Does institution matter for human capital development?

graph_human capital

A fundamental proposition of new growth theories is that human capital is a key driver of economic growth. Development of human capital for the people of a country encompasses not only the diffusion and assimilation of available knowledge, but also the generation of new knowledge – the source of innovation and technological change – which boosts economic growth.

It is rather a challenging task to measure a country’s stock of human capital. Popular indicators, used to measure human capital, include adult literacy rate, school enrolment rates, average years of schooling, quality of schooling etc. The Penn World Table version 8.1 provides a dataset on an index of human capital (HCI) for 134 countries over a period of 6 decades. HCI is an index of human capital per person which is related to the average years of schooling and the return to education. In 2010, United States had the highest HCI value (3.62) and Mozambique had the lowest one (1.27). In that year, among the 134 countries, 33 countries had HCI values higher than 3; 48 countries had values between 2.5 and 2.99; 28 countries had values between 2 and 2.49; and 25 countries had values less than 2. In South Asia, in 2010, the HCI values for Bangladesh, India, Nepal, Pakistan and Sri Lanka were 2.07, 1.93, 1.71, 1.99 and 3.16 respectively.

Why do some countries have higher level of human capital than others? Empirical literature have looked at different factors such as spending (both public and private) on education and health, and differences in income levels; but hardly there has been any emphasis on differences in institutional capabilities among the countries. However, quality of institution, as it affects economic growth process, can also have a bearing on the quality of human capital. Therefore, a valid question can be asked: does institution matter for human capital development? Of course there could be a bi-directional causality between human capital and quality of institution, where quality of institution could also be influenced by the level of human capital. Nevertheless, leaving aside the causality, here we are more interested to know about the association between these two.

The scatter-plot, as presented in the graph, has been generated using the data of index of human capital and index of institution for 93 countries over a period of 1984-2010 with over 2500 observations. We have constructed the index of institution using the data of six major ICRG (www.prsgroup.com) variables, namely bureaucracy quality, control of corruption, investment profile, democratic accountability, government stability, and law and order. As values of these six ICRG variables have different scales, we have rescaled them between 0 and 10. The aggregate institution index is the average of these six indicators with the range between 0 and 10, where 0 and 10 respectively indicate the lowest and highest levels of quality of institution.

The scatter-plot suggests a very strong positive association between quality of institution and level of human capital, which signifies the importance of better institution for higher level of human capital. Interestingly, if we compare Bangladesh with Malaysia, levels of both institution and human capital of Bangladesh in 1990 (1.62 and 1.52 respectively) were much lower than those of Malaysia in 1990 (6.05 and 2.31 respectively). Despite the fact that during 1990 and 2010, Bangladesh made some notable progresses in both fronts, by 2010, the levels of these two indices of Bangladesh (5.52 and 2.07 respectively) were below than what Malaysia had in 1990!

Results from a more sophisticated cross-country panel econometric regression reinforces this association. In this regression, the index of human capital has been considered as the dependent variable. We have also created two institutional indices: economic institution and political institution. The economic institution index is comprised of three ICRG indicators – bureaucracy quality, control of corruption and investment profile; whereas the political institution index consists of other three ICRG indicators – democratic accountability, government stability and law and order. Other explanatory variables include initial GDP per capita, public expenditure on education as a percentage of GDP, and under-five mortality rate. The regression results indicates that after controlling for initial GDP per capita (which has a positive significant association with human capital index), public expenditure on education has a statistically significant positive association and under-five mortality rate has a statistically significant negative association with the human capital index. The highly significant and positive coefficients of both economic and political institution indices suggest strong positive associations between these institutional variables and human capital index. The z-score regression analysis, however, refers to larger importance of political institution over economic institution in human capital development.

The aforementioned analysis points to the fact that better economic and political institutions matter for human capital development. While countries need to make critical spending for human capital development, improvement in institutional environment is unequivocally essential.

Published at the Thinking Aloud on 1 July 2016

Published at The Financial Express on 18 July 2016

Advertisements

How pervasive is regional disparity in primary education in Bangladesh?

Selim Raihan and Mansur Ahmed

CEDI14_1

Sound knowledge on educational performances of different regions across the country can be helpful in the decision making process for better resource allocation and policy formulation. A multidimensional composite measure of educational development, that captures many dimensions such as access, inputs, quality, gender-parity, and outcome, would enable policy makers to target and to channel scarce resources in lagging regions more efficiently.

