Thursday, November 11, 2010

Morality play and probability in development discourse

This post will highlight two strands of the prevailing development discourse in India which, I am inclined to believe, comes in the way of effective public policy formulation and implementation.

1. Development is not morality play. Any discussion on a development issue is generally overlaid with powerful moral overtones. Accordingly, policies fail to yield the desired results for a variety of moral failures - officials and politicians easily fall prey to corrupt practices, capitalist businessmen exploit ignorant consumers, bureaucracy does not empathize with the plight of the poor, supervisory officials lack character, and so on.

This normative discourse suffers from a serious disconnect with the dynamics of the real world in which the policy has to be implemented. In simple terms, it refuses to acknowledge the reality of the world-as-it-exists. The challenge then is to design a policy implementation environment for a world which is characterized by some or all of the aforementioned moral failings. How do we structure the incentive framework so as to align the preferences and urges of the various (corrupted, partly or fully) stakeholders towards achievement of the desired objective?

A pre-requisite to effectively answering this question is the acceptance of the reality of a world with pervasive moral failures. The policy environment has to be then micro-founded by borrowing insights from the fields of public choice theory, behavioral psychology, micro-economics, and political economy. The framework so arrived can be effectively translated into action using the latest developments in Information Technology.

For example, setting up an effective framework for supervisory officials has to acknowledge the reality of the lowest common denominator - corrupt, inefficient, apathetic and truant official. They have to be incentivized into working towards achievement of the desired objective by appropriate structuring of their work environment. This can range from carefully designed reporting formats to innovative use of IT tools, all of which liberate the official from diversionary and routine activities and converge his energies on his critical responsibility.

2. Binary evaluation of policies and implementation strategies. The prevailing discourse on development policy-making is shaped, almost exclusively, in terms of achievement of the final policy objective. Framing the terms of the debate on the outcome of a policy in such restrictive manner has important pitfalls.

Policy interventions or development strategies get judged on a binary scale - success or failure in achievement of the final goal. This means that the full-spectrum of possibilities between success and failure gets edged out from this binary evaluation framework.

The science of behavioural psychology informs us that framing of questions and its terms of reference play a crucial role in determining the course of debate about any issue. A binary evaluation framework therefore restricts policies to the standard straitjacket of solutions and strategies.

This is unfortunate since social policy interventions are inherently unsuited to such binary evaluation and are more accurately judged on a probabilistic scale. Accordingly, a more appropriate method to assess the effectiveness of a particular strategy could be in terms of how far it has increased the chances of achieving the pre-defined objective. What is the probability of achieving the objectives with this strategy? Has the strategy increased the likelihood of success? If so, by how much?

In other words, the touchstone for the acceptance of a particular implementation strategy would require a subtle re-formulation of the question. Today we ask whether the proposed strategy would ensure achievement of the objective? As we know, generally the answers to such sweeping questions in social sector, whatever be the policy objective, are in the negative. Instead, the query should be whether the proposed implementation strategy (or policy itself) significantly improves the likelihood of achievement of the objective?

Yes, a new approach will increase the likelihood (than is the case now) of the ANM or teacher attending to hospital and school respectively. But it will still leave open the possibility that some of them will game the new system and continue to play truant. But the net increase in the probability of achieving the objective, weighed against the relative increase in cost, merits adoption of the new strategy.

Both these biases in the development discourse have profound impact on policy-making and implementation. They exert a strong influence on the minds of the decision makers at all levels. The result is that the choices tend to get restricted to the already existing set of strategies. As a corollary, innovative or different approaches get crowded-out. Taken to extremes, they breed a culture of cynicism and status quo preference.

It is therefore important that any evaluation of available choices on development policies and their implementation strategies be made on a stochastic scale and without morality overtones.

5 comments:

KP said...

Dear Gulzar,

I am in complete agreement of evaluating policy choices to arrive at the one with the best chance of implementation success.

It is also creditable that your blog is trying to inculcate a policy discipline to all who may care to view policy implementation as continuous improvment and course correction.

That said, I think the blandness that we see in India's politics is naive faith in market based solutions that precludes the need to pave the way through political engagement - where politicians behave like bureaucrats on steroids and do not serve to inspire a vision.

I am quoting Michael Sandel in his book Justice here .. “Justice is inescapably judgmental,” he writes. “A politics emptied of substantive moral engagement makes for an impoverished civic life. It is also an open invitation to narrow, intolerant moralisms. Fundamentalists rush in where liberals fear to tread.”

And that includes market fundamentalism.

Development is a morality play, its implementation may needs hard headed evaluation as you rightl recommend.

regards, KP.

sashi said...

Your point about strategies getting judged on a binary of success or failure is well taken. Also the need for a judgement along a scale. But your recommendation of a probabilistic scale really does not address this problem because it can just point to the likelihood of success or failure and not to assessments of partial successes or failures - which are more likely to be the case.

a way out of this could be creation of indicators such as 'stages of progress' as suggested by anirudh krishna.

another problem that deserves to be discussed is that of attribution. you might have a success but unless you establish the causality in a rigorous manner its link with the policy may remain tenuous.

gulzar said...

thanks KP and sashi,

i absolutely agree with KP that the big development decisions will always have to be judgemental - because a sense of fairness in distribution and outcomes is what underpins development objectives, and fairness is morality play. but reaching that goal requires looking at the world as-it-exists and then designing solutions that stands chance of succeeding.

but, as you point out, the danger is with market fundamentalism, the sort espoused by the right-wing thinktanks, who grossly mis-interpret the teachings of Adam Smith.

sashi, the moment the discourse breaks out of the binary scale and embraces the probabilistic approach, then the spectrum of possibilities emerge as possible solutions. "stages of progress" markers would be one approach to manage such changes.

about the causation-corelation issue, much has been written about. even with rigorous research, it is really difficult to have generic causation analysis to problems, because of the important role played by environments. however, it is possible to tease out the causative factors for outcomes in specific environments. but then, such analysis have limited relevance.

sai prasad said...

Policy evaluation is a very old issue. The brave start with the binary question. Once implementation is started and money spent, the policy makers and implementers are desperate to ensure a modicum of success.

Hence they move from from binary to decimal to base 1000 etc. Already happening since the beginning of bureaucracy.

Danger with this -- bad policy escapes under the guise of non-binary evaluation.

gulzar said...

thanks sir. i agree with the slippery slope in moving over to a non-binary scale. but, it is inevitable, and we need to be careful about it...

i am only saying that the discourse should be open to the possibility of interventions and implementation strategies which, while not being fool-proof, are likely to be more effeective than the current option. however, the rigor in the evaluation of those outcomes, which finally is what we want, should not be compromised.