Tackling Corruption Overseas Inquiry

(U4 Senior Advisor, Aled Williams, was called as witness to the UK Parliament Oral Evidence Session, 22 March 2016 See UK Parliament video.

See also the committee's report (October 2016) which includes several key messages put forward by Williams, and a letter to the Secretary of State for International Development from the Chair of the International Development Committee on 26 April 2016, quoting evidence including statements by Aled Williams.)

Executive Summary

• Aid agencies experience problems squaring zero-tolerance to corruption with realities on the ground, where hard choices have to be made in terms of which corruption risks should be prioritised;

• DFID’s operational work on anti-corruption depends on a still-evolving field where evidence gaps, challenges in translating policy into practice, and limitations in modes of aid delivery remain significant;
  
  
• DFID could benefit from an improved approach to the identification, assessment and mitigation of various corruption risks in its programmes (discussed in greater detail under Theme One below);
  
  
• Although aid agencies have become better at acknowledging that “country context matters”, the meta-context for DFID’s anti-corruption programming is that there is still too little research on how anti-corruption interventions should best take into account political and social contexts;
  
  
• While there is some existing evidence for the effectiveness of certain types of anti-corruption interventions (e.g. public financial management reforms), we still know too little about why certain combinations of interventions are effective given particular preconditions;
  
  
• Corruption challenges are difficult enough to address in well-institutionalised, affluent countries with strong civil societies and legitimate governments. They are even harder to address in fragile settings where pursuing reforms can end up doing more harm than good;
  
  
• The principle of “do no harm” in fragile settings can be interpreted by aid agencies in ways that discourage risk taking and innovation, overwhelm recipient governments, or avoid confronting the political nature of anti-corruption interventions.

About U4

This submission draws on research and learning through the U4 Anti-Corruption Resource Centre’s activities. U4 is a center for applied research and training on anti-corruption based at the Chr. Michelsen Institute, Norway, an applied social science research institute focused on global development issues. U4 works in partnership with eight bilateral aid agencies: BMZ-GIZ (Germany), Danida (Denmark), DFAT (Australia), DFID (UK), Finland’s MFA, Norad (Norway), SDC (Switzerland), Sida (Sweden), who provide core funding and guide our work via a Steering Committee. More information about U4 is available on our website.

Theme One

Should DFID have a zero tolerance policy towards corruption in the countries where it is working or is a more nuanced approach needed to tackle corruption over the long-term?  How can DFID manage the risks associated with corruption and reconcile them with its value-for-money agenda?

1.1. Aid agencies experience problems squaring zero-tolerance to corruption with realities on the ground where hard choices must be made in terms of which corruption risks should be prioritised. U4 has examined corruption risk management in development aid in order to identify how risks can be better identified, assessed, and mitigated (Johnsøn 2015).

1.2. In development aid, corruption risks are treated differently than other risks because there is a moral dimension to corruption and great reputational risks involved. A minor fraud case in a project implemented by an aid agency may pose a significant reputational risk for the agency, but the loss may be insignificant for the average citizen in a developing country where grand corruption makes headlines routinely. In other words, perceptions of and tolerance for corruption risk can differ depending on perspectives. The aid community does not currently have a systematic way to decide on the right level of investment in risk mitigation for different types and magnitudes of corruption in different settings. Measurement and diagnostic tools have improved, however, and zero-tolerance policies are becoming more operational and pragmatic (Sequeira 2012; De Simone and Taxell 2014).

1.3. Any basic risk management process has a minimum of three stages: (i) risk identification; (ii) risk assessment; and (iii) risk mitigation. The problem with most governance and corruption diagnostic tools available for development practitioners is that they rarely move beyond risk identification to risk assessment. For example, Public Expenditure and Financial Accountability (PEFA) reports and Fiduciary Risk Assessments (FRAs) identify weaknesses in public financial management systems. This is risk identification, not risk assessment. The following is an explanation of how U4 believes corruption risks could be better identified, assessed and mitigated across the project cycle.

