Can Bangladesh develop without decentralising? Implementation lessons from East Asia

April 1, 2019

By  Ahmad Ahsan

This article is the last in a series of three articles that tries to answer the title question drawing on lessons from East Asian countries. The answer is that it is highly unlikely that Bangladesh, a large country of 161 million people, can sustain its impressive development record with its current highly centralised government. Future development will become more knowledge intensive and institutionally demanding, needing strong governments, knowledge, and initiative at both the central and local levels. Across the world, countries decentralise as they develop. But it is the experience of East Asia’s spectacular growth and some of their highly decentralised governments that draws out most clearly the linkages between development and decentralisation and provides the most relevant lessons.
This first article in the three-article series laid out the context for decentralisation in Bangladesh and drew a contrast between a highly centralised Bangladesh and a highly decentralised East Asia. Drawing on international data it showed how Bangladesh’s extreme centralisation – political and economic – adversely affects overall urban development. It then provided evidence to show that some of the dynamic East Asian countries are among the most decentralised-not only among developing countries but also among all countries. The second article of the series explored how decentralisation has helped East Asian development. Decentralisation boosted growth and development in East Asia through five channels. First, because local leaders and the people owned the local economy and reforms, they had powerful incentives to develop. Second, this ownership led to intense competition among local governments to achieve faster development- not only external global competition, but also domestic sub-national competition to develop stimulated growth. Third, decentralisation and local governments piloted policy reforms, promoted experimentation and limited risks. Fourth, decentralisation helped find and train leaders. Fifth, it facilitated high-quality urban development that is essential for sustaining growth and structural change.
The third article will draw on the lessons of East Asian decentralisation, of both successes and failures, and suggest a course to implement decentralisation in Bangladesh.

The design of decentralisation will be critical to manage risks

Decentralisation is not a panacea; it has risks. Without careful design, decentralisation can lead to several adverse consequences. These include local elite capture by the locally wealthy and politically powerful who can indulge in corruption and rent seeking. Another consequence can be that instead of a race to the top there can be a race to the bottom as local governments give away tax exemptions and land too cheaply to investors and neglect environmental consequences. Another serious consequence can be vertical fiscal imbalances and local government debt that can lead to a decline in public services and threaten overall macroeconomic management. Poor functional assignments and inadequate arrangements for coordinating among local governments can lead to fragmentation and decline in government services. Finally, decentralisation can lead to greater regional inequality by divergence in development between richer and poorer regions. Faulty design such as ambiguous assignment of responsibilities, weak transparency and accountability systems, decentralising to the wrong tier of government, inadequate fiscal resources leading to imbalances between functional responsibility and financing, fiscal irresponsibility and borrowing at the local levels can lead to such outcomes. In addition to lessons of success, East Asia’s experience also provides examples of how poor design can lead to some of these adverse outcomes. These examples thus give important lessons on designing decentralisation to address these risks.
Let us start with some principles and then use East Asian examples to illustrate them. First, a superior design will delineate, as clearly as possible, government functions among the different tiers of governments. This delineation should correspond to how wide the effects of government functions considered for decentralisation are. Or to use economic jargon, the delineation should depend on the level at which the benefits of the functions are ‘internalised’. For this reason, defense, foreign affairs, foreign trade and inter-state commerce, the conduct of monetary policy – critical matters that affect the entire country – are always kept as central subjects. Similarly, economic activities such as providing national transport connections, national grid systems, standardisation of measures and services, research and development activities and higher educational institutions that have nation-wide benefits should be assigned to the central government. Other subjects can then be made provincial, district, and lower level assignments again depending on the how wide the impact of the subject is.

Poor functional assignments and inadequate arrangements for coordinating among local governments can lead to fragmentation and decline in government services.