This paper develops a multidimensional composite index for the primary education development for 483 upazilas (sub-districts) in Bangladesh and identifies the lagging regions for potential policy intervention. More specifically, this paper constructs the Education Development Index (EDI) for the primary education sector of Bangladesh. This index facilitates cross-sectional analysis of the levels of attainment in education among different regions of Bangladesh. Furthermore, it draws policy attention to crucial parameters for achieving equity in access and attainment in educational development.

Bangladesh has one of the largest primary education systems in the world with an estimated 16.4 million primary school aged children (6 to 10 years). This study uses the data from a census which was carried out in 2011, which covered all 11 types of primary schools with a total number of more than 80,000 schools. Education Management of Information System (EMIS) division of Directorate of Primary Education (DPE) under Ministry of Primary and Mass Education (MoPME) undertook the census.

Five broad parameters and 19 sub-parameters (individual indicators) are used in the construction of EDI. The broad parameters are (i) Access, (ii) Infrastructure, (iii) Quality, (iv) Gender Equity, and (v) Outcome. This study has applied the Principal Component Analysis (PCA) method for each broad parameter and calculated weights for each of the indicators within the broad parameter. (For detail methodology of EDI construction and list of indicators, see Raihan and Ahmed, 2016). The objective of PCA is to reduce the dimensionality (number of indicators) of the data set but retain most of the original variability in the data. The overall EDI constructed for this analysis is again a weighted summation of five broad EDIs – access EDI, infrastructure EDI, quality EDI, gender equity EDI and outcome EDI, with weights derived from the PCA on these five EDIs. The index value of 1 indicates the highest educational development with 0 as the lowest development.

Analysis of the aforementioned census data suggests that, despite indicators related to accessibility of schools showed good scenarios, still about 20% schools were not easily accessible to the neighboring residents. Astonishingly, only 20% of schools enjoyed electricity access. Class rooms at the primary schools in Bangladesh were quite crowded as the student-room ratio was 38. Student-teacher ratio was also very high, implying crowded class rooms with low degree of interaction between students and teachers. Still a significant proportion of teachers in primary schools were without bachelor degree. In terms of gender parity at the primary school enrolment, not all upazilas achieved the gender parity. Though Ministry of education set a target that the ratio of female to male teachers should be above 60%, the observed female to male teachers ratio in the census data was about 53%, which suggests need for renewed efforts to reach at that goal. Another important indicator related to gender equity is the percent of schools with girls’ separate toilet. The census data shows that only 40% of schools had separate toilets for girls. Despite Bangladesh achieved remarkable success in primary school enrollment, average pass rate at grade V and school attendance rates were 87% and 85% respectively with wider variations among the upazilas. On average, 1 out of 10 students needed to repeat the same class and 1 out of 20 students dropped out from school.

In terms of access EDI (constructed using two sub-parameters – schools per thousand populations and accessibility of schools), most upazilas performed in the mid-range (0.4 – 0.6), suggesting a significant scope of improvement in terms of accessibility of schools. However, the upazilas around the ‘haor’ (large water bodies) regions in Sylhet division and in Mymensingh division and the upazilas from Chittagong Hill Tracts (CHT) lagged behind other upazilas badly in terms of accessibility. Some other upazilas along the Jamuna River and the Padma River (the ‘char’ lands) also performed poorly. While improvement of accessibility of schools is necessary for most upazilas, these lagging upazilas warrant special attention for their geographical locations. Upazilas located in the metropolitan areas, in contrast, performed well in terms of accessibility.

The patterns of infrastructure EDI (constructed using five sub-parameters – school with safe water, school with electricity, school with toilet per 100 students, average room condition of the school, and student-room ratio) were similar to those of the access EDI. However, the performance of upazilas in terms of infrastructure EDI was worse than that of access EDI. A large number of upazilas were in the lower mid (0.2 – 0.4) of infrastructure EDI, while most of them belonging to the Chittagong Hill Tracts and Mymensingh division. Upazilas in the south west coastal region and along the upper Jamuna River in the Rangpur division also performed poorly.

In terms of quality EDI (constructed using three sub-parameters – students-teacher ratio, qualification of teachers, and availability of teaching-learning materials), though some number upazilas performed in the upper middle range (0.6 – 0.8), only few upazilas were in the top quintile. In fact, quite a few upazilas were in the lower middle range (0.2 – 0.4). Most of the top ten performing upazilas were from metropolitan areas.

In terms of equity EDI (constructed using four sub-parameters – ratio of girls among total students, ratio of female among teachers, schools having separate toilet for girls, and gender equity in dropout rate), most upazilas in Bangladesh performed in the lower middle of the ladder (0.4 – 0.6). Some upazilas performed even poorly. Therefore, despite the ‘satisfactory’ level of gender equity in primary education at the national level, gender parity in primary education is a serious issue in a large number upazilas. Like the access EDI, upazilas from ‘haor’ regions performed poorly in gender equity. Quite understandably, upazilas from urban areas were among the top performing upazilas in terms of equity EDI.

