The world has been witnessing varying degrees of economic distress since the western Atlantic financial crash of 2008. You may have been hearing people grumbling that economics as a serious discipline has failed.

I think this is because economics has run into a peculiar problem. It’s claiming a degree of scientific status that it cannot, as a social science, possibly have.

The problem now is that when the data supports something, the theory doesn’t support it and when the theory supports it, the data doesn’t. Economists, however, refuse to admit this.

Theory was discredited around the early 1980s when more and more data started becoming available and when econometrics and its analytical tools became sharper. It was a perfect combination and it knocked theory off its prestigious perch.

Until then nearly all of economic theory was aimed at explaining the behaviour of economic agents, whether individuals or economic groups like firms. Data, however, has focused on finding patterns which help predict the behaviour of individuals and groups. Thus, while theory had tried to explain, data analysis tries to predict.

Simple vs complex

There were two types of economic theory: simple and complex. Likewise, there are two kinds of data: simple and complex.

Simple economic theory said things like demand comes down when price increases. Or, that if the demand for cars increases, so will the demand for petrol. And so on. It was mostly common sense.

Also, simple data was essentially bipolar, which means any two variables could be plotted on a simple graph with two axes.

If you produced more doctors (X axis) you would need more nurses (Y axis).

Then complex economic theory came along and started getting into things like accumulation of capital and rates of economic growth and the relationship between money supply and the price level and so on. It was trying to be multivariate without the supporting econometrics. Pure maths made up for the deficiency.

Complex data, likewise, happened because with digitalisation all sorts of data started getting recorded automatically. But it was just a shapeless mass. Then artificial intelligence slowly started to make sense of it.

But this had a perverse outcome. By 2000, economists had become addicted to this data cornucopia and completely forgot that, just as a theory without data was useless, all this data without a theory was also useless. It was the old complementary goods thing.

As a result economics started floundering and flailing. It’s not taken very seriously any longer.

A few examples

Take three random examples: GDP measurement, the behaviour of firms and labour market functioning. In each case there is a divergence between theory and data.

In the 1940s an English economist called Richard Stone developed a framework for measuring GDP and refined it over the next 15 years. In 1984 they gave him the economics ‘Nobel’. The previous 16 had gone to theorists.

Stone’s central contribution was to introduce double entry accounting into national accounts. He thus showed how to disaggregate and measure production. But he emphasised that his method wasn’t meant for forecasting because his model was static and forecasting is dynamic.

But what’s the result now: we do a lot of forecasting that’s not based on anything stronger than the quicksands of production data. No one anywhere gets these fully right.

Or take firm behaviour. All useful theories about it had been written by the mid-1970s. They essentially have to do with competition and strategy.

But what useful data have we got about it? Close to zero and again for the same reason: non-linearity, which means you can’t predict the final outcome from initial conditions. Explaining how firms behave (theory) is a far cry from forecasting what they will do (data).

Yet the whole stock market depends on this parrot wala and millions take it on faith. Failures are explained away by saying “oh, there was ‘asymmetry in information’!” Of course there was. That’s what non-linearity is.

Finally, take labour market theory which is about labour allocation and returns. This means decisions regarding who will do what and how much he or she will be paid.

But what does the data show? Nothing more useful than how many people have a job at any given time. That’s all. Period. So from being an economic construct, it’s become a political one.

The failure is because the data is about jobs, and not work. Jobs are about certainty and income smoothing. Work is about uncertainty and stochastic earnings.

I can go on but let’s stop here. I want to wait for an economist to say I am talking nonsense. It would be another example of the divergence between theory and data.