Economics is changing. It’s relying less on theorising about how the economy works, and more on testing to see whether there’s hard empirical (observable) evidence to support those theories.
Advances in digitisation and the information revolution have made much more statistical information about aspects of economic activity available, and made it easier to analyse these new “data sets” using improved statistical tests of, for instance, whether the correlation between A and B is causal – whether A is causing B, or B is causing A, or whether they’re both being caused by C.
But another development in recent decades is economists losing their reluctance to test the validity of their theories by performing experiments. Let me tell you about two new examples of empirical research by Australian academic economists, one involving data analysis and the other a laboratory experiment.
We see a lot of calls for reform that take the form: change taxes or labour laws in a way that just happens to benefit me directly, and this will make “jobs and growth” so much better for everyone.
These reformers always convey the impression that the changes they want are backed by long established, self-evident economic principles. And they can usually find professional economists willing to say “yes, that’s right”.
But what gets me is that, when the self-declared reformers get their “reform”, it’s rare for anyone to bother going back to check whether it really did do wonders for jobs and growth. Wouldn’t there be something to learn if it was a great success, or if it wasn’t?
Do you remember back in 2017, when employers were campaigning for a reduction in weekend penalty rates? The retailers and the hospitality industry told the Fair Work Commission that making them pay much higher wage rates on Saturday and Sunday was discouraging some businesses from opening on weekends, to the detriment of the public’s convenience.
If only penalty rates were lower, more businesses would open on weekends, or stay open for longer, meaning consumers would spend more, and more workers would be employed for more hours, leaving everyone better off.
The employers got strong support from the Productivity Commission and some economist expert witnesses. So the commission decided to reduce the Sunday and public holiday penalty rates in the relevant awards by 25 to 50 percentage points, phased in over three years.
Associate Professor Martin O’Brien, of the University of Wollongong’s Sydney Business School, commissioned a longitudinal survey (looking at the same people over time) of about 1830 employees and about 240 owner-managers or employers, dividing the workers between those on awards and a control group of those on enterprise agreements (and so not directly affected).
The economists’ standard, “neo-classical” model of the way demand and supply interact to determine the market price, with movements in the price feeding back to influence the quantity that buyers demand and the quantity sellers want to supply, does predict that a fall in the price of Sunday labour will lead employers to demand more of it.
So what did the survey find? It could find no effect on employment in the retail and hospitality sectors. This is consistent with a growing body of mainly American empirical evidence that, contrary to neo-classical theory, increases in minimum wages have little effect on employment.
But here’s an interesting twist: a majority of employers reported not making the reduction in penalty rates and a majority of employees reported not receiving any reduction.
One explanation for this is that employers didn’t pass on the cuts because they valued staff loyalty and commitment. If so, this fits with the judgment of many labour economists that the relationship between a firm and its workers is far more nuanced than can be captured by the neo-classical assumption that price is the only motivator.
An alternative explanation, however, is that those employers didn’t cut the Sunday penalty rate because they weren’t paying it in the first place.
Turning to the laboratory experiment, it tests the much more theoretical assumption that the behaviour of people engaged in economic activities is guided by their “rational expectations” about what will happen in the future.
Economists have come to care about what people expect to happen because this affects the way people behave, and so affects the future we get. In recent decades, many mathematical models of the macro economy have used the assumption that people form their beliefs about the future in a “rational” way to make the maths more rigorous.
By “rational” they mean that people respond to new information by immediately and fully adjusting their expectations – beliefs – about what will happen to prices, the economy’s growth or whatever. Which is a lovely idea, but how realistic is it?
Dr Timo Henckel, of the Research School of Economics at the Australian National University, Dr Gordon Menzies, of the University of Technology Sydney, and Professor Daniel Zizzo, of the University of Queensland, analysed the results of an experiment conducted by Professor Peter Moffatt, of the University of East Anglia, involving 245 students answering questions.
On receiving each piece of new information, the subjects had first to decide whether to adjust their beliefs and then, if so, by how much. The experimenters found that the subjects reacted very differently.
They found that, in general, people don’t update their beliefs with each new piece of information. And when they do, they tend not to adjust their beliefs by as much as they probably should. In other words, people display a kind of belief conservatism, holding on to a belief for longer than they should.
They found that this conservatism is explained to some extent by people’s inattention – they were distracted by other issues – and to some extent by the complexity of the issue: it was “cognitively taxing”.
It turns out that very few people – just 3 per cent of the subjects – display the rational expectations economists assume in their model-building. Most people’s behaviour, the authors say, is better described as “inferential expectations”.
Now, you may not be wildly surprised by these findings. But, in the academic world, common sense doesn’t get you far. You must be able to demonstrate things the academic way.
Even so, Henckel says that the responses of the experiment’s subjects extend to many parts of life, from the behaviour of investors in the share and other financial markets – this is how bubbles develop – to people’s political convictions, where they hold on to beliefs for far too long, ignoring much contrary evidence.
Indeed, inferential expectations apply even to scientists, who form a view of the world which they will revise or overturn only if there is overwhelming evidence to the contrary. So don’t expect economic modellers to abandon their convenient assumption of rational expectations any time soon.
