Should the World Bank become more adaptive by weakening its safeguards?

September 26, 2017

     By Duncan Green     

The World Bank wants to become more agile, to speed up its grant/loan-making, be less bureaucratic, leap on the ‘adaptive management’ bandwagon etc. In its rush to change direction, it hasn’t had too many discussions with NGOs, so I thought I’d raise some of the issues on the blog. Perhaps the lack of discussion is because the Bank sees NGOs as a potential roadblock to reform. By insisting (and getting agreement) on a range of ‘safeguards’ on issues such as consultation or impact assessment, we are seen to have slowed down Bank procedures (and it’s probably true, to some extent, though we’d argue for good reason).

I spent an hour on the phone with some Oxfam colleagues last week kicking these issues around – I do like conversations about trade-offs, tensions and shades of grey. At the end of it we agreed I’d consult the FP2P hivemind for advice.

So who could possibly be against agility? Well, me, in some cases. What if speed comes at the expense of properly consulting local communities, or doing a rigorous impact assessment before you build the road or the hydroelectric dam on their land? One person’s bureaucracy can be another’s livelihood protection.

First of all, what’s the Bank’s motive for wanting to be more ‘agile’? Is it to speed up disbursements, recognize that flexibility and adaptive management is the key to working in complex systems, or to reduce the heavy transaction costs on staff and partners of all its procedures (see DFID’s transition to Smart Rules)? It’s probably all three, but the balance varies depending on who’s talking. It’s worth bearing in mind because different motives have different consequences for how ‘agility’ is interpreted.

From the point of view of systems thinking, this is about adaptive management: you can’t completely understand the system in advance, you will learn more about it as you go, and in any case it keeps changing. So what matters is building in feedback loops and mechanisms to respond/adapt your work as you learn by doing/failing – compliance approaches and tickbox thinking are so last year. In adaptive management you’re only as good as your feedback, but what makes for a good feedback loop?

Oxfam itself is keen to become more agile, and where we trust the Bank, we ought to support it to do likewise. But the upfront safeguards are there for a reason, to understand if a project is actually viable, to ensure appropriate resources are available for managing risk etc, but also partly because of lack of trust. If you don’t trust an institution or its officials, you pile on the regulations and procedures to try and stop them doing bad stuff.

Trust also determines how we see the role of the recipient government. If CSOs trust the government, they will support devolution of power and decision-making – having a responsive, accountable government in charge is infinitely preferable to a Washington bean counter, however well intentioned. But if the government has a track record of brutalising the poor, then we want to keep strong oversight in the Bank’s hands, where civil society can at least exert some influence over what happens.

And some of the things NGOs care about do not lend themselves to fast-and-funky startup-style agility. Big projects in infrastructure or mining need to consult communities, find ways for them to participate in a genuine way in making sure the project helps (and doesn’t harm) local people. That takes time – shoving a load of data on a website may look like participation, but is more likely to be lip service.

That links to a second point. Adaptive Management sounds good – but who decides what needs to be adjusted and adapted? If it’s just some technocrats in Washington, what is to say the adapted project will help local communities more than the original?

The optimal balance of agility and procedural rigidity also depends on the topic. If the Bank’s dollars are going to help build a dam, a road or a gigantic mine, then if things go wrong, they may well be irreversible. That argues for sticking to the current system of high levels of upfront analysis and safeguards. If the project is less immediately disruptive, and can be tweaked before it does lasting damage, then adaptive management based on the right kinds of feedback loops could be a better option.

It also depends on the government receiving the loan. If it has a track record of accountability and responsiveness to community needs, then adaptive management and devolution of decision making could make a lot more sense than if it routinely crushes feedback (and people) it doesn’t like.

Given my fondness for 2x2s (though to my dismay, not all my colleagues’), you may already have guessed where this is going. I’m trying to construct a 2×2 which helps frame the situations in which you would want to stick with the current system of upfront safeguards, and others in which it would make more sense to jump into a more adaptive, iterative approach. How about this?

Reversibility of change against level of trust in bank/government

Unpacking the quadrants:

High irreversibility/low trust (due to track record of Bank or Government): Stick with current approach

Examples: Dams, roads, extractives, PPPs that saddle governments with decades-long contracts

High Irreversibility/high trust: Still needs proper impact assessment, but maybe spread consultation and participation out more into the implementation phase

Examples: ?

Easily Reversed/low trust: The safeguards need to be on process – how can we be sure the consultation is genuine, and the project/government/Bank will respond to the results?

Examples: results based financing

Easily Reversed/High Trust: get stuck in, pass power and decision making to local actors, keep an eye on feedback loops to make sure they are working

Examples: Citizen engagement; teacher training; financial inclusion; women’s economic empowerment

Thoughts? Anyone got better suggestions for the axes?