Part I - The Economic and Institutional Background of Bangladesh’s Development

2 Bangladesh’s Development Achievements and Challenges

Selim Raihan and François Bourguignon

I Introduction

Bangladesh’s economic growth and development experiences over the past five decades, since independence in 1971, have generated much interest among academics and development practitioners both home and abroad. From its war-torn economy of 1972 until now, Bangladesh has been able to increase its per capita GDP in real terms¹ by 3.7 times (from US$ 460 in 1972 to US $1,700 in 2018), cut down the poverty rate from as much as 71% in the 1970s to 20.5% in 2019, become the second largest exporter of ready-made garments (RMG) in the world, and registered some notable progress in social sectors. In 2015, Bangladesh graduated from the World Bank’s classification of low-income country to the lower middle-income country category. Also, in 2018, the country met the first review of the three criteria required to graduate from the least developed country (LDC) status and is on track to meet the criteria under the second review in 2021 to finally graduate out of LDC status by 2024. At the same time, Bangladesh’s aforementioned development has happened in a context of a widely recognised weak institutional capacity. Bangladesh has almost always been ranked in the bottom part of most international rankings of governance indicators, as summarised in the Worldwide Governance Indicators database. Also, up to the late 2000s, its political climate was extremely tense, unstable, and often violent. All these factors have prompted some to term Bangladesh’s development the ‘Bangladesh paradox’ or the ‘Bangladesh surprise’.²

Over the past five decades, the major factors behind Bangladesh’s growth and development achievements have been both internal and external. The major internal factors include an overall stable macroeconomy, large expansion of the private sector, robust growth in exports driven by the performance of the RMG sector, robust growth in remittances, resilient growth in the agricultural sector, a reasonably ‘working’ political climate over the last 12 years, some expansion of social protection programmes, and a wide coverage of social needs by non-governmental organisations (NGOs). The major external factors include favourable market access in major export destinations, reasonably stable economic conditions in Bangladesh’s major trading partner countries, Bangladesh’s stable political relations with neighbouring countries, some degree of regional cooperation in South Asia (especially with India), and Bangladesh’s ‘weak’ financial linkages with the global economy, which cushioned Bangladesh from the Global Financial Crisis. In 1990, Bangladesh was the 50th largest economy in the world (in international dollars). Impressively, by 2018, Bangladesh improved its position in this ranking to 33rd. According to PricewaterhouseCoopers (2019), it should become the 28th largest economy by 2030 and close to the 20th by 2050. The main question addressed in this chapter is whether the internal and external factors just mentioned will indeed keep pushing Bangladesh’s economy up at the same speed, or possibly even faster. More precisely, the issue considered is what constraints the economy may face in the future and whether further development can be achieved without a significant new direction in policy.

There are concerns that the weak institutional capacity of the country may work as a binding constraint as Bangladesh attempts to meet the stiff targets of the Sustainable Development Goals (SDGs) by 2030, and as it aspires to become an upper middle-income country by 2031. Moreover, the dividends from the so-called ‘Bangladesh surprise’ are likely to be on a decline as the country is confronted by several serious economic challenges. These include the slow progress in the structural transformation of the economy, the lack of export diversification, the high degree of informality in the labour market, the slow pace of formal job creation, the poor status of physical and social (i.e. education, healthcare) infrastructure, the slow reduction in poverty and rising inequality, and the consequences of the COVID-19 pandemic.

Against this backdrop and keeping in mind the ambitious development targets the country wants to meet in the next two to three decades, this chapter analyses the major development achievements of the Bangladeshi economy until today and seeks to identify the major challenges it will have to address in the future. Whether these challenges can be overcome, and which reforms can be undertaken for this to happen, depends in turn on the institutional context of the country. This particular aspect will be considered in the institutional diagnostic chapter at the end of the volume, after a deeper reflection on Bangladesh’s politico-economic institutions, through the analysis of key economic sectors and major socio-economic issues.

Focusing exclusively on economic and social issues, this chapter first analyses the sources of growth and possible limitations of the present development regime (Section II). It then evaluates the financing constraints faced by the economy in general, and by the public sector in particular (Section III). The next sections (Sections IV–VI) focus on sources of concern in social and environmental areas and the COVID-19 crisis. The conclusion summarises the main results of this chapter.

II Sources of Growth and Possible Limitations of the Present Development Regime

In this section, economic growth in Bangladesh is first analysed at the aggregate level, before considering the structural evolution of the economy, the key role played by trade, and the constraints arising from lagging infrastructure and progress in the business environment.

A Aggregate Growth

The long-term trend in the GDP growth rate shows that Bangladesh has steadily increased its rate of growth over the past 47 years, since independence in 1971 (Figure 2.1). Starting from a highly volatile growth rate in the 1970s, growth became more stable and slightly faster in the 1990s and has accelerated since the turn of the millennium. From an average rate equal to or below 4% per annum in the 1970s and 1980s, growth accelerated and shot up to over 5% in the 2000s, then exceeded 6% for several years during the 2000s, and has crossed the 7% mark in recent years. Bangladesh has been able to increase the average GDP growth rate by 1 percentage point in each decade since the 1990s. In 2018, the country achieved its highest growth rate in the past four decades: 7.9%.³ In Figure 2.1, the steps indicate the average growth rates of the decades, and the highlighted years indicate the years of political transition in Bangladesh.⁴

Figure 2.1 Real GDP growth rate in Bangladesh.

Source: World Development Indicators (WDI)

GDP per capita has grown less rapidly because of population growth, but because the latter has significantly slowed down over the last 40 years, the growth acceleration is even more noticeable for GDP per capita. The 10-year average growth rate increased from 1.5% in the 1980s to more than 5% over today.

Compared to other developing countries, in international purchasing power (purchasing power parity 2011 US$), Bangladesh’s per capita GDP was US$ 3,880 in 2018, around 1.5 times the average for LDCs, 57% of the average for lower middle-income countries, 50% of the South Asian average, and 23% of the average for upper middle-income countries. In terms of growth, however, Bangladesh has done substantially better than the average country in all of these groupings.

Interestingly enough, the vigorous acceleration of GDP growth since 1990, from 4% to 6.6%, is in line with the increase in the share of investment in GDP. The investment to GDP ratio was 17.5% around 1990. It increased regularly since then and reached 31% in 2018. However, two remarks are in order here. First, further growth acceleration, as targeted by the present government, may require a substantial increase of the investment to GDP ratio. The Perspective Plan 2041 targets an 8.5% real GDP growth rate by 2025. Given an average incremental capital output ratio of 4.3 for the period 2014–2018, the investment to GDP ratio should be more than 36% by 2025, which represents a growth rate for investment that is much faster than what has been observed in the recent past. Second, an important aspect of Bangladesh’s investment regimes in the 1990s and 2000s is that the major contribution to the growth of the investment to GDP ratio came from the rise in private investment and its share in total investment. However, the GDP share of private investment has remained stagnant in recent years, so the growth of the overall GDP share of investment has been mainly due to its public component.⁵ The stability of the ratio of private investment to GDP might make it possible to sustain present growth rates but may be a concern for further acceleration.

Conventional growth accounting⁶ is helpful in regard to obtaining insights into the aggregate sources of GDP growth, that is what portion is derived from capital accumulation, the growth of the labour force, and total factor productivity (TFP) in general. Figure 2.2 shows the result of that decomposition on an annual basis since 1980, although the 1980s may not be relevant given that this was a very tumultuous period from both a political and an economic point of view.

A. Growth decomposition.

Source: Authors’ calculation using the Penn (World Table 9.1) database

B. Percentage points contribution to growth rate.

Source: Authors’ calculation using the Penn (World Table 9.1) database

Figure 2.2 Growth decomposition in Bangladesh (1980–2017).

Over the next three decades, capital accumulation was, on average, the most important factor of growth. This dominance increased over time, in line with the acceleration of investment just mentioned. The contribution of labour steadily declined over time, in part because of slowing population growth, but above all because capital grew much faster. TFP growth is the residual: it represents the productivity gains that were independent of the accumulation of capital, making up a little less than one-third of total GDP growth.

Subtracting the contribution of labour to GDP growth in the preceding decomposition is equivalent to considering the growth of the average productivity of labour: that is, GDP per worker. It has steadily increased over time.

B Structural Transformation

Part of the TFP growth in the preceding decomposition is caused by structural changes taking place in the economy. Over time, the allocation of production factors changes with net movements across sectors of activity. If productivity is not the same in the sectors of origin and in the sectors of destination, these movements affect the overall level of productivity in the economy. This mechanism has been seen as a major driver of development ever since the work of Lewis (1954). However, it has recently been found that, since the 1990s, this process has worked in the opposite way. Working on sub-Saharan African and Asian countries, McMillan and Rodrik (2011) found that structural change in those countries had a negative impact on overall labour productivity. However, Bangladesh did not follow such a pattern in the past, and this remains the case today.

Table 2.1 shows the evolution of the GDP and employment structure by sector of activity between 1991 and 2018. The contribution of agriculture, both to GDP and employment, sharply declined during 1991 and 2018, and those of non-agricultural sectors, especially the services sector, increased. Comparing the structure of GDP and that of employment, it is readily apparent that in 1991, productivity was the lowest in agriculture and the highest in ‘other industry’, followed by services. This means that net labour movements have been from low-productivity to high-productivity sectors, that is mostly from agriculture to services and ‘other industry’. Structural change has thus had a positive effect on the overall labour productivity in the economy, in line with Lewis (1954).

Table 2.1 Structural change in the economy (percentage share of total)

Broad sectors	GDP		Employment
	1991	2018	1991	2018
Agriculture	31.7	13.1	69.5	40.1
Manufacturing	13.9	17.9	12.4	14.2
Other industry	7.2	10.6	1.2	6.3
Services	47.2	58.4	16.9	39.4

Source: World Bank, WDI.

At the same time, Figure 2.3 shows that in the 1990s, labour productivity increased overall in all sectors of activity except ‘other industry’, possibly because of structural changes within that sector. These sectoral productivity increases thus provided a second source of overall productivity gain, in addition to structural change.

Figure 2.3 Bangladesh’s labour productivity by sector and for the whole economy (GDP per worker 2018 = 100).

Source: Authors’ calculation, from World Bank, WDI

Figure 2.3 also shows some interesting short-run variations in productivity gains. For instance, the 1991–1995 period exhibited a huge increase in manufacturing productivity – a doubling in four years – apparently precisely at the time the RMG sector was taking off. It is also striking that productivity did not change in the four sectors of activity between 1995 and 2005, but overall productivity increased because of net labour movements from agriculture to other sectors, that is structural change.

Avillez (2012) provides an interesting method for decomposing aggregate labour productivity gains⁷ into structural change and within-sector productivity growth. This distinguishes three components: (1) the within-sector effect (WSE) reflects the overall impact of productivity growth within individual sectors on aggregate productivity; (2) the static reallocation effect (SRE, Denison effect) captures the impact of the reallocation of employment from less productive to more productive sectors (i.e. it arises from sectoral differences in initial productivity levels); and (3) the dynamic reallocation effect (DRE, Baumol effect) describes the impact of reallocating employment to sectors with fast- or slow-growing productivity (i.e. this results from asymmetric productivity growth rates across sectors).⁸ The last two effects correspond to the impact of structural change on growth.

Table 2.2 presents the decomposition of labour productivity growth in Bangladesh into WSE, SRE, and DRE using data for the four sectors appearing in Table 2.1 and three approximately 10-year periods. It appears that while total labour productivity growth doubled after 2000 compared to what it was in 1991–2000, the contributions of WSE, SRE, and DRE changed quite dramatically across the periods. WSE increased by more than 10-fold between the first and the last sub-periods, while structural change, that is the sum of SRE and DRE, which was strong and approximately constant in the first two sub-periods, declined significantly in the last decade. Interestingly, the dynamic productivity effect, DRE, was strongly negative in the 1990s, suggesting that labour shifts during that period were towards sectors with higher but slow-growing productivity. That effect then practically vanished.

Table 2.2 Decomposition of labour productivity growth in Bangladesh

	1991–2000				2001–2010				2011–2018
	Total	WSE	SRE	DRE	Total	WSE	SRE	DRE	Total	WSE	SRE	DRE
GDP per worker	0.166	0.031	0.246	–0.110	0.356	0.111	0.261	–0.016	0.338	0.240	0.077	0.020

Note: GDP per worker is in purchasing power parity constant 2011 international $.

Source: Authors’ calculation, from World Bank, WDI.

Figure 2.4 suggests that in the recent decade, the major contribution to labour productivity comes from WSE (71%), followed by SRE (23%) and DRE (6%). This suggests that productivity growth within individual sectors, rather than the reallocation of employment from less productive to more productive sectors and reallocating employment to sectors with growing productivity, has been the driving factor behind labour productivity growth during the 2011–2018 period. What is surprising in that evolution, and in the difference compared to the previous period, is that with agriculture still accounting for 40% of the labour force and having a substantial productivity gap, there would seem to still be a huge potential for structural change. Productivity has increased very quickly in the service sector during the last decade, with that sector being responsible for the high WSE effect in that sub-period. But it follows that it created less jobs than in the previous sub-periods, thus weakening labour movements and the structural change effect on GDP per worker, or closely related GDP per capita.

Figure 2.4 Percentage contribution to labour productivity growth in Bangladesh.

Source: Authors’ calculation from World Bank, WDI

To understand better the reason for this slowing down in structural change requires disaggregating the services sector, which comprises very different types of activity, ranging from financial services to informal retail trade or ancillary jobs. It cannot be excluded that what is being observed is a mounting productivity gap within that sector that hides a falling gap between agriculture and low-productivity service activities, and thus a weakening of incentives to move away from agriculture, while huge productivity gains and faster growth take place in high-productivity services like finance. Unfortunately, employment data do not allow for a more detailed analysis that would make it possible to test that hypothesis.

C Trade

After independence in 1971, Bangladesh adopted a highly restricted trade regime that was characterised by high tariffs and non-tariff barriers and an overvalued exchange rate system, in support of the Government’s import-substitution industrialisation strategy. This policy was pursued with the aim of improving the country’s balance of payment position and creating a protected domestic market for manufacturing industries (Bhuyan and Rashid, 1993). Then, in the mid-1980s, the trade regime registered a major shift, when a moderate liberalisation reform was initiated. Yet the boldest transition from a protectionist stance to a freer trade regime took place in the early 1990s, which led to a drastic reduction in the average tariff rate, from as high as 105% in 1990 to 13% in 2016 (Raihan, 2018a).

The liberal import policies during the 1990s and onwards led to a fast growth of imports. Figure 2.5 shows that the import to GDP ratio increased from 13% in the early 1990s to 23.5% today. The export to GDP ratio had started to rise by 1990, from a very low level of slightly above 5% of GDP in the 1980s, but it then closely followed imports, reaching an all-time peak of 20% in 2010. The trade deficit is endemic to the Bangladesh economy, even though it has been rather stable over time, fluctuating at around 5%, with some widening since the mid-2000s. In this long-run ascending trade perspective, there may be some concern with respect to the substantial decline in the GDP share of both imports and exports over the last five years. Is this the result of temporary shocks or the sign of deeper structural changes?

Figure 2.5 Foreign trade.

Source: World Bank, WDI

Over the years, the composition of imports has also changed in Bangladesh. There has been a move away from the heavy dominance of food imports in the early 1970s to industrial raw materials and machinery today. The change in favour of industrial raw materials and capital machinery is partly linked to the rapid expansion of the manufacturing sector, particularly RMG exports, which is the dominant feature of the Bangladeshi economy over the last four decades.

Bangladesh’s fast export growth since the late 1980s has indeed been overwhelmingly driven by the dynamism of the RMG sector alone. While the export basket was heavily dominated by jute and jute products in the early 1970s, the composition of exports in Bangladesh has evolved steadily in favour of RMG products. Today, they constitute more than 80% of export earnings. The drastic change in the import mix, as well as the spectacular surge of RMG manufacturing exports, bear witness to the structural transformation of Bangladesh’s economy and the role played by foreign trade.