In terms of outcome EDI (constructed using five sub-parameters – gross enrolment ratio, pass rate at grade five, attendance rate, dropout rate and repetition rate), most upazilas performed in the mid-range , implying a room for improvement for all upazilas. Upazilas from the ‘haor’ region, Chittagong Hill Tracts, and poverty-stricken North-Bengal were the worst performers.

The map presented depicts the spatial distribution of composite EDI (constructed using five EDIs – access EDI, infrastructure EDI, quality EDI, gender equity EDI, and outcome EDI). The map shows that very few upazilas were in the highest range (0.8 – 1.0) of EDI. In fact, not many upazilas were in the range of 0.6-0.8 of EDI score. Most upazilas were concentrated in the range of 0.4-0.6. Most of the top ten upazilas were from large metropolitan areas such as Dhaka, Chittagong or Khulna. Lowhajong of Munshiganj and Shibpur of Norsingdi were only the exceptions and these upazilas were also located in close proximity to the capital city Dhaka.  Upazilas in the ‘haor’ region of Sylhet division and Mymensing division and upazilas from the CHT were fatally lagging behind all other upazilas in terms of primary education development. All the bottom ten upazilas were either from the ‘haor’ region or from the CHT. Though the population density in the CHT is low, the upazilas in the ‘haor’ region are home of a sizeable portion of population of the country. Thus, these lagging regions warrant special attention to improve the overall development of the primary education in Bangladesh.

The aforementioned analysis suggests that despite many achievements during the past decades, major improvements are still needed in Bangladesh in order for all children to receive the benefit of quality primary education. Opportunities for good quality primary education in Bangladesh are limited by inequalities associated with wealth, location, ethnicity, gender, and other factors. The major challenges thus include addressing poor quality of education, high dropout rates, promotion of equity and accessing education, and targeted programs for lagging regions.

 Reference:

Raihan, S. and M. Ahmed. (2016). Spatial divergence of primary education development in Bangladesh through the lens of Education Development Index (EDI). MPRA paper 71177.

First published at the Thinking Aloud on 1 July 2016

Published at The Financial Express on 4 July 2016

Why do some countries trade more than others?

Theoretically, trade liberalization results in productivity gains through increased competition, efficiency, innovation and acquisition of new technology. In particular, the changing relative prices induced by trade liberalization cause a re-allocation of resources from less efficient to more efficient uses. Trade liberalization is also thought to expand the set of economic opportunities by enlarging the market size and increasing knowledge spillover effects. Empirical research on international trade also shows that, in general, larger trade-orientation and freer trade, with supporting policies and institutions, can lead to higher welfare for a country than otherwise.

However, a major question remains some way unclear – why do some countries trade more than others? More specifically, does country size matter? How does differences in per capita income affect trade-orientation among countries? Does human capital make any difference? How does tariff liberalization promote trade-orientation? Moreover, does foreign direct investment (FDI) affect trade performance? Furthermore, does geographical location have a bearing, i.e., being an island country or a landlocked country? Also, does membership of the GATT/WTO raise trade-orientation? Finally, does institution matter in trade-orientation?

In order to answer these questions, fixed effect panel regressions using a database covering the period between 1981 and 2014 for 128 countries were conducted. We have defined country’s trade to GDP ratio as the country’s trade-orientation. We want to explain why some countries have higher trade-GDP ratio than others. The explanatory variables are the size of population (to represent country size), per capita real GDP, an index of human capital, domestic average applied tariff rate, and FDI to GDP ratio. Data for all these variables, except human capital, are taken from the World Bank’s WDI, and the data of the human capital is taken from the PWT-8.1. All variables are expressed in natural logarithm. The regression results show that all explanatory variables are statistically significant.

The negative coefficient estimate of the size of population reveals that larger countries tend to be less trade-oriented than their counterparts, as 1% rise in the size of the population is associated with 0.2% fall in the trade-GDP ratio. The reason is that countries with a large population find a ready domestic market and can substitute imports by producing for the internal market. The positive coefficient of the per capita GDP shows that a rise in the real GDP per capita by 10% is associated with a rise in the trade-GDP ratio by 2.2%. The reason behind such an association could be related to domestic producers, with the rise in per capita GDP, becoming more efficient in competing and integrating with their foreign counterparts in the world market. As expected, domestic tariff liberalization is positively associated with higher trade-GDP ratio, as a cut in tariff rate by 10% is associated with a rise in trade-GDP ratio by 0.7%.