Second, fiscal decentralisation should follow functional assignments and provide incentives for good governance. Ideally, fiscal resources for local governments should come mostly as unconditional transfers from the central government and from their own tax and non-tax revenue sources. It is the ability to raise their own tax revenues that increase the accountability of local governments to the people. In practice, however, especially in initial stages, local governments do not have the capacity to raise these taxes and depend on central government transfers. Central government fiscal transfers, in turn, need to be designed to meet a few objectives. They need to:

  •  Adequately fund government responsibilities assigned to sub-national tiers in a vertically balanced way to avoid fragmentation and unfunded mandates;
  • Share public resources among the subnational entities horizontally with reasoanble equity. Otherwise subnational governments which start off with wealthier populations, greater natural resources, better infrastructure, and natural advantages such as being located in coastal areas will become richer and inequality will increase. Central government ‘equalisation transfers’ to poorer regions are often used to reduce regional inequality;
  • Incentivise local governments by rewarding good governance and development performance. This can start by linking central financing to revenue collections by the local government and good financial management practices, such as preparing timely and transparent budgets and audit reports;
  • Ensure that at least in the initial stages and for smaller governments, local governments face hard budget constraints. That is, local governments should be prohibited from running fiscal deficits or borrowing money. Exceptions can be made in the case of economically well-developed governments that have adequate revenues and assets of their own. But here too central governments should have approval authority and transparent rules for borrowing;
  • Encourage local governments to fund national priority programs, e.g. in education and health, by providing conditional transfers that are tied to sub-national government spending.
    East Asian countries provide both lessons of successes and failures in their design of local governments. As we have seen earlier, the fiscal decentralisation system of China was crucial in providing provinces incentives to raise revenues for development. The Chinese, Vietnamese, and the Indonesian governments all transferred substantial resources to provincial and prefecture/county governments to carry out their functions.

A clear lesson is resource transfers should be well balanced among provincial and lower level governments to avoid fragmentation. In China, provinces and prefecture/counties shared fiscal resources in a balanced way. However, it was not so in Indonesia. Within a span of a few years of the big-bang decentralisation in early 2000s more than a third of the fiscal resources had been transferred to local governments. Most of these resources were spread thinly across to the nearly 300 districts and some 98 municipalities while neglecting the 34 provinces. Moreover, the number of districts quickly increased after decentralisation from 294 at the end of the Suharto era to 510 in mid-2013. As a result, both fiscal resources and administrative and technical capacity became too thinly spread across districts leaving the provinces with few resources to carry out their tasks and coordinate districts. This lack of coordination led to what is known in Bahasa as ’aspemekaran’ or fragmentation of government functions. Not surprisingly, the performance and development impact of local governments have been muted even though they have varied across regions. Two lessons emerge from the Indonesian experience: first, decentralisation needs to devolve to the appropriate level. If responsibilities are devolved too far down – the 510 districts in the case of Indonesia- where capacity and accountability mechanisms are weak, development outcomes will be disappointing. Second, there is a need to phase in decentralisation and avoid a big-bang.
In the case of the Philippines, the problem lay in both directions: inadequate resources were provided to local governments and they neglected the provinces. Only about a fifth of the national budget was allocated for city and municipal governments to the neglect of provincial governments. Cities and towns were made responsible for roads and water supply but without adequate technical support. Further regional services such as regional roads and water-works connecting cities had no clear responsibilities assigned.
From our neighbors, India, and Pakistan, comes an important lesson in the opposite direction. India and Pakistan are federal countries where significant fiscal resources have been transferred to the states and provinces, especially in the case of the former. However, fiscal decentralisation mostly stops at the state and provincial level, i.e. at a too high level. In India, the state of Uttar Pradesh has a population of 210 million people while in Pakistan the state of Punjab has a population of 110 million people. On average states and provinces in India and Pakistan have more population than most countries in the world. Thus, decentralisation to the state or provinces in these countries did not go far enough.
In China and Vietnam where fiscal and functional responsibilities have been most decentralised, coordination and fiscal balances have also posed challenges. There has been duplicative and sometimes excessive investment in infrastructure. Further, local governments have sometimes resorted to excessive borrowing either directly or through the enterprises they own. Overall, however, the problem of overinvestment in these countries, rather than an underinvestment, is perhaps preferable.
Another lesson is local governments should have authority over trained civil servants and technically skilled people. More than 90% of Government servants in China work in sub-national governments. Indonesia decentralised more than 2 million staff to local governments after the big-bang decentralisation in early 2000s. However, in practice implementation was far slower and local governments had little authority over the size and composition of the civil service; over changing recruitment practices to fit in with the priorities of the governments, and in general holding the civil service accountable to them. As a result, the administration of local governments suffered from ambiguities in responsibilities and accountabilities. The Philippines transferred some 70,000 employees from central ministries to local governments following the Local Government Code of 1991. In both countries however, the central government retained responsibility for the capacity building of the civil service staff. Further, little attention was paid to the training of the elected councilors.
Local governments must be made accountable and transparent in their operations. This needs to happen from both sides: through monitoring by the central government from the top and by their constituents. It is the important task of the upper-tier central government to make the people aware of responsibilities and budgets of the local governments and make the results of monitoring available to the people. The preparation of timely, transparent budgets and good quality audits certifying probity in the use local government funds should be made preconditions for at least part of central government transfers. There should also be arrangements for the local population to air their grievances against local governments and even a process to change the local government through recall should the majority want to. Governments should also encourage both regular and social media to play the role of watchdogs.

It is the important task of the upper-tier central government to make the people aware of responsibilities and budgets of the local governments and make the results of monitoring available to the people.

Interestingly, it is China and Vietnam that do a far more thorough monitoring of provincial and county governments than the governments of Indonesia and the Philippines. China’s central government not only monitors the performance of local governments using a variety of indicators, they also encourage the population to monitor local governments and express their views using social media. Thus, a district official in Nanjing was sacked and put under criminal investigation after he was photographed smoking expensive cigarettes and wearing a very expensive watch. In Yunnan province when the local people exposed that jail officials have used violence inside the prison, this led to the removal of officials from their posts. As noted earlier, the central government tacitly encouraged citizens to demonstrate against mis-governance and corruption. Monitoring and evaluation of government performance by the central government have been weaker and irregular in the Philippines and Indonesia. Although there is a strong civil society movement in the Philippines that monitor government performance, general awareness of local government performance is low. In Indonesia, awareness and monitoring have become stronger in recent years. Hundreds of local government officials there have faced corruption charges and research and public awareness about provincial performance has increased.
In closing, it is worth noting that despite limitations, decentralisation has been popular in Indonesia and the Philippines. These shortfalls have included imbalanced decentralisation with overly weak provincial governments and more empowered district and city governments, but which had inadequate technical capacity to use these powers. Overall fiscal transfers have also been low in the case of the Philippines. As a result, the development impact of the decentralised entities has been uneven. However, there is little evidence that decentralisation has led to greater elite capture, corruption, and inequality compared to centralised government rule. Nevertheless, interestingly, surveys of the people in recent years consistently point to the satisfaction of most people with their local governments. In Indonesia, a respectably large survey’s results found that 78%, 90%, and 85% of respondents were, at least, somewhat satisfied with the quality of local administrative, health and education services respectively (Hill, 2013). In the Philippines, nearly 40% of the population supported the federal system of government while only 30% opposed it. It seems that not only have local governments ‘owned’ their regions, but also the people there have ‘owned’ their governments.

