MfDR – what’s the problem with impact oriented program steering?

May 25, 2011 | Adrian Gnägi | Learning Elsewhere |


Rating: none

 Adrian picture for sdclan



by Adrian Gnägi


A few years ago, when I was posted in Amman, we frequently visited my wife’s family in Beirut. That made for long rides on monotonous Middle Eastern desert highways. The deal with the kids was that they could wish for stories to be told. One of their favorites was the illustrated book “Beaver, give us a ride”. The story goes like this: Beaver uses a hollow log as boat on the river. His friends are not impressed: nice, but small. So beaver builds a large raft and invites one after the other of his friends to come on board. When bear joins after all the others, there definitely is no space left. The friends try to prevent an uninvited butterfly from landing on the overloaded raft in the rapids, but …

The story is a beautiful explanation of complexity theory for children. Our kids used to discuss for hours whether and how the wreckage could have been prevented. They asked to review the pages where the different friends joined the party, discussing who could have done what differently at which moment to prevent the accident. The better they knew the story and the pictures, the more weak signals they discovered. In the end they realized there would have been steering potential in every single scene, right from the beginning.

But why was this steering potential not realized, why was the catastrophe not prevented from happening? Standard MfDR (managing for development results) thinking explains impact as the end of the result chain: impact happens in a distant future, when all outputs have been produced, when outputs have interacted with other factors into outcomes, and when other forces have diluted outcome influence in the attribution gap. 

Beaver’s story shows why the MfDR impact model is not useful for development program steering:

  • When impact is conceptualized as happening in the distant future, all impact induced steering opportunities are forgone, because they lie in the past. At the moment when the friends are swimming to the shore, they cannot prevent the wreckage any more.
  • The attribution gap prevents from knowing what kind of steering should have happened – the cause-effect chain is broken. When the friends discuss who was to be blamed, they agree none of them had caused the result – it was the butterfly’s fault!


My favorite part of the story comes after the guilt attribution scene. After having attributed responsibility to the butterfly, the friends burst out in laughter and declare: “That was great fun, let’s do it again!” It is at this moment that they discover beaver has left them and returned to his little log boat. There obviously had been two simultaneous projects:

  1. a beaver project with safe trip as intended result, and 
  2. a friends project with partying as intended result.

The two projects may or may not have co-existed for long. It is possible that the wreckage was just a funny accident for the friends, an emergent result they could easily live with. They are happy with the result – beaver is not happy.


Beaver did everything right: he built the raft, he floated it on the river, he invited his friends on board, he steered the raft through the rapids. He did everything right, but he did not produce his intended result: not only because steering opportunities were forgone when impact happened, not only because steering potential could not be assessed due to the attribution gap, but very fundamentally because he carried on with his project when his friends started interfering in the result chain.


In MfDR terms, there was a problem with assumptions and risks: under the assumption that there were no political economy interferences (no separate friends’ agenda) and that external risks could be managed (no butterfly effects), beaver’s result chain might have worked out. But the story is well constructed to support its morale. Emergence is the name of the game. Beaver sits on his raft in the current, his friends jump on board and start partying, the raft is torn into the rapids, the butterfly approaches – there is no time and there are no venues for political economy stakeholder negotiations, and there definitely is no hidden magical hand streamlining external influences. Beaver has a plan based on assumptions. He pushes his raft into the river and from there on he goes with the flow.


This is the critique we currently hear from our superiors: you are too naïve, you go with the flow. We are asked to do more serious political economy and risk analyses (see new templates for SDC cooperation strategies, entry proposals and credit proposals). But would better risk awareness have helped beaver? From the moment he is on the river with his raft, this is mission impossible. When the assumptions turn out to be naïve, he cannot change – not his intended result, nor his path of action. He can steer his raft, but he cannot steer events. He goes down with the flow.


Risk awareness might have helped him in one way: he might have abandoned his project and stayed with the log boat right from the beginning.  Complex tasks need simple approaches is the textbook recipe for handling such situations. “Stay away from complex programs – simple approaches for complex tasks” currently is no good advice to SDC staff. SDC staff cannot abandon the raft and return to their little log boat. With the new rules for cooperation strategies, complex approaches are declared to be SDC’s standard: all SDC mid-term programs are asked to produce results on the domain (sector) level, as contributions to partners’ results. I assume this is mission impossible in many cases. If the approach chosen to achieve those results is informed by MfDR result chain logic, I am sure this is mission impossible in most cases. If, for a four-year cooperation strategy period, the statistical frequency of government/policy changes, major economic shocks, disasters, and conflicts (just to mention four major risks) are added up, it is clear that program planning assumptions over four years are high risk business. SDC decided for pushing its rafts into the current. Given the attribution gap and prior-to-impact problematic related steering difficulties associated with result chains, my best bet is that most SDC rafts will go with the flow.

Beaver’s inability to effectively interfere with the unfolding chain of events definitely was the most vexing element of the story for our kids. They kept asking questions like:

  • why aren’t they talking about what is happening?
  • why are they inviting bear on board even though the raft is overloaded already?
  • why is nobody preventing the freinds from rocking the raft?
  • and why, why, why for God’s sake, is beaver sticking to steering the raft, instead of…. 

… of course they loved to propose advice: the friends should have met and agreed before venturing onto the river; when the raft got stuck after bear got on board, they should have paused and reflected instead of just pushing into the river again; beaver should have told his friends that he was scared when they started partying, instead of keeping silent; etc.  My kids were just kids. Not development professionals. They have no idea what plans and result chains did to uor brains.

In complexity theory terms, my kids were saying: the actors should have agreed on a common vision and on a plausible theory of change; they should have defined progress markers to know whether they were moving towards their vision; they should have paused in-between, reflected on what was happening, and adapted their change strategty accordingly; and they should have stoped their project when it became clear the situation was turning from complex into chaotic, before change became erratic. I have argued in an earlier post that a “theory of change” based approach is more realistic than a result chain approach for handling complex programs. HIVOS/UNDP just published a beautiful handbook explaining how this can be done: Theory of change – a thinking and action approach to navigate in the complexity of social change processes.

If you live in Switzerland and are interested in the results controversy: the debate in SDC starts on June 30th, 3pm, with a forum discussion on “effectiveness vs measurability”. Guests are most welcome to join the debate!


Comments to“MfDR – what’s the problem with impact oriented program steering?”

  1. Anne Bichsel says:

    I couldn’t agree more! Our contexts are not simple and linear. They are emergent, dynamic, adaptive and uncertain. While Theories of Change and Impact Logics help us to structure and sharpen our thinking, they don’t map reality one on one. I think we need to move beyond the narrow mindset of filling in boxes and aggregating in a linear chain. We should be constantly analysing the evidence with a systems mindset and continuously adjusting to changing realities in order to see the tipping points coming and avoid shipwreck! That is why I fell that Monitoring and Evaluation should not be conceived as add ons but rather fully integrated in program management systems to ensure the feedback loops for adjustin. For more on this, I highly recommend Michael Patton’s new book: Developmental Evaluation: Applying Complexity Concepts to Enhance Innovation and Use.


Leave a Reply