The positive coefficient of the FDI-GDP ratio suggests that greater FDI orientation is positively associated with greater trade orientation, and a rise in the FDI-GDP ratio by 10% is positively associated with a rise in the trade-GDP ratio by 0.3%. FDI is assumed to have a positive impact on the export-orientation of any economy, as much of FDI is directed towards the export-oriented sectors. The success stories of East and South East Asian countries have suggested that FDI is a powerful tool of export promotion because multinational companies, through which most FDI is undertaken, have established-contacts and up-to-date information about foreign markets. FDI may also lead to increasing imports in the recipient country as foreign owners tend to have a higher propensity to obtain their inputs from abroad than do their domestically owned counterparts.

Finally, in the case of human capital variable, a rise in the index of human capital by 10% is associated with a rise in the trade-GDP ratio by 9%. This is not surprising! A higher level of human capital is likely to have a positive impact on the perception of the people, as well as on the policy making of the government, in integrating their economy with the world market.

The findings of the LSDV models show that landlocked countries and island countries are 194% and 284% respectively more trade oriented than their counterparts. Both for island and landlocked countries, international trade plays a crucial role in their economic lives as most of these countries are dependent, to an unusual degree, on imported goods and services, including foodstuffs, fuel, equipment and industrial material as well as a wide range of manufactured products. However, interestingly, being a member of the GATT/WTO doesn’t make any difference in terms of trade-orientation.

We have also explored the association between trade-orientation and different institutional variables. The data of these institutional variables are derived from the ICRG database. The fixed effect regression results suggest that countries with better bureaucracy quality, larger democratic accountability, and sounder investment profile are associated with higher trade-orientation. These results are also consistent with findings from studies on the determinants of trade flows which argue distortions or costs placed on firms under inefficient institutions and poor governance can negatively affect trade flows.

Bangladesh’s trade-GDP ratio was only 19.2% in 1981, which increased to 44.5% by 2014. Despite the fact that Bangladesh has been able to raise its trade-GDP ratio by more than two-fold during this period, in 2014, out of the 166 countries, Bangladesh ranked 150th in terms of higher trade-GDP ratio. This suggests, greater trade-orientation in Bangladesh would require further cut in tariff rates, larger FDI-orientation, investment in human capital and improvement in institutional quality.

Published at the Thinking Aloud on 1 June 2016

Published at The Daily Star on 13 June 2016

Unearthing Bangladesh’s Comparative Advantages

Selim Raihan and Md. Jillur Rahman

The analysis of comparative advantage is important from the policy perspective. Trade policies of a country should be tuned to promote export items where the country has comparative advantage.  The Revealed Comparative Advantage (RCA) analysis, suggested by Bela Balassa in 1965, is an ex post analysis of comparative advantage and has been used in many studies. RCA index is used to calculate the relative advantage, disadvantage and trade potential of a certain product in a country.

The RCA index is measured as the ratio of a product’s share in the country’s total export relative to its share in the world’s total export. The formula for the RCA is equal to (Xij/Xit)/(Xwj/Xwt) where, Xij and Xwj are country i’s export and world export of product j respectively, while Xit and Xwt are country i’s total export and world total export respectively. If RCA is greater than unity, the country is said to have comparative advantage in that product; and if RCA is less than unity, the country has comparative disadvantage in that product. The RCA index is popular because of its simplicity, availability of data and for cross-country comparisons. The index is consistent with country’s factor endowment and productivity.

In this article, we are interested to know in which products Bangladesh has comparative advantage, and the dynamic changes of its comparative advantage. We have calculated RCA at 6-digit level of the harmonized system (HS) of classification for the periods between 2001 and 2013. RCA indices for Bangladesh are calculated using the data of export volumes of Bangladesh and the world from the Trade Map database.

Before going into the RCA analysis, let’s first explore how many products Bangladesh exports. At the 6-digit HS code level, there are approximately 5300 products. Figure 1 shows that in 2001, Bangladesh exported 896 products, which, by 2013, increased to a number of 2038. In 2012, Bangladesh exported 2126 products which was the highest among the years under consideration. This suggests that, not only in terms of volume but also in terms of number of products, Bangladesh’s export capacity increased by more than double during 2001 and 2013. On a year-to-year basis, some new products were added to the export basket and some were ceased to be exported. However, there were 375 common products which Bangladesh exported all the years under consideration.

