Transcript | 4. lecture: Are the economical laws the same as the laws of nature? (Miroslav Beblavý) - University for modern Slovakia

The topic of today’s lecture will be the economical “laws” and findings. These can be based on deduction or induction.

Deduction is the derivation of conclusions from purely theoretical bases. For example, stemming from the assumption that people maximize their personal benefit, I can build certain models of the human behaviour on this with the help of other simple assumptions. The power of the deductive models lies in their completeness. Their weakness is, that they are independent from the reality and it’s really hard to evaluate them without further verification.

The second possibility is induction, the drawing of general laws from concrete examples. Such thinking is in the social sciences often based on correlation – that the occurrence of some two things is somehow linked. For example after the increase of interest rates generally the demand after loans decreases. This is often used to derive that the increase of interest rates causes the decrease of demand after loans. Does the correlation really mean causality? When yes? When no?           

Let’s start with an example from the American practise. This graph can serve as an example of how the correlation and causality don’t always have to affect each other. It shows the comparison of how many places for employment were created when the congress was managed by the republicans and when the majority were democrats. During the management of the republicans was the number of working places increased, after the arrival of the democrats it had a decreasing tendency. Therefore- the republicans are good for employability and the democrats are bad....And as we will see alter, it’s not necessarily this way.

Let’s start with a real study done at the Pennsylvania university published in 1999 in the prestigious magazine Nature. Found a correlation between short-sightedness and whether parents leave the light on in the room where their kids sleep. From the correlation comes the logical causality – who leaves the light on for their children damages their eyesight.

At later study at the Ohio State University didn’t replicate this relationship between these two phenomena. However it found a strong correlation between the parents who suffer from short-sightedness and the short-sightedness of their children. In other words, short-sightedness is to a great amount hereditary. At that time they noticed that the short-sighted parents leave the light on for their sleeping children much more often.

Generally it means, that just because two things happen at the same time, it doesn’t mean that they are in a a direct or indirect relation.

What can be done about this?

Already a few centuries ago was the Scottish philosopher, historian and economist David Hume (known for his strict empiricism) dealing with these problems in more detail. Hume claimed that the correlation on its own is never a  sufficient proof of the causality of a phenomenon and therefore we can’t gain certain knowledge based purely on empirical resources. This is how he thought that for example there is no insurance, that the sun will appear on the sky today based only on what happened yesterday. This was later further developed by the German professor Immanuel Kant, who said that in order to tell which correlations are correct and which are only coincidences we will use reason. Naturally, many other have been dealing with the methodology of science, however for the purpose of today’s lecture this is sufficient.                           

So in the first place we need some theory, how the world works in a certain aspect and a from it stemming hypothesis – a provable statement – and then we try to prove or disprove it using empirical facts.

Economists are, compared to the natural scientists, much more reliant on the observance of phenomena that occur in the society without the possibility to experimentally work with them- so to repeat them in laboratory under the same and different conditions and thanks to it disprove or prove the various hypotheses.

In the economy- on contrary to the natural sciences – its very difficult to carry out laboratory experiments, even if the development has advanced even in this aspect in the last decades. The process of economical laboratory processes has been described in detail by Louise L. Wilde. The laboratory experiment in economy tries to create and subsequently study the smaller and artificially created micro-economical environment. The purpose is to reveal the systematic relationships between individual preferences, parameters and results. The whole experiment is based and falls on the experiment leader. Firstly, the authenticity of the micro-economical environment has to be secured, sufficient control has to be held over the individual preferences and parameters in order to find any relationship between them and the final results.

The effort to generalize laboratory findings is criticized by many economists. For example it happened with the study from Stephena D. Levitta a Johna A. Lista, who claimed, besides other, that such theoretical progress in the laboratory is based on the assumption, that the findings can be extrapolated into the outside world – however this isn’t really true in the economy according to Levitt and List. If we confirm some physical (gravitation) or biological (photosynthesis) law in a laboratory, we are certain that it will work in the real world. However were missing this certainty in the economy, because to simulate real world markets into a small micro-economical sample is impossible. Their application to the real markets would therefore be a problem. At the same time they claim that good ideas can be gained through the laboratory experiments, however we can never be 100% sure, that they will really react this way after their application to practise.

So if we step out of the laboratory, we are left with the statistical techniques. Before year 1980 most economists used the system of linear regression on non-stationary time series, which is a very complicated way, how to say, that the verification of theories was in many things based on the verification of correlations.

However Clive Granger (University of Nottingham) a Robert Engle (New York University) found, that such system is responsible for many deceptive relationships. They created new economical tools, with which I won’t torture you here with, but which are much more adequate to reveal  relationships between various variables. One of them is co-integration – today we can thanks to Granger and Engle find out, whether different time series are co-integrated, so whether they move hand in hand. This, naturally, also doesn’t prove the causal relationship, but it shows a much closer relationship. They also came up with the concept of the so called Granger causality, where a statistical test makes it possible to find out, whether there is a phenomenon following another in the data (for example the already mentioned increase of interest rates before the decrease of the demand after loans). Granger and Engle got a Nobel prize in economy in 2003 for these findings.

The statisticians and economists gradually develop more and more sophisticated tools in this field, however they will not bring us complete certainty.

And that’s why at the end the economical “laws” and findings will be left to be based on the interpretation of the empirical data / the subject to possible various interpretations.