Monday, January 28, 2019

Give economists a PC and they start making more sense

Economies turn down and back up, but one of the biggest, long-running economic stories of our time is the way the digital revolution is disrupting one industry after another. So let me tell you how it’s changing the academic study of economics.

You probably imagine the economic research carried out in universities is terribly theoretical and impractical. It used to be, but not anymore.

You can trace the progress of academic economics by looking at who’s been awarded the Nobel memorial prize in economic sciences from about 2001 onwards, and what they did to deserve it. Of course, there’s usually a long delay between when you make your seminal contribution and when you get your gong.

Until the turn of the century, the prize usually went to people elaborating on orthodox neo-classical theory, particularly by shifting to mathematical reasoning.

It may surprise you that the man who wrote the most popular introductory textbook of the post-war years, Paul Samuelson, was also the individual who did most to turn economic reasoning from words and diagrams to equations.

The development of the first mathematical “econometric” models of the macro economy was another important advance.

It was about 30 years ago that the frontier of economic research took a more realistic turn by shifting to the study of “imperfect competition”, where the idealised assumptions of the simple neo-classical model of markets were critically examined.

In 2001, for instance, the prize was shared by three American economists – George Akerlof, Michael Spence and Joseph Stiglitz – for their demonstration that, rather than being perfectly shared by everyone in a market, information is usually asymmetric – with sellers knowing more than buyers – and, rather than being costless, is expensive to acquire.

Another example: Paul Krugman got his gong in 2008 for demonstrating that there’s more to international trade than just each country pursuing its “comparative advantage”, as mainstream theory assumes.

It was about 40 years ago that the psychologist Daniel Kahneman (gonged in 2002) and the rebellious economist Richard Thaler (2017) began formulating behavioural economics, an advance on the neo-classical assumption that all decision-making is rational. Robert Shiller got his in 2013 for his study of non-rational behaviour in financial markets.

But recent studies of articles in the world’s top economic journals (mainly American) have shown that, since about the turn of the century, theoretical papers have largely been replaced by empirical studies of particular relationships in particular markets (competition between male and female drivers in Japanese speedboat races, for instance).

This shift from deducing conclusions from assumption-based theory to examining the relationships between real-world variables, to see how the theory measures up, is a big improvement. But why has it happened?

I give most credit to the information revolution. Computerisation has hugely increased that number of “data sets” of business information waiting to be discovered and subjected to statistical tests by academic economists checking hypotheses or just looking for interesting relationships.

All of which is easily done using programs on your personal computer, rather than waiting your turn for time on the main-frame. And it fits with economists’ modern addiction to using stats and maths for “academic rigour”.

As part of their greater interest in empirical evidence rather than what theory tells us should be the case, economists have started doing something they long believed was impossible: economic experiments – including searching out “natural experiments”, such as the famous study of two nearby cities in different US states, where one state raised the minimum wage and the other didn’t.

By the standards of real mathematicians, economists’ maths isn’t that fancy, but it’s more advanced than used by others in the social sciences. Economists have made more progress in moving from finding correlations to establishing causal relationships than the psychologists have – which probably means they get more research funding.

It also means there’s less resistance from international journals to publishing research about that uninteresting and unimportant place called Australia. I’m told doctoral students come to Oz because they’ve heard we have good data sets.

The risk, however, is that research projects are chosen because good data are available, rather than because the questions being answered are important to our understanding of how the economy works and to finding better solutions to our economic problems.

We don’t want academic economists losing interest in their theory, we want them using their empirical evidence to improve it. Making it more realistic and thus more reliable in its predictions about what happens if you do X, or whether policy A or policy B is more likely to improve things.