Economists have never shied away from exploring the big questions that historians, sociologists and political scientists often claim exclusive rights to. What can we say about the long-term consequences of slavery in contemporary American society? Why is the level of mutual trust in some communities higher than in others? How to explain the rise of right-wing populism in the last few years?
In dealing with issues that go beyond the domain of their discipline, economists go beyond the usual preoccupations with the dynamics of supply and demand. Such crossing of boundaries between disciplines is not always approved. Representatives of other disciplines (often quite rightly) note that economists do not invest enough effort to inform themselves about existing achievements in relevant fields. They complain (and rightly so) about their inhospitable academic culture. Due to frequent interruptions, interruptions and aggressive questioning, meetings of economists often look more like a session of the Inquisition than a forum of colleagues who exchange results and consider new ideas.
However, probably the most important source of tension is the method that economists use. To show that the outcome they are examining is "caused" by a certain factor, economists rely on statistical tools. Often misunderstood, their method is the source of endless and useless conflicts between economists and representatives of other disciplines.
Understanding the strengths (and limitations) of their methods could illuminate the benefits of the insights that economists bring to the analysis of non-economic issues. Also, such an understanding would finally show that their approach can complement, but not replace, the qualitative methods of other disciplines.
Here it is best to start from the very idea of causality. We acquire knowledge about cause-and-effect relationships in two ways. We start from the causes and try to determine their effects, or we start from the effects and try to determine their causes. Andrew Gelman, a statistician at Columbia University, calls the first method "forward causal inference" (moving from causes to possible effects), and the second "backward causal inference" (moving from effects to causes).
Economists are obsessed with the first of these two approaches - causal inference. The most respected empirical studies are those that show that exogenous variation causes X produce a predictable and statistically significant effect on outcome Y.
In the natural sciences, causality is established by performing laboratory experiments that isolate the effects of certain variations in the physical environment on the outcome under investigation. Economists sometimes imitate this method with randomized social experiments. For example, households are randomly selected to receive cash assistance - while some other households will not - in order to examine the effects of increased income.
But history and social processes are not a laboratory in which the effects of every change in the environment can be precisely determined and measured. That is why economists resort to imaginative statistical techniques.
For example, based on the documented statistical association of exogenous factors such as rainfall and the possibility of civil conflicts, they conclude that a change in the level of income (due to fluctuations in agricultural production) causes civil wars. Note the key element of this solution: since civil wars cannot affect weather conditions, the relationship between them must be unidirectionally causal in the opposite direction.
Well-conducted research of this kind provides useful insights - to the extent that causal claims are generally acceptable in the social sciences. But historians and political scientists are not impressed.
The reason is that the economic model cannot offer an answer to the question "what causes civil strife" (which would be backward causal inference). The economic model provides insight into only one from the cause (fluctuation of income), which does not have to be decisively important. Worse, since they are trained only in causal inference in advance, economists tend to present the partial results of their research as inclusive and final, which further irritates representatives of other disciplines.
Economists use other tricks that create problems. In an attempt to statistically "identify" causality, they often resort to techniques that ultimately provide an answer to a significantly narrowed or even differently formulated question compared to the question that originally motivated the research.
The results of randomized social experiments in certain regions, for example in India or Kenya, may not be valid for other regions and countries. Investigating variation in space does not provide correct answers to questions that require examining changes over time: for example, what will happen if there is a barren year. Also, the exogenous shock used in the research may not be representative: for example, a drop in income caused by some other reasons will affect conflicts differently compared to a shock caused by drought.
That is why economists' research cannot replace comprehensive synthetic works that take into account a greater number of causes, weigh their likely effects and consider spatial and temporal variations of cause-and-effect mechanisms. Such an approach is characteristic of historians and representatives of non-quantitative social sciences.
In this type of research, value judgments play an important role, which leaves room for disputes about the validity of the conclusions. No such synthesis can offer a complete list of possible causes, even if their relative importance could be correctly assessed.
But such works are fundamentally important. Economists would not know what questions to ask if it were not for historians, ethnographers and representatives of other social sciences who provide them with rich narratives about various phenomena and hypotheses about their possible causes, without claiming that the causal relationships they propose are irrefutably established.
Economists can pride themselves on the strength of their statistical and analytical methods. But they must not lose sight of their limitations. Ultimately, both approaches contribute to our understanding of society. Economists and researchers in the social sciences must embrace the diversity of their approaches, not dismiss and denigrate the contributions of neighboring disciplines.
(Project Syndicate; Peščanik.net; translation: Đ. Tomić)
Bonus video:
