Reviewing “Modelling Bitcoin’s Value with Scarcity” —Part II: The hunt for cointegration

Does OLS regression of natural logarithms of bitcoin price and stock-to-flow ratios result in spurious regression, or are we dealing with the exceptional case of cointegration?

In my first review of the work of PlanB, I concluded that the relation between stock-to-flow and bitcoin price as pointed out by the author was invalid because the general assumptions of ordinary least squares regression were not met. When two variables are non-stationary and we estimate a regression model, there is a good chance we find highly autocorrelated residuals and a significant value for the coefficient. This phenomenon is well known as spurious regression. But, spurious regression isn’t always the case. Sometimes the variables might be cointegrated, which would imply that the estimated relation is super consistent. Another review by Nick pointed out that in this specific case we could be dealing with the exceptional case of cointegration. For a better understanding of cointegration, I would recommend to have a look at a very good visual introduction of the concept here.

In this article, I will investigate if the log of bitcoin price and the log of its stock-to-flow ratio are indeed cointegrated. If cointegration applies, it turns out that the OLS estimates of the coefficients are consistent. If this is the case, I would have to reject my…