Bangladesh: Selected Issues and Statistical Appendix

This Selected Issues paper and Statistical Appendix examines Bangladesh’s nonfinancial state-owned enterprises. The paper argues that, given the overall policy environment and external vulnerabilities, the usefulness of the fixed exchange rate system in Bangladesh has run its course. Greater exchange rate flexibility is needed to ensure that the exchange rate sends appropriate market signals, and to enhance the authorities’ ability to address more effectively and timely both domestic imbalances and external real shocks arising from a rapidly changing global environment. The paper also outlines the structure and recent performance of the commercial banking sector.

Abstract

This Selected Issues paper and Statistical Appendix examines Bangladesh’s nonfinancial state-owned enterprises. The paper argues that, given the overall policy environment and external vulnerabilities, the usefulness of the fixed exchange rate system in Bangladesh has run its course. Greater exchange rate flexibility is needed to ensure that the exchange rate sends appropriate market signals, and to enhance the authorities’ ability to address more effectively and timely both domestic imbalances and external real shocks arising from a rapidly changing global environment. The paper also outlines the structure and recent performance of the commercial banking sector.

IV. Growth Requirements for Poverty Reduction in Bangladesh1

A. Introduction

1. Reducing poverty in a visible way within a reasonable period of time remains Bangladesh’s main policy challenge. The 2000 Household Expenditure Survey shows that poverty in Bangladesh has declined significantly during the 1990s owing to a relatively good growth performance (about 5 percent per year on average; Table IV.1). In 2000–01, 34 percent of Bangladesh’s population was very poor (under the lower poverty line) and 50 percent was poor (under the upper poverty line) as compared, respectively, to 43 percent and 59 percent in 1991–92.2 Rural poverty experienced a greater decline during the period, but it remained higher than urban poverty. Notwithstanding this progress, much more remains to be done. Bangladesh still has the highest incidence of poverty in South Asia, the third highest absolute number of poor in the world after India and China, and its per capita GDP ($366 in 2000) is one of the lowest in the world.

Table IV.1.

Trends in Poverty: Headcount Rate

(In percent)

article image
Source: Bangladesh: Poverty Assessment, World Bank, 2001.

2. A faster pace in the reduction of poverty would require putting Bangladesh’s economy on a higher sustainable growth path. The simulation exercise presented in this chapter shows the growth path required to achieve the IDA ceiling of $885. The implied poverty paths for both the rural and urban areas are also computed.

B. Growth Scenario for Achieving the IDA Operational Ceiling by 2027

3. The growth scenario presented below is based on the sources of growth accounting framework. A production function has been estimated for Bangladesh. The description of the data and details of the estimation method are given in Annex IV.1. Assuming constant returns to scale, the sources of growth accounting framework used is based on the following aggregate production function:

Yt=AKta(HtLt)1α(1)

where Yt is real output, A is total factor productivity (TFP), Lt is total employment, and Ht is an index of human capital and therefore HtLt, should be interpreted as skill-augmented labor.3 The parameter (1-α) yields the share of labor. Taking logs and totally differentiating yields:

y^t=a^+αk^+(1α)(h^+l^)(2)

where the lower case variables with a “hat” represent the growth rate of the corresponding variables described above. Equation (2) implies that three factors contribute to the growth rate of real GDP: TFP growth, the growth rate of physical capital, and the growth rate of skill-augmented labor. The key parameter is the share in GDP of physical capital (α). As outlined in Annex I, its point estimate in Bangladesh is 32 percent, which implies that the share of labor is 68 percent. These values are within the range of values found in the literature (see references in Annex IV. 1).

4. In the simulation exercise, the path for ŷt (is given by the underlying objective to achieve the IDA ceiling ($885 per capita in 2000) within 25 years.4 The level of GDP per capita in Bangladesh in 2000 was $366, which represents 40 percent of the target level. The interest of this exercise is to assess the feasibility of this objective and examine the combination of factor accumulation, both in physical and human capital, and TFP growth that will achieve the target per capita income level. Annex IV.2 shows the methodology used for the projection of population and labor force.

5. In the benchmark simulation, the population growth path is given by UN projections.5 These projections assume a stepwise decline in population growth from 2 percent in FY00/01 to 1.1 percent in FY26/27. The simulation can be divided into two periods. Most of the reforms and transitional dynamics take place during the first period (FY02/03–FY06/07). During the second period (FY07/08–FY21/27), the economy reaches a steady growth path for GDP per capita of 4.8 percent per year necessary to achieve the IDA ceiling level of income per capita by FY26/27. Note that real GDP growth during that period varies slightly because of the declining population growth.6

6. The simulation results are summarized in Table IV.2 and Figure IV.1 and Figure IV.2. During the adjustment period (FY01/02–FY06/07), growth first declines sharply in FY01/02 due to the current unfavorable external environment and the required fiscal and monetary adjustments. Subsequently, growth picks up progressively as appropriate macroeconomic policies and structural reforms help build confidence and a favorable environment for domestic investment. Among the prerequisites for buoyant private investment are improvements in infrastructure, and education and training of the labor force. Thus, the projections make allowance for a modest increase in public investment during the adjustment period despite the necessary fiscal adjustment. The projections also show increased spending in education and on poverty reduction. As the fiscal constraint relaxes, owing to higher growth, public investment reaches 8½ percent of GDP. Private investment increases steadily but its pace picks up only after the adjustment period once the investment environment has significantly improved.

Table IV.2.

Selected Economic Indicators Simulation of Achieving IDA Income Level in 25 Years Using WB Population Projections

article image
Figure IV.1
Figure IV.1

Selected Economic Indicators, 2002-27 Simulation of Achieving IDA Income Level in 25 Years Using World Bank Population Projections

Citation: IMF Staff Country Reports 2002, 114; 10.5089/9781451804096.002.A004

Figure IV.2
Figure IV.2

Poverty Reduction Implied by the Growth Scenario

Citation: IMF Staff Country Reports 2002, 114; 10.5089/9781451804096.002.A004

7.TFP growth from the implementation of the structural reform program is the main engine of growth during the adjustment period. Table IV.2 shows TFP steadily increasing during the adjustment period to reach 2.1 percent by FY05/06.7 While this implies important efficiency gains, the Bangladesh economy is still very far from the efficiency frontier and therefore has significant room for improvement. However, as the economy approaches the efficiency frontier, TFP growth will decelerate toward a more sustainable level. In this scenario, it is assumed that after TFP growth reaches a peak of about 2 percent per annum in FY05/06, it starts declining progressively toward its steady-state level of 1¼ percent per annum. To sustain growth, the fall off in TFP growth rate has to come from an acceleration in the growth of factor accumulation. Thus, as TFP growth decelerates, investment both in physical and human capital increase to maintain the constant annual growth rate of 4.8 percent for real GDP per capita.

8. Investment as a share of GDP reaches a maximum of 3.5 percent by FY 18/19 before starting to decline. Even with this ambitious investment plan, the growth of contribution from human capital (as measured by the index H) will have to increase from 0.3 percent in FY00/01 to 2 percent per annum by the end of the simulation period. Obviously, lower levels of human capital growth would have to be compensated either by higher investment or higher TFP growth.

9. An interesting insight from this framework is that TFP growth and improvements in human capital have a much more powerful effect on growth than physical capital accumulation. The reason is twofold. First, physical capital accumulation is subject to decreasing returns while TFP is not and, while human capital is also affected by decreasing returns, returns on human capital decrease at a much slower speed than that of physical capital. Second, what matters for growth is the growth rate of physical capital not the growth rate of investment or investment as a share of GDP. Because the stock of physical capital is about ten times larger than investment, even a significant increase in investment translates into only relatively small increases in the stock of physical capital.

10. The execution of the ambitious plan shown in this scenario requires significant additional resources. So far, Bangladesh has benefited from significant transfers, totaling 5 percent of GDP in FY00/01 alone. About ½ percent of GDP is from official sources and 4½ percent from private sources, of which almost 4 percent originate from workers’ remittances. There is a high degree of uncertainty about the projection of these flows over long periods of time. The view reflected here is that these flows may taper off over time as the capacity of neighboring countries to absorb Bangladeshi workers may decrease, and donors’ commitment may fade away as the economy reaches a higher sustainable growth path. Consequently, it is assumed that total transfers decline over time to reach 4 percent by FY26/27. However, it is expected that domestic saving will fill the gap. Indeed, there is strong empirical evidence of a negative relationship between consumption as a share of GDP and GDP per capita, to a large extent reflecting the typical decline in household income share as their income increases. Because public consumption is already relatively low (about 4½ percent of GDP), it is assumed that the contraction in the share of consumption in GDP will come from the private sector for which consumption declines from 78 percent of GDP in FY00/01 to about 68 percent in FY26/27. Consequently, domestic saving (excluding transfers) as a share of GDP increases from 15 percent in FY01/02 to 28 percent in FY26/27. The current account deficit increases steadily to reach 4.4 percent of GDP by FY18/19, before gradually declining to less than 1 percent by the end of the simulation period.

