Substack

Wednesday, February 6, 2019

The 'experience' discount in development

Knowledge about a subject is of two types - theoretical and experiential. 

The former is accumulated through reading, observing, and discussing the issue or thing (learnt knowledge). You could learn about traffic congestion, including its relationship with network theory etc at say, Massachusetts Institute of Technology or Madras Institute of Technology. Here a logic or hypothesis or theory is used to draw inferences about a problem or a situation. This application of deductive logic or positivism is also what characterises a computer.

The latter is accumulated by doing the thing in question (lived career) and/or being part of the thing itself (lived life). You experience traffic congestion either as a daily commuter or as a practitioner trying to resolve traffic congestion in a particular place/city. This accumulation is the experience of what one comes across in daily life. Here priors arising from countless empirical observations is used to draw inferences about a problem or a situation. This application of inductive reasoning is also what characterises a human being.

In the world of impact investing and innovation-focused development, one comes across very smart people, with limited or no experiential knowledge (nor even reference to priors), trying to draw inferences about a problem (and its solutions) using purely deductive logic.

So, the wonder App which gives information (weather, agronomy, market etc) to farmers can boost farm productivity and incomes. The adaptive learning solution can transform the learning outcomes of students in public schools. The fintech solution can cut transaction costs and increase the savings of the poor or provide micro-pensions and micro-insurance products. The blockchain solution can address the problem of poor land records in developing countries. The smart meter or GIS mapping or SCADA can cut the persistent distribution losses in water and electricity utilities. The nifty labour market matching solution can connect households with cooks, housemaids, and drivers. 

This approach glosses over the context as well as the complex practical challenges associated with the actual realisation of impact. It is motivated by the neatness of logical reasoning. In other words, it overlooks priors or experiential knowledge. It overlooks the reality that the theory of change to realise impact in each of the aforementioned examples is very complex. In simple terms, it is the deductive analysis of a computer for a problem which requires inductive analysis. 

So, theory says that farm productivity is constrained by informational factors. Relax them, and you have increases in farm incomes. Theory informs that children have differential learning trajectories. So an algorithm can customise instruction to the specific needs of each child and lead the child up the most optimal learning trajectory. But in practice, the pathway from theory to realisation of impact is filled with numerous risks and uncertainties.

The academic trends of our times exacerbate the problem. Consider the example of RCTs. It gives power to the appealing belief that one could juxtapose a solution against the business as usual counterfactual and establish efficacy without any consideration for the priors, experience, and history surrounding that intervention. So you could publish an RCT study in a reputed journal about the efficacy of an Agtech App which ostensibly delivers agriculture extension services, without any exploration of the rich history of the programmatic delivery of such extension services (and its failures/struggles) and the associated priors.

This also parallels the general marginalisation of the study of economic history and philosophy in development economics courses of the day and the elevation of data analysis using statistical techniques. It manifests in the form of important debates on critical issues that face us today like inequality (which Thomas Pikkety triggered off with his book in 2014) being reduced to technical debates on an inequality or the relative importance of different contributory variables

This also has the effect of distorting the debate in other ways. Consider the exploration of whether rural roads and rural electrification provides value for money. Some researchers have come out in the negative, even suggesting that the expenditure could have been better made on some of the more fashionable 'kinky' development initiatives. Leaving aside its morality, it is astonishing that well into the 21st century, we are debating the merits of all-weather roads and three-phase electricity, the two basic requirements to be able to meaningfully engage with modern civilisation, and are debating making choices between them and 'kinky' stuff like cash transfers or providing chicken or kicking Soccket balls!!!

No comments: