In his 2013 volume explaining why donor-supported reforms often go awry in developing states, Kennedy School Professor Matt Andrews lays the blame on the failure to appreciate how political imperatives, patronage networks, cultural practices, and other elements of local context affect the way reforms are implemented. While Andrews offers telling examples of how ignorance of context doomed reforms in Argentina and Malawi, the failure to stamp out corruption in Uganda’s revenue collection service provides an even more vivid illustration of the way the very different context in a developing state can cause “best practice” reforms to fail. The analysis is taken from Odd-Helge Fjeldstad’s classic account of the attempt to reform tax collection in Uganda, “Corruption in Tax Administration: Lessons from Institutional Reforms in Uganda,” chapter 17 of Susan Rose-Ackerman’s 2006 edited volume, International Handbook on the Economics of Corruption.
In 1991 revenue from taxes and customs duties in Uganda were seven percent of GDP, an astonishing low figure even on a continent where tax evasion was the norm. Under pressure from the IMF, the World Bank, and other donors the then recently installed government of Yoweri Museveni took decisive action. Following what was then considered best practice for boosting revenues and cutting corruption in a revenue service, the government made the revenue department of the Ministry of Finance into an autonomous agency. Independent agency status allowed the Uganda Revenue Authority to implement a number of reforms to reduce corruption. Salaries were raised above civil service levels and strictures on firing non-performing workers removed. As a new agency, all employees were considered new hires and had to prove themselves during a probationary period; as a result almost 250, or 15 percent, of the old revenue department staff were weeded out. In addition, “clean” expatriates were hired into senior management positions, and measures were taken to improve morale: offices were upgraded, working conditions improved, and training provided. All in all, the Uganda Revenue Authority was considered a model for how to create an efficient, non-corruption revenue collection agency.
During the first years of its existence, the authority’s performance suggested these reforms were succeeding. Revenue collection as a percentage of GDP improved and perceptions of corruption declined. These early indicators of success, however, soon began to decline. Forty-three percent of businesses surveyed in 1998 reported paying a bribe to a Uganda Revenue Authority employee; in March 2000 President Museveni termed the authority a “den of thieves,” and in 2003 its former head listed corruption as “problem number one” in the organization. A Commission of Inquiry of C corruption in the Uganda Revenue Authority was appointed in 2002, and although its report was never released, leaks suggest the commission found massive corruption in the ranks.
Fjeldstad’s careful review of the reasons for the authority’s failure to lick corruption identifies several reasons including creeping political interference, job insecurity producing a “take it now” attitude, and the failure to rid the organization of all the “bad apples” it inherited from the predecessor revenue department. As Fjeldstad explains, one “best practice” reform imported from OECD countries actually increased the chances revenue staff would become corrupt. Employees were given raises to reduce their incentive to take bribes, and rather than furnishing them with housing, meals, vehicles and other non-cash benefits, the Uganda Revenue Authority began paying all compensation in cash, thus allowing employees to choose what to spend their salaries on.
How could higher pay and monetized benefits encourage corrupt behavior? The reason lies in the traditional pattern of social relations that still obtained in Uganda. In a country with little or no social safety net, those in need rely on tribal and kin networks for support; the better off within these networks are expected to care for those less well-off. Full monetization of revenue officers’ pay, along with the salary hikes, increased their obligations to help those in their social networks. Indeed, the raises increased employees’ obligations to others, “forcing” them to take bribes to fulfill them. As Fjeldstad writes: “What looks like corruption from the outside is undertaken by some tax officers in a context where the reciprocal obligations of kinship and community loyalty require such behavior in order to be regarded as a ‘good person.’”
Uganda’s experience with its revenue authority drives home Andrews’ point: a thorough analysis of context is a critical prerequisite to a reform project. Had reformers had a deeper understanding of how social networks in Uganda operated they might have done several things differently. They might, for example, have followed the advice of the Belgian development economist Jean-Philippe Platteau and raised salaries by increasing in-kind benefits, which are often less visible than straight salary hikes and harder to share with one’s kin. But however reform advocates might have addressed the social network issue, it would surely have been an improvement over the ignorance shown by disregarding it altogether.