C. Poverty Reduction Implied by the Growth Scenario

11. Recent empirical studies have attempted to estimate the link between GDP per capita growth and poverty by estimating the elasticity of poverty with respect to GDP per capita growth.8

12. Unfortunately, these are only estimates for the elasticity of poverty with respect to the growth of consumption per capita (Table IV.3). All the elasticity estimates are larger than one which implies that poverty would decline rapidly if the economy were to sustain higher growth levels of consumption per capita. To obtain a direct link between poverty and GDP per capita growth, we multiplied the elasticities in Table IV.3 by two-thirds which is the ratio of GDP per capita growth to consumption per capita growth during the 1990s (the estimation period).

Table IV.3.

Elasticity Estimates of Poverty with Respect to Growth for Bangladesh Using the Headcount Rate

article image
Source: Bangladesh: Poverty Assessment, World Bank, 2001. These estimates were obtained from the 2000 Bangladesh Household Expenditure Survey. All these estimates are significant at 5 percent level.

13. Assuming that the growth rates required to achieve IDA operational level of income by FY26/27 can be achieved and the elasticities implied by Table IV.3 prevail, poverty, as measured by the headcount rate, would decline rapidly, reaching less than 8 percent for both the rural and urban areas by the end of the simulation period under the lower poverty line, and 11 percent and 7¾ percent for the rural and urban areas, respectively, under the upper poverty line.9 Under this scenario, Bangladesh could achieve the United Nations International Development Goals (IDG) by FY07/08.10 These results are consistent with recent World Bank findings.11

14. Having said this, these results need to be interpreted with caution. In particular, the simulation assumes that there are no changes in the distribution of income as the economy grows (i.e., the net elasticity of poverty with respect to GDP per capita growth remains constant. It is likely, however, that such elasticity will decrease over time. Consequently, the degree of confidence of this result is likely to decline as the forecasting horizon lengthens.

D. Conclusions

15. A simulation exercise, based on the sources of growth accounting framework, suggests that Bangladesh would need to sustain an average annual growth rate of about 6–6½ percent over the next 25 years to reach the present IDA operational per capita income ceiling of $885. Achieving this objective will require both significant improvements in productivity as well as much higher factor accumulation, which can best be attained under a stable macroeconomic environment, rapid implementation of the structural reforms, improved governance, and adequate infrastructure.

16. Using recent elasticity estimates of poverty with respect to GDP per capita growth, the simulations show that if such growth path were to be reached, poverty would be significantly reduced. Indeed, Bangladesh could achieve the IDG of reducing extreme poverty earlier than the target date of 2015.

ANNEX IV. 1 Estimation of The Production Function

The share of labor in aggregate output has been estimated using the following production function:

Yt=AKta(HtLt)1α(1)

where Yt is real output, A is total factor productivity (TFP), Lt is total employment, and Ht is an index of human capital. The parameter (1-α) yields the share of labor. Expressing Yt and Kt in per capita (in efficiency units) and taking logs yields:

log(Yt/(HtLt))=log(A)+αlog(Kt/HtLt)(2)

The estimation of the production function raises the issue of whether to estimate it in levels or in first differences. As is well known, the first difference operator removes all the long-run information in the data. One important insight from the cointegration literature is that we know much more about the long-run than the short-run relationship between macroeconomic variables. Consequently, by differencing, we disregard the most valuable part of information in the data. In the context of production function estimation, this point is particularly relevant because the business cycle frequencies of the production process may be dominated by variations in capacity utilization factors that are difficult to measure, especially for developing countries. Therefore, for this study, the estimation of the production function is estimated in levels.

It will be shown below that the two variables entering the production function cannot reject the unit root assumption at conventional significance levels. Using a residual cointegration test shows that the variables are cointegrated. The cointegrating vector has been estimated using the Fully-Modified (FM) estimator which takes into account potential endogeneity of capital and the skill-augmented labor (that is, HtLt). Furthermore, this method is also robust with respect to the likely autocorrelation of the error term. The FM estimator is an optimal single-equation method based on the use of ordinary least squares with semi-parametric corrections for serial correlation and potential endogeneity of the right-hand variables. The FM estimator has the same asymptotic behavior as the full systems maximum likelihood estimators.1

Source of data

The remainder of this section briefly describes the series Yt, Kt, Lt, and Ht. Kt was constructed using the perpetual inventory method based on the investment series. The geometric depreciation rate was set at 4 percent per year, a value widely used in literature for developing countries. The initial capital stock was derived by assuming an initial capital-output ratio of 2. Because the estimation period starts in 1985 while the construction of K starts in 1977, reasonable variations in the initial capital stock do not change the results significantly. Indeed, using alternative values of 1.5 and 2.5 did not significantly alter the results. Real GDP(Yt) and labor force (Lt) were taken from the IMF’s World Economic Outlook (WEO) database.

