Photo Credit: Adobe Photoshop

Written by Hans Lind, Professor of Real Estate Economics, formerly at the Department of Real Estate and Construction Management, KTH Royal institute of Technology, Stockholm, Sweden

What distinguish Economics from other social sciences is the much larger use of mathematical models. It is therefore important to understand the role of models in Economics, especially in theoretical economics where economists work with building new models and prove that certain relation exist or do not exist in a specific model.

A basic idea behind this book is that if you want to understand the role of theories and models in Economics you need to have a systematic empirical foundation. Two large datasets are the main material used in this book. The first is texts published when Nobel prizes are awarded. This material is primarily used for identifying what is seen as pathbreaking contribution in Economics but also for comparing Economics and Physics as similar documents about the Nobel prize winner are available for both sciences.
The second larger material used is 200 full-length articles published in American Economic review, 100 from 1990 and 100 from 2020. The articles were classified into different types and each of the types were analysed more in detail from a number of perspectives. Here I will focus on the “theoretical articles” of the type described above, where a mathematical model is built and analysed. It can however be noted that the share of these articles fell from around 50% 1990 to 30% 2020.

A clear majority of these articles starts with a description of an anomaly: observations about what happens in a specific market or in an economy that earlier models have not been able to explain in a credible way. The new article presents an idea about what can explain this anomaly (that A leads to B, that A can explain B) and shows that in a specific model A actually leads to B. It is argued that the models becomes more realistic when A is introduced. It can for example be a new behavioural assumption or a factor left out in earlier models. By including the new factor it is argued that the new model is more realistic and more credible.

The situation can be described in the following more general way: There is an hypothesis that A leads to B (A can explain B) in a certain real market/economy (in country C at time T). The economist builds a mathematical model and shows that A leads to B in the model, that A can explain B. In the book there a discussion about whether this result in the model is a strong argument that A is the cause of B or not?
From an informal Bayesian perspective, the question is how much the new information that A leads to B in the model will change the probability that A is the cause of B in country C at time T.

It is argued that the following is important:

How probable is it that B is observed even if A is not the case? The weight of the new information that A leads to B in the model increases if there are no credible models where for example the factor A2 leads to B. If there are competing credible theories then new empirical information is need to evaluate which is the most credible.

In the book there is an illustration of how a small simple and partly very unrealistic model can be very influencial. The example concerns Krugman new trade theory. That the results in his early very simple models were seen as very important depended on
(1) the hypothesis presented (that heterogenous preferences, economies of scale and low transportation costs leads to trade in similar goods) seemed generally credible in relation to everyday knowledge, (2) initially it was hard to see that the simplifying assumptions was important for the result and the result in the models was shown to be robust to a number of changes, for example in market structure and (3) No one could produce a credible model where A did not lead to B or where some other set of credible factors led to B.

A general conclusion from a Bayesian perspective is that any kind of information that affects the probability of the empirical statement is important. This information can be anything from statements from people working in the sector to results in models that are seen as credible And of course in order to evaluate the credibility of a theoretical model you need institutional knowledge to be able to argue that central assumptions are credible. In my view, economics needs a broad spectrum of methods, and the challenge is both to be able to synthesize knowledge from different kinds of studies and to evaluate what kind of study it is that can change the probabilities of the empirical statement at hand most.

I have here only discussed specific mathematical models that is built to help us understand a specific observed phenomena. The book also discusses broader theoretical frameworks like microeconomic theory, game theory of contract theory and what role they play.

The realism of the crucial assumptions. There should be some kind of empirical information supporting A. This can be anything from informal information from actors in the specific market to more systematic empirical studies.

The robustness of the model. It is still a simplified model and there typically remains many unrealistic assumptions, even though it is argued that the model is more realistic in some important respects, If it can be argued that these assumptions are “harmless” in the sense that A probably leads to B in a model where these unrealistic assumptions are not made, then it increases the weight of the result from the model. Authors often end the article with a discussion about things that should be improved in future theoretical models. It is noted that theoretical articles of this type seldom ends with saying that the next step is an empirical study – and also noted that empirical articles seldom starts with references to specific theories or theoretical models.

Theories and Models in Economics
An Empirical Approach to Methodology

Hans Lind, Professor of Real Estate Economics, formerly at the Department of Real Estate and Construction Management, KTH Royal institute of Technology, Stockholm, Sweden is available now.

Read the introduction and other free chapters on Elgaronline.