GAB | The Global Anticorruption Blog 
Very interesting analysis, and a useful cautionary tale. A couple of questions, though:
First, do we know what happened to tax collection efficiency after the reform. You note in your second paragraph that prior to the reform it was 7% of GDP. Did it increase, decrease, or stay about the same after the reform? What is it today?
Second, and more generally, while it seems clear that corruption is currently very bad in the Ugandan revenue authority, do we have a clear sense of whether it is better or worse than it was before the reform?
Third, while I find the argument about cash v. in-kind benefits intriguing, I’m not sure I quite follow the logic. Let’s say I’m a Ugandan tax collector. Members of my kinship network are constantly asking me for (material) favors, and taxpayers are constantly offering me bribes. If the government pays me a much higher salary, I can do my kin group more favors (buying stuff for them from my official salary), so at the margin I may not need to raise as much money through bribe-taking to meet those obligations. If the government gives me a bunch of non-transferrable in-kind benefits, then my personal material well-being may be higher, but I still face the same kin-group demands, and now to meet those demands I need to take more bribes (so I have the ready cash to spend on whatever it is I’m supposed to provide). That logic would suggest that paying tax collectors in-kind rather than in-cash would make bribe-taking worse. What am I missing?
First, do we know what happened to tax collection efficiency after the reform. You note in your second paragraph that prior to the reform it was 7% of GDP. Did it increase, decrease, or stay about the same after the reform? What is it today?
—
Tax rose to 13% of GDP after recent revisions to GDP figures. Was 15.7% before the revisions.
Still below the sub-Saharan average and should grow if not for the many issues.
—
Second, and more generally, while it seems clear that corruption is currently very bad in the Ugandan revenue authority, do we have a clear sense of whether it is better or worse than it was before the reform?
—
It became better, then worse. A commission of inquiry was instituted which led to deeper restructuring even though not communicated as the reason. Corruption situation improved and has gone back to bad again
—–
Third, while I find the argument about cash v. in-kind benefits intriguing, I’m not sure I quite follow the logic. Let’s say I’m a Ugandan tax collector. Members of my kinship network are constantly asking me for (material) favor ….
—
This argument is without merit, whichever way you look at it. It is like saying there can be no corruption in barter trade!
There are staff that have worked 20+ years in URA, earning the same salaries as others, yet they have remained incorruptible. They also have relatives.
—-
The path to leadership in URA has a role to play. Staff who do not see a clear career path tend to find ways to reward themselves when the opportunity presents.
URA autonomy is on paper only. During the period covered by the article, there were instances of “saved” people picked from churches and sent to take up jobs in URA. They turned up to be more corrupt than those recruited using regular processes. The act of interference in URA whether at recruitment, procurement, deployment, etc.. is enough to endanger any positive gains in the fight against corruption. Remember, all staff in URA with the exception of drivers, cleaners and office attendants are graduates.
The environment in which URA operates is Very Very corrupt. It is people in this environment who are transacting with URA. Probably, the only way to create a clean URA is to create a URA village, so that staff who get in never have to get out, and their affairs can easily be observed.
Here is a fact:
All the corruption deals that take place are made before a payment is made into the banking system. Once that happens, the risk to such money is almost ZERO. All funds that have ever been paid into a URA bank account by the taxpayer have never been lost. The cases I remember involved taxpayers entrusting URA staff to complete the transaction for them. The staff then manipulated the documents in order to account for the full amount to the taxpayer and a reduced amount for Government. In each such case the processes of URA caught up with the staff. But where the taxpayer initiates a deal and agrees to pay a reduced amount to the bank – the risk is greater and more difficult but not impossible to trace. Several staff have lost their jobs on this account.
Reality
URA operates in a very corrupt environment and the temptations for staff are immense. Measures are put in place which enhance integrity, but they are soon lost because of the pressures from the environment.
URA salaries were meant to be high enough to outweigh corruption tendencies but they have been eroded over the years, and staff barely manage.
URA autonomy is not complete – a few Govt officials interfere with this autonomy during recruitment, procurement, deployment, and other operations. Each such action undermines the integrity of URA and foments corruption.
De-monetizing benefits cannot solve anything! In fact doing so would push the cost of administration through the roof.