The financial collapse in 2008 following the collapse of Lehman Brothers was enthusiastically prepared by the political and economical decision makers. In the 70s and 80s, in the name of more efficiency and free marked, regulations were more and more seen as a restraint on the development of companies. With less regulation, the market would be more efficient with less transaction costs. The firewalls between savings and investment were torn down, as the memories of the thirties were deemed irrelevant and “this time it is different”. Indeed economic growth followed. Financial markets seemed more efficient. The business cycle seemed to have disappeared.

Companies became more and more interlinked and financial products became more sophisticated. Risk was shared among more actors, all of them with a AAA rating. The risk vaporised in the system. Until it suddenly was there again. The dew point was reached. And the lack of firewalls took down the system, including some governments who believed the hype, like in Iceland and Ireland. As I am not an economist, I would like to refer to Tim for a better explanation of the crisis. His explanation in his book “Adapt” is enlightening and suits my point well.

So the financial system became so integrated that risk became systemic. All actors were linked up so much that the failure of one hurt all of them. The financial innovation went so fast and the system became so complex that nobody could assess the overall risk anymore.

Development is a risky complex system.
Development is a risky business. Success is elusive and failure frequent. Moreover the “transaction costs”, the difference between what the donor wants to give, and what the ultimate recipient gets, is high. There are important inefficiencies: unaccountable partners, overlaps, gaps, lack of results, lack of knowledge what works and what not, not forgetting stubbornness in repeating things that failed over and over again like swedow.

The way to make changes in a complex system is best described by “do no harm”: a prudent and evolutionary approach. Make small changes, and with a short feedback-loop check on the effects on the system as a whole. Make another change. Innovations should be never too big too fail. Innovation should be tested before bringing it to scale. Indeed, this is our world, our ecosystem. We should not take systemic risks with the lives of poor. As it is impossible to predict what will work or not, it is better to have a lot of initiatives and not to pre-empt the outcome.

A development system based on these principles should be expected to be very conscious of the risks that go with large-scale intervention, and focus on the value chain from innovation to bringing to scale.

The development system that exists however has the Paris agenda and the humanitarian cluster approach. The 3D approach (development, diplomacy and development), linking relief to development and integrated missions. The items on the international agenda are aiming to link the different systems to each other. Does this lead to more efficiency or to unacceptable systemic risk?

Some examples where this clustering of agendas seems to have led to collapse due to systemic risk: :

• In Afghanistan the West has tried the 3d approach, it did not seem succesful.
• In Uganda, donors have linked themselves into a budget aid logic, meaning that to punish the parliament for a gay law, children will probably not get their vaccine anymore.
• Madagascar textile lost their jobs, because the duty free regime for their country was withdrawn after the politicians squabbled too much.
• In DRC and Sudan, the UN integrated missions make the UN-humanitarian agencies de facto not neutral, affecting their efficiency in a serious way.
• In Somalia the mixing up of the humanitarian and anti-terrorism agenda was an element in the current crisis.

From these examples I would like to conclude that systemic risk for the whole aid effort in a country can exist if agendas are lumped together. If an approach does not work, the most logical explanation could be that the approach is not a good one, and different options must be explored. The alternative narrative, that it did not work because we did not try rigorously enough, seems dangerous.

An alternative: nimble aid (agile aid, mindful aid)
Nimble aid would consist of independent interventions, each very limited in its objectives and conscious about the unpredictability of externalities. Like a bird in a flock, each programme would be able to steer itself in full consciousness of the effects it has on its environment. It is the evolutionary approach to development. If every objective is diluted in a wider technocratic programme, nothing is really happening. Trying to be responsive in a wider programme just leads to more meetings and more lemming thinking on development.

Vaccination programmes in humanitarian settings do just that: saving the life of thousands of children, leading to healthier and more productive lives for the beneficiaries. An effect that can be traced over up to 100 years.
New agricultural practices are tried one by one on a small scale, until there is one that works well, and everybody adopts it. Like the use of maize in Africa, long before colonisation, and varieties resistant to plagues now.

It is a development agenda which is less ambitious in the short term, but revolutionary in the long term.

A prerequisite for this approach to work is information sharing. Not information about what happened, but information on what is happening. So all actors know what others do by reading this information and can adapt their interventions to what is already happening around them. The actors can only correct their actions if they know what is happening around them. This is very much like obliging public companies to publish essential information , development actors should to post essential information on their activities. Information sharing should not be confused with going to meetings nor coordination.

Keeping the firewalls
In the consensus thinking, a compounded indicator will tell you about success of failure (like the human development index). This means that each value, each objective is not important enough to fight for on its own. You want to raise the overall index. In a nimble approach there is a tension between overall goals and each objective. There is a specific programme assuring that child mortality is down. Even if the government does not have the institutions or intentions to do it themselves yet. Good child mortality work will always strengthen the local institutions when the objectives are both long-term and short-term.

This is why I would like to argue to keep the firewalls between the values that are important. This does not mean to work in an ivory tower, It means that there are some objectives you just don’t negotiate away.

Moreover: fund what works
There is a lot of evidence on interventions that do work, even on a large scale. Getting rid of funds is not a problem for the agencies even if they would limit themselves only to evidence based interventions. Just a few examples that work in some circumstances if done well:

• Basic economical infrastructure
• Basic health services
• Cash support for the poorest
• Water and sanitation
• Most humanitarian interventions if done according to the Sphere Standards.

And so much more. In the cases of decent accountable local institutions, even some forms of budget aid seem to work.

In conclusion

Of course, coördination and coherence are important. But they are only means for reaching results. Sometimes other means, like innovation or competition, might work better. When coherence becomes an objective of the aid, however, the level of systemic risk for the development system might just be too high.

Finally, transparency will always be beneficial for planning interventions, as well in a coherent as in a competitive environment.