reader economicalresponding to my post The picture of M2 divided by real GDP and the GDP deflator severely criticizes my use of traditional monetary aggregates:
M2? This is a sadly dishonest post – those who see the value of QTM use the divisia index as a monetary aggregate. you know this.
I understand the theoretical basis for using divergent monetary measures. However, I am not aware that the use of these measures validates the quantity theory of money in a robust way.
I restate the alleged Figure 1 Offending postadd M2M and MZM (Figure 1 and Figure 2, respectively).
figure 1: Log M2 divided by real GDP (blue), log GDP deflator (tan), log M2M divided (pink), all Q1 2000=0. Recession peak-to-trough dates as defined by NBER are grayed out. Source: BEA, Fed via FRED, Center for Financial StabilityNBER, and author’s calculations.
figure 2: Log M2 divided by real GDP (blue), log GDP deflator (tan), log MZM divided (red), all Q1 2000=0. Recession peak-to-trough dates as defined by NBER are grayed out. Source: BEA, Fed via FRED, Center for Financial StabilityNBER, and author’s calculations.
Now, indeed, if one uses MZM or M2M instead of M2 in the maximum likelihood estimation of the co-integration vector for the period 1968-2023, one can find more evidence of co-integration (10% msl rejected using trace and maximum) eigenvalues test, asymptotic critical value, and allowed cointegration vector and intercept in VAR), but the sign of the price coefficient is wrong. So, I’m happy to use Divisia’s exponents instead of traditional exponents, but I’m not sure you’re resurrecting an empirical version of quantity theory by doing so.
For completeness, I plotted the velocities of M2 (blue), MZM (red) and M2M (pink).
image 3: Log nominal GDP divided by M2 (blue, left scale), divided by M2M divisor (pink), divided by MZM divisor (red), all Q1 2000 = 0. Recession peak-to-trough dates as defined by NBER are grayed out. Source: BEA, Fed via FRED, Center for Financial StabilityNBER, and author’s calculations.