Fig1-RCA

Figure 2 presents the numbers of products at 6-digit HS code where Bangladesh had comparative advantage during 2001 and 2013. In 2001, the number of products with RCA>1 was 316, which, with some year-to-year fluctuations, increased to 382 by 2013. The highest number of RCA>1 was observed in 2007 consisting 483 products. Figure 2 also suggests that the percentage share of RCA>1 products in total number of products declined over time: from 35% in 2001 to 19% in 2013. However, as a percentage of total exports, throughout those years, Bangladesh enjoyed comparative advantage in more than 97% of its total export. Furthermore, over those years, comparative advantage had been consistent for 130 products at the 6-digit level among which 115 products were from readymade garment industries. All these suggest that although Bangladesh was able to expand its export basket during 2001 and 2013, the number of products it had comparative advantage didn’t increase proportionately, which indicates escalated concentration of RCA in certain products.

Fig2-RCA

The escalated concentration of RCA in certain products during the period under consideration is manifested by the fact that Bangladesh’s RCAs had been concentrated around the products in the HS codes 03 (fish and shrimp), 41 (raw hides and skins and leather), 52 (cotton yarn), 53 (raw jute), 61 (knitted readymade garments), 62 (woven readymade garments) and 63 (home textile and jute hessian bags). However, a close look at Figure 3 suggests that Bangladesh’s comparative advantage has been highly concentrated around the readymade garments sector. In 2013, number of products with RCA>1 under the HS codes 61, 62 and 63 accounted for 57% of the total number of products with RCA>1. In 2007, such number was 43%. It should also be mentioned here that, readymade garments account for more than 80% of total export earnings of Bangladesh in recent years.

Fig3-RCA

Although RCA had been concentrated around the readymade garments sector, the average value of RCA declined. The maximum value of RCA in the readymade garments was 495 in 2001, which declined to 184 by 2013. Bangladesh had also been losing the very high comparative advantage it had in garments exports. Figure 4 suggests that, in 2001, Bangladesh enjoyed very high RCA (RCA>100) in 18 garments products, which declined to only 3 in 2013. In contrast, the number of products with RCA less than or equal to 30 increased over time: from 142 in 2001 to 181 in 2013.

Fig4-RCA

Similar analysis, with respect to the leather and leather goods, suggests that there had not been much variations in the number of products having RCA in this sector. And, as in readymade garments sector, Bangladesh had been losing very high comparative advantage it had in this sector. In contrast, Bangladesh had been enjoying consistently very high comparative advantage in jute and jute products, where, in all of 6 products, RCA ranged between 53 and 1068.

The aforementioned analysis shows that during the period under consideration, Bangladesh’s comparative advantage had been concentrated around low-skilled labor intensive readymade garments exports. However, in recent years, compared to early 2000s, there had been some products where Bangladesh gained comparative advantage. These include edible fruits, animal and vegetable fats and oil, preparations of cereals, flour, starch or milk and pastry cooks’ products, preparation of vegetable, fruits, nuts, residues from food industries, rubber and rubber products, copper and copper products, and furniture. However, Bangladesh lost comparative advantage in fertilizers, printing industry’s products, articles of iron and steel, and miscellaneous manufactured articles.

Finally, we are interested to know how tariff rates, both at home and partner country, affect Bangladesh’s revealed comparative advantage at the sectoral level. For this exercise, we have constructed a panel data at 6-digit HS code level for the period between 2001 and 2013. The dependent variable is the RCA which is a binary variable, where it takes a value of 1 if RCA is greater than unity and zero otherwise. The first explanatory variable is the domestic tariff rate at 6-digit HS code level, which is the effectively applied tariff rate and its data is taken from the WITS database. The second explanatory variable is the partner country’s tariff rate, which is calculated as the weighted average of simple tariff rates imposed by top export destination partners of Bangladesh namely USA, EU, Canada and India. Data of partner countries’ tariff rates are taken from the WITS and OECD-WTO database. The fixed effect panel logit regression results suggest that domestic tariff rate is negatively associated with RCA and the coefficient is statistically significant. This suggests that a cut in domestic tariff raises the likelihood of RCA greater than unity among the sectors. In contrast, the coefficient of the partner countries’ weighted tariff rate is not statistically significant. The reason behind the non-association between the RCA and partner countries’ tariff rate could be because of the fact that the large part of Bangladesh’s export to its major partner countries are under different preferences schemes; for example, Bangladesh’s exports enjoy the duty free and quota free market access in the EU market.

Published at the Thinking Aloud on 1 June 2016

Why do countries differ in total factor productivity?