The growth of Bangladesh’s RMG exports had its origins in the international trade regime in textiles and clothing, which, until 2004, was governed by Multi-Fibre Arrangement (MFA) quotas. This quota system restricted competition in the global market by providing reserved markets for numerous developing countries, including Bangladesh, where textiles and clothing items were not traditional exports. The duty-free access of Bangladesh’s RMG products to the European Union (EU) has also greatly supported the growth of the sector. Yet the surge in RMG exports from Bangladesh took place precisely at the time of the extinction of the MFA regime and its successor Agreement on Textile and Clothing: that is, at the time the international market was liberalised and the RMG Bangladeshi sector appeared as particularly competitive relative to other providers in developing countries (outside China).

The growth of RMG exports has been one of the main growth drivers of Bangladesh’s economy over the past three decades. Although the sector directly contributes to only a little more than 6% of GDP, its indirect contribution is much larger: accounting for both backward and forward linkages, it accounts for at least twice as much. With exports increasing at an average annual rate of 12% over the last two decades, and probably more for its RMG component, the sector may have directly contributed to close to 1.4% GDP growth during that period. Yet such an estimate does not include the indirect effects on aggregate demand, on easing the foreign currency constraint, and on investment incentives. The econometric exercise reported in Annex 2.3 suggests an even larger contribution when all these effects are accounted for. With an estimated elasticity of GDP to the volume of exports of 0.22, a 12% growth of exports generates a 2.6% increase in GDP – a little less than half the average growth rate since 2000.

Despite this impressive growth record, it bears emphasis that the export base and export markets have remained rather narrow in Bangladesh, which is a matter of great concern. Undiversified exports, both in terms of market and product range, are likely to be much more vulnerable to external and internal shocks than well-diversified exports. Despite various incentives provided by trade policy reforms, Bangladesh’s manufacturing sector seems to have failed until now to develop a diversified export structure. Bangladesh’s export markets have been highly concentrated – North America and the EU being its major clients. Bangladesh’s growth is thus heavily dependent on economic activity in these two parts of the world. As far as the export product range is concerned, UNCTAD’s export concentration index suggests that Bangladesh’s export concentration has increased over the last two decades. In fact, it is much higher today than the averages for LDCs, lower middle-income countries, upper middle-income countries, and South Asian countries.⁹

Despite a very high share of manufacturing exports in total merchandise exports, the export basket of Bangladesh thus remains highly concentrated around low-complexity products with lower growth prospects in world markets. A measure of the complexity of the economy is the Economic Complexity Index (ECI) of the Centre for International Development at Harvard University, which measures the knowledge intensity of an economy by considering the technical knowledge that is incorporated into the products it exports. In this respect, Bangladesh performs very poorly, with an ECI that has deteriorated over time. Following the current view on the relationship between economic complexity and the level of income, Bangladesh’s growth prospects would not seem favourable, as a low ECI would be compatible with a relatively low level of GDP per capita.¹⁰

On the import side, it must be stressed that, despite the liberalisation of tariffs, Bangladesh’s average applied tariff rate in 2016 was the highest in South Asia, and much higher than those of the countries in Southeast Asia. Also, in 2016, the share of tariff lines with international peaks (rates that exceed 15%) in total tariff lines was as high as 39%, which was much higher than most of the South Asian (except Nepal and Pakistan) and Southeast Asian countries.¹¹ Given this scenario, there seems to be room for further tariff liberalisation in Bangladesh as part of a broader trade policy reform aimed at accelerating export growth and export diversification (Raihan, 2018a; Sattar, 2019).

The need to diversify exports thus appears as a key policy agenda in Bangladesh. It is essential to sustain long-term growth and employment creation, as it is far from clear that Bangladesh will be able to keep increasing its global market share in low value–added RMG products as fast as it has done in the past. Low-wage competitors are appearing in sub-Saharan Africa, for example, Ethiopia, and the whole industry is mechanising so that low labour costs may not be as strong a comparative advantage as before, and, domestically, the sector will provide less jobs. Foreign clients of Bangladeshi RMG firms are also more and more attentive to the labour conditions among their suppliers, particularly after the 2013 Rana Plaza accident that cost the life of a thousand employees. At the same time, the pressure for higher wages and better labour conditions is increasing in the country, making the RMG sector less competitive internationally.

D Remittances and Foreign Direct Investment

Remittances can be considered the second major driver of growth in Bangladesh. Figure 2.6 shows the evolution of remittances sent back home by Bangladeshi workers employed abroad since the mid-1970s. Starting from a low base, remittances increased at slow rates until the turn of the millennium. Since then, however, they have increased at very sharp rates, reaching US$ 14 billion around 2013 but then remaining roughly constant. In terms of GDP, they went up from 1% around 1995 to 10% in 2008–2012, falling back to 6% today. A reduction of transaction costs and remitting delays, but above all the fast-growing demand of foreign workers in the Gulf, where the net inflow of migrants multiplied by 5 between 2000 and 2010, made a major contribution to the increase in the flow of remittances.

Figure 2.6 Remittances and foreign direct investment.

Source: World Bank, WDI

Remittances contribute to domestic growth essentially by increasing aggregate demand, that is domestic spending, while at the same time eliminating foreign currency bottlenecks that could constrain an increase in production. Other things being equal, the same increase in spending would not have resulted in more production if it had a purely domestic origin, because of the constraint arising from the financing of imported raw materials and equipment required by any increase in domestic output. Thus, one may estimate that, with an average annual growth rate of 11% between 2000 and 2018, remittances have contributed to roughly 1 percentage point of annual GDP growth.¹² The econometric exercise reported in Annex 2.3 suggests a long-run elasticity of GDP to real remittances of 0.14. When all indirect effects are accounted for, the contribution of remittances to GDP growth is thus 1.5% per annum on average. Together with export growth, they thus generate roughly two-thirds, that is 4.1% annually, of overall growth.

Foreign direct investment represents another regular inflow of foreign currency that has the potential to accelerate domestic growth. In the case of Bangladesh, however, that contribution has been only marginal. If the amount of foreign direct investment has increased since the mid-1990s, it has represented only around 1% of GDP after the mid-2000s, and a little more than that over recent years. This does not compare well with what is observed in LDCs (3.3% on average), or in Southeast Asian manufacturing exporters such as Cambodia, Laos, and Vietnam – all RMG competitors of Bangladesh – where the foreign direct investment to GDP ratio is above 6%. At the same time, Bangladesh’s domestic investment effort is larger and more dynamic.

E Infrastructure and Megaprojects

If Bangladesh does not rank high either in the availability or quality of its infrastructure according to the World Economic Forum 2019 Global Competitiveness Index, its overall score is nevertheless comparable to those of African countries at a comparable level of GDP per capita (Ghana, Kenya) and not much below dynamic middle-income industrialising Asian countries except Vietnam. However, the situation differs depending on the kind of infrastructure being considered. If Bangladesh compares well with other lower-middle income countries in transport infrastructure, the gap is more pronounced with respect to power and water. True, 80% of the population now has access to electricity, a spectacular progress mostly achieved over the last 10 years. But with an average consumption of 330 kWh per person and per year, Bangladesh keeps lying well below the rest of South Asia – except Nepal –, industrialising Asia, and even some African country like Ghana. Moreover, outages in urban areas are still frequent, which entails substantial costs. As far as access to safe drinkable water is concerned, on the other hand, the situation is definitely worse in Bangladesh than in most countries at a comparable level of development.

As in other developing countries, considerable efforts are still to be made to provide to economic agents the infrastructure that development, especially industrialisation, imperatively requires. Lately, these efforts have taken the form of a few ‘mega-projects’ meant to accelerate the pace of development and make it more transformative. These essentially are major infrastructure projects, with complex logistics, high technological requirements, and very substantial funding needs. Undertaken in key sectors such as transportation, energy, seaport, airport, mining, etc.,¹³ they are intended to have a significant impact on the country’s development through job creation, enhanced connectivity, regional trade facilitation, geographic integration, and energy provision.

These mega-projects also raise several concerns, which relate to: (i) their financing; (ii) their implementation, including construction delays and their final cost; and (iii) their political significance within the context of Bangladesh’s political economy.

The total cost of the seven largest mega-projects has been estimated at USD 31.2 billion,¹⁴ of which only 9.9 billion was funded on government funds. The rest was mostly funded by loans, part of which from foreign donors. The foreign debt of Bangladesh was significantly affected by these projects, and the issue arises of whether their return will cover reimbursement. In this regard, the systematic over-running of both construction delays and costs may be a worry. They stem from a general lack of experience in dealing with such large-scale projects but also to a lack of transparency and political pressure in selecting the implementing agency, and in procurement. In effect, many of implementation issues reflect complex dynamics of rent generation and sharing which are typical of large economic projects (Hassan and Raihan 2018).

Even though their economic utility is not in doubt, the multiplication of these mega-projects owes much to political factors. These include vested interests created by a range of stakeholders benefiting from these projects, their multiplier effects, the clout and image associated in public opinion with such prestigious project but also the relative ease in spending money quickly, including the rent-sharing opportunities that it creates. In this respect, the key role that the Bangladesh Army has played in some of the projects may be underscored, as it may have been a way of distracting it from politics.

In summary, the recent and spectacular expansion in infrastructure projects has the potential to deliver significant economic development impacts, alleviating some key binding constraints, in terms of power, transport, and connectivity. Yet, management flaws and rent-seeking behaviour made those projects notably more costly than they should have been.

F Bangladesh’s Growth Engines and the Possible Limitations of Its Present Development Regime

Such are Bangladesh’s achievements and challenges on the economic growth front, as they appear from a thorough review of the pace of the accumulation of essential production factors, their allocation, and productivity gains. The main lesson to retain from this review is the major role of RMG exports and remittances in explaining the overall satisfactory growth performance over the last two or three decades. Exports and the RMG sector have been particularly important in making the economy more dynamic, accelerating structural change, increasing productivity, and incentivising investment. Remittances have contributed to the growth of aggregate demand and the response of the domestic production machinery. Both exports and remittances have eased the foreign currency constraint that most developing countries at Bangladesh’s development stage are confronted with. Together, RMG exports and remittances may have been responsible for two-thirds of the overall growth in GDP per capita over the last 20 years or so, the remaining third consisting of productivity gains in other sectors and the net movement of workers away from low-productivity agriculture.

The role played by manufacturing RMG exports in Bangladesh is in many respects remarkable. At the same time, the present situation hides weaknesses. On the one hand, there seems to be some anomaly in observing such a high concentration of exports in a country where apparel manufacturing export is already a mature although still growing activity. On the other hand, the international context, which has been favourable to Bangladesh RMG exports until now might become less so in the years to come.

On the first point, a common observation among countries whose development is based on manufacturing exports is precisely the fast diversifying array of products they offer. The rationale behind this diversification is the inherent complexity of manufacturing products in comparison with raw materials or commodities, which leads to progress taking place at both the intensive and the extensive margin that is exporting more of the same but also closely related products upstream or downstream as well as integrating into global value chains related to initial exports.¹⁵ Such an evolution does not seem to take place in Bangladesh, and it will be important to understand why.

On the future of RMG exports, several obstacles were already mentioned: the appearance of low-labour cost competitors in both East Asia and sub-Saharan Africa, the pressure to improve labour conditions and its impact on cost-competitiveness or the automation of production which also reduces Bangladesh’s comparative advantage. To these, should be added the forthcoming graduation of the country from the LDC status, which granted to it some trade preferences in both the EU and the United States.

If not able to increase its global market share, the RMG sector will be, at best, growing as foreign demand, the same being true of migrant remittances. This means that, if Bangladesh wants to maintain its fast pace of development, it has to substantially diversify its economy, and its exports in the first place.

Another limitation that needs to be mentioned, which has been absent from this review: land. It indeed bears emphasis that, except for a handful of tiny countries, Bangladesh is the country with the highest population density on earth: Singapore and Hong Kong have a higher population density, but Bangladesh has 20 or 30 times more inhabitants. Even though it is difficult to quantify the impact of this on growth, it is difficult to imagine that the (un)availability of land is not exerting a severe constraint on Bangladesh’s development. This makes especially important the procedures for the allocation of land among its various economic uses. This point will be dealt with in some detail in the chapter on land in this volume.

III The State and the Financing of the Economy

State capacity is not a factor of production, properly speaking, and it would be difficult to provide a quantitative measure of it. Yet the ability of the state to coordinate the activity of private economic actors and, most importantly, to provide the public goods that are needed for the smooth working of the private sector is essential. This section reviews the financial aspects of the public sector in Bangladesh: that is, both its capacity to cover the public expenditures necessary for the development of the country and its ability to finance state-owned and private enterprises through public financial entities or the supervision of the private banking sector. It will be seen that, in both cases, the diagnostic is rather unfavourable. When considering the overall financing of the economy, however, a rather favourable feature is Bangladesh’s autonomy with respect to foreign financing.

A Public Finance and State Capacity

The key function of the government sector is public revenue and expenditure management, with the aim of providing the public goods needed by the population and required for accelerating economic growth, reducing infrastructure gaps, promoting investment, and ensuring an efficient redistribution of resources to alleviate poverty and reduce inequality.

Table 2.3 presents the performance of key fiscal indicators in Bangladesh between fiscal years 2013/14 and 2017/18. Focusing first on the revenue side at the top of the table, two features are striking. First, government revenues are extremely low. Second, they show no sign of growth, relative to GDP. With a 10% tax to GDP ratio, Bangladesh is among the countries with the lowest taxes in the world. Unlike some countries that can count on other sources of public revenues, for instance through royalties on exports of natural resources, non-tax revenues add only marginally to the fiscal space in Bangladesh. On several instances, the Government announced major tax reforms that would drastically increase the tax/GDP ratio. Until now, however, such reforms have failed to materialise.

Table 2.3 Government revenue and expenditure in Bangladesh

Fiscal year	2013/14	2014/15	2015/16	2016/17	2017/18
	(as % of GDP)
Revenue and grants	10.92	9.77	10.09	10.22	9.66
Total revenue	10.45	9.62	9.98	10.18	9.62
Tax revenue	8.64	8.48	8.76	9.01	8.64
National Board of Revenue (NBR) tax revenue	8.29	8.17	8.44	8.69	8.31
Non-NBR tax revenue	0.34	0.32	0.33	0.33	0.32
Non-tax revenue	1.81	1.13	1.22	1.17	0.99
Grants	0.47	0.15	0.11	0.04	0.04
	(as % of GDP)
Total expenditure	14.01	13.46	13.76	13.64	14.30
Non-development expenditure including net lending	9.60	9.27	9.06	9.18	8.87
Non-development expenditure	9.01	8.53	9.05	8.90	8.51
Revenue expenditure	8.23	7.84	8.33	8.33	7.95
Capital expenditure	0.78	0.69	0.71	0.58	0.56
	(as % of GDP)
Net lending	0.60	0.74	0.01	0.28	0.37
Development expenditure	4.40	4.19	4.70	4.45	5.43
Annual Development Programme (ADP) expenditure	4.12	3.98	4.58	4.26	5.31
Non-ADP development spending	0.28	0.22	0.12	0.20	0.12
	(as % of GDP)
Overall balance (excl. grants)	–3.56	–3.85	–3.78	–3.45	–4.68
Overall balance (incl. grants)	–3.09	–3.69	–3.67	–3.42	–4.64
Primary balance	–0.99	–1.65	–1.76	–1.62	–2.78
	(as % of GDP)
Financing	3.09	3.69	3.67	3.42	4.64
External(net) (including market borrowing)	0.25	0.32	0.74	0.59	1.14
Loans	0.89	0.79	1.13	0.95	1.47
Amortisation	–0.64	–0.47	–0.39	–0.36	–0.33
Domestic	2.84	3.37	2.93	2.83	3.50
Bank	1.35	0.03	0.29	–0.42	0.52
Non-bank	1.49	3.34	2.64	3.25	2.98

Note: Net financing includes market borrowing. Bank includes secondary market.

Source: Ministry of Finance.

