“Progress in the social sciences is less tangible than in the natural sciences”

Why are there so many competing theories in economics?

Most research fields are simply extremely controversial. Only in a few areas is there a clear-cut, uncontroversial position. Most studies are not compatible with each other. That does not necessarily mean that they directly contradict each other. However, they arrive at different answers to the same research question and cannot be reconciled with each other.

But couldn’t this competition between theories also be seen in a positive light, because ideally it brings plurality, diversity, competition and a culture of debate into economics without any one ideology dominating the discourse?

Yes, you could see it that way. Competition invigorates research. But this often entails long and tough debates. In physics, for example, there is no one who takes a view that is fundamentally opposed to Newton’s. After all, Newton’s mechanics can be interpreted as a particular aspect of Einstein’s theory of relativity. The natural sciences comprise many more stable theories and much more canonised knowledge. This has the advantage that research can be based on these theoretical foundations, which does not exclude the possibility that these foundations will be improved at the margins over time. Without such stable foundations, scientific progress is difficult. We engage in endless debates and go round in circles. The extreme example here is philosophy where, even today, the debates revolve around topics that Plato and Aristotle addressed almost 2,500 years ago.

Would one solution be to do more experiments in economics, as is common practice in the natural sciences?

In principle, experiments are a good thing and economics has made great progress in this area in recent years. But experiments do not solve the fundamental problem facing the social sciences. Although psychology has been doing experiments for decades, it is not making progress similar to the natural sciences either. The crucial point is not that there are fewer experiments in economics but that the natural sciences mostly deal with physical technologies that can be seen and touched directly, such as cars, robots, smartphones and machines. Physical technology enables true feedback to be obtained because you always know where you stand. Either it works or it doesn’t. Since the natural sciences – especially physics, as the natural science par excellence – are based on physical technology, they can develop very quickly. In economics, on the other hand, you don’t deal with physical technology, which creates huge scope for false feedback. It is working with data – rather than working with non-experimental data – that creates these problems. Experimenting with objects is required in order to succeed. Whenever you only have data – even if it is experimental data – and nothing physical to hand, the problems of false feedback arise.

What role do institutional factors play in the slow scientific progress in economics?

There is a disincentive in the way we economists publish. The fact is that journals usually only publish studies which produce statistically significant findings. This gives researchers an incentive to search relentlessly for significance in their data, so-called ‘p-hacking’, which is statistically unacceptable.

Would it therefore be a good idea to move away from data analysis in research and work more theoretically?

No. You need theory and empirics. Without empirics we revert to an area such as string theory in modern theoretical physics. Although string theory is highly interesting, it cannot be empirically tested and again leads to endless discussions. Without theory, on the other hand, we have no guidance on the direction in which research should go. Ideally, there should be a dynamic interplay between theory and empiricism.

Would one solution be to work more on qualitative case studies rather than on numbers and data?

Case studies are not common in economics. They are more usual in business administration or political science. I do not believe that case studies would solve the problem. You generally only have very few observations in case studies, for example when comparing two countries with each other. We economists prefer to use large samples, for example comprising 100 or even more countries. The use of statistics in economics is a good thing – it’s just a matter of implementing the whole thing better.

Your book is subtitled The Case for Replication. Why is replicating studies so important?

When researchers know that their studies are being replicated by independent third parties, they become more careful and put more effort into them. Well-performed replications are an effective tool against the sort of p-hacking mentioned above. Crucially, a study carries more weight when several researchers independently reach the same conclusion. This can result in canonised knowledge, thereby creating more stability in economic research.

Why are so few studies replicated nonetheless?

You can’t achieve much recognition by conducting replications. If you want to make a career in economics, you have to use new methods and work in new fields of research. But replication is, by definition, not new. Consequently, there is no incentive for researchers to replicate studies conducted by other researchers.

Can science ever achieve truth?

Truth is a helpful ‘regulative idea’, even though we will probably never achieve it. For post-modern theorists, such as Thomas Kuhn and Jean-François Lyotard, there is no such thing as truth. However, many renowned philosophers, such as Karl Popper and the logical empiricists, retain this concept. Personally, I find the concept of ‘truth’ useful and would not wish to discard it. Truth remains the goal to strive for.