Theoretical and empirical literatures on sources of economic growth emphasized on factor accumulation and factor productivity as two major sources of growth. Though factor accumulation can explain a significant part of economic growth, it can’t explain the sustained long run economic growth, as sustained long run economic growth is attributable to growth in productivity. Productivity is the cornerstone of economic growth. Increases in productivity allow firms to produce greater output for the same level of input, and thus result in higher Gross Domestic Product (GDP).

We should make clear the difference between labor productivity, which is output per worker, and Total Factor Productivity (TFP), which is the ‘ability’ with which all factors are combined to produce outputs. TFP is the part of output which is not explained by the amount of inputs used in production. Essentially, its level is determined by how efficiently and intensely the inputs are used in the production process. TFP growth is usually measured by the Solow residual. TFP plays a key role in economic fluctuations, economic growth and cross-country differences in per capita income.

The scatter-plot using the data for 110 countries in 2011 shows a very interesting association between TFP and log of per capita GDP. The TFP data are derived from the Penn World Table version 8.1 (PWT 8.1) with some required adjustments and extensions. Here the TFP level of USA is considered as 1 and other countries’ TFP levels are indexed against USA’s TFP level. For example, among the South Asian countries, the TFP levels of Bangladesh, India, Nepal, Pakistan and Sri Lanka in 2011 were 0.15, 0.48, 0.11, 0.28 and 0.42 respectively. Similarly, TFP levels of Malaysia and Thailand were 0.47 and 0.65 respectively. Singapore’s TFP level (1.1) was higher than that of USA. The trend line shows a very strong positive association between TFP and log of per capita GDP (the correlation coefficient is 0.9). Nepal and Bangladesh, though on the trend line, are at the lower end of the association. A straightforward policy suggestion for these countries is that a rise in the TFP level is required to raise their per capita GDPs.

Fig1_prod

Why do countries differ in TFP? How to improve the level of TFP? In order to answer these questions, we have run fixed effect cross-country panel regressions for 110 countries for the period 1995-2011 considering log of TFP as the dependent variable. The explanatory variables include log of human capital (an index of human capital per person which is linked to the average years of schooling and the return to education), log of public expenditure on health as % of GDP, and log of trade-GDP ratio. The data source of human capital is the PWT 8.1, and the data of health expenditure and trade-GDP ratio are taken from the World Bank WDI. The logic behind the formulation of this model is that we want to explore how cross-country differences in statuses of education, health and openness affect the cross-country differences in the TFP. The regression results suggest that all three explanatory variables are statistically significant with expected signs. 1% rise in the human capital index is associated with 0.39% rise in the TFP. Also, 1% rise in the ratio of public health expenditure to GDP is associated with 0.03% rise in the TFP. Finally, 1% rise in the trade-GDP ratio is associated with 0.03% rise in the TFP.

In extended models, we have found that the ratio of FDI to GDP is positively and significantly associated with the TFP. 1% increase in the FDI-GDP ratio is associated with 0.01% rise in the TFP. Furthermore, institutional variables like bureaucracy quality and investment profile (from the PRS database) are positively associated with the TFP with statistical significance. Improvements in the bureaucracy quality and investment profile by one unit are associated with the rise in TFP by 0.03% and 0.01% respectively.

The aforementioned analyses point to some very obvious policy suggestions. Countries like Bangladesh need to attach decisive emphasis on improving their currently low levels of human capital. This can happen by enhancing investing on education and health quite a lot in order to increase the efficiency in using inputs in the production process thus raising the level of TFP. Also, larger trade and FDI orientations and improvement in the quality of institutions are indispensably important.

Published at the Thinking Aloud on 1 May 2016

Published at The Financial Express on 10 May, 2016

Does export orientation lead to higher productivity? Firm-level evidence from Bangladesh

Selim Raihan, Nafiz Ifteakhar and Mir Tanzim Nur Angkur

For long, empirical studies on the role of exports in promoting growth in general, and productivity in particular, used data at the country or industry level to test whether exports promote productivity growth or vice versa. However, a series of empirical studies since early 1990s started using firm level data to look at differences between exporters and non-exporters in various dimensions of firm performance, including productivity.

Two alternative but not mutually exclusive hypotheses can be mentioned why exporting firms can be expected to be more productive than non-exporting firms. The first one relates to the fact that firms which are considered to be more productive than others are likely to participate in export markets – the so called ‘self-selection’ of the more productive firms into export markets. The second one relates to the notion of ‘learning by exporting’ hypothesis which suggests that after entering the export market, firms are able to acquire new knowledge and adopt new expertise which eventually leads to higher level of productivity. Though there is sizeable evidence that exporters perform better than non-exporters, the issue of the direction of the causality between exports and productivity is still debated. While in the contexts of more advanced countries most studies find evidence that the export premium is due to a self-selection process, a number of recent studies on less developed countries tend to endorse the learning effect.

