World-System (1900-1950) The USGD Measurement Model

 

Untangling the causes of the Great Depression in the 1930s United States will not be an easy task. There are many assumed causes (see the Notes below) and equally many theories: Monetary,  Keynesian, Marxist, Austrian School, Debt-Deflation, Structural, International (Gold Standard), and Policy Failures (see the Summary Tables in the Notes). What has not been tried is a System Theoretic explanation.

To avoid any misunderstanding, if that is possible, what I mean by a "System Theoretic Explanation" is developing a State Space Representation of the US Economy and the World-System in the period (1900-1950) and then experimenting with the model using computer simulation. The difference between a State Space model and an Economic Model (discussed in more detail here) is that the State Space Representation studies attributes of the "System" (attributes such as stability, controllability, mechanization, etc. more information on model properties can be found here) rather than assuming structure for the economy and understanding the model's response to shocks. In economics, the attributes of the system are assumed (stability, for example) and the state of the system is assumed to be known. In Systems Theory, the state of the system has to be estimated and analyzed.

I have been working on this project since the early 1980's (starting with a Systems Model of German Nineteenth Century Development) and this post starts describing how the model was constructed. My approach was to read the existing Great Depression literature and note the variables that were being discussed. I came up with thirty one (more reading would probably have produced more variables but the ones I have chosen do seem to group within the "Contributing Factors" listed in the Notes).

One immediate problem with my list, emphasized by Marxist Analysis (Wolf and Resnick, 2012), is that "Everything is related to everything."  There are no zero correlations among the thirty-one indicators I have chosen to cover the Great Depression (here are just a few as examples, definitions are in the Notes):


Systems Theory has an answer to the "Correlation Problem" and it is the Dynamic Components Model (DCM). The first step in constructing a DCM model is to develop the Measurement Matrix (H) which connects the state variables with the thirty one  output variables. The Measurement Matrix is computed from Historical data (see data definitions below) using Principal Components Analysis (PCA). The resulting state variables (rows of H), are the minimal collection of variables that explain the most variance (usually above 80%) of the measurement variables.


In the USGD Measurement Matrix (above) three component state variables explain 86% of the variation in the thirty one indicator variables. As a practical matter, for macroeconomic systems, the PCA result means that picking out one or two variables (such as the Money Supply, M1 and M2) is potentially misleading. However, the weights in H show how important an indicator is to the component sate variable (M1 is important to US3, the third component state variable but, along with M2, does not correlate with other indicators).

Over time, the component state variables can be plotted (see the graphic at the beginning of this post). Typically, the first state variable is a balanced growth component while the second and third (or more) state variables are cyclical, feedback components. You can barely see the Great Depression in US1 (the  balanced growth path), but you can see the Roaring Twenties in US2 and another cyclical component peaking after the Stock Market Crash in 1929 in US3. What do the state variable time paths mean?

The definition of the State Variables depends on the weightings of the indicators variables in the measurement Matrix (H) above. I've underlined the larger weightings and the key to the variable names is in the NOTES: 

        US2 = (Banks-Unemployment)

         US3 = (Unemployment-GNP-G)
 
In words: US1 = (Overall Balanced Growth), US2= (Banking System Controller) and US3 = (Unemployment and Government Controller). These controllers track balanced growth Theory, the Austrian School and Keynesian theories. I'll discuss the two Feedback Controllers more fully in a future post. But for this post, the PCA decomposition has divided 0.855% of the variance up into  balanced growth and feedback control components. The decomposition allows us to analyze the US Great Depression USGD) Economy as a General System, a topic that will involve a number of future posts.

You can run and investigate the USGD model here.



Notes

Google AI Contributing Factors  (double-click to enlarge)



ChatGPT Summary Table





Definition of the variable names  (double-click to enlarge)



Feedback Controller Details


   US1 = (Overall Balanced Growth - 0.1338 P.CPAPER - 0.151 IMM.US.)
    US2 = (0.398 O.B. + 0.326 P.S.P.DPR + 0.261 P.FUELS. + 0.2417 P.WPI. + 0.213 GNP.I. - 0.272 L.U.)
    US3 = (0.4216 L.U. + 0.3847 P.FED.FUNDS + 0.331 V.NYSE. + 0.2174 K.US. - 0.2549 GNP.G. - 0.2181 M1 - 0.2419 P.W.AG. - 0.248 IMM.US)

Expanded Measurement Matrix (double-click to enlarge)



It is interesting to scan the columns of H to determine how important indicators were in explaining the Great Depression. 

Important indicators: L.U., GNP.I., GNP.G, P.FED.FUNDS., P.SP500., V.NYSE. P.S.P.EPR., Q.H.Starts, P.WPI., O.B., P.FUELS., P.W.AG., and IMM.US.

Unimportant Indicators: L.US.E, GNP.US., GNP.C., P.CPI., P.GDP., P.S.P.DPR., M2, Q.A., Q.I., P.W.MFG., Q.OIL., N.US., and U.US.

There are some surprises here: the Labor Force, overall GNP, Consumption, Deflation, Agricultural and industrial production, Population and Urbanization simply grew with the rest of the economy and were not affected by the Great Depression.

Comments

Post a Comment

Popular posts from this blog

A Theory of Great Depressions

Image
  I had wanted to study the Great Depression as a graduate student at the University of Wisconsin-Madison in the late 1970's. My pitch was that the Great Depression was a "Systemic Event" and therefore should be studied from the perspective of Systems Theory . I received two responses: (1)  Systems Theory  had a "bad reputation" in the Social Sciences and (2) the    Great Depression  was a topic reserved for elder statesmen in Economics and was not appropriate for a graduate student. I went on to apply  Systems Theory   to Late Nineteenth Century Development in Germany and emphasized the systemic issue of Stability . Today, I am at the end of my career, I'm still interested in the Great Depression and I don't need anyone's approval to study it. What I have found is that there is really no consistent theory of Great Depressions and that Systems Theory in the Social Sciences is, unfortunately, poorly defined (it is clearly defined in Environm...
Read more

About

Image
  The title of this blog is based on a book by Charles Kindleberger World in Depression (1929-1939) . In the book, Kindleberger argues that (1) Financial markets are prone to contagion ( Economic Bubbles ), (2) the bubbles spread across countries and (3) Hegemonic leadership is important to stabilizing these dynamics. Markets pass through cycles of boom, crash, panic, repulsion and finally depression (sometimes called the Minsky-Kindleberger model ). In this blog, I will present my own state space (statistical) models (see the Boiler Plate ) of the period for as many countries in the World-System as time permits. I have another blog, The Economic Bubble Machine,  where present models of  Economic Bubbles  in other countries and other time periods . To get an idea of my approach, check the post  W orld-System (1900-1950): Did the Smoot-Hawley Tariff Cause the Great Depression?  The Smoot-Haley Tariff is particularly relevant to current Economic Policy in t...
Read more