The tax system generates little revenue and, on top of that, it is highly distortive and regressive. Less than a third of government revenue comes from direct taxes and only half of that is from personal income tax. As far as the latter is concerned, a large number of potential taxpayers, including many ultra-rich people, remain outside of the tax net or pay a small amount of taxes. Also, several economic sectors that are capable of paying taxes are either fully exempted or enjoy substantial tax rebates. By contrast, indirect taxes (value-added tax, excise taxes, and import duties) that fall on the whole population and are on balance regressive account for the bulk of government revenues.¹⁶ Overall, these features result in a tax system that is unable to raise enough revenue to finance the country’s development, as non-tax revenues are marginal; that is inefficient and distortive of economic activity because tax privileges are granted to specific sectors or firms; and that is inequality enhancing.

The situation is not different on the expenditure side of the general government account. Even though the total spent is substantially higher than the revenue figure, Bangladesh’s government expenditures are lower, in relation to GDP, than in many countries. Recurrent expenditures, for instance, amount on average to 9% of GDP, approximately the same level as revenues. Over the recent period, only 10 countries in the world have exhibited such a low level. Combined with the level of GDP per capita, this suggests that the provision of public services, as measured by spending per capita, is most likely to be of worse absolute and relative quality in Bangladesh than in most countries in the world. Such a state of affairs can only have negative implications for both current economic growth and poverty reduction but also for future growth and economic welfare, as human capital formation is necessarily affected by this lack of resources.

It is difficult to make a judgement about the size of development expenditures in Bangladesh in comparison to other countries, and similarly about public investment, as no recent dataset with comparable international data is available. From what can be gathered from work referring to the mid-2000s, Bangladesh would seem close to the international norm, but somewhat below it.¹⁷ Yet what is most striking is the fact that the financing of the investments scheduled in the Annual Development Programme, a multi-year development plan, is almost fully funded by government debt. Indeed, over the last five years, development expenditures represent on average 4.7% of GDP, whereas the deficit of the general government is slightly below 4%.

This deficit remains in relatively reasonable territory, yet it contributes in normal times to an increase in the public debt relative to GDP: as Table 2.3 shows, the Bangladesh Government exhibits a primary deficit, that is after taking into account the payment of interest on the outstanding debt, which is above 2% of GDP and has been clearly increasing over recent years. It is also interesting to note that one-quarter of the loans to the Government originate abroad and three-quarters originate domestically. Because of this heavy reliance on domestic lending, it cannot be discarded that public investment and the lack of fiscal space due to a low tax to GDP ratio is crowding out private investment.

To conclude this brief review of the government sector, it is worth mentioning that a recent study by the General Economics Division of the Planning Commission of Bangladesh (2017), estimated that the tax to GDP ratio in Bangladesh needs to be progressively increased to 16–17% by 2030 in order to achieve the major SDGs. This figure shows the huge revenue increase that the country needs to reach its announced development goals. The comparison with today’s resources shows that such a target is ambitious but not unachievable if growth continues at the current pace. However, three years after this statement of goals, no real change is observed.

B External Financing of the Economy

The evolution of the external financing of the Bangladeshi economy is summarised in Figure 2.7. Since 1990, the balance of the current account has fluctuated over time in a rather narrow interval around equilibrium. Over the last two decades, it has been mostly on the positive side, except for the last two years, where it shows a more pronounced deficit. Overall, it is thus fair to say that the financing of the Bangladesh economy has been mostly domestic. This is well reflected in a debt to GDP ratio which is today around 20%, after peaking at close to 50% in the tumultuous 1980s and early 1990s.

Figure 2.7 Current account, foreign debt, and foreign aid as a share of GDP, 1990–2018.

Source: World Bank, WDI

This does not mean that the economy is free from financing from the rest of the world. Actually, it can be seen in Figure 2.7 that the trade balance has systematically been negative ever since independence, with a deficit ranging from 5% to 10% of GDP. Historically, the financing of that deficit has been covered by the remittances of Bangladeshi workers abroad and foreign aid. Over time, foreign aid dwindled, and current grants are today almost negligible – although the country still benefits from concessional loans from donors. Aid grants became unnecessary when remittances surged after the turn of the millennium, as was seen earlier, whereas concessional loans mattered more for technical assistance than for financing. At the same time, however, Bangladesh has become extremely dependent on the economy of those countries that host its migrant workers, led by the Gulf countries. However, as remittances started to fall rather drastically around 2015, precisely at the time oil prices plummeted and the Gulf countries cut down on their public spending. Two years later, this entailed a higher than normal deficit of the current account.

The rather balanced external position of Bangladesh’s economy does not mean that the country did not have to request the help of the IMF on a few occasions over the past two decades. The first time was in 2001 after several years of a negative balance of the current account that had led the country to run down its reserves of foreign exchange. A stabilisation programme was then signed with the IMF for the 2003–2006 period, which came with some conditionalities relating to the management of state-owned companies and state-owned banks, tax revenues, and more general governance issues.¹⁸ In 2011, an ‘extended credit facility’ was requested due to the impact of the European crisis on Bangladesh’s balance of payment. It is thus not the case that the country is fully autonomous with respect to foreign financing. If they are strong and persistent, external shocks require payment facilities of the type provided by the IMF.

C Domestic Financing: The Banking Sector and the Weak Regulation of the Financial Sector

Two indicators can be used to evaluate the degree of development of the financial sector of a country: the broad money (M2) to GDP ratio, which measures the depth of the financial sector; and the domestic credit to the private sector by banks as a share of GDP, which measures the lending activity of the banking sector. These indicators were, respectively, 65% and 47% in Bangladesh in 2018. They doubled over the last two decades and have now reached the same level as other South Asian and lower middle-income countries. Yet they are substantially below levels observed among emerging manufacturing exporters in Southeast Asia (e.g. Cambodia or Vietnam), which were at comparable levels of financial development around 2000. The past two decades have also seen efforts to increase the quantity and the quality of banking services in Bangladesh. A broad-spectrum digitalisation, including electronic, has enabled the country to expand the coverage of its banking sector and the range of banking products not only in urban areas but also in rural areas. Yet, despite notable success, the banking sector in Bangladesh has some inherent weakness. One of its major weaknesses is the recurrent high level of non-performing loans (NPLs). The situation had become unsustainable at the end of the 1990s when NPLs reached the level of 40% of all outstanding loans. Some reforms in the regulation of the financial sector, as part of the 2003–2006 IMF stabilisation programme, were able to bring NPLs down during the 2000s. The share of NPLs in outstanding loans even reached 6% in 2011. In recent years, however, this share has started increasing again and is now above 10% – much higher than most comparable countries in South Asia and Southeast Asia, and undoubtedly damaging for the efficient functioning of the banking sector.

Even at 10%, the NPL rate is a real drain on the development capacity of the country and a powerful factor in relation to inequality. To measure its effect, the following rough calculation is instructive. It is estimated that approximately two-thirds of the NPLs can be recovered at the end of a long and costly litigation procedure. The other third must be recapitalised by the Government through drawing on fiscal resources. As NPLs are unlikely to finance the enlargement of the production capacity of the economy and are more likely to increase the wealth of defaulters, they may be considered as a sizeable net transfer – that is 3.3% of GDP – from investible resources paid by taxpayers to top income scammers.¹⁹

A lax regulation of the banking sector plays an important role in creating and piling up NPLs, particularly in state-owned banks and other public financial institutions. Cases of loans approved due to political considerations or family connections are frequent. Loans are then often disbursed without a proper or adequate credit assessment or sanction procedure, either in terms of the viability of the project they are supposed to fund or proper valuation of collateral. When default occurs, nepotism in the sanctioning procedure in favour of politically connected people makes recovering the lost money difficult.²⁰

The lack of independence of the central bank, the Bangladesh Bank, and the fact that it has no control over the sizeable state-operated financial sector, are responsible for the weak regulation of the overall banking sector, that is private and state-owned. As an example of the limited regulating power of the central bank, an amendment to the Bank Company Act was recently passed that clearly facilitates the control of private banks by family or political interests.²¹

Beyond increasing the frequency of NPLs, the weak regulation of the financial sector has two major consequences for the real economy. On the one hand, it allows the financial sector not to strictly obey capital adequacy requirements. This increases the probability of crises and the periodic refinancing of some banks by the state. This is especially the case for state-owned banks, where the NPL rate is the highest, as is well documented in the press.²² On the other hand, weak control of the lending activity of banks is responsible for an inefficient allocation of investment funds. It is certainly the case that valuable projects are being financed, and the past industrial growth of Bangladesh is testimony to this, but it is also the case that weak projects are being financed, thus depriving much better projects of funding. Evidence of this is provided by the causal effect of NPLs on banks’ availability of funds, their cost, and the interest rate being charged.²³ Unfortunately, measuring the consequences of this inefficient use of those funds which have not leaked through fraudulent NPLs is extremely difficult and does not seem to have been attempted in the academic literature on Bangladesh.

Summarising, this section on the financing of the economy suggests two different views. The favourable one is that Bangladesh has been able to finance a successful industrial development in the RMG sector and is not excessively relying on foreign financing, except perhaps through the remittances of its migrant workers, which raises different issues. The less favourable one is that the exceedingly low revenue raised by the state through taxes and other means, as well as the very lax regulation of the financial sector, especially in regard to the state-owned institutions, leads to a sizeable waste of resources, and most likely to an inefficient allocation of investment.

IV Social Matters

Even though analysis of employment and the labour market would seem to belong to an analysis of output growth, it will be handled under the heading ‘social matters’ because of the deep consequences for the whole society of the pace of decent job creation in the economy, and, by contrast, the evolution of informal and often precarious employment. In particular, the state of the labour market and its evolution has direct implications for the distribution of economic welfare within the population and for the pace of poverty reduction. Human capital policies, that is education and healthcare, also affect the future level of employment and earnings, and present welfare, respectively.

A The Labour Market

The major labour market and employment challenge in Bangladesh, notably in the context of achieving the SDGs by 2030, is to provide enough jobs to the population of working age, particularly to women and youth.

A difficulty common to most emerging and developing countries is the measurement of employment and what is actually meant by a ‘job’. Without an unemployment insurance system, the notion of being unemployed is ambiguous, and in many cases irrelevant. A person without resources will always ‘do something’ to try to survive and will be recorded as ‘employed’ by enumerators. In fact, only people with some resources can afford to be truly unemployed. To a large extent, the level of employment is thus practically determined by the size of the labour force.

However, it is not the case that if most people report themselves as employed then there is no employment problem. If almost all people ‘have a job’, then Raihan (2015) is right to make a distinction in Bangladesh between ‘good enough’, ‘good’, and ‘decent’ jobs, where the first category corresponds roughly to jobs in the informal sector – self-employment or wage work without a formal labour contract – and the last two categories to the formal sector. In the latter, however, a distinction must be made between ‘good’ jobs, with a labour contract and possibly some social insurance – healthcare, pensions – and ‘decent jobs’ that would fit the International Labour Organization definition. In the case of Bangladesh, that distinction is important. The Rana Plaza collapse on thousands of workers in 2013 showed that RMG jobs are not always ‘decent’ jobs, given the lack of security of workplaces. Other features defining a decent job are also often missing in the RMG sector and in other formal firms.²⁴

Based on that distinction, a good measure of the employment performances of a country is the creation of jobs in the formal sector, and within them the proportion of decent jobs. Based on labour force surveys over the last two decades, Rahman et al. (2019) found that the share of informal jobs in total employment probably increased a bit between 2000 and 2006 – from 75.2% to 78.5% – and very slightly declined between 2010 and 2016: the problem being that no direct comparison is possible between 2006 and 2010 because the nature of the survey questions used to decide whether a job is formal or informal changed. According to the post-2006 surveys, the degree of informality was 87% in 2010 – indeed a big difference compared to previous estimates – and 86% in 2016. Overall, it would thus seem that: (a) the degree of informality is extremely high in Bangladesh; (b) it has varied only marginally over time. This means that formal employment grew approximately alongside the labour force, at an annual rate around 1% over the last two decades, but substantially faster since 2010, with a rate slightly above 2%. Over the recent past, Raihan and Uddin (2018) find similar results for ‘decent’ jobs, the employment share of which increased from 10% in 2010 to 12% in 2018: that is, an annual growth rate of 3.3%. However, two or three times the rate of growth of the labour force does not make a big difference in terms of the degree of informality, given the overwhelming weight of the informal sector in total employment. It is interesting that formal or decent jobs have tended to increase at a faster pace lately, but, at that pace, it will take many years for the growth of the formal sector to make a dent in informality. This is a major challenge for Bangladesh: formal and decent job creation needs to proceed much faster.

There are bad and good signs with respect to this challenge. On the bad side, it may be stressed that between 2013 and 2016/17, and despite average annual manufacturing growth being above GDP growth, at 6.6%, the number of manufacturing jobs declined by 0.77 million, and by 0.92 million for women. This represents a 10% drop in total employment and a strong substitution of female by male jobs.

This drop in the level of employment in the manufacturing sector despite RMG exports still increasing in volume is the result of a strong automation drive that has started to displace jobs. It is a sign that the manufacturing sector might not contribute to employment growth in the future as much as it has done in the past, and that Bangladesh’s low-wage low-skill labour comparative advantage may be weakening. Even though the country may succeed in keeping, and even possibly increasing, its share of the global RMG market, the favourable social consequences of that development through the labour market will probably decline.

However, there is also a good sign in the fact that the share of the formal sector in total employment has not fallen despite an adverse evolution in manufacturing. Other formal sectors have created jobs, and presumably decent jobs, thus compensating for the loss in manufacturing. One may expect a large part of these jobs to be in the service sector, including transport and information and communication technology, and to concern workers with higher skill levels. If so, this points to the need to equip the labour force with more human capital and to invest more in education than has been done in the past.

Female employment may also be an issue. Over the past three decades, female labour force participation has increased, possibly in part as a response to the growth in RMG-based demand. Nevertheless, female labour supply has remained stagnant since 2010. Raihan and Bidisha (2018) explored both labour supply- and demand-side factors affecting female labour force participation in Bangladesh. Their analysis suggests that custom factors like child marriage, early pregnancy, or reproductive and domestic responsibilities have not changed much with the economic progress of the country and continue to constrain female work. But the demand side also plays a role, including the stagnation, and then the recent drop, in female employment intensity. Firm-level data from the World Bank’s Enterprise Survey of 2007 and 2013 suggest that the ratio of female to male employment declined in major manufacturing and service sectors during that period, mostly due to the impact of innovation and technological upgradation.

Youth employment may also suffer from these changes. The share of youth not in education, economic activities and training (NEET) significantly increased from 25% in 2013 to 30% in 2016/17, with 87% of the youth NEET being female, possibly affected by the loss of jobs in the RMG sector.

Migrant work might be taken as a possible equilibrating mechanism of the labour market, lack of dynamism at home being compensated by more migrants. However, the driver of migration is unclear. Is it the domestic labour market pushing Bangladeshis abroad, or the labour demand pulling them towards destination countries? The recent drop in remittances, apparently linked to the slowdown in economic activity in the Gulf countries, would suggest that foreign demand matters most. Having said this, outmigration has undoubtedly had a huge impact on the domestic labour market, by reducing the supply of predominantly young, unskilled male workers, although a non-negligible share of migrants is skilled. According to World Bank statistics, some 7.7 million Bangladeshi work abroad,²⁵ which represents a little more than 10% of the domestic labour force.

B Poverty and Inequality

Bangladesh has made important progress in reducing poverty over the past one and half decades. According to the national estimates, the overall poverty headcount was halved between 2000 and 2016, from as high as 49% to 24%. Extreme poverty, defined by the international poverty line of 1.90 2011 international US$ per person and per day, fell still more drastically from 34% to 13% during the same period.

According to the United Nations Development Programme (UNDP) multidimensional poverty index, based not on income or consumption expenditure per capita but on various types of deprivation (nutrition, child mortality, school attendance, sanitation access to drinking water, etc.), Bangladesh has also made significant progress. The poverty headcount fell from 66% in 2004 to 47% in 2014, with particularly strong progress in child mortality, school attendance, and access to electricity, and more modest gains in access to drinking water and housing.²⁶ Overall, however, Bangladesh remains in the bottom third of emerging and developing countries. As this rank is somewhat below its rank in GDP per capita ranking, this suggests that Bangladesh does not do as well as other countries in the social area.

One area of concern relating to the poverty headcount is that its rate of decline seems to be slowing down. The average annual (monetary) poverty reduction has declined gradually over the past one and half decades, the same being true of the growth elasticity of poverty, which measures the capacity of economic growth to reduce poverty.²⁷ There are good reasons to expect such a slowdown in the reduction of poverty when poverty is already very low, as the average poor person is further and further away from the poverty line. But poverty in Bangladesh is not yet at this stage, which suggests that growth is not as inclusive today as it was in the past, and not as inclusive as it could be.

A possible explanation for the decline in the pace of poverty reduction may be the steady increase in income inequality that has been observed over time in Bangladesh. Such an increase means that better-off households benefited more, and/or worse-off households less, from economic growth. Inequality markedly increased during the 1990s. It then increased again, especially since 2010, as growth was accelerating. According to the Bangladesh Bureau of Statistics, based on its Household Income and Expenditure Survey, the Gini coefficient of income rose from 0.458 in 2010 to 0.482 in 2016. The poorest 10% of the household population saw its share of the total household income fall from 2% in 2010 to 1% in 2016. By contrast, the share of the richest 10% increased from 35.8% to 38.2%.

While the preceding figures refer to household income, it is worth emphasising that a different conclusion is reached when considering the inequality of consumption expenditures per capita, as is done in the Povcalnet database maintained by the World Bank, which relies on the same household surveys as the BBS. There, no noticeable change in inequality, as measured by the Gini coefficient, seems to have taken place since 2000, and particularly between 2010 and 2016. A possible explanation could be that top incomes saved a higher fraction of their income in 2016 than was the case in 2010, but the reason for such a behavioural change is unclear. On the other hand, this hypothesis is consistent with the sizeable drop in the share of aggregate consumption expenditures in GDP since 2010. It is indeed little likely that low- and middle-income households, whose saving rates are very low, were responsible for such a fall in the aggregate propensity to consume.²⁸

The inequality issue also involves regional disparity in development. While Dhaka and a few metropolitan cities have been the major beneficiaries of development so far, many regions in the country are lagging behind. There also are genuine concerns that large discrimination prevails when it comes to budgetary allocation for social sectors and physical infrastructure, with Dhaka and a few other metropolitan cities benefitting as against many other regions in the country.²⁹ With two to seven times more development spending per capita in the Dhaka region in comparison with other regions, it is predicted that the country’s inequality situation will worsen in the future, despite the small improvement observed in the 2000s in the east–west gap.³⁰

C Education

As measured by the number of years of schooling, the level of education in Bangladesh’s population above 25 years old was 5.8 years in 2017. Among South Asian countries, this was higher than Pakistan (5.2) but lower than India (6.4), and far behind Sri Lanka (10.9) and leading Southeast Asian countries like Malaysia (10.2), Thailand (7.6), and Vietnam (8.2). In the recent past, however, Bangladesh has made remarkable progress in primary school enrolment, with near universal enrolment attained by 2010, as well as in secondary enrolment, with a rate now reaching 63%. It is thus to be expected that the average years of schooling of the Bangladeshi adult population and labour force will increase at a fast rate in the one or two decades to come.

Despite considerable progress in primary school enrolment, however, the country is seriously lagging in terms of the quality of education. Numerous studies point to low learning achievement of children who have gone through primary education. According to a recent evaluation, 59% of Grade 3 students and 90% of Grade 5 students were below the required level in mathematics at the end of the respective grades.³¹ It is thus not clear that the recent increase in years of schooling will soon entail a higher productivity of the labour force. Serious efforts are now needed to improve the education system.

Regrettably, Bangladesh is among the countries in the world with the lowest ratio of public expenditure on education to GDP. This ratio has fluctuated around 2% over recent years,³² a level that is much lower than most sub-Saharan countries, although these countries generally are poorer than Bangladesh. The contrast with the recommendation by the United Nations Educational, Scientific and Cultural Organization that countries should target educational expenditures amounting to 6% of GDP cannot be starker.

It is important to mention that Bangladesh’s education sector also suffers from huge disparities. There is a high degree of inequality with respect to access to quality education, depending on where people live. Consequently, major spatial differences are observed in educational performances among primary schools, with schools closer to metropolitan areas performing much better than others (Raihan and Ahmed, 2016).

D Healthcare

Bangladesh has made considerable progress in basic health indicators over the last decades. Advances in life expectancy, infant mortality, and maternal mortality are noteworthy. Over the last 20 years, life expectancy has risen by eight years, to reach 72 by 2016, higher than the average for lower middle-income countries and for South Asia. Infant mortality was reduced by a factor of almost 3, being today 30 per 1,000, lower again than lower middle-income and South Asian countries. Finally, maternal mortality was brought down from more than 300 to 170 for 100,000 live births. However, to achieve the targets under SDG Goal 3 by 2030, Bangladesh still has to make significant efforts: both infant and maternal mortality must still be divided by 2.5.

There are numerous challenges for Bangladesh in achieving these targets. In particular, the public health budget is only 0.39% of GDP, which is one of the lowest in the world. For this reason, the share of out-of-pocket health expenditure in total health expenditure is much higher than in other countries, reaching 71.8%. Overall, including NGOs, it is estimated that total health expenditures amount to around 3% of GDP. In other words, the public sector covers only 13% of the cost of healthcare.

With such a low level of health expenditure compared to GDP, especially in the public sector, how has Bangladesh achieved so much in terms of the health indicators mentioned above? There is evidence that over the past few decades, Bangladesh has successfully opted for low-cost solutions to some vital health-related problems. Also, widespread activities of NGOs created a deeper awareness of health issues among the population (Sarwar, 2015). A study of the use of remittances also showed that they played an important role in increasing the capacity of households to pay for health expenditure (Raihan et al., 2017a).

However, in the future, such options are likely to be limited as the health system in Bangladesh is increasingly facing hard and multifaceted challenges. These result from new pressures originating from an ageing population, the rising prevalence of chronic diseases, and the growing need for more intensive use of expensive and still critical health-related equipment (like advanced scanners, MRI machines, etc.). On the other hand, financing health-related problems through out-of-pocket expenditures increases inequality, as this places a huge cost burden on poorer people and feeds the vicious disease–poverty cycle (World Health Organization and World Bank, 2019). More investment in healthcare is thus not only a desirable but also an essential policy priority. It is hard to imagine that this could be done without a substantial increase in government revenues.

E NGOs and Microcredit

Bangladesh is famous around in the world for the number and the dynamism of its NGOs. It is said they have their roots in the intense solidarity movements that developed after a deadly cyclone and the war of independence that left the country devastated and during the terrible famine that killed more than a million people during 1974–1975.³³ Thanks to them, Bangladesh has shown strong progress on several social indicators, mostly due to a multifaceted service provision regime. The expansion and multiplication of NGOs made it possible to scale up innovative anti-poverty experiments into nationwide programmes. Some notable programmes include innovations in providing access to credit to previously ‘unbanked’ poor; the development of a non-formal education system for poor children (particularly girls); and the provision of door-to-door health services through thousands of village-based community health workers. On a different level, it must also be stressed that NGOs have notably contributed to the empowerment of women within a strongly patriarchal society. The large portion of NGO beneficiaries who are poor women is evidence that a cultural change that is taking place in regard to the position of women in society.

The delivery of social services and pro-poor advocacy are not NGOs’ only activities. They have also developed commercial ventures aimed at creating a bridge between poor subsistence farmers and markets, as well as an internal revenue generation model for the NGOs themselves. Their largely self-financed pro-poor services have become an integral part of achieving national poverty reduction targets.

NGOs differ in their size and coverage. There are about 2,000 development NGOs, some of which are among the largest of such organisations in the world. NGOs such as BRAC, Grameen Bank, ASA, and PROSIKA have tens of thousands of employees and multi-million-dollar budgets, with their operations spreading throughout the nation. Other NGOs are smaller in scale and function with limited managerial and staff capacity.

Microfinance is a key factor behind the growth of NGO programmes. Some 55% of rural households have resorted to microfinance at some stage in their lives, and almost 46% hold the status of current borrowers (Raihan et al., 2017b). The sector is dominated by the Grameen Bank, BRAC, ASA, and PROSHIKA, which between them cover around 90% of microfinance operations. Even though there is some ambiguity regarding whether micro lending has a long-run transformative impact on household income,³⁴ it reduces current poverty and provides insurance against weather shocks or other accidents. Improvements in social indicators like female empowerment, children’s schooling, and health status in part reflect the complementary social mobilisation, training, and awareness building activities developed by NGOs alongside microcredit.

Door-to-door health services are provided by NGOs through village-based community health workers, focusing mostly on preventive care and simple curative care for women and children. NGOs have also achieved notable success in promoting behavioural change at community level by providing water and sanitation services. Their community work also includes programmes on child nutrition and tuberculosis treatment, in collaboration with the Government.

BRAC is the pioneer in launching primary education and has become the single largest NGO in the world working in primary education, social enterprises, microfinance, health services, housing, and water and sanitation programmes all over the country and also abroad. Today, BRAC provides primary education to over 1 million children in 22,000 education centres nationwide. It lends half a billion US dollars a year to 7 million people and employs more than 100,000 people for an annual budget of 750 million dollars.³⁵ Other NGOs are smaller but altogether they may amount to two to three times BRAC’s size. Their role is thus far from marginal. Although not quantified, their contribution to poverty reduction, in the multidimensional sense used by UNDP, has been and continues to be crucial.

V Environment and Climate Change Challenges

Bangladesh is highly vulnerable to climate change impacts because of its vast low-lying areas, large coastal population, high population density, inadequate infrastructure, and high dependence on agriculture. For Bangladesh, climate change is manifested as both changes in the severity of extreme events and in greater climate variability. Climate variability involves haphazard wet and drought years, whereas extreme weather events take the form of violent tropical cyclones which also generate powerful storm surges, and whose effects are amplified by rising sea levels. About 20% of the population lives in the low coastal zone and any increase in sea level will have disastrous effects. Because of the flat topography, even small increments in sea level rise will affect large areas, directly through inundation and salt intrusion. In the Global Climate Risk Index 2019, Bangladesh has been ranked seventh among the countries most affected by extreme weather events in the 20 years since 1998.³⁶

Climate change–linked problems are likely to act as a drag on the nation’s growth prospects. The Asian Development Bank (ADB, 2014) concluded that climate change poses big economic and development challenges for Bangladesh. The study pointed out that the country will face annual economic costs equivalent to about 2% of its GDP by 2050, widening to 9.4% by 2100. The reasons for these losses include an immense decline in crops, land loss and salinity, and internal migration, among other things. Climate-related disasters regularly cause migration movements and changes in poverty patterns. Because of major climate shocks, poor people from the south of the country are migrating to urban areas. Analysis at the district and sub-district levels shows that there is a strong positive correlation between the incidence of poverty and the intensity of natural hazards. On average, districts that are ranked as most exposed to natural disasters also show poverty rates that are higher than the national average. Strikingly, of the 15 most poverty-stricken districts, almost 13 of them belong to high natural hazard risk categories.

VI The Covid-19 Crisis

Bangladesh is currently the second most affected country by COVID-19 in South Asia (after India) with 1.5 million confirmed cases and 28,000 confirmed deaths between January 2020 and December 2021. However, excess mortality estimates released by The Economist Magazine suggest a much higher death toll, probably above half a million. Despite measures intended to attenuate its impact, the pandemic led to a serious economic and social crisis, and one may wonder whether the growth potential of the country has not been affected. GDP kept growing but its growth rate fell by 3–4% in 2020 and was likely still below trend in 2021. As far as poverty is concerned, on the other hand, it is estimated that at the heart of the crisis, in the spring of 2020, the headcount may have doubled, reaching maybe 40% of the population.³⁷ It went down, probably close to its initial level since then.

The drop in GDP growth is due to three major causes: (i) the pandemic itself, which has affected the health workers and may have forced some of them to temporarily cease working even outside lockdown periods; (ii) the lockdown episodes, especially the longest one from 23 March to 30 May 2020; and (iii) the fall in global demand. A stimulus package, consisting mostly of subsidised loans to support furlough strategies by firms, and cash/food transfers to needy households, for 3.5% of GDP, was meant to attenuate the shock.

It is difficult to disentangle the role of these various factors. Yet, global demand has undoubtedly played a major role. The quasi-stagnation of manufacturing output for the whole of 2020 is almost entirely due to the RMG sector, despite the fact that it partly defied the lockdown by resuming work a month before it ended. Other sectors are much less exposed to foreign demand. They were much less affected by the lockdown or recovered quickly when it was lifted. Over the whole year, rather than the few months of the lockdown, the main shock on the Bangladeshi economy turns out to be a 20% drop in exports, which mechanically generated a drop of GDP around 3%, a bit less than the observed slowdown of growth. This confirms the dominant role of RMG exports in the economy and the vulnerability of the latter to any kind of shock in that sector.

That GDP kept growing in Bangladesh during the pandemic is quite remarkable when comparing it to India where GDP fell by 8% in 2020. The pandemic has been more severe there, at least by the relative number of estimated casualties with respect to the population, but lockdown, social distance and school closure measures were very much the same. It would be interesting to know whether this divergence is due to differences in the socio-economic structure of the two countries or in their reactivity.³⁸

As far as poverty is concerned, a difference must clearly be made between the impact of COVID-19 measured at the time of the lockdown or averaged over the whole year in a way that is consistent with GDP growth measurement. On the spot, it was seen above that the shock was violent, mostly because the lockdown affected more those low-income people whose occupation in the informal sector required some spatial mobility. Unlike formal workers, they could not benefit from the furlough system based on subsidised loans to formal firms, or work from home. Food distribution and cash transfers only moderately attenuated the income shock that they suffered. Estimated over a longer period, or a few months after the end of the 2020 lockdown, the situation is likely to be different. The calculation of the change in poverty, in line with the observed GDP growth and what is known about the growth elasticity of the poverty headcount, indicates that the proportion of poor people may have been between 0.5 and 1 percentage point below trend by December 2020, except if the recovery from the crisis came with a strong increase in income inequality.

Such a calculation does not take into account the potential permanent effects of the crisis on those people who were the most affected at the peak of the pandemic and during the lockdown. Those damages may be related to health, either through the pandemic itself or as a result of undernutrition, on productive economic assets sold to face up to income losses, or on the education of the children. As schools have been closed for more than a year it is expected that a higher proportion of students will drop-out than in normal times, whereas it is not clear that, without an effective remedial programme, others will be able to make-up for the time lost. This deficit of human capital is unlikely to have a major impact on future aggregate growth because it will affect only a few cohorts. However, this may not be the case with poverty and inequality as students from distressed families are likely to bear a higher burden from the disruption.³⁹

A final word should be said about what the crisis revealed about the capacity of the government to handle such shocks. From that point of view, the existing evidence is far from satisfactory. The crisis not only exposed a lack of administrative capacity within the health and social protection sectors, but also difficulties in co-ordinating across sectors and public entities. They have been compounded by the under-development of the public health sector, and issues surrounding finance, equity, quality and efficiency. Similarly, underlying weaknesses in the social protection system, such as the inability to quickly and efficiently disburse funds and other forms of support to those most vulnerable, exacerbated the social consequences of the pandemic.⁴⁰ On the economic side, on the other hand, the policy response has been uneven, benefiting some groups far more than others.

These shortcomings have deep institutional and political economy roots, since they reflect how certain vested interests benefit from the status quo, and how policy decisions mirror the political and lobbying power of particular groups. Micro, small and medium enterprises have seen little benefit from the stimulus packages. In contrast, larger firms with powerful lobbying and useful political links, despite the fact that many of them may not need the stimulus package, dominated the scenario, as shown in Raihan et al. (2021b). As a consequence, the inherent bias towards large firms, especially in the RMG sector, that has been a hallmark of Bangladesh’s industrial development in the past, has been replicated in the manner in which the government has sought to respond to the economic harms due to the pandemic.

VII Conclusion

This in-depth analysis of the achievements and challenges of the Bangladeshi economy may be summarised in three basic points.

First and foremost, the economy has experienced, and may still be experiencing, a very successful development episode, based on an extremely vigorous export manufacturing sector born in the 1980s out of the opportunities offered by the MFA and then expertly exploited and amplified by a dynamic and influential entrepreneurial class. It can be estimated that, directly and indirectly through backward and forward links, the RMG sector may have been responsible for 30–40% of Bangladesh’s very solid growth performance over the last two or three decades. The role played by the remittances of migrant Bangladeshi workers which peaked at 10% of GDP around 2010 must also be mentioned as a powerful factor affecting growth, through their major impact on domestic demand. Together, manufacturing exports and remittances are estimated to explain two-thirds of Bangladesh’s growth over the last 20 years. The fact that both RMG exports and remittances have slowed down in the recent past is a source of concern – and makes the GDP growth rate estimates for the last few years a bit of a puzzle.⁴¹

Second, Bangladesh’s strong development performance is all the more surprising given that, except for the forward and backward linkage effects of RMG exports and the multiplier effects of remittances, the diagnostic regarding the rest of the economy is not very favourable. An underdeveloped public sector and public services, an infrastructure deficit, among the worst investment climates in the world, and a rather corrupt banking sector may all explain a lack of dynamism outside the RMG sector. This is more apparent in other manufacturing export activity. Domestic-oriented sectors, including a few import-substitution sectors, benefited directly from the expansion of RMG exports and remittances, through their impact on aggregate demand. All in all, Bangladesh may not be very different from those developing countries exporting a natural resource that feeds the development of domestic-oriented sectors. The difference is that export prices in Bangladesh do not fluctuate as much as those of commodities, and the demand for its exports have grown at a steady and fast rate over time, in part thanks to a comparative advantage and in part based on low-wage labour.

The third point is that, when abstracting from the successful RMG export sector, Bangladesh exhibits most of the features of low-income countries, despite having graduated from the LDC status and accessing the group of lower middle-income countries: extended poverty, a huge traditional farming sector, overwhelming informality, low human capital, and weak public services. With respect to the latter, however, one feature of Bangladesh is the presence of a buoyant NGO sector that partly compensates for the failings of the state. Besides the role of the NGO and RMG sectors, and despite favourable past growth performance, Bangladesh thus faces many of the structural obstacles that hinder the development of countries at the same level of income, except the strong dependence generally observed on foreign financing.

Looking now to the future, the main challenge is whether the RMG sector will continue to be as dynamic as it was before the recent slowdown, or whether the latter has some premonitory meaning. Competition from other low-wage countries; the mechanisation of the industry, which reduces its employment capacity, notably of women, and which may trigger some reshoring policy among some clients of Bangladesh; the loss of trade preferences that comes with graduating from the LDC status – will all affect the main growth driver of the economy. Given the high product concentration of exports, manufacturing export diversification, including the upgrading of RMG production, is the strategy that should be pursued in order to maintain the rate of growth of the economy. This requires not only an adequate investment programme, but also progress in infrastructure, banking, human capital accumulation, and state capacity in general – all areas where Bangladesh is presently under-performing. The present slowdown in RMG exports and in remittances suggests that there may be an urgent need to launch such an ambitious strategy.

It remains to be seen whether Bangladesh’s institutional context is conducive to such an inflexion of its economic development strategy. Chapters 4–9 analyse in more detail this crucial aspect of development.

Annex 2.1 Decomposition of Growth in Bangladesh

The methodology⁴² consists of decomposing output growth into a weighted average of the rate of growth of the factors of production, labour, and capital, and a residual, termed TFP (or ‘Solow residual’).

A neoclassical production function represents output at time t, $Y_t$, as a function of the economy’s capital stock, $K_t$, its labour force, $L_t$, and the economy’s TFP, $A_t$. With a Cobb–Douglas specification, this function is written as:

$Y_t=A_t{K}_t{L}_t^{1-\alpha }$

where α and 1-α are, respectively, the output elasticity of capital and labour. It is well known that at a competitive equilibrium of the economy, α and 1-α also are y, the share of capital and labour income in GDP.

Within this framework, output changes can only be caused by changes in the capital stock, the labour force, or in TFP. Taking logarithms and differentiating leads to the identity:

$\begin{array}{c}Ln{Y}_t-Ln{Y}_{t-1}=Ln{A}_t-Ln{A}_{t-1}+\alpha \left(Ln{K}_t-Ln{K}_{t-1}\right)\\ +\left(1-\alpha \right)\left(Ln{L}_t-Ln{L}_{t-1}\right)\end{array}$

Or:

$\begin{array}{c}GDPgrowth=S_k.Capgrowth+(1-S_k).Labgrowth+TFPgrowth\\ ObservedGDP\downarrow ObservedCapgrowth\downarrow ObservedLabgrowthResidual\\ ObservedGDPObservedGDP\\ shareofshareof\\ capitalincomelabourincome\end{array}$

However, it should be mentioned that there are some limitations of this exercise. TFP growth factor productivity is calculated as a residual, and therefore any measurement error in the variables that measure labour, capital, or the GDP share of labour and capital income, are mechanically imputed to TFP. Also, growth accounting is a descriptive tool, and it does not provide insights into the nature of TFP growth (technological, structural, and/or institutional change).

The above growth accounting identity may be generalised to any production function that exhibits constant returns to scale. A crucial assumption is that the economy is competitive so that the weight of capital and labour growth in the accounting identity are their income shares in GDP. This may be considered to be a rather strong assumption.

Annex 2.2 Decompositions of Aggregate Labour Productivity

From Avillez (2012): For the aggregate economy in period t define $X_t\left(Y_t\right)$ as aggregate real (nominal) output, N_t as labour input, and $A_t:=X_t/N_t$ as aggregate labour productivity. Variables with subscript i refer to corresponding variables in sector i, of which there are M, for example $A_{it}:=X_{it}/N_{it}$ is productivity in sector i in period t. Define, furthermore, sector i’s labour input share as $S_{it}:=N_{it}/N_t$ and denote overall (sector i’s) productivity growth as follows:

$G_t:=\frac{A_t-A_0}{A_0}\left(G_{it}:=\frac{A_{it}-A_{i0}}{A_{i0}}\right).$

Assume that real output is measured in constant prices – more specifically, using fixed-base Laspeyres quantity and Paasche price indexes – so that aggregate real output corresponds to the sum of sectoral real output:

$X_t={\sum }_{i=1}^M{X}_{it},i=1M,\cdots ,.$

Then aggregate labour productivity is equal to the weighted sum of sector labour productivity across all sectors, where the weights are the sectoral labour shares:

$A_t=\frac{{\sum }_{i=1}^M{X}_{it}}{N_t}=\frac{{\sum }_{i=1}^M{A}_{it}{N}_{it}}{N_t}={\sum }_{i=1}^M{A}_{it}{S}_{it}.$

Accordingly, for a given year t and base year 0, we can decompose Gt as follows:

$\begin{array}{c}G=t\frac{{\sum }_{i=1}^M\left(A_{it}{S}_{it}-A_{i0}{S}_{i0}\right)}{A_{i0}}\\ =\sum _{i=1}^M\frac{A_{i0}}{A_0}\left(S_{i0}{G}_{it}+\left(S_{it}-S_{i0}\right)+\left(S_{it}-S_{i0}\right)G_{it}\right)\\ =\underset{\text{WSE}}{\underbrace{\sum _{i=1}^M\frac{X_{i0}}{X_0}{G}_{it}}}+\underset{\text{SRE}}{\underbrace{\sum _{i=1}^M\frac{A_{i0}}{A_0}\Delta S_{it}}}+\underset{\text{DRE}}{\underbrace{\sum _{i=1}^M\frac{A_{i0}}{A_0}\Delta S_{it}{G}_{it}}}.\end{array}$

Here, $\Delta S_{it}:=S_{it}-S_i{}_0$ represents the movement of labour across sectors. Labour productivity growth can thus be decomposed into three distinct effects:

WSE – reflects the impact of productivity growth within individual sectors on aggregate productivity;

SRE (Denison effect) – captures the impact of the reallocation of employment from less productive to more productive sectors (i.e. it arises from sectoral differences in productivity levels); and

DRE (Baumol effect) – describes the impact of reallocating employment into sectors with growing productivity (i.e. this results from asymmetric productivity growth rates across sectors). The Baumol effect detracts from aggregate productivity growth when labour moves towards (away from) a sector with negative (positive) labour productivity growth. Note that the magnitude of the DRE depends not only on the sectoral productivity rates and sectoral employment shifts but also on the ratio between the sector’s labour productivity level and the aggregate labour productivity level.

Annex 2.3 Factors Affecting Growth in Bangladesh

To analyse the sources of growth in the Bangladesh economy, we use a time series econometric model. The model aims to estimate the relationship between economic growth and the potential drivers of growth analysed in the main text, which are exports and remittances. We begin with the Solow model specification with a Cobb–Douglas production function, along the lines employed by Luintel et al. (2008) and Rao et al. (2008):

$y_t=a_t{k}_t^{\alpha },1>\alpha ,\beta >0$(1)

where y = real GDP per worker, a = TFP or accumulated stock of technological knowledge, k = capital stock per worker, and t = time period.

The Solow model assumes that the evolution of technology is exogenous and TFP grows at rate g:

$a_t=a_0{e}^{gT}$(2)

where a₀ = initial stock on knowledge.

As suggested by Rao et al. (2008), it is assumed, in addition, that the growth drivers affect growth essentially through their impact on TFP, for instance, by facilitating technological import. Thus, assuming Cobb–Douglas functional forms, TFP may be specified as:

$a_t=a_0{e}^{gT}E{X}_t^{\delta }RE{M}_t^{\gamma }$(3)

where EX = real exports of goods and services per worker, REM = real migrants’ remittances per worker, and $\delta$ and $\gamma$ are constants.

Substituting into (1) and taking logarithms leads to the following linear equation:

$Ln{y}_t=Ln{a}_0+gT+\alpha .Ln{k}_t+\delta .LnE{X}_t+\gamma .LnRE{M}_t+{\epsilon }_t$(4)

where ${\epsilon }_t$ stands for random deviations from the theoretical relationship.

As some variables in this model may be non-stationary, a cointegration approach is adopted. The method of estimation is the auto regressive distributed lag (ARDL) framework by Pesaran and Shin (1999).⁴³ The ARDL approach to cointegration involves estimating the conditional error correction version of equation 5 and allows for combining non-stationary variables – supposedly I(1) – and stationary variables – that is I(0). An F-statistic makes it possible to test the null hypothesis of no cointegration.

Prior to the testing of cointegration, we conducted a test of the order of integration for each variable in the model, using augmented Dickey–Fuller (ADF) and Philip–Perron (PP) techniques. Even though the ARDL framework does not require pretesting of the variables of the model, the unit root test could inform us whether the ARDL model should be used. The results show that there is indeed a mixture of I(1) and I(0), with trends.

The ARDL model was applied to equation (5) using Akaike information criterion (AIC) as information criterion and time trend. The number of lags is selected by the Pesaran and Shin (1999) and Narayan (2004) techniques. Two lags were introduced in the procedure. The calculated F-statistic of 26.47 made it possible to reject the null hypothesis of no-cointegration at the 1% level of significance.

The empirical estimates of the long-run relationship between the variables of equation (4) are shown in Table 2.4-A. The regression results suggest that the long-run elasticity of real GDP per worker with respect to capital stock per worker is 0.64. In addition, a 1% increase in real export per worker leads to a 0.123% increase in real GDP per worker, and a 1% increase in real remittance per worker leads to an increase in real GDP per worker of 0.137%. The short-run coefficients (in first differenced form), not shown here, for capital per worker and export per worker are positive and statistically significant, while that for remittance per worker is not statistically significant.

Table 2.4 Results of ARDL estimation (long-run model)

A. Dependent variable: log of real GDP per worker	Coef.	Std. err.	t	P > t
Log of capital per worker	0.642	0.076	8.35	0.000
Log of real export per worker	0.123	0.032	4.85	0.000
Log of real remittance per worker	0.137	0.057	2.39	0.024
T (time trend)	0.000034	0.0008	0.04	0.968
Constant	4.15	1.99	2.08	0.047

B. Dependent variable: log of capital per worker	Coef.	Std. err.	t	P > t
Log of real export per worker	0.396	0.099	4.02	0.000
Log of real remittance per worker	0.132	0.063	2.59	0.044
T (time trend)	0.0018	0.0011	1.54	0.138
Constant	0.391	0.282	1.39	0.178

C. Dependent variable: log of real GDP per worker	Coef.	Std. err.	t	P > t
Log of real export per worker	0.220	0.022	10.18	0.000
Log of real remittance per worker	0.143	0.022	4.46	0.000
T (time trend)	0.00145	0.00078	1.84	0.074
Constant	3.2821	1.222	1.69	0.088

Source: Authors’ estimation.

As explained in the text, exports and remittances may also affect directly the accumulation of capital, by providing the foreign currency needed for importing equipment and by incentivising investment in domestic production through their impact on aggregate demand. There are two ways for estimating these effects. The first one consists of analysing the relation between the stock of capital and the two growth drivers, EX and REM. The second one consists of estimating the reduced form model, where the capital stock is omitted in the output regression. In that case, the estimation should reveal the overall impact of the growth drivers, through both TFP and the accumulation of capital.

These two models are respectively:

$Ln{k}_t=Ln{b}_0+hT+\lambda .LnE{X}_t+\mu .LnRE{M}_t+{\eta }_t$(5)

$Ln{y}_t=Ln{c}_0+mT+\theta .LnE{X}_t+\pi .LnRE{M}_t+{\omega }_t$(6)

where $b_0$ and $c_0$ are constants and ${\eta }_t$ and ${\omega }_t$ are random deviations from the corresponding theoretical relationship. The estimation results obtained with the same procedure as before are shown in Table 2.4-B and Table 2.4-C.

Together, models (4) and (5) provide a ‘structural view’ of the way the growth drivers EX and REM affect output growth. They do it in two ways: through the TFP effect in equation (4), and through their impact on capital accumulation in equation (5). The overall elasticity is obtained by combining the coefficients of the two equations. The EX elasticity is thus: $\delta +\alpha .\lambda$ and is worth: 0.12+0.64*0.40 = 0.37, whereas the REM elasticity is $\gamma +\alpha .\mu$ or 0.14+0.64*0.13 = 0.208.

The problem of the preceding estimation of the overall effect of the growth drivers on GDP growth is that, by not taking into account the dependency of capital accumulation on the growth drivers, the estimated coefficients of the variables Ln EX and Ln REM in (4), already take into account their effect on TFP as well as on the accumulation of capital. There thus is some double-counting in preceding calculation, which over-estimates their effect.

To correct for this, Ln k in (4) should be instrumented through equation (5), adding to the later equation an instrumental variable orthogonal to both the growth drivers and to growth itself. In the absence of such a variable, only the reduced form model (6) can be estimated, with the time trend possibly accounting for unobserved output determinants that increase with time. Thus, the estimates of $\theta$and $\pi$ combine the TFP and the capital accumulation effect.

Based on the estimations in Table 2.4-C, the overall effect of a 1% change in exports results in a 0.22% increase in GDP, the corresponding figure being 0.14% for remittances. Between 2000 and 2018, EX and REM both increased at an annual rate of 12%, thus entailing an annual growth of, respectively, 2.6% and 1.5% of GDP on average, for an overall growth rate of 6%.

