As Mark Anthony might have said, I come to praise the standard economic paradigm and academic or practitioner networks, not to bury them. But I do so on one condition: that the standard paradigm is complemented by other paradigms, and that key networks are dedicated to modelling pluralism.
In a nutshell, I am going to argue that professional networks may well lead to dangerous monocultures, but only if they are homogenous and dedicated to pursuing the illusion of best practice even in conditions of radical uncertainty. In the right hands, networks can be the foundation of disciplined eclecticism.
The standard economic paradigm is the most successful social-science paradigm in history, but it has three besetting sins:
First, its proponents are frequently guilty of a form of academic imperialism that seeks to explain all problems – economic, political, and social – with the same basic set of models based on assumptions of rational expectations, rational choice, and the market tendency towards an optimal equilibrium.
Second, it is obsessed with being a predictive science based on parsimonious assumptions. It is only willing to take account of complicating behavioural factors and the messiness of reality if it can bolt on predictive amendments to its core rational expectations/rational choice models.
And third, despite taking behavioural economics and framing biases seriously, the profession is blind to the way its models, assumptions, and working practices bias as well as focus its own analysis.
To simplify greatly, three intellectual errors lie behind these three sins:
First, standard economics ignores the central importance of radical uncertainty or indeterminacy caused both by incessant innovation and novelty and by frequent changes in political trade-offs between incommensurable values. By ignoring the highly creative and necessarily political nature of economic systems – it puts too much faith in the ability of modellers to predict behaviour on the basis of revealed preferences, objective probability forecasts, systematic behavioural biases, and optimization among given factors and constraints. In conditions of uncertainty, there is no optimal choice, there are no objective probabilities, there is no single correct model, and behaviour changes.
Second, standard economics ignores the deep complexity of socio-economic systems. We need theories and models to make sense of the chaos around us in the same way we need a light to see in the dark. But if we only have access to one theory or set of models – one source of light – then our field of vision is limited and our analysis biased. This is the danger of analytical monocultures. I would argue that we frequently need to use multiple perspectives and models if we want to make sense of the different facets of a problem; and – rather than seeing models as encapsulations of the truth – we should view them as diagnostic tools for teasing out such systematic tendencies as do exist and for spotting newly emerging patterns.
The third mistake economists and policy practitioners make is to ignore the extent to which their analysis is embedded in particular modelling practices. As economists, our mental maps and our modelling practices are mutually reinforcing. As regulators, our ways of looking at problems and the data we use are embedded in the shared analytical routines mandated by homogenous best practice rules and standards.
This brings me to the key point about the impact of professional networks: If they are dedicated to the propagation and projection of homogenous best practice, to the entrenching of common professional methods, and the coordination of a single analytical approach, then these networks do enable a monoculture that is every bit as dangerous as agricultural reliance on a single crop strain.
The reinforcement through conference and journal-based networks of a single global set of professional and modelling practices in economics homogenises the way economists think about issues. Similarly, global best practice regulations homogenise how regulators and other economic agents think as well as how they act. The result is not only high (and potentially destabilising) correlations in behaviour, but also widely shared analytical blind spots.
This is not to deny the possibility of analytical or policymaking progress: we can and must learn from each other and from past mistakes. But in conditions of radical uncertainty, it is impossible to know ex ante what best practice will be, and we can no longer assume that competition will ensure that good theory pushes out the bad as we converge on the one true model. When dealing with rapid change and indeterminate futures we need diverse models to spot newly emerging problems.
The only sustainable basis for such disciplined eclecticism – and, indeed, for modelling innovations – is a supportive network of experts and academics specialising in different disciplinary and policymaking approaches. What we need above all is inter-disciplinary networks devoted to particular subject areas; or networks of professionals from different walks of life exchanging ideas.
I want to leave you with three final pointers for discussion:
First, I am not arguing that economists can do without any shared assumptions and professional ground-rules. Rather, I am arguing that they need to maintain sufficient cognitive diversity to enable on-going judgments about the trade-offs between diversity and homogeneity of standards and modelling practices.
Secondly, in arguing against convergence on best practice methodologies, I am not arguing against coordination of research across departments and disciplines. My point is rather that the problem of coordination should be seen as how best to use networks to coordinate heterogeneous approaches to common research problems rather than instil a communism of models.
Finally, my plea for disciplined eclecticism – where the choice of method used is defined by an open-minded assessment of the nature of the problem – is not a plea for anything goes. Rather, I am arguing for careful pre-analytical assessment of the nature of a problem from multiple perspectives and judgement calls about the evolving boundaries of applicability of different modelling techniques.
You may also like…
11 December 2017