The index Ht was provided by Barry Bosworth for 1985 to 1994. The series was extended to 2000, using a quadratic trend. This series was constructed following Barro and Lee’s (1994)2 methodology based on educational attainment. H is defined as follows:

Ht=Σj=17WjtPjt

where Pjt, represents the share of the population that completed the level of education j, where j varies from 1 (corresponding to the share of the population with no schooling) to 7 (corresponding to beyond secondary education). Wjt represents aggregation weights based on the observed relative earnings of the different educational groups.3

Estimation results

The unit-root hypothesis is tested using the Augmented-Dickey-Fuller (ADF) test, which amounts to running the following set of regressions for the two variables entering the production function (equation (2)):

yt=μ+γt+Φ0yt1+Σi=1k1ΦiΔyt1+ξt,k=1,,5(3)

The lag length (k) in the ADF regression is selected using the Schwarz Information Criterion (SIC). The following table presents the test results.

ADF Test Results

article image

The critical values for the 1 percent, 5 percent, and 10 percent levels are - 4.89, -3.82, and -3.36. Therefore, the unit-root hypothesis cannot be rejected even at the 10 percent level. The estimated production share of the physical capital (a), using the FM estimator, is 0.32 with a t-statistic of 6.71, and thus the estimate is statistically significant at the 1 percent level.

ANNEX IV.2 Projection of Exogenous Variables

Population

Interpolated using the World Bank 5-year figures for 2000-50. For 1990-99, growth rates from WEO population series were used to extend World Bank series backward.

Labor force

WEO figures for 1990–06 were used. For 2007–26, a logistic function for the ratio (labor force/population) was estimated and the estimated function was used to project labor force.1 The estimated logistic function is:

lft/popt=exp[1.10182+.00442(t2006)](4)

where lft/popt is the ratio of labor force to population in year t.

STATISTICAL APPENDIX

Table 1.

Bangladesh: Gross Domestic Product at Current Prices, 1996/97–2000/01

(In billions of taka)

article image
Source: Bangladesh Bureau of Statistics
Table 2.

Bangladesh: Sectoral Growth Rates at Constant Prices, 1996/97–2000/01 1/

(Annual percentage change)

article image
Source: Bangladesh Bureau of Seatistics

Data based on 1995/96 benchmark and prices.

Table 3.

Bangladesh: Gross Domestic Product by Expenditure at Current Prices, 1996/97–2000/01

article image
Source: Bangladesh Bureau of Statistics.

GDE minus consumption.

GDP minus consumption.

National saving minus gross investment.

Table 4.

Bangladesh: Agricultural Production, 1996/97–2000/01

article image
Sources: Bangladesh Bureau of Statistics; Ministry of Agriculture; Ministry of Food; and BADC.

Provisional.

Table 5.

Bangladesh: Jute Goods Statistics, 1996/97–2000/01

article image
Sources: Bangladesh Jute Mills Corporation; and Ministry of Jute.

Figures shown do not correspond to balance of payments data; some private mill production of jute yarns and twine is not included.

Table 6.

Bangladesh: Public Food Distribution, 1997/98–2001/02

(In thousands of metric tons)

article image
Source: Ministry of Food.
Table 7.

Bangladesh: Fertilizer Statistics, 1997/98–2001/02

article image
Source: Bangladesh Chemical Industries Corporation.

Figures as of December 2001.

Table 8.

Bangladesh: Manufacturing Production by Main Categories, 1996/97–2000/011/

article image
Source: Bangladesh Bureau of Statistics

Excluding mining and electricity.

Table 9.

Bangladesh: Energy Statistics, 1997/98–2001/02

article image
Sources: Data provided by the Bangladesh authorities; BPDB; BOGMC; and BPC.

Including 774 million cubic feet for use by the production and transport and distribution companies.

Table 10.

Bangladesh: Retail Prices of Petroleum Products, 1996/97–2000/01

(End of period)

article image
Source: Bangladesh Petroleum Corporation.
Table 11.

Bangladesh: Consumer Price Index, National

(Based on 1985/86 weights; percentage change)

article image
Source: Bangladesh Bureau of Statistics.
Table 12.

Bangladesh: Labor Market Statistics, 1990/91–1999/00 1/

article image
Source: Bangladesh Bureau of Statistics.

Based on labor force surveys.

Natural growth rate (crude birth rate minus crude death rate).

Aged 10 years and above.