3 Critical Institutional Areas for Bangladesh’s Development

Selim Raihan , Rafiqua Ferdousi , and Towhid I. Iqram Mahmood

I Introduction

This chapter presents analyses using information from a variety of sources in order to identify areas where in-depth research can identify institutional challenges that are most critical to Bangladesh’s economic development. Two approaches are employed. The first approach uses a variety of institutional measures available in international databases to examine how a country, in this case Bangladesh, differs from a set of comparators. A questionnaire survey of various types of decision makers and academics is used in the second approach, as well as a set of open-ended interviews with senior policymakers and decision makers.

II Bangladesh’s Position in the Global Ranking of Institutional Indices

Since the pioneering work of North (1990), there has been widespread agreement that institutions matter for development. Narratives have described some features of the relationship between institutions and development and theoretical models of that relationship have been proposed that fit some stylised facts, often drawn from history. Numerous authors could be cited, but Acemoglu and Robinson (2012), Khan (2012a, 2018), or more recently Pritchett et al. (2018) are prominent examples of the first approach, while Acemoglu and Robinson (2008) are a good example of the second. Going beyond this approach and getting into more detail on the nature and the quality of institutions requires the availability of qualitative or quantitative indicators describing them. Such country-level indicators and indices have been developed over the last two or three decades, which has given rise to an empirical cross-country literature exploring the relationship between institutions (as described by some of these indicators) and particular characteristic of economic development (primarily the level and growth rate of gross domestic product (GDP)). Knack and Keefer (1995), Acemoglu et al. (2001), and Rodrik et al. (2004) were the first notable attempts in this direction.

While relying on the same type of data, that is the existing databases of institution-oriented indicators, the objective of this exercise is somewhat different. Focusing on a single country, Bangladesh, its main objective is to characterise its institutional profile as reflected in available indicators and to see what its absolute and relative strengths and weaknesses are. This will be done in three ways. First, relying on the most complete repository of indicators, the University of Gothenburg’s Quality of Government database (Dahlberg et al., 2020), six aggregate indicators will be defined, and countries, both advanced and developing, will be ranked according to each of them. The quality of Bangladeshi institutions will then be analysed according to each aggregate indicator taking into account each of the individual indicators that make up that aggregate indicator. Because all of these indicators are closely related to economic development, as measured for instance by GDP per capita, the second question that will be asked is how far away Bangladesh is from what could be considered an international norm: that is, the level of each aggregate indicator that corresponds to Bangladesh’s level of GDP per capita. To some extent, this is equivalent to comparing Bangladesh to countries with more or less the same level of income. The same comparison will be made with geographical neighbours or countries that have outperformed Bangladesh over the last two or three decades, despite being initially at the same level of development. Finally, the time evolution of the institutional quality of Bangladesh will be analysed by relying on a database that makes it possible to cover the last three decades.

Analysing the various findings, Bangladesh’s institutional profile as indicated by institutional indicators will be summarised in Section IV. The general diagnostic is that Bangladesh ranks uniformly rather badly in many institutional dimensions. Given its high-growth performance, the so-called ‘Bangladesh paradox’ or ‘Bangladesh surprise’ of this combination of under-performing institutions and over-performing economy underlined by several observers (World Bank, 2007b, 2007c, 2010; Mahmud et al., 2008; Asadullah et al., 2014) is worth serious investigation. It should be kept in mind, however, that the institutional part of this paradox relies on indicators that are essentially imprecise and that can only give a rough description of the nature of institutions in a given country.

A Constructing Synthetic Institutional Indices

There now are many databases with sets of indicators that seek to describe the quality of various aspects of a country’s political, sociological, and economic institutions. Well-known databases of this type include the Worldwide Governance Indicators, the Logistics Performance Index, Doing Business, the Global Competitiveness Index, and the International Country Risk Guide (ICRG), or Polity IV. Several single indicators have also become a key reference, for instance the Transparency International corruption index. The Quality of Government is a repository of institutional indicators present in all these databases. As such, it comprises more than 2,000 indicators over a period that extends from 1949 to 2018 for some indicators and some countries. However, it would not make sense to use every indicator to study the profile of one specific country in comparison to others. Moreover, there are many missing observations. Instead, the technique used here has been to develop a small number of synthetic institutional indices that aggregate individual indicators in the database with similar distributions across countries at a given point of time – the year 2016. A method of clustering a subset of indicators simultaneously available for the largest number of countries into a pre-determined number of groups – that is clusters – was used. The data selection procedure ended up with a set 97 indicators available in 105 countries – both developed and developing. The clustering method is based on the correlation across indicators in the cluster using the country values of indicators as observations. It thus consists of minimising the variance across indicators within clusters and maximising the variance between clusters. A synthetic index is then associated with the cluster by using a linear combination of all indicators in the cluster. The coefficients of the first axis in a principal component analysis (PCA) of all indicators in the cluster were used. They thus maximise the cross-country variance explained by the synthetic index.

The main parameter in the hands of someone using clustering methods is the number of clusters. In the present case, it was decided to stay with six clusters, and thus six synthetic indices, for both practical reasons and to ensure consistency. The practicality requirement refers to the need to be able to visualise and compare observations across a multidimensional space, which requires minimising the number of clusters. Consistency requires differentiating as much as possible the synthetic indices, while making it possible to give some clear indication of their meaning. Indeed, each cluster may include very different indicators, without an obvious common link between them, although the fact that they are correlated suggests that such a link must exist. However, it turns out that if the number of clusters is increased, it makes it increasingly difficult to identify such a link. In the present case, it also turned out that the six synthetic indices were in rough agreement with the main themes of the institutional diagnostic survey undertaken in this research project, the results of which are analysed in the next section.

The six clusters or groups of indicators that are selected by the procedure just described are Democracy, Rule of law, Business environment, Bureaucracy, Land, and Human rights. Number of indicators used by the synthetic indices of Democracy, Rule of law, Business environment, Bureaucracy, Land, and Human rights are 22, 14, 23, 9, 8, and 11, respectively. Furthermore, the variance captured by the first principal component within the group of each synthetic index of Democracy, Rule of law, Business environment, Bureaucracy, Land, and Human rights are 57.21%, 73.46%, 68.47%, 79.30%, 38.72%, and 54.84%, respectively.

Under the heading democracy are found indicators describing the political regime, its effectiveness, pluralism, stability, or transparency. The rule of law heading comprises indicators describing the effectiveness of the legal framework, the judiciary system, the control of corruption, and the quality of economic regulation. Business environment, not surprisingly, includes the quality of business infrastructure and the market context in which firms operate. Bureaucracy describes the quality of the administration and some public services. Land does not cover many indicators because it turns out to be more focused than other synthetic indices. Finally, human rights comprise indicators of a more social nature, that is education, healthcare, and civil liberties, including freedom of expression.

Each individual indicator was linearly normalised for its value to range between 0 and 100, but of course their distribution across countries, including their mean and median, is not the same. It appears that the mean and median of the democracy, land, and human right indices are above those of rule of law, bureaucratic quality, and business environment. To the extent that the value of individual indicators is not necessarily comparable among themselves, this result is not of major importance for our purposes. Instead, we now focus on the relative position of Bangladesh across the six-dimensional space of the synthetic indices.

B How Does Bangladesh Compare to Other Countries According to the Synthetic Institutional Indices?

This section summarises Bangladesh’s relative position in the synthetic institutional indices compared to the top and bottom performing countries of the world. According to Figure 3.1, Bangladesh’s relative performance in the global ranking, established on the basis of the synthetic institutional indices, is rather uniformly mediocre, as it systematically ranges in the lowest quartile – as a matter of fact, even in the lowest quintile of the global ranking. The situation is even worse for the rule of law, bureaucratic quality, and land synthetic indices, where Bangladesh ranks in the bottom 5% or close to it. Its position on human rights is only slightly less disastrous, as it still lies at the upper limit of the bottom 10%. In short, it is only on democracy and business environment that Bangladesh gets somewhat away from the very bottom of the global ranking. This is an interesting finding since it allows us to differentiate the relative quality of Bangladeshi institutions with respect the nature of these institutions. It will be shown later that this conclusion resonates rather well with other evidence or judgements about Bangladeshi institutions.

Figure 3.1 Distribution of the synthetic indices.

Note: The star indicates Bangladesh’s position. For each synthetic index, the figure shows the limits of the four quartiles of its distribution among countries, the bottom and top whiskers corresponding to the bottom and top quartiles, and the horizontal segment within the central box, the median, separating the second and third quartiles.

Source: Authors, based on the synthetic institutional index

Table 3.1 shows the countries that are ranked close to Bangladesh in the various synthetic indices, the idea being to see whether they share some common features besides their institutional ranking. Diversity is clearly the dominant factor here. There is little regional alignment, except the presence of Pakistan in democracy and land, something that can be linked to the common past with Bangladesh, first as British colonies and then as two parts of the same political entity. Several Middle Eastern and North African countries appear in the list, with no obvious geographical, historical, or political similarity with Bangladesh. Finally, many low-income sub-Saharan countries are present, but this may perhaps reflect more the relatively large number of countries in that region of the world, their low income, and their absence of efficient institutions.

Table 3.1 Ranking of the countries around Bangladesh for each summary index in 2016

Democracy		Rule of law		Business environment		Bureaucracy		Land		Human rights
84	Kuwait	98	Zimbabwe	82	Zambia	94	Argentina	99	Haiti	93	Zimbabwe
85	Jordan	99	Ukraine	83	Senegal	95	Lebanon	100	Algeria	94	Liberia
86	Nigeria	100	Madagascar	84	Jamaica	96	Dominican Republic	101	Madagascar	95	Tanzania
87	Bangladesh	101	Bangladesh	85	Bangladesh	97	Bangladesh	102	Bangladesh	96	Bangladesh
88	Pakistan	102	Myanmar	86	Guyana	98	Zimbabwe	103	Guinea	97	Algeria
89	Lebanon	103	Haiti	87	Iran	99	Madagascar	104	Nigeria	98	Egypt
90	Algeria	104	Guinea	89	Paraguay	100	Guinea	105	Pakistan	99	Venezuela

Note: This ranking is performed for 105 countries.

Source: Authors’ own calculation, based on the synthetic institutional indices.

The most striking feature of Table 3.1 is the absence of countries with a growth record as strong as Bangladesh’s over the last few decades: on the contrary, several countries show rather inferior performance. Likewise, only one country (i.e. Thailand) would qualify as a manufacturing exporter (like Bangladesh). All other countries are typical commodity exporters, except Jordan and Lebanon, and four of them are major oil exporters – Algeria, Nigeria, Kuwait, and Iran. These observations reinforce the idea that there is a ‘Bangladesh paradox’: a fast-growing manufacturing exporter with institutional quality comparable with slow-growing commodity exporters, including oil exporters. It will be seen later in this study that the latter analogy echoes the fact that ready-made goods (RMG) manufacturing exports in Bangladesh may indeed play a role in the economy and the society similar to that played by raw commodity exports in other developing countries.

C Major Institutional Weaknesses of Bangladesh in the Synthetic Institutional Indices

Box 3.1 shows those individual indicators in each synthetic cluster on which Bangladesh performs substantially less well compared to the others, that is the mean of the cluster. For instance, in democracy it performs particularly poorly on the following indicators: the presence of ‘fractionalised elites’, the lack of ‘public trust in politicians’, and the strength of the ‘political competition’. Likewise, in the rule of the law, it can be seen that the ‘corruption perception index’ plays an important role in bringing Bangladesh’s overall score down, the same being true of the overall evaluation of the ‘judicial independence’ and the ‘inefficiency of the legal framework’.

Box 3.1 Major areas of weaknesses in each synthetic institutional index

1. 1. Democracy: Political stability; Government effectiveness; Public trust in politicians; Transparency of government policymaking; Factionalised elites; State fragility; Political pressures and controls on media content; Political competition

2. 2. Rule of law: Efficiency of legal framework in challenging regulations; Efficiency of legal framework in settling disputes; Judicial independence; Strength of auditing and reporting standards; Corruption perception

3. 3. Business environment: The efficiency of the clearance process by border control agencies, including customs; Quality of trade and transport-related infrastructure; Competence and quality of logistics services; Ability to track and trace consignments; Taxation on investment; Financial market development; Labour market efficiency; Production process sophistication; University–industry collaboration in R&D; Capacity for innovation; Company spending on R&D; Venture capital availability; Intellectual property protection

4. 4. Bureaucracy: Public services; Favouritism in decisions of government officials; Irregular payments and bribes; Wastefulness of government spending

5. 5. Land: Land administration and management; Registering property

6. 6. Human rights: Voice and accountability; Freedom of expression; Protection of minority investors’ rights; Ethical behaviour of firms

Source: Authors, based on the synthetic institutional Indices.

Given the clustering procedure that was applied in defining the synthetic institution indices, it may be the case that some individual indicators do not fit the label attributed to the cluster very well. For instance, in business environment, some indicators refer more to the behaviour of firms, like ‘spending on research and development (R&D)’ or ‘production sophistication’ than their environment, although particularly negative indicators there include ‘customs’, ‘infrastructure’, and ‘lack of competition’. In the same way, it might be considered that ‘irregular payments and bribes’ would belong more to the rule of the law than bureaucratic quality – but the fact that it appears in the latter cluster clearly shows that this infringement of the rule of the law is closely linked to unsatisfactory ‘public services’ and ‘favouritism by government officials’, and therefore to an under-performing bureaucracy.

Box 3.1 could also have shown the individual indicators with scores above the mean of the synthetic indicator. It is worth stressing that, the low government militarisation index and the ‘autonomy’ of the government do not do as badly as other indicators. However, they do not necessarily do well either. Transparency or press freedom may be above the mean score of ‘democracy’, but that score is low, and those indicators are simply less low in the global ranking. Yet it may be worth keeping this kind of nuance in mind.

Another interesting point is the relative lack of consistency of various sources on the same topic. For instance, ‘rule of law’ as evaluated by Freedom House¹ is above the mean in the synthetic rule of law index, whereas ‘rule of law’ as evaluated by the Quality of Government² falls below the mean. Clearly, this kind of discrepancy shows the unavoidable imprecision of these individual estimators – sometimes themselves based on several sources – and underlines the need to be cautious in interpreting these results.

Is Bangladesh an outlier in the institution–development nexus? The preceding comparisons of Bangladesh with other countries were based on ad hoc criteria, whereas the analysis of its global ranking is biased because of the presence of so many countries at higher level of development. A relevant comparison may be to match Bangladesh with countries at similar levels of development and to see whether it does so badly, and in what dimension of the synthetic institutional indices. To do this, a simple approach consists of running a regression of the various institutional indices on a development index of the countries and to test whether Bangladesh is an outlier on the negative side, that is exhibiting a negative gap greater than 2 standard deviations, as usually defined in econometric work. Two definitions of the level of development have been used: GDP per capita – measures in international 2011 dollars – and the Human Development Index (HDI), used by the United Nations, which comprises not only GDP per capita after normalisation but also measures of education and health. To avoid this procedure having to depend too much on the relationship between institutions among advanced countries, or on the difference between developing and advanced countries, the estimation is performed on developing countries only.

Figure 3.2 shows the scatter plot of the democracy synthetic index against the log of GDP per capita for developing and emerging countries, with a trend line that represents the predicted value of the democracy synthetic index on the basis of GDP per capita. It can be seen that Bangladesh lies below the line, which means that, conditionally on its level of GDP per capita, Bangladesh underperforms on that index. Yet the gap with respect to the trend line is not sizeable, which means that Bangladesh cannot be considered an outlier in comparison with other observations. In other words, there is nothing exceptional in such a deviation from the trend line. This would not be true, however, of China, Iran, or Egypt, because their gap with respect to the trend line is larger than twice the standard deviation of that gap among all observations.

Figure 3.2 Scatter plot of the democracy synthetic index against (log) GDP per capita.

Note: Democracy synthetic index as a function of (log) GDP per capita (2011 US$).

