.812.213.014.294.1120.917.830.234.748.521.01006
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
Gov 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 Index4.64.64.64.24.24.23.84.84.84.84.32.56.04.51008
External factors
Terms of trade change2.22.52.30.70.70.70.93.97.70.92.12.02.61.7998
Terms of rrade Change, lagged2.22.52.30.80.80.80.94.08.01.22.50.93.62.3997
Variance of ToT change17.721.020.915.816.415.619.520.838.412.218.59.536.78.71001
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 US$45.454.155.348.350.228.262.625.831.011.216.031.59.218.4900
Aid per capita to GDP per capital8.713.013.011.712.07.615.96.19.94.38.115.94.110.4899
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−0.2−0.4−0.4−0.2−0.2−0.2−0.4−0.2−0.5−0.1−0.7−0.6−0.4−0.71001
Voice and accountability−0.2−0.4−0.3−0.3−0.3−0.5−0.4−0.2−0.6−0.6−1.1−1.0−1.2−0.71001
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−0.3−0.6−0.6−0.3−0.3−0.3−0.6−0.3−0.6−0.2−0.9−0.9−0.6−0.9798
Sources: WEO, WDI, and Kaufman, Kraay and Zoido-Labaton (1999).

Some variables are for 1970–2001.

Number of observations refers to number of 5-year period averages per variable.

Sources: WEO, WDI, and Kaufman, Kraay and Zoido-Labaton (1999).

Some variables are for 1970–2001.

Number of observations refers to number of 5-year period averages per variable.

Annex Table 2.List of Countries Included in the Analysis
AfricaAsiaMiddle East
AngolaBangladeshAlgeria
BeninBhutanBahrian
BotswanaCambodiaEgypt
Burkina FasoChinaIran, I.R. of
BurundiFijiJordan
CameroonIndiaKuwait
Cape VerdeIndonesiaLebanon
Central African RepublicLao.PD.R.Libyan Arab Jamahiriya
ChadMalaysiaMalta
ComorosMaldivesOman
Congo, Dem. Rep. ofMyanmarQatar
Congo, Rep. ofNepalSaudi Arabia
Cote d’IvoirePakistanSyrian Arab Republic
DjiboutiPapua New GuineaTurkey
Equatorial GuineaPhilippinesUnited Arab Emirates
EthiopiaSamoaYemen, Republic of
GabonSingaporeWestern Hemisphere
The GambiaSolomon IslandsAntigua & Barbuda
GhanaSri LankaArgentina
GuineaThailandThe Bahamas
Guinea-BissauTongaBarbados
KenyaVanuatuBelize
LesothoVietnamBolivia
MadagascarEuropeBrazil
MalawiAlbaniaChile
MaliBulgariaColombia
MauritaniaCroatiaCosta Rica
MauritiusCyprusDominica
MoroccoCzeck RepublicDominican Republic
Mozambique, Rep ofEstoniaEcuador
NamibiaHungaryEl Salvador
NigerLatviaGrenada
NigeriaLithuaniaGuatemala
RwandaMadedonia, FYRGuyana
Sao Tome and PrincipePolandHaiti
SenegalRomaniaHonduras
SeychellesSlovak RepublicJamaica
Sierra LeoneSloveniaMexico
South AfricaFormer Soviet UnionNetherlands Antilles
SudanArmeniaNicaragua
SwazilandAzerbaijanPanama
TanzaniaBelarusParaguay
TogoGeorgiaPeru
TunisiaKazakhstanSt Kitts and Nevis
UgandaKyrgyz RepublicSt Lucia
ZambiaMoldovaSt Vincent and Grenadine
ZimbabweMongoliaSuriname
RussiaTrinidad and Tobago
TajikistanUruguay
UkraineVenezuela
References

    BosworthBarry andSusanCollins2003“The Empirics of Growth: An Update,” (unpublished; (Washington: Brookings Institution andGeorgetown University).

    • Search Google Scholar
    • Export Citation

    BarroRobertJ.May1991“Economic Growth in a Cross-Section of Countries,”Quarterly Journal of EconomicsVol. 106pp. 407-33.

    BarroRobert andXavierSala-i-Martin2004Economic Growth (New York: McGraw-Hill, Inc.)

    FischerStanley1993“The Role of Macroeconomic Factors in Growth,”NBER Working Paper No. 4565 (Cambridge, Massachusetts,National Bureau of Economic Research).

    • Search Google Scholar
    • Export Citation

    International Monetary Fund1999Economic Adjustment and Reform in Low Income Countries, (Washington: International Monetary Fund).

    International Monetary Fund2003aWorld Economic Outlook,April2003;Growth and Institutions World Economic and Financial Surveys (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

    International Monetary Fund2003b“How Can Economic Growth in the Middle East and North Africa Region be Accelerated?”in World Economic OutlookSeptember2003;Public Debt in Emerging markets World Economic and Financial Surveys (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

    KaufmanDanielAartKraay andPabloZoido-Lobatón1999“Aggregating Governance Indicators,”World Bank Policy Research Working Paper No. 2196 (Washington: World Bank).

    • Search Google Scholar
    • Export Citation
1Prepared by Wafa Fahmi Abdelati (APD)
2LICs are defined as the group of PRGF-eligible countries. ASEAN excludes Brunei-Darussalam due to data limitations.
3It should be noted, however, that Cambodia’s GDP per capita growth over the longer period is misleading as it reflects the sharp reduction in population in the 1970s.
4Seizing the Global Opportunity: Investment Climate Assessment and Reform Strategy. World Bank, 2004.
5Further 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 decomposing the sources of growth.
6Reasons for using equation (5) are explained in the Annex.
7Similarly, although the coefficient for weather, proxied by the number of years with a large drop in crop yield, was large and significant, Cambodia’s share of bad weather is similar to that of other ASEAN countries and better than the average for all non-fuel exporting LICs (Annex Table 1).
8Early papers that spurred research include Barro (1991) and Fischer (1993).
9Bosworth 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 on initial capital stock contained in the dataset from a World Bank 1993 study.
10Economic Adjustment and Reform in Low-Income Countries, 1999.
11Chapter on Growth and Institutions, in World Economic Outlook, April 2003.
12Chapter on “How Can Economic Growth in the Middle East Be Accelerated”, in World Economic Outlook., September 2003.
13This 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.
14In the study on the impact of institutions in the September 2003 WEO, geographic latitude and ethnolinguistic diversity were used as instruments for institutions, but instruments were not used for macroeconomic variables.
15Datasets 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.
16The 2003 WEO study used “economically active population growth differential” measured as the rate of growth in the labor force minus the population growth rate. This measure yielded a wrong sign in our analysis as well, possibly as advance economies are excluded.

Other Resources Citing This Publication