Against this backdrop, this paper explores how export orientation affects firm-level productivity by looking at the range of determinants of productivity of manufacturing firms in Bangladesh. Review of empirical studies suggest that there could be several factors, i.e. firm size, firm age, share of firm’s output in the industry, export orientation measured as percentage of total firm’s output that is exported, which may affect firm’s productivity. Our measure of firm’s productivity is the total factor productivity (TFP) which is derived using the Cobb-Douglas production function framework. Specifically, we have regressed log of output (calculated as total sales of firms) on log of capital (measured as netbook value of fixed assets of the firms) and log of labor (measured as total number of employees) to get the output elasticity of capital and labor which are then used to estimate the total factor productivity (TFP).  To get unbiased estimates of those elasticities in the presence of industry fixed effects, we have included industry dummies in the above regression. We have used the dataset of “The World Bank, Enterprise Survey-Bangladesh” for 2007 and 2013 and have only considered firms belonging to the manufacturing sector. Table 1 shows the industry descriptions along with the distribution of firms for both 2007 and 2013. We have estimated TFPs of firms separately for 2007 and 2013 by following the same procedure described above.

Tab1_export

In order to explore the effect of export orientation on the productivity of firms we have run cross-section regressions for 2007 and 2013. For both years we used the same model and Table 2 shows the estimated results. The dependent variable of our model is total factor productivity. The main explanatory variable is the export orientation while the set of control variables include firm size, firm age, firms’ output share and internet connection. In the regression models, export orientation of the firm is represented by a dummy variable, where the dummy variable takes the value of 1 if the firm exports 25% or more of its total output.

Tab2_export

For firm size, we have also taken three dummies- large, medium and small based on the number of employees. For capturing the effect of technology on productivity, we have taken internet connection dummy. Internet connection dummy will take the value of 1 if the firm communicates by e-mail.

The cross section regression result of 2007 suggests that firm age has positive and significant effect on productivity, while the result of 2013 indicates no such relationship. For 2007, it is estimated that an increase in firm age by one year would lead to a rise in productivity by 1%. For both years, firm size has an effect on productivity. In particular, both medium and small sized firms tend to be less productive than large firms. The firm’s output share is found to have a positive and significant effect on productivity for both 2007 and 2013 respectively. For 2007, one percentage point increase in firm’s output share would lead to a rise in firm’s productivity by 31%, while for 2013 such productivity rise would be by 12%. Now considering the effect of internet connection on productivity, firms with internet connections are found to be more productive than firms with no internet connection for both 2007 and 2013.

Our variable of interest is the export orientation which is found to have a positive and significant effect on productivity for 2007 and 2013. For 2007, we have found that on average productivity of a firm that exported 25% or more of its output was 156% higher than a firm that exported less than 25% of its total output. Such productivity difference was however reduced in 2013, as productivity of a firm that exported 25% or more of its output was 112% higher than a firm that exported less than 25% of its total output.

From the aforementioned analysis, it can be said that larger firms are more productive as compared to small and medium sized firms. Larger firms, due to economies of scale, are able to reap some benefits which help them to utilize resources more efficiently. Firms which started earlier in an industry also tend to be more productive than firms which entered in the industry later. This is due to the fact that already established firms have advantages over new firms in case of production, marketing, etc.

Also output share of the firm belonging to an industry (measured by the proportion of sales of firms in total industry sales) may influence the firm’s productivity. Firms with higher output share can positively affect productivity, as dominant firms hold the necessary resources and technical skill and expertise as compared to firms with low output share. It is also found that firms which have access to internet connections can benefit from lower communication cost and can also communicate with its clients and suppliers timely and thus leads to higher productivity.

Finally, the regression results confirmed that the exporting firms in Bangladesh are more productive than their counterparts. There could be several reasons for this. The learning process may work through technical supports from external buyers, and/or through the exposure to competition in the international markets, which can result in knowledge, technology, and efficiency gain from exporting.

Published at the Thinking Aloud on 1 May 2016

Published at The Financial Express on 4 May, 2016

Political economy of regional integration: Where do we stand in South Asia?

The aspiration for deeper regional integration is high on the political agenda of most of the leaders in South Asia. Since the early 1980s South Asian Association for Regional Cooperation (SAARC) has been working as an economic and geopolitical organization for South Asian countries with the aim of deeper regional integration and cooperation in areas of economic, trade and other common regional issues. Until now, there have been some achievements. Still, frustration prevails, as actual implementation of agreements often does not match the declared ambitions. The resulting implementation gap is most commonly attributed to the lack of political will and leadership, institutional weaknesses and capacity and resources constraints.