Source: Authors, based on the synthetic institutional index

Table 3.2 summarises the results obtained for the six synthetic indices using GDP per capita or the HDI as normalising device. Because the deviation of Bangladesh from the norm never exceeds 2 standard deviations, it cannot be said that Bangladesh is an outlier in any institutional dimension. What is striking, however, is that, conditionally on its level of development, Bangladesh always underperforms. In other words, it cannot be said that Bangladesh’s bad position in the global institutional ranking shown in Table 3.1 is due to its level of development, as measured by GDP per capita or the HDI. Even controlling for this – that is, even comparing it with countries at a comparable level of development – Bangladesh is under-performing. This is true for all institutional indices except one, business environment, for which Bangladesh is slightly above the norm. Indeed, it was on this index that it reached the highest position in the global ranking discussed earlier.

Table 3.2 Normalised deviation of Bangladesh from predicted synthetic indices based on GDP per capita and Human Development Index

	Democracy	Rule of law	Business environment	Bureaucracy	Land	Human rights
Deviation from GDP norm	–0.44	–0.82	0.24	–0.72	–1.19	–0.74
Deviation from HDI norm	–0.53	–0.93	0.04	–0.84	–1.43	–0.91

Note: GDP (HDI) norm = predicted value of the regression of synthetic indices on log GDP per capita (HDI).

Deviations are standardised by standard deviation of residuals.

Source: Authors, based on the synthetic institutional index.

D What Have We Learnt from the Cross-Country Comparison?

Bangladesh has gone through several phases of crisis in the past. Despite numerous challenges, most indicators describing the institutional environment and the political and socio-economic conditions have significantly improved over the last three decades, very much in line with the stabilisation of the political scene since the mid-2000s. The overall socio-economic condition has improved. Even an indicator like control of corruption is still gradually improving today.

The situation looks less positive when comparisons are made between the current institutional context in Bangladesh and that in other countries, even when the comparison is restricted to developing countries. The synthetic institutional indices, based on a large number of individual indicators available in databases on governance and the quality of institutions, paint a broad picture of Bangladesh’s institutional context that is not positive. Bangladesh is found to be in the bottom 20% of global rankings based on these indices and, in some institutional dimensions, even in the bottom 10%. As a matter of fact, despite its development achievement over the last two decades, Bangladesh is even outperformed on all institutional dimensions by several developing countries, including poorer countries.

This outperforming is not uniform, and much can be learned for an institutional diagnostic from disparities across the various institutional indices. Bangladesh appears as particularly weak in areas like bureaucratic quality, rule of law, land issues, and, to a lesser extent, human rights. However, the situation is noticeably better, though still far from outstanding, when considering the democratic functioning of the country and the business environment it offers. It is interesting that these relative institutional strengths relate to two key features of Bangladesh’s development over the last 20 years or so: the relatively stabilisation and pacification of the political game and the surge of manufacturing exports in the RMG sector.

This kind of ranking must nevertheless be treated with caution. On the one hand, Bangladesh does not appear as an outlier when the ranking is made conditional on the level of development of a country. It is still the case that it often underperforms other countries in several areas, though mostly by a narrow margin. It does better with respect to the business environment. On the other hand, it must be kept in mind that individual indicators of governance and institutional quality are necessarily rough and may miss important details that might change the overall judgement to which they lead. Relying only on them to establish a diagnostic would thus be extremely restrictive. Hence the alternative approach of surveying different types of decision makers on their perceptions of the institutional strengths and weaknesses in the context in which they operate, as is discussed in Section III.

III The Country Institutional Survey (CIS) in Bangladesh

Although many thinkers throughout history have thought of societies as organisms with some similarity to the human body, simple diagnostic tools of the kind that are available to detect human diseases do not exist for societies and the institutions that govern them, even when the investigation is restricted to what may weaken their economic development. Economic development per se, and its relationship with institutions, are such complex topics that only in-depth analyses can possibly shed some light on them. Even the ‘growth diagnostic’ tool proposed by Hausmann et al. (2008) identifies ‘binding constraints’ on growth that are contingent upon the ‘economic and institutional environment’ of a country and does not say much about the institutional roots of these constraints. This was very much the approach followed in Chapter 2. Yet the complexity of the relationship between institutions and development should not prevent an analysis from relying on simple diagnostic tools, provided that the limitations of these tools is kept well in mind when trying to go deeper in an institutional diagnostic exercise. Simple tools can help us find the way to search for the bigger picture. This was done in the preceding section by trying to extract information from existing cross-country indicators of the quality of governance and institutions. Another simple tool is reported on in the present section: the survey responses of decision makers of various types who were asked about the institutional features that hinder Bangladesh’s development.

Two approaches were followed in this survey. The first was a questionnaire survey that was administered among a selected sample of people who regularly confront Bangladesh’s institutional context in their activities. This survey was copied from the CIS, which has been used in other countries³. The second approach consisted of conducting open-ended interviews with a few key informants in political, business, social, and academic circles.

A The Survey: Background and Design of the Questionnaire

The CIS is a sample survey tool developed as part of the institutional diagnostic activity of the Economic Development and Institutions (EDI) programme. Its aim is to identify institutional challenges as they are perceived by the people in a country who are most likely to confront them on a regular basis. These challenges are then made the subject of deeper scholarly analysis. Being based on a broad sample of respondents, the CIS intends to yield a more diverse view of the country than the numerous institutional indicators that rely most often on the opinion of a few experts.

At the beginning, a pilot for the CIS in Bangladesh was held in late 2018. Those who took part in the pilot occupied top decision-making positions at this time. The reason for choosing respondents from top decision-making positions was to get a lucid idea of the institutions in Bangladesh, since such respondents either interact with the institutions on a regular basis or they work as an active part of the institutions. As decision makers, they have in-depth knowledge of institutions and their weaknesses. Of course, these opinions are quite different than the opinions of the general mass of the people, as they are based on direct experiences with the institutions and/or rigorous analysis of the institutions from their vantage point. The insights gathered from the pilot helped conducting the Bangladesh CIS between December 2018 and February 2019. The remainder of this section will discuss the design of the questionnaire and the execution of the survey, respectively.

The questionnaire had three primary components: a section on the personal characteristics of the respondent; another on the institutional areas seen as most constraining by the respondent; and the last one, a long section on the respondent’s perceptions of the institutions and the functioning of institutions in Bangladesh.

The first section was split into two parts. The first part initiated the discussion and asked general questions such as the respondent’s name, gender, and sector of affiliation, including political affinity and sub-sectors that the respondent was associated with. The other part compiled more sensitive information on the past and present occupation of the respondent, the location of their work, their family size, and their religion.

The second section of the questionnaire was composed in such a way as to gather information about the most constraining institutional areas in Bangladesh. In this part of the questionnaire, the respondents were provided with the details of institutional areas that we had focused on for the survey. This comprised seven broad institutional areas: ‘Political institutions – executive’; ‘Political institutions – system’; ‘Justice and regulations’; ‘Business environment’; ‘Civil service’; ‘Land’; and ‘People’ (Box 3.2). Respondents then had to select two institutional areas that, according to them, most constrain development in Bangladesh. The chosen areas were not only important for the analysis but were also important for the subsequent part of the survey since they determined the set of questions presented to the respondent in the third section of the survey.

Box 3.2 Institutional areas and description

1. 1. Political institutions – executive: Effective concentration and use of power; type of governance; relationship with parliament, judiciary, local governments, media, and civil society

2. 2. Political institutions – system: Functioning of elections; voice of opposition parties, civil society, and media; checks and balances on the executive

3. 3. Justice and regulations: Fairness, independence, and effectiveness of the judicial system; regulation of public and private monopolies

4. 4. Business environment: Relationship between the private sector and public administration; protection of property rights and labour contracts; business registration and licensing; taxation; availability of infrastructure

5. 5. Civil service: Efficiency, fairness, effectiveness, and transparency in the management of social and economic policy, including customs, taxation, education, health, etc.

6. 6. Land: Provision of ownership, protection of tenants and small holdings, promotion of commercial ventures

7. 7. People: Sense of solidarity, discrimination practices, security, trade unions

Source: CIS, Bangladesh.

The core section of the CIS comprised 415 unique questions on the perception of institutions in Bangladesh. The collection of information relied on a Likert scale, ranging from ‘not at all’ and ‘little’ to ‘moderately so’, ‘much’, and ‘very much’. Responses were then converted into discrete numbers, ranging from one to five, for the analysis. The CIS questionnaire was unique in several dimensions, mostly with the aim of making it as close as possible to the specific context of Bangladesh.

There are particular challenges that come with surveying top-tier executives, not only with access but also because their time may be limited. Our survey had a high volume of questions and sought to gather information on a broad spectrum of institutional issues. If we had asked every respondent every question, the survey would be far too long to be of practical use. Keeping these constraints in mind, the survey was conducted in a dynamic way. As mentioned earlier, in the second section of the questionnaire, respondents were asked to identify the most constraining institutional areas according to them, from the list of seven broad institutional areas. Then they were first asked to answer both the primary and secondary questions related to the two institutional areas they had selected, as well as only primary questions related to the other five institutional areas that they did not choose. Notice also that, given the overlap between institutional areas, respondents had to answer about 70–80% of the full set of questions on average.

Changes in institutions are infrequent and most happen over the course of time rather than suddenly and abruptly. Even though there are a few examples of institutional changes which have happened overnight, most institutions persist. At the same time, human psychology works in such a way that people tend to react to the most recent events associated with a certain entity. For that reason, it is quite possible that the perceptions of the respondents were biased towards the present. However, the current de jure institutional authority in Bangladesh has not changed much over the last decade and so it was expected that perceptions about the overall context would be reflected in their responses. In addition, in-depth discussions with top decision makers on the institutional constraints shed light on the changes over time. Second, the enumeration took place right around the time when a general election was taking place in Bangladesh. The election thus had impacts at several levels, in terms of survey responses that commented specifically on recent institutional characteristics. Last but not least, very recent changes in institutions do not explain the past economic trajectory, so the questions about the more stable aspects of institutions were relevant.

The survey covered the views of people who were either affiliated with institutions or in close contact with them. The survey also aimed to capture the view from the top down, where decisions are made or where policies are generated. To do this, it was very important to select respondents from the first- or second-tier position of any institution. These decision makers had experienced the impacts of changes in certain institutions first-hand and were concerned about the functioning of the country’s institutions. As a consequence, a pure random sampling in the overall population was not an option. The selection of respondents had to be based on an arbitrary stratification of groups of expert respondents, to make sure various sectors, occupations, and individual profiles would be present in the sample. This implies a strong selection bias with respect to the Bangladesh population, but, of course, this was deliberate.

B Execution of the Survey

The Bangladesh CIS was conducted between December 2018 and February 2019 in a collaborative effort between EDI researchers, Oxford Policy Management (OPM), and South Asian Network on Economic Modeling (SANEM), a think-tank from Bangladesh. A total of 355 individuals were sampled in a purposively stratified sample. The selection process followed two steps. First, researchers listed strata in terms of occupation, position level, geographical constraints, and tentative gender balance. Samples were surveyed in major cities in the country, like Dhaka, Gazipur, Chattogram, Sylhet, Rajshahi, Bogra, Rangpur, Barisal, and Khulna.

SANEM, in cooperation with OPM, determined a list of target respondents who satisfied the occupational, geographical, and gender considerations. Next, these probable respondents were contacted, and if they gave their consent they were interviewed. The sample is divided into five sectors: politicians, bureaucrats, business executives, academics, and civil society members. The respondents included politicians from ruling party and opposition; current and ex bureaucrats; business executives from agriculture, fishing, livestock, manufacturing, construction, Information & communication, wholesale and retail, health, transport, bank; academics from teaching and research professions; and people from non-governmental organisation (NGO). Responses of a total of 48 politicians, 51 bureaucrats, 131 business executives, 76 academics, and 49 civil society members were collected.

Table 3.3 provides details regarding the characteristics of the survey respondents. It is most unfortunate that, though the initial target was for at least 31% of the sample to be female, the enumerators struggled to contact or arrange interviews with female respondents. This may be linked to the fact that in Bangladesh, only a few of the top-tier positions are held by women. This in fact was observed to be the case when conducting the survey and can be considered a finding of the study. Thus, only 14% of the respondents were female.

Table 3.3 Composition of the sample

Respondent’s main characteristics		Occupation history (number of respondents)
Number of female respondents	50	Politician	48
Number of respondents: Married	324	Bureaucrat	51
Average family size	4.15	Business executive	131
Average age in years	47	Academic	76
Average education: university degree or above	318	Civil society	49
Average years of experience	21	Total number of respondents	355
Average years of experience at the current institution	15

Source: CIS, Bangladesh.

The main goal of the CIS in Bangladesh was to capture an amalgamation of viewpoints about the institutions in the country. As mentioned earlier, the survey targeted respondents from the top tier; thus, the mean level of education for these respondents was well above the national average. As we can see in Table 3.3, about 90% of the respondents had a university degree or above. The same argument regarding choosing respondents from the top tier applies to the age distribution of the respondents. Since it takes years of experience to reach a top-tier position, respondents tended to come from older age brackets. The average years of experience of the respondents explains the spectrum of their experiences with institutions in the country, and it also indicates the way in which the survey captures the respondents’ perceptions of institutions in a dynamic way: as most of them had worked under varied circumstances, each of them had a unique experience with the institutions in Bangladesh which the survey intends to capture.

It is also important to point out that 23 respondents declared a political affinity with the ruling party and 25 with the opposition, and the rest of the respondents declared no political affinity. This enables us to compare the responses of respondents with ruling party or opposition affiliation with respondents without any declared political affinity, to assess whether party affiliation had any bearing on the responses given. In terms of geographical diversity, most respondents lived in an urban area. It is in fact not surprising to see that most of the respondents resided in urban centres, as the survey targeted the elite in the country, who tend to live in or close to the cities. Since most head offices or main branches of public and private organisations in Bangladesh are located in Dhaka, the region around the capital is overrepresented.

C Results of the CIS in Bangladesh

1 Critical Institutional Areas for Bangladesh’s Development

According to the respondents, the major constraining institutional areas for the development of Bangladesh are the political institutions and public administration. The ranking of these two areas depends on the measure chosen to aggregate individual opinions. However, it can be seen in Figure 3.3 that they are very close to each other in number of occurrences chosen by respondents. Notice also that, conditionally on being chosen, political institutions were selected by the respondents. Justice- and regulation-related institutions come in third position in the ranking of the most critical institutional areas for development in Bangladesh. On the other side of the spectrum, only 5.1% respondents chose land as one of the two most constraining institutional hurdles in Bangladesh’s development, possibly because the respondent assumed that this area needs specific knowledge of land administration. However, due to the design of the questionnaire, most of the respondents had to answer questions related to land, and it has one of the lowest average scores. This will be discussed later in detail.

Figure 3.3 Choice of institutional areas.

Note: For a description of the institutional areas, see Box 3.2.

Source: CIS, Bangladesh

The probability of framing bias must be considered, with the first areas in the list appearing more frequently than the other choices of respondents. It is possible that the respondents intentionally chose the areas with which they were affiliated as most constraining for Bangladesh’s development. However, since all respondents answered most of the questions (through primary and secondary questions), survey responses should be independent of biases and should have provided a robust idea about each of the constraints being discussed.

The choice of the top two constraints to development, according to respondents’ opinions, is a piece of information in itself, but it also determined the number of questions asked of each respondent. Given the explicit choices made during the selection of institutional field, the fact that respondents faced detailed questions about their top two choices, and not about other areas, raises a concern. Considering choices of institutional areas by sector of affiliation, it is possible that choices might be biased towards the sector of affiliation of the respondents. Additionally, it is quite possible that some less important areas are left out because there is no information about them. Alternatively, some institutional areas might be left out because of a perception that they were working well, or they could work poorly but be considered unimportant for economic development. For these two reasons, it was important to gather information about all areas. It was thus decided to ask all primary questions in relation to all areas. For this, even the less critical institutional fields were covered by all respondents.

Analysis of the survey results show that choices of institutional areas were somewhat different for male and female respondents. The top choice of male respondents was political institutions, whereas for female respondents, it was mainly public services. This is in line with general norms since women mostly experience discrimination at the public service level. The rest of the institutional areas received an almost equal degree of preference.