Towards a more decentralised Bangladesh: some initial thoughts

Local governments in Bangladesh already have a firm constitutional basis. Chapter three of the constitution states: ‘Local government in every administrative unit of the republic shall be entrusted to bodies, composed of persons elected in accordance with law’ (Article 59) and ‘Parliament shall, by law, confer powers on the local government bodies to impose taxes for local purposes, to prepare their budgets and to maintain funds’ (Article 60). At present several legislations have been passed upholding the constitutional articles: the Hill District Local Government Parishad Act (1989), the Zila Parishad Act (2003), the Local Government (Municipality) Act (2009), Local Government (Union Parishad) Act (2009), Local Government (Upazila Parishad) Act (1998) and Amendment (2009) and the Local Government (City Corporation) Act 2009. Based on these acts a local government structure of 64 District Parishads (councils), 11 city corporations, 329 municipalities, 492 Upazila Parishads (councils), and 4573 Union Councils now operate. Under these legislations, de jure these governments have, to varying degrees, the power to levy taxes and service charges, oversee public health services, education and social welfare by city corporations and municipalities, public libraries, roads and the implementation of development projects.
But implementation of these laws has lagged and in practice Bangladesh is one of the least decentralised countries. Under current legislations, decentralisation focuses on a system of strong upazila governments underlain by a network of union council governments. Nominally, District Councils are upper tier government of upazilas, but both in law and current practice districts are the weakest part of the local government structure. The functional assignments of the District Councils are unclear and residual: i.e. often defined to be those not assigned to the upazilas or to the National governments. Consequently, resources flows are most limited. The weakness of district governments is evident in that elections to District Councils never took place before December 2016.
However, despite upazilas being envisioned as the principal local government hub, fiscal resources allocated to both upazila governments and union council governments are highly inadequate. Their meager resources are almost entirely used for current expenditures for paying the salaries of the staff. The salaries of the staff were also pitiably meager until about a year ago when they were significantly revised upwards. Further, elected local government leaders have no authority over civil servants who work as officials of the local government but are accountable almost wholly to their parent Ministries in the central government.
There are important political-economy reasons for the ineffectiveness of upazila governments. The upazila government’s sphere of interests overlaps almost directly with the interests of the Member of Parliament of that area. Members of Parliament are far more influential and prestigious than the chairperson of upazila governments. In such a scenario, the members of Parliament effectively take over the running of the local government by colluding with the civil servant executive officer or with the chairperson of the upazila. Beyond politics, there are also capacity constraints due to which upazila governments do not function well. They are too small in scale to provide an adequate tax base to make the government function properly.
The experience with upazila governments and lessons of East Asia that argue against decentralising to overly small governments suggest that we need to consider decentralising to a higher level, the district level. Such a higher, more consolidated level would exceed the authority of any single member of Parliament and provide greater prestige to the elected District Council and the chairperson of the district. Such a higher district government level would also have a higher revenue base from which to derive revenues. Government by 64 such district governments with significant autonomous authority over the administration and the budget would in effect work like the provinces of Indonesia, the Philippines, and Vietnam. We can also declare our districts to be provinces. Such district or provincial governments would also help to realise the vision of the founder of the country of setting up powerful district governments.

Government by 64 such district governments with significant autonomous authority over the administration and the budget would in effect work like the provinces of Indonesia, the Philippines, and Vietnam.

While the aim of this article is not to suggest a path for decentralisation but to make a case for it, a few policy milestones are reasonably clear. First, if we agree that strong district governments should be the main hub for decentralisation then current functional assignments between national, district, upazila, and union councils will need amending. At present district functions are only residual with most powers assigned to upazilas and the central government. A Decentralisation Commission can be set up to reexamine the current functional assignments and make recommendations to reallocate key functions to district governments. It is likely that current legislations will need amendments to support such a reassignment of functions.
Second, a Finance Commission will have to be set up to recommend reallocation of fiscal resources vertically among the different tiers of government and horizontally across districts and upazilas. This commission will need to devise a formula for the allocation of resources among different districts and upazilas. This Commission should also allocate tax bases among the different tiers. A particularly important task will be to strengthen property and land tax bases for local governments.
The third task of the Decentralisation/Finance Commissions will be organising stronger administrative capacity in local governments. A local government civil service cadre accountable to local governments can be set up. To encourage superior performance, the upper echelons of this cadre should also have an opportunity to join the national civil service if their performance is strong. Local governments, however, will need to have authority – jointly with the central government – over the national cadre staff assigned to them. Such control should include a say on their appointments and their tenure.
Finally, robust, transparent arrangements for monitoring, accounting, auditing, and evaluation of local government performance will need to be set up. Such arrangements will need to include monitoring and evaluation by the upper-tier central government, independent government agencies such as the CAG’s office, by civil society groups and their constituencies.
All these tasks are complex and likely to be contentious, but we have our own history as well as international experience to guide us. It will be extremely important that these two Commissions have widely respected and technically able members drawn from various parts of government and society. They will also need to have strong technical support. The Commissions should also consider piloting and experimenting with decentralisation designs in different districts of the country. One clear lesson is decentralisation must be gradually phased in with a focus on ‘provision’ or expenditure decentralisation in the beginning.

Ahmad Ahsan

Ahmad Ahsan is a Director at Policy Research Institute of Bangladesh (PRI.) He was formerly Lead Economist of the World Bank, Consultant to the United Nations, New York, Food and Agricultural Organization, the Bangladesh Institute of Development Studies and a Dhaka University faculty member. He holds a PhD from Columbia University. He has publications in the Journals of Comparative Economics, Bangladesh Development Studies, World Bank Research Working paper series, books and as the co-author of the book International Migration and Development in East Asia and the Pacific, World Bank, 2014. He led World Bank teams in policy dialogues and development policy reform and technical assistance operations, regional programs on economic integration and economic policies in Africa, East Asia and South Asia regions.