World-System (1800-1950) Forecasting the Great Depression in France

Dec 24, 2025. The New York Times is reporting (here) that "for the second consecutive year ...[France has failed]... to agree to a new national budget by the end of the legislative calendar." It is difficult to understand what is going on in France without some historical background, specifically how France responded to World War I and the Great Depression. ChatGPT reports that 

In this post, I am going to use France's economic statistics in the Long 19th Century to forecast France's economic performance in the early 20th Century. The multi-model forecasts based on different input assumptions for the FR19 Model are presented in the graphic at the beginning of this post. The results are somewhat surprising!

I looked at the following possible drivers for the French Economy: (1) None (Business-as-Usual, BAU, and Random Walk, RW), (2) Technical Productivity, TECHP, (3) the World System (W), (4) Western Europe (WE), (4) Germany (DE) and the United States (US). The best model, using the Akaike Information Criterion was US input ([-1215 < AIC = -1138 < -1077]). All the models were unstable (some were cyclical, see the Eigenvalues below in the Notes with the AIC Statistics) and  collapsed (except the RW) at some point in the Early 20th Century. 

At first, you might think that the models are predicting the Great Depression. They are not! We know from the historical record that the French Economy did not collapse as a result of shocks from WWI, the Great Depression and WWII (according to ChatGPT, in any event).

Then, how are we to understand the forecasts from the 19th Century Models? To understand the French Economy in the 19th and early 20th Century, we have to look back further before the first (1878-79) and the second (1848) French Revolutions. I'll do that in future posts.

For this post, let's take the German perspective as embodied in the Schlieffen Plan to invade and neutralize France before the First World War. Assume (counterfactually) that Field Marshall Alfred von Schlieffen had output from the FR19 BAU model (here) or at least understood that France was a collapsing power. The idea of a lightning first strike to neutralize France before fighting Russia on the Eastern front would seem attractive and necessary. 

Of course, everything changed for France after WWI, during the Inter-War Years. France reacted very conservatively to the Great Depression and stabilized the economy (or at least turned it into a Random Walk).

Returning the present Budgetary Crisis in France, it would be reasonable to predict that any fundamental changes would require formation of a New Republic (see history here). A budget crisis is, historically, not significant enough to trigger major political change.

You can stabilize the FR19 BAU model with instructions in the code (here). For more information about data sources and how the models were constructed, see the Boiler Plate. You can follow a Blog Roll of other postings I have done on the French Economy here.

Notes

FR19 AIC Statistics

The Akaike Information Criterion statistics show that (1) all the models were unstable and (2) the best model was input from the US19 BAU model.

FR19 Measurement Model

Three components from the FR19 BAU model explain almost 100% of the variance in the indicator variables. The components are (1) FR1=(Overall Growth), (2) FR2=(0.5633 Q + 0.5979 X - 0.2781 N - 0.4070 HOURS) Malthusian Export Employment Controller, and (3) FR3=(0.75876 XREAL - 0.3849 X - 0.4924 HOURS) Export-Price Employment Controller.

FR19 System Matrix

The FR19 BAU System matrix is unstable with two strong negative feedback components, F[1,3] = -0.05130896 and F[2,3] = -0.05297408, both from the Export-Price Employment Controller (recall that World Trade Collapsed during the InterWar years).