Austrian School of Economics emerged in Vienna at the end of the XIX century. In a certain way, it meant a rediscovery of old ideas endorsed by the Salamanca School.
On the one hand, Luis de Molina determined that the value of goods is a reflection of the use that people find in them. Furthermore, Diego de Covarrubias said that those valuations depend on estimations from economic agents, even if those calculations are foolish. On the other hand, Domingo de Soto and Luis de Molina stated that the value of goods are not determined by their cost, labour or risk, but both the abundance or scarcity of them.
Carl Menger expanded the above ideas on his book “Principles of Economics”. For example, the author states that, on this discipline, we can only use the logical-deductive method. This idea meant a clash with the inductive method used by marxist economists.
Later, Ludwing von Mises and Friedrich von Hayek fiercely criticized the role of the State in economy. For them, demand for goods and services depends on multiple and unpredictable factors. As a consequence, it would be impossible for a State to calculate the quantity of different items to be produced, and any attempt to do so would eventually cause a collapse. Contrarily, business can do this job because they focus on niches, so the information needed is substantially less.
Mises and Hayek also developed the Austrian Business Cycle Theory, where they explain how business cycle are created and how they could be avoided.
Many investors claim to use the Austrian Business Cycle Theory to make investment decissions. Some of them even link this framework to Value Investing philosophy. So I wanted to really understand it and used statistic tools in order to do it.
Stages of the Austrian Business Cycle Theory
The intention of this post is not to make an extensive explanation of the Business Cycle theory, but to summarize the key ideas so everyone can get a clear idea:
- The business cycle starts when the central bank (ECB, FED, BoJ, among others) decide to lower the interest rate below its natural level. This fact creates a mismatch between the price of money and agents preferences between savings, investment and consumption rates.
- The recently printed money has two effects when put into circulation:
- It clearly impacts the saving and consumption rates. Because money is now cheaper, agents feel they can consume more. Furthermore, cheaper money discourage savings due to a lower remuneration.
- Additionally, entrepreneurs undertake new projects that appear to be profitable, starting an expansion period.
- At this point of the cycle, stock prices start growing due to recently printed money. Specially, during the last crisis where households solvency did not improve and banks did not invested the money in real economy.
- Undertaking new projects in earlier stages of the economy means that industrial prices will tend to go up at a faster pace than consumption prices and so do industrial stock prices.
- As time goes on, consumer prices will look cheaper in contrast with production prices. This fact will act as a signal for entrepreneurs, who now will want to reallocate resources in the consumption stage. Demand for consumption goods exceeds supply and this type of stocks benefit from good prospects.
- The increase in money demand will push interest rates up. Initially, a central bank can print more money but ultimately demand will exceed supply.
- Once interest rates start rising more and more projects will become less profitable, and some of them will face bankruptcy, giving way to a recession period. At this point is when stock índices plument, and the creative process start purging unprofitable projects and starting different ones.
Can this theory explain reality?
It is well known that austrian economists reject the use of mathematical and econometric methods in Economics. For them, if a proposition is derived from a true synthetic “a priori” judgement, it does not need to be tested empirically. Nevertheless, there are many schools and banks that use such methods in finance. For that reason, I have tested the Austrian Business Cycle Theory with econometrics and see if the logical-deductive method can lead to a true explanation of reality.
The analyzed variables are the following:
- Interest rates set by the central bank.
- One-year interest rates
- Industrial price index
- Consumer price index
- Spread between short-term interest rates (3 months) and 10-year bond rate.
- Gross Domestic product index.
The techniques used for this test are the following:
- Since we are working with time series, we firstly need to test for stationarity. The reason is that a model built with non-stationarity variables can produce misleading results. For that reason, we use the Augmented Dickey-Fuller test.
- Secondly, once that we know we are working with stationary series, we need to make sure that variables are not cointegrated. Here, we use the Engle-Granger test.
- Thirdly, we run the causality test in order to determine if a variable is, statistically speaking, causing a second variable. The Granger test is used for this purpose.
As stated above, we first look for stationary variables. In the following table we can see the different variables analyzed and the ADF test as well as the order. In all series, stationarity is reached with first differences:
In the first column of the second table, we can see that all pairs of series analyzed are not cointegrated. In the second column we see the p-value and lag of causality tests.
I find interesting how interest rates granger-causes industrial prices with a lag of one period (once interest rates have been lowered, entrepreneurs firstly invest new money in earlier stages) and consumer prices with a lag of 5 periods (once that consumer prices look cheaper in comparison with industrial prices). Similarly, rising interest rates granger-causes a slowdown in gross domestic product (pib) with a lag of 5 periods. For the rest of pairs of variables, the impact is almost immediate.
Once that we have checked for stationarity, cointegration and causality, I have run a multiple linear regression model to assess the impact that interest rates set by central bank (i_ref), ratio of consumption to savings (constosave), ratio of industrial prices to consumer prices (ppi/cpi) and spread between 10 years bond and 3 month interest rate have over GDP (pib). The results are the following:
As we can se from the previous table, all components look statistically significant at 95% with the exception of the cons/save, which is only significant at 90% of confidence. The adjusted squared-R of the model is 0.738, which means that about 73,8% of the variance of GDP (pib) can be explained by the chosen predictors.
After the previous analysis, we can conclude that statistically speaking, Austrian Business Cycle Theory works in real life. But, does my analysis mean a true validation? I personally think that, as stated by many Austrian authors, mathematics and statistics do not truly verify neither cause-effect relationships among variables nor their intensity. If fact, the inverse experiment could have been conducted (gdp granger-causing spread; spread granger-causing ppi/cpi movements, an so on) and the results could have been statistically significant as well. For that reason, I see that the logical-deductive method needs to be used in order to reach true validation.
We can see the superiority of logical-deductive method in order to explain business cycles and discover investment opportunities. So yes, we can benefit from this framework in order to make better investment decissions. And the good new is that Carl Menger gave the same message a long time ago without the need of building any econometric model.