Thursday, February 17, 2022

Public Debt vs M2 Velocity

Maybe debt matters after all. A little at first is OK, even a lot. But eventually, there appears to be an impact on the economy. Correlation is not causation, but if there is causation here, the trend is not our friend. It's a song of fire and ice, and winter is here.

Based on a simple linear regression, velocity should slow as the debt continues to grow. At 50 trillion in debt, the predicted M2 velocity is 1.0129424.

Here is what M2V/Public Debt looks like with a linear regression line (produced with scikit-learn and matplotlib).

I tried various regression models but many of them stopped making predictions outside the training data. One that didn't was Gaussian process regression. That one predicted that M2 velocity would go negative at 35 trillion in debt. I am not sure if a negative velocity makes sense in the real world. There are a half dozen hyperparameters for Gaussian process regression and I don't understand it well enough to have a high confidence in the predictions.