With the recent claims by the masses that QE2 will not achieve its goal of lowering unemployment and jump starting economic growth, I decided to conduct a simple regression test to see if expanding the money supply does affect at least one of the Fed’s concerns: The unemployment rate.
I collected the yearly change in the money stock (using the M2 seasonally adjusted numbers) and compared it to the yearly change in unemployment to see if there really was any cause and effect relationship here.
In my study, the change in unemployment would be the dependent variable and the money supply would be the independent variable. The data I used dates back to 1960, a period which has seen some rapid swings in monetary policy and the unemployment situation. So I believe it would be good enough with which to formulate some kind of opinion on QE2.
Note: When I refer to % change in unemployment below, I refer to the change in the number of people unemployed in the economy, not to the number of people unemployed as a percentage of the labor force.
Since 1960, there have been many periods of volatility in both data sets. For example, during the 1970s and 80s, when at times the money supply was being raised by 12-13% on a yearly basis to combat worsening economic conditions. After taking the yearly data and plugging it in to Eviews, I was able to generate the following equation: %Change in unemployment=7.963982-0.735844 (%Change in M2) + Ui. This means a 0% change in M2 would result in a 7.96% rise in unemployment. A 1% change in unemployment would result in a 7.23% rise in unemployment. One key thing to note about this regression equation is that it generated an R squared of .020254, indicating the regression line to be very weak and that there is a large number of unobservables that effect the change in unemployment.
Now to test the claim that quantitative easing (raising the money supply) does not have any affect on lowering the number of unemployed people (and thus lowering the unemployment rate), I will construct a two sided alternative hypothesis test, where the null hypothesis is B1 (or the coefficient in front of % change in M2) = 0 vs B1=/= 0. The null in this case means that we are claiming the change in M2 does not affect the number of people unemployed.
To do this, I calculate the t-statistic by taking the expected value of B1, and subtracting it from from the value of the null, and taking that value and dividing it by the standard error of the coefficient (calculated by Eviews to be 0.746512). In this case the t-stat is calculated out to be -0.98571, much smaller than what it would need to be to reject the null hypothesis at the 99%, 95%, and 90% significance level.
So the result of my hypothesis test supports at least one of the claims by many Americans and International politicians and economists that QE2 will not bring down the unemployment rate. Since the t-stat is lower than the critical values of the 99%, 95% and 90% significance level, or 2.58, 1.96, and 1.64 respectively, the null hypothesis is not rejected that a rise in the money supply lowers the unemployment rate, AT LEAST with the data provided in my sample set (the effects of monetary policy in the last 50 years).
Along with that, as mentioned before, the low value of R squared indicates that there is a strong number of unobservables in the model that account for the change in the number of people unemployed rather than the change in the money supply.
Disclosure: No securities mentioned