Step 1: Identify corruption risks and determine risk tolerance

1.4. Aid agencies care about corruption risks for many reasons. There are risks to development overall, direct fiduciary risks, and more indirect reputational risks. Corruption poses a development risk when resources are diverted from public services, leading to lost public revenue, poor-quality infrastructure, wasted resources spent on underperforming public institutions or employees, an unattractive investment environment, erosion of public trust in government, and fuelling of conflict. Corruption also poses a direct fiduciary risk to aid agencies. These agencies have a responsibility to ensure that development funds are spent efficiently and effectively, without misuse. Finally, corruption can pose a reputational risk for aid agencies. Misuse and waste of development funds damages the reputation of aid providers, and perhaps more importantly, may delegitimise and undermine public support for the entire enterprise of development assistance, both domestically and in partner countries. There is no clear demarcation between development risks, fiduciary risks, and reputational risks. An analysis focused purely on fiduciary risks is only partial. Reputational risks are often intangible but will nevertheless be a key concern for decision makers. Still, the risks that different types of corruption pose for development outcomes must be the central analytical point of departure. Aid agencies tend to focus on financial fraud and bribery, often using these concepts as synonyms for corruption. Fraud and bribery cases, however, may not always present the greatest risk to aid operations. All relevant types of corruption should be identified and their potential risk assessed, even if some types are easier to measure than others.

Step 2: Assess the level of risk

1.5. Effective risk management requires good analysis of the likelihood that a risk will occur, and of how damaging this occurrence will be. We want to move from a situation where we are aware of risks but uncertain of how they will affect us to a situation where we estimate with some confidence the effects of risks on programming and outcomes. It is useful to consider a risk as an event whose exact likelihood and outcome are uncertain but nonetheless can be estimated. The magnitude of a risk ultimately depends on (a) the probability of the event occurring, and (b) the expected impact of the event. In the financial sector, risk models can be fed huge quantities of financial and economic data, but there is a scarcity of data in most areas of development aid. A move from uncertainty to a calculated risk approach will depend on the feasibility of quantifying a certain type of corruption risk (e.g. bribery is probably easier to quantify than nepotism) and on the resources available. If bribery is thought to be an important risk that threatens achievement of a programme’s overall development objective, then one can make strong estimates of the probability and impact of this risk.

Step 3: Compare actual risk with tolerable risk, decide whether mitigation is required

1.6. It is not until the third step that we argue one should decide whether or not a specific type of corruption should be actively mitigated. To be clear, this is not a choice about whether corruption in general should be tackled or not, but about which types of corruption in which sectors should be targeted. Trade-offs are necessary in all development assistance. For each type of corruption risk – informal payments, procurement fraud, embezzlement by programme staff, and fraud by partner NGOs, for example – predefined “tolerable” risk levels should be compared to the actual or estimated levels. It may be that informal payments, for instance, are expected to occur, but very sporadically and involving only small sums, considered a tolerable risk.

Step 4: Choose the most cost-effective corruption risk mitigation tools

1.7. Once a decision has been made to engage in mitigation, the most cost-effective tools must be found. The basic principle for decision-making is to consider the costs of a specific type of corruption on the one hand, and the effectiveness of tools to target that type of corruption on the other hand. It is unrealistic to perform a full cost-effectiveness analysis (CEA) or cost-benefit analysis (CBA) every time, but the principles behind these analyses are useful to keep in mind when making the decision. A menu of different corruption risk mitigation tools is available. The basic point is that none of these tools should be chosen by default. It is not until one has identified the corruption risks, analysed them, and decided whether they should be actively targeted that one looks at which tools may be best for the job. Currently most aid agencies rely by default on regularity/compliance audits and/or due diligence, only rarely selecting other tools to either complement or replace them

Theme Two

In 2013 DFID published anti-corruption strategies for all the countries it works in; how effective have these been? Do they adequately take into account the political and social context of the countries in question?  Do certain sectors require particular attention? How should success be measured?

2. Without having conducted a systematic review of DFID’s anti-corruption strategies for all countries, we cannot directly comment on their effectiveness. However, having worked in partnership with DFID and observed its anti-corruption policies and practices in multiple locations, we share below some general reflections. We recognize that DFID’s operational work on anti-corruption is dependent on a still-evolving field where evidence gaps, challenges in translating policy into practice, and limitations in modes of aid delivery remain significant.

2.1. Some sectors and institutions pose greater corruption risks than others. If an aid agency wants to engage in, for example, a highly corruption-prone infrastructure sector, then it would be cost effective to invest substantially in corruption risk management. Often, effective risk mitigation will require a redesign of a policy or programme to incorporate performance measures, procurement designs, expenditure controls, reporting options, and so on. As aid agencies frequently work through implementing agents, risk mitigation will entail helping these organisations incorporate anti-corruption measures. They may need to simplify their systems and procedures, publish information, standardise work processes, introduce electronic application or financial management systems, and so forth. Spot checks, field visits, and third-party monitoring (evaluations, audits, community monitoring) are also essential risk mitigation tools. Testing and comparing the effectiveness of different tools, and keeping the focus on results, are sound principles in nearly all cases. Effective risk management will not entail one integrated solution but rather a myriad of smaller solutions that address specific risks effectively. Monitoring such a patchwork of solutions requires financial investments, and a business case should show how these costs will be lower than the anticipated gains. Success should be measured in terms of the overall developmental impact of an intervention.

