For quite some time, I had been meaning to write about the seemingly new “admitting failure” phenomenon in development policy. That idea, however, had been pushed back by other, more pressing ideas.
Thankfully, Ed Carr, my favorite online sparring partner, wrote the following last week:
(…) I wonder about the utility of these admitting failure efforts that I see coming from groups like Engineers without Borders. I had the good fortune to catch up with Tom Murphy (or, as the twitterati know him, @viewfromthecave) the other day while he was here in DC, and we started talking about learning from failure. In the course of our conversation, we came around to two key problems. First, really admitting failure requires reframing the public image of development as an inherently do-no-harm effort, where just doing something is better than nothing. Second, given this first problem, when we really start talking about what failure means, even in the most constructive of settings, we will call the entire development enterprise into question.
I don’t have much to say about the two points Ed and Tom brought up — that is really their area, and Ed summarizes their discussion nicely. But I can offer my economist’s take on admitting failure.
My first contact with the idea of admitting failure (in development policy, that is) came when I met an Engineers Without Borders (EWB) representative at a tweetup in Washington, DC in January.
Indeed, EWB were the ones who set up AdmittingFailure.com, a website where development practitioners can submit their stories of failure (see the list of failures here). And not only have EWB received a lot of attention for their website, it also appears that that the idea of admitting failure is spreading, as even World Bank folks now talk about the importance of admitting failure. [UPDATE: After I wrote this post, it came to my attention that GiveWell was also admitting failure when I read about it in a post on Andrew Gelman's blog.]
“Corporate” Social Responsibility?
Why admit failure? Here is where I get a bit more cynical. A few weeks ago, in response to the Aid Blog Forum, I discussed corporate social responsibility (CSR) and wrote:
To me, it’s all marketing. If a firm adopts some CSR practice, there is a real cost to it. But once that cost is incurred, it’s pretty much all upside. A CSR practice does not turn any potential client off from buying from the firm, nor does it turn any potential employee off from working for the firm, but it may attract potential clients and quality potential employees for whom it is important that the firm they buy from or work for adopt CSR practices.
Again, there is a cost to adopting a given CSR practice. If there is no cost to some practice, it is presumably not worthwhile. But in cases where that cost is offset by the expected benefits just discussed, the firm will adopt the CSR practice. Alternatively, in cases where that cost exceeds the expected benefits, the firm does not adopt the CSR practice. Like any rational decision, this one is taken at the margin: if “good for the community” means “good for the bottom line,” why would a firm willingly leave dollar bills on the sidewalk?
Likewise, even though a nongovernmental organization (NGO) admitting failure might incur a small cost for it (i.e., some donors might decide to withhold their funding), something tells me that admitting failure is also pretty much all upside. In other words, there are many more donors who will applaud the NGO for admitting failure than there are donors who will withhold funding. More importantly, admitting failure will push new donors to fund the NGO, which will eventually mean more funds for the NGO.
I thus see “admitting failure” as playing for NGOs a role similar to that played by CSR for firms. In other words, a lot of it is plain old marketing.
This is similar to a politician breaking the news of a scandal he is involved in on his own terms before the press does it — something that was characteristic of the Clinton presidency. I was reminded of that earlier this week when I caught a rerun of an episode “30 Rock” in which Jack Donaghy, the right-wing NBC executive played by Alec Baldwin, was told by none other than James Carville himself to break the news of his relationship with Celeste Cunningham, a Democratic congresswoman from Vermont who had attempted to sue NBC’s parent company.
Between the various development NGOs, however, there is a distinct first-mover advantage to admitting failure. In other words, I would expect the NGOs who are among the first to admitting failure to reap more benefits from it than the latecomers would.
If the “admitting failure” movement is indeed a trend, I expect there will come a moment where NGOs don’t have a choice but to admit failure because everyone else does it — in microeconomic theory parlance, a pooling equilibrium.
The Epistemology of Failure
Last but not least, I would like to throw what is an important methodological question out there.
As an applied microeconomist, I am all too familiar with how difficult it is to make causal statements. It is considerably easier to say that more of X is correlated with more of Y than it is to say that more of X causes more of Y. This is something that I spend a great deal of time on in the development seminar I teach in the fall.
I recently gave a problem set to my students in which I gave them a table of empirical results and asked them whether, for a specific variable in the table, the results were causal. In other words, did that specific variable (i.e., the presence of a title on a given plot) cause changes in the dependent variable (i.e., agricultural productivity)?
The tricky part was that the variable of interest did not have a statistically significant impact on the dependent variable in that table of results, which in turn led students to discuss whether they were presented with evidence of an absence of impact or with an absence of evidence on that impact. It can be extremely difficult to determine which is which in practice.
As regards admitting failure, which is really about admitting a demonstrable absence of impact, for which development practitioners don’t have rich data sets, how is one to demonstrate the failure of a given intervention, let alone credibly identify the causes of said failure?