Another interesting observation from the CIS results relates to the choices of institutional areas by political affinity. The choices of institutional area of respondents from both the ruling and opposition parties are skewed towards political institutions. Politicians from the ruling party did not consider the business environment as involving any institutional constraints. On the other hand, given the current context and circumstances, it is surprising to see that supporters of the ruling party considered public services as one of the constraints. The institutional choices of respondents with no political affinity are almost equally distributed across institutional areas.

The choices of institutional areas made by respondents from the business sector differed depending on the specific sector they were affiliated with. For respondents affiliated with agriculture and manufacturing, the top choice was ‘Business environment’. However, for those affiliated with the service sector, it was ‘Justice and regulations’. It is very interesting to see that the choice of ‘Land’ as an institutional area was more common for respondents from the service sector than for those from the other two sectors. Respondents affiliated with agriculture chose ‘Public services’ as a constraint more frequently than respondents affiliated with manufacturing or the service sector.

2 The Perceived Functioning of Institutions in Bangladesh

Within and across areas, the CIS aimed to identify, as precisely as possible, which specific institutions were perceived as constraining by respondents. The subsequent analysis evaluates questions by their mean response on a scale ranging from 1, ‘very negative’, to 5, ‘very positive’. For questions asked in a negative way, the Likert scale is inverted to make sure that a higher value always means a better perception. Questions are then divided into clusters and sub-clusters to closely identify the core problems of the institutions in Bangladesh. This section first discusses the state of Institutional areas captured by the CIS and what are the underlying state of each of the institutional areas. Then the underlying problems of the institutions in Bangladesh are discussed. Finally, the choice of institutional constraints is discussed from the perspective of respondents’ gender and political affiliation, to identify any differences correlated with respondents’ characteristics.

The negative perception of institutions in Bangladesh can be observed if we consider the distribution of the average scores of each of the questions. The mean score is 2.81, slightly below the mid-point of the Likert scale, lying at 3. This is not unusual in opinion surveys and may simply reflect the ways in which respondents answer questions. It is therefore more interesting to look at the tails of the distribution: namely, questions with clearly positive or negative answers. Figure 3.4 plots the distribution of questions by average score. It shows that the left tail (negative perception) is fatter than the right one (positive perception). A total of 131 questions has an average score below 2.5, while only 39 score above 3.5.

Figure 3.4 Distribution of questions by average score – all questions.

Note: Total number of questions: 415. Options in the Likert scale: 0 = no opinion; 1 = very negative; 2 = negative; 3 = neither negative nor positive; 4 = positive; 5 = very positive.

Source: CIS, Bangladesh

Looking at the perception of the institutions by sector of affiliation, we can see that the average score for each of the sector is either 3.00 or below (Figure 3.5) the threshold level. It is not very surprising to see that politicians and bureaucrats consider institutional quality to be slightly better than do academics and civil society members. Business executives on average gave a score of 2.77 for institutional quality in Bangladesh, a relatively low result. Figure 3.5 suggests that, in general, the no response rate was between 4.7% and 6.1%. While, on average, only 35.4% of respondents expressed positive views (4 and above) about the functioning of the institutions, leaving aside politicians and bureaucrats, all other three categories of respondents held much lower opinions. The dominance of ‘negative’ views (1 and 2 together) is the highest and is almost the same for business executives (52.7%) and academics (52.8%).

Figure 3.5 All questions – analysis by sector of affiliation.

Note: Options in the Likert scale: 0 = no opinion; 1 = very negative; 2 = negative; 3 = neither negative nor positive; 4 = positive; 5 = very positive.

Source: CIS, Bangladesh

3 Distribution of Average Score by Cluster

Figure 3.6 depicts the percentage distribution of scores by each theme. As the figure shows, on average, the ‘no opinion’ view had a share of only 5%; the ‘very negative’ perception had a share of 4.6%; the ‘negative’ view had a share of 45%; the ‘indifferent’ view had a share of 8.6%; the ‘positive’ view had a share of 35.1%; and the ‘very positive’ view had a share of only 1.8%. The worst situation is observed in the case of the theme related to ‘Land’, where 57.7% of the responses were ‘negative’ (1 and 2), followed by ‘Civil service’ with 56.8%. The general picture drawn from these figures offers a pessimistic view of the institutions in Bangladesh. However, in the discussion of each institutional area by cluster and sub-cluster in the previous sub-sections, we identified specific problems associated with the institutions in Bangladesh. This is more useful than generalising perceptions of the institutions in Bangladesh based on these scores.

Figure 3.6 Percentage distribution of scores by theme.

Note: Options in the Likert scale: 0 = no opinion; 1 = very negative; 2 = negative; 3 = neither negative nor positive; 4 = positive; 5 = very positive.

Source: CIS, Bangladesh

4 Identification of the Major Areas of Institutional Weaknesses

The previous section discussed the distribution of scores across institutional areas. We saw that, on average, the scores are well below 3. However, generalising these scores can be misleading. In this section, the scores for all the sub-clusters are plotted in Figure 3.7. Although we have previously discussed specific clusters and sub-clusters, a graph like this provides us with the broader viewpoint regarding the institutions in Bangladesh. This also gives us incentives to study further, and concentrate on, those thematic areas that we intend to study for the growth diagnostic for Bangladesh.

Figure 3.7 Major areas of institutional weaknesses.

Note: Options in the Likert scale: 0 = no opinion; 1 = very negative; 2 = negative; 3 = neither negative nor positive; 4 = positive; 5 = very positive.

Source: CIS, Bangladesh

It is clear from Figure 3.7 that institutional anomalies in Bangladesh are prevalent in relation to the judiciary, the business environment, the efficiency of public services, the efficiency of tax administration, and land. The discussion of each of these areas in the previous sub-sections illustrated that political institutions and conflict and discrimination are cross-cutting and are associated with these broader thematic areas, depending on the context of the discussion.

5 Perceptions of Institutions from a Gender Perspective

An interesting insight of the analysis is that the survey asked some questions related to discrimination based on gender. The majority of the respondents, regardless of gender, agreed that discrimination on the basis of gender is prevalent in Bangladesh. As the responses to these questions show the extent of discrimination spans both public authorities, society, and the workplace.

Figure 3.8 reports the responses where the largest percentage difference in responses (in absolute term), related to the sub-clusters, were found between male and female respondents. The figure shows that the largest difference in opinion between male and female respondents was in terms of long-term planning, central bank independence, evaluation of policies, corruption in electoral process, civil liberties, and quality of public policymaking. For these sub-clusters, the percentage difference in opinion was at least 10%. However, it is interesting to see that, though there were vast differences in opinion in the context of so many sub-clusters, both males and females agreed on the fact that discrimination exists in the society, especially in the labour market, as differences in responses were very low for these sub-clusters.

Figure 3.8 Percentage difference (absolute) in responses by gender.

Source: CIS, Bangladesh

6 Perceptions of Institutions Based on Political Affinity

The previous sub-sections have discussed in detail the institutional areas based on sector of affiliation. However, as the survey gathered information on the political affinity of the respondents, it is interesting to look at any differences in opinions based on such political affiliation.

First, we plot the differences in opinions between the respondents who were affiliated with the ruling party and those affiliated with the opposition. This is depicted in Figure 3.9-A. The figure shows there was a vast difference of opinions between these two types of respondents. The average percentage of difference in responses lies around 20%.

A. Percentage difference (absolute) between the scores of ruling affiliation and opposition affiliation.

Source: CIS, Bangladesh

B. Percentage difference (absolute) between the scores of ruling affiliation and no affiliation.

Source: CIS, Bangladesh

C. Percentage difference (absolute) between the scores of opposition affiliation and no affiliation.

Source: CIS, Bangladesh

Figure 3.9 Percentage difference (absolute) in responses by political affinity.

Next, we plot the percentage difference between the scores of respondents with affiliation to the ruling party and respondents who did not express their affiliation to any political party (Figure 3.9-B). This is almost identical to the previous figure, with a low average percentage difference.

However, the most interesting point to note is that depicted in Figure 3.9-C, where we plot the differences in opinion between respondents with an affiliation to the opposition party and those without any political affiliation. This shows that the average percentage difference in opinions is very low – almost close to zero – showing the similarity between the opinions of affiliates of the opposition parties and the opinions of respondents with no revealed political bias, as regards the institutional areas. Those who stated they had ‘No affiliation’ are likely to often include opposition people who do not dare say so, or who do not want to be involved in politics.

7 Open-Ended Interviews with Top Decision Makers and Policymakers

Parallel to the CIS we conducted several open-ended interviews with top decision makers and policymakers in Bangladesh. These decision makers naturally were not interviewed in the same way as the other respondents to the CIS. Nor were they selected based on a stratified sampling technique. These interviewees were chosen simply because they had been working with the institutions in Bangladesh and/or were closely affiliated to and associated with the functioning of the institutional areas under discussion. These stakeholders (politicians from the ruling and opposition parties, bureaucrats – current and retired, business executives from different sectors, academics – teaching and research, NGO members, and other activities) were carefully chosen to avoid any kind of bias. They were asked several questions about the institutions and institutional diagnostics for Bangladesh. In their responses, they pointed to several sectors it may be useful to concentrate on in order to come up with diagnostic tools for Bangladesh. The anomalies in these sectors were then discussed in detail with these stakeholders. The main aspects of the institutional areas mentioned by these stakeholders are discussed below.

The failure of institutions to diversify markets and exports is causing Bangladesh to lose a huge sum of revenue. The major problems relating to market diversification include: a lack of comparative advantage; the inefficient use of available resources; poor capacity to ensure product diversification; high trade costs; poor physical connectivity; political patronage and bias; and the size of the importing country’s economy, etc. In Bangladesh, there are still no proper studies that have been conducted to understand the institutional failures in relation to market diversification. To this end, it is important to understand the dynamics of the vast concentration exports around the RMG sector and the neglect of other potential industries.

The key feature of the fiscal sector is the public revenue and expenditure management, with the aim of reducing infrastructure gaps, promoting private investment, generating employment opportunities, and ensuring the efficient redistribution of wealth through a pro-poor and inclusive fiscal policy. Data show that tax revenue is the major source of income or revenue for the Government of Bangladesh. Low administrative capacity and strong lobbying by businesses can be seen as the prime institutional failures as regards revenue generation for the Government. The Government should bring the target group under the tax net and make it mandatory to submit income tax returns, whether an entity is taxable or not. However, another important challenge to progress is mismanagement in expenditure in a weak institutional environment. Delayed funds disbursement, delay in land acquisition, and the lack of skilled project directors are also identified by the Implementation Monitoring and Evaluation Division (IMED) of the Ministry of Planning as important reasons behind mismanagement in expenditure.

With a continual wealth transfer from the general public to the corruption-ridden and seemingly incompetent state-owned banks (and ultimately to defaulters), the non-performing loan (NPL) situation has worsened. In Bangladesh, the main source of total NPL is state-owned banks. The experts interviewed identified a few factors behind this situation. Of course, they mentioned systemic corruption, but they also went further and mentioned the appointment of corrupt officials to important positions at the state-owned banks as a fundamental reason for this. A few of the symptoms of political patronage in the sector are: the Finance Ministry’s overreach in licensing private banks while exercising political considerations; injecting incentives without the recommendation of the Bangladesh Bank; not complying with the suggestions of the central bank; and influencing the decisions of the autonomous central bank. Though it is an independent regulatory authority, the central bank cannot completely monitor these private banks as they are owned by politically influential people. Historically, there have been many regulations in the banking sector, especially relating to the entry/exit mechanism of banks and their governing bodies. The Bank Company Act, 1991, has been amended six times since its formation. Recent laws have sought to bolster political hold over the governance of these banks. These new laws have brought in changes in directorship positions, triggering a state of panic among depositors and other stakeholders. At the current point in time, from the discussion, we can see that the independence of the central bank is not producing its intended benefits, due to the presence of political pressure. Thus, the institutional efficiency of the banking sector hinges upon diversification of the sector, to control the ongoing political pressure place upon it.

Land litigation procedures and land management in Bangladesh are convoluted. With land being the most valuable asset in the country, the institutions associated with land management are susceptible to bribes. The influence of political patronage has made the transfer of land and land availability for businesses a complex issue. For better and more sustainable economic growth, the availability of land is crucial. However, the current situation in Bangladesh suggests that the problems associated with land are much politicised. Weaknesses in the institutions associated with land are related to the prolonged time required to obtain approval for transfer of land, politicisation in allocating land, and bribery in the transfer or approval of land use. As stakeholders mentioned, the convoluted nature of the problem and the failure of the judicial system to ensure justice in cases related to land may have a long-term impact on Bangladesh’s economic system.

The judiciary in Bangladesh is faced with many problems: a low number of judges compared to the number of cases; an unregulated system and laws; and the questionable independence of the system. All of this calls for an elaborate study on the subject. Complex procedures, case backlogs, and a lack of effective case management are also key constraints to the court system in Bangladesh. An independent judiciary is the sine qua non of democracy and of good governance. However, though the Constitution requires the separation of the judiciary from the executive, no steps whatsoever have been taken by the legislative or executive branch of the government in this regard. The independence of the judiciary is a must for any democratic country but attempts to influence the judiciary and steer it for political benefit are prevalent in Bangladesh.

D What Have We Learnt from the CIS?

The CIS and the open-ended discussion with top decision makers and policymakers has provided some interesting insights about the institutional functioning and mechanisms in Bangladesh. Key findings regarding the institutional strengths and weaknesses, and the recommendations of the stakeholders in the open-ended discussions, are consistent, even though the latter were able to go into more detail than the CIS. The following paragraphs summarises the most salient points that have come out of this double exercise.

In the first place, it should be stressed that the CIS yielded a ranking of problematic institutional areas similar to the one derived from exploiting cross-country institutional indicator databases, as reported on in the preceding section. Namely, the two areas found to be the most favourable (or perhaps the least unfavourable) to Bangladesh’s development are the business environment and the political system – an area that roughly fits the ‘democracy’ synthetic index in the preceding section. This convergence between insiders, that is local decision makers, and the experts behind the cross-country indicators, reinforces the view that other institutional areas than the preceding ones are problematic.

Second, it turns out that the general appraisal of institutional areas most likely to hinder development is not very informative, except perhaps in regard to the low weight put on land issues, something that is surprising given the emphasis of key informants on this aspect of Bangladeshi institutions. By contrast, in the CIS, the detailed evaluations were much more informative. They clearly put the civil service and land issues at the top of the list of poorly functioning institutional areas, closely followed by the state of political and administrative management exercised by the executive.

The sub-cluster analysis, within each institutional area, yielded still more interesting, because more precise, information. Of particular importance are the following weaknesses it pointed to:

• ubiquitous corruption (election, business, and recruitment in civil service);

• executive control of legal bodies, media, judiciary, and the banking sector;

• inadequate coverage of public services;

• the number and intensity of land conflicts; and

• gender discrimination.

A few institutional aspects were also found to be rather satisfactory, though not always without some contradiction as regards other judgements. These include the general development of a middle class, the national feeling, and the quality of public policymaking. Of very special importance for the subsequent analysis in this volume is also the relative satisfaction regarding informal arrangements with the administration and as a way to secure contracts, as an efficient way of avoiding the ineffective formal channels. This may seem a bit paradoxical when evaluating institutions, but this opinion is quite revealing of what may be an important trait of the institutional context in Bangladesh.

Finally, the opinions expressed by the key informants generally confirmed the views of the CIS respondents: as, for instance, when they emphasised the low administrative capacity of the Bangladeshi state, corruption, or the ineffectiveness of the judiciary. But they added to the survey by pointing to sectors of activity where those weaknesses may be more salient. Of special importance from that point of view is their emphasis on industrial policy and the lack of diversification away from the RMG sector, to which this lead, possibly because of the over-influence of the RMG entrepreneurial elite. Key informants’ insistence on the severe failings in the regulation of the banking sector, the corruption behind the huge and increasing NPLs, and the lack of regulatory power on the part of the central bank are also deeply revealing of the way several institutional weaknesses generate deep inefficiency in a key sector of the economy.