2.2. Although aid agencies have become better at acknowledging that “country context matters” for anti-corruption interventions, the meta-context for DFID’s anti-corruption programming is that there is still too little research on how anti-corruption interventions should best take into account political and social contexts. The current anti-corruption evidence base rarely accounts for outside influences on individual programme performance, even if we know that institutional, geographic and temporal context matters. Individual evaluations and academic case studies often do not have the scope or sophistication to go beyond an assessment of one intervention. Large research programmes, on the other hand, have tended to focus on cross-country comparisons (using economic regressions) or proving the independent impact of a single intervention (using randomised control trials). There are very few comparative case studies undertaken with the specific aim of identifying the effects of various combinations of anti-corruption interventions under the imperfect conditions present in most developing countries (Søreide and Williams 2014). It is positive therefore that DFID is investing in the generation of new research evidence on what works in anti-corruption.

Theme Three

What should the balance be between seeking to tackle corruption top down at institutional level and bottom up at the grass roots? What works and what is not working as well and why?

3. Corruption involves a variety of practices, behaviours and mechanisms. There is therefore no one-size-fits-all solution, and specific anti-corruption measures appear most effective when other contextual factors support them and when they are integrated in a broader package of reforms. Despite the accumulation of extensive experience via policy and practice in aid agencies and other institutions and the generation of large quantities of research publications, the practical outcomes of anti-corruption interventions have been disappointing and the overall evidence base on the effectiveness of anti-corruption interventions remains weak (Rocha Menocal et al 2015, Johnsøn et al 2012). Until quite recently there was no overview and assessment of the strength of evidence across the whole spectrum of anti-corruption reforms. Reviews of donor-supported anti-corruption efforts had highlighted the need for stronger evidence (Hanna et al 2011, Norad 2011), while, operationally, anti-corruption practitioners were requesting evidence for what works, when, where, and why. U4 undertook in 2012 a systematic review of the evidence for the effectiveness of anti-corruption interventions pursued by aid agencies (Johnsøn et al 2012). We subsequently worked with colleagues at the Overseas Development Institute (ODI) to produce a DFID Evidence Paper on corruption published in 2015 (Rocha Menocal et al 2015). These two publications provide a summary of evidence on the effectiveness of various anti-corruption interventions. Rather than attempt to provide a detailed summary of these studies here, we highlight the main directions these studies point us towards in furthering work on effective anti-corruption approaches:

• While many corrupt practices have been unpacked and studied in depth (e.g. patronage politics, rent seeking, petty bribery), we lack a broader understanding of how they interact within given contextual settings;
• There is some existing evidence for the effectiveness of certain types of anti-corruption interventions (e.g. public financial management reforms). We still know too little, however, about why certain combinations of interventions are effective given particular preconditions;
• While we know interdependencies among anti-corruption interventions are important for anti-corruption outcomes, we need to know more about how to draw out the maximum anti-corruption value from these interdependencies.

Theme Four

Corruption and poor governance can be a key cause of instability in fragile states. Is the UK Government appropriately prioritising and managing anti-corruption strategies in these settings? What are the challenges in practice and what could it be doing better?

4. Corruption challenges are difficult enough to address in well-institutionalised, affluent countries with strong civil societies and legitimate governments. Many countries in greatest need of reform are, however, fragile. They are deeply impoverished, scarred by natural disaster, or socially divided; they may have only recently emerged from war, dictatorship, or internal conflict. Adding the stress and uncertainty of significant reforms to those sorts of problems may end up doing considerably more harm than good.

4.1. In recent years, such concerns have begun to coalesce around the principle “First, do no harm.”  Aid agencies, increasingly seeing their interventions in fragile states as state building, find that the “do no harm” concept provides a useful analytical lens (OECD 2010). The concept may be defined as avoiding premature or poorly-thought out reforms that can do more harm than good (Johnston 2010). There is a general discrepancy (Johnsøn 2014) between stated policy and implementation practice of aid agencies in anti-corruption programming, and this extends to the “do no harm” concept. Aid agencies often integrate the concept into their official policies but have problems operationalising it. Alternatively, they may choose to apply a narrow understanding of “do no harm” in order to defend a retrenchment of the anti-corruption agenda to one that focuses mainly on their own resources and internal integrity systems, instead of working with civil society and government.