The dominant literature has looked primarily at the narrow economic factors influencing regional integration. However, to have a better and systematic assessment of the factors driving and constraining regional integration, it is important to explore the political economy dimensions. While policy makers and stakeholders are often aware of such political economy dimensions, they are generally discussed only informally or in ad hoc manner. A systematic discussion of the political economy factors around the regional integration agenda can generate a broader awareness among stakeholders that may ultimately lead to more realistic and effective regional policy design and processes.

From a political economy perspective, there could be three interconnected drivers for a deeper regional integration. These are economic drivers, political economy drivers and extra-regional drivers.

PE of regional integration

The economic drivers include four integration processes: market integration, investment integration, growth integration and policy integration. ‘Market integration’ emphasizes on the integration in trade in goods and services through the removal of tariff and non-tariff restrictions. ‘Growth integration’ is the integration of economic growth processes of the respective countries in a way that growth in one country benefits growth processes in other member countries. The ‘investment integration’ calls for promotion of regional investment and trade nexus. Finally, the ‘policy integration’ is the harmonization of economic and trade policies of the countries for a deeper regional integration.

However, the aforementioned four integration processes need favorable political economy (PE) drivers. The political-economy perspective considers how various players influence the national and regional decision-making context, and what impact their actions (or lack of action) have on the integration agenda. The first PE driver is the ‘primary institution’ which are the official institutions at the regional level and in respective countries entrusted to carry out the agenda of regional integration. In South Asia, the SAARC Secretariat and relevant ministries in the member countries are such institutions. The second PE driver is the ‘secondary institution’ which are private sectors, private sector associations, civil society organizations and media. Primary and secondary institutions are a combination of market and non-market actors that govern economic and political environments in the region. The third PE driver is the ‘regional public good’ which includes regional infrastructure and the status of regional trade facilitation. In South Asia, status of such ‘regional public good’ is very weak. ‘Structural factor’ is the fourth PE driver which includes historical processes and geographic factors that shape the types of political, economic and socio-cultural institutions. In South Asia, landlockedness of Nepal, Bhutan and Afghanistan, political rivalry between India and Pakistan, and huge differences in the sizes of the countries where India accounts for around 80% of the regional GDP, trade among the South Asian countries primarily through land borders are such structural factors. The final PE driver is the role of the ‘political elite’. Strong and visionary leaderships are needed from the political elites to eliminate any ‘trust deficit’, which can emerge as a result of a variety of the ‘structural factors’ mentioned above. In South Asia, such ‘trust deficit’ is often highlighted as one of the major barriers for a deeper regional integration. Also, there are concerns with regard to hesitant and inconsistent leaderships from the political elites of these countries, especially from India, in taking the regional integration agenda to a higher level.

Finally, the extra-regional drivers include a wide range of global economic and political factors that can have influence over the region. In South Asia, countries are at different levels and with different patterns of integration with the extra-regional drivers.

There are now convincing evidences that a deeper regional integration is needed for generating and sustaining economic growth and reducing poverty in South Asia. Intra-regional trade in South Asia has been low, but there are signs of huge potentials. For a deeper market integration in goods, full implementation of SAFTA is needed with emphasis on further liberalization of intra-SAARC tariffs, reduction in the sensitive list, and establishing effective mechanisms to deal with the NTMs/NTBs.

Intra-regional services trade and intra-regional investment are also low in South Asia. Regional and sub-regional efforts have to be promoted for different trade and transport facilitation measures, for cooperation in energy generation and transmission, and for linking energy cooperation and trade and transport facilitation to investment and growth processes of these countries. Promotion of intra-regional investments and attracting extra-regional FDIs in goods and services sectors in general, and energy and infrastructural sectors in particular will be very crucial for South Asia to integrate further. There is a continued need for a greater integration in trade, macro, financial and industrial policies in the region.

A deeper regional integration in South Asia requires clear and visible leadership from the political elites in this region, especially from India, in taking the agenda forward. The political elites have to be convinced and act accordingly to reduce the ‘trust deficit’. Regional institutions, like SAARC Secretariat, have to be institutionally reformed and reoriented with much stronger engagements from the respective ministries and relevant organizations of the member countries. Business associations, civil society organizations and media have to pursue the regional integration agenda in South Asia more pro-actively than ever.

Published at the Thinking Aloud on 1 April 2016

Published at The Daily Star on 12 April 2016