4.2.In 2014, U4 conducted interviews with aid and development professionals working in fragile situations (Johnston and Johnsøn 2014). Our aim was to assess what aid officials and practitioners understand “do no harm” to mean, how (if at all) they follow the principle, and how they judge whether they are at, or near, a point at which they must reconsider their activities in order to avoid negative consequences.

4.3. All but two of our 23 respondents agreed that corruption control measures can cause damaging stress in fragile societies. In general, they demonstrated good understanding of both the stabilising and destabilising effects of anti-corruption interventions in the settings where they work. Practitioners reported that their agencies try to gain a better understanding of the political economy of countries where they work in order to fine-tune existing programmes. However, less than a third of respondents believed that such minimal responses were likely to succeed in reducing stresses to acceptable levels in aid programmes. Knowing the context is not enough: aid agencies also must respond to it by changing strategy and implementation to more effectively make national institutions resilient to stresses.

4.4. Concerns that the “do no harm” principle might encourage self-protective responses within aid agencies are not misplaced: respondents unmistakably ranked the protection of their own agencies’ funds as the highest priority overall. More than twice as many respondents ranked agency self-protection as a top priority compared to those who gave top priority to society’s capacity to use aid constructively or to preventing the capture of anti-corruption initiatives. Those concerns – which lay at the core of the original “do no harm” argument – were low priorities. When respondents were asked whether aid agencies and host governments should go ahead with anti-corruption controls even when significant stresses seemed likely, five said “yes” outright; no one said “no.” The remaining 18 said that they did not know. Most of those who said “yes” defended their responses by arguing that the long-term consequences of not fighting corruption outweighed the immediate damage that might be done. This view risks conflating “do no harm” with doing nothing. “Do no harm” means choosing the least harmful among a range of possible interventions; this sometimes means doing nothing, but not always.

There is a potential that aid agencies may interpret “do no harm” in ways that discourage risk taking and innovation, overwhelm recipient governments, or avoid confronting the political nature of anti-corruption interventions.

References

De Simone, F. and N. Taxell. 2014. Donors and “zero tolerance for corruption”: From principle to practice. U4 Brief. Chr. Michelsen Institute. Bergen.

Hanna, R., S. Bishop, S. Nadel, G. Scheffler, and K. Durlacher. 2011. The effectiveness of anti-corruption policy: What has worked, what hasn’t and what we don’t know. EPPI-Centre. Social Science Research Unit. Institute of Education. University of London. London.

Johnston, M. 2010. First, Do No Harm – Then, Build Trust: Anti-Corruption Strategies in Fragile Situations. Background paper for World Development Report 2011. World Bank. Washington, D.C.

Johnston, M. and J. Johnsøn. 2014. Doing the wrong things for the right reasons? “Do no harm” as a principle of reform. U4 Brief. Chr. Michelsen Institute. Bergen.

Johnsøn, J. 2014. Cost-Effectiveness and Cost-Benefit Analysis of Governance and Anti-Corruption Activities. U4 Issue. Chr. Michelsen Institute. Bergen.

Johnsøn, J. 2015. The basics of corruption risk management: A framework for decision-making and integration into the project cycles. U4 Issue. Chr. Michelsen Institute. Bergen.

Johnsøn, J., D. Zaum and N. Taxell. 2012. Mapping evidence gaps in anti-corruption: Assessing the state of the operationally relevant evidence on donors’ actions and approaches to reducing corruption. U4 Issue. Chr. Michelsen Institute. Bergen.

Norad. 2011. Joint evaluation of support to anti-corruption efforts 2002-2009. Norad Evaluation Department. Oslo.

OECD. 2010. Do No Harm: International Support for Statebuilding. OECD. Paris.

Rocha Menocal, A., N. Taxell, J. Johnsøn, M. Schmaljohann, A. G. Montero, F. De Simone, K. Dupuy, and J. Tobias. 2015. Why Corruption Matters: Understanding Causes, Effects and How to Address Them: Evidence Paper on Corruption. DFID. London.DFID. London.

Sequeira, S. 2012. “Advances in Measuring Corruption in the Field.” In New Advances in Experimental Research on Corruption, edited by D. Serra and L. Wantchekon, 145–76. Emerald. Bingley, UK.

Søreide, T. and A. Williams. 2014. Corruption, grabbing and development: Real world challenges. Edward Elgar Publishing. Cheltenham and Northampton (MA).