
Monetarists blame fluctuations in inflation on excessively volatile growth in monetary aggregates. In 1982, Friedman defined monetarism in an essay on defining monetarism as follows:
Like many other monetarists, I have concluded that the most important thing is to keep monetary policy from doing harm.
We believe that a steady rate of monetary growth would promote economic stability and that a moderate rate of monetary growth would prevent inflation
The U.S. data supported this hypothesis about the volatility of monetary growth and inflationuntil 1982, but since 1983 monetary aggregates have been essentially uncorrelated with subsequent inflation in the U.S.
Levis Kochin pointed out in 1979 that a well designed monetary policy would lead to zero correlation between any measure of monetary policy and subsequent inflation. The reason for this is the correlation between any variable and a constant is zero.
If monetary growth is stable, say, a constant growth rate of 4% per year, as advocated by Milton Friedman, monetary growth will have no correlations with any variable:
Poole (1993, 1994) and Tanner (1993) also argue that one predictable consequence of optimal monetary policy is that the correlation between monetary policy instruments and policy goals will be driven to zero.
Poole further contends that it is obvious to any careful reader of Theil (1964) that optimally variable policy will give rise to a zero correlation between policy and goal variable…
In 1966 Alan Walters, a U.K. monetarist, observed:
If the [monetary] authority was perfectly successful then we should observe variations in the rate of change of the stock of money but not variations in the rate of change of income… [a]ssuming that the authority’s objective is to stabilize the growth of income.
Milton Friedman in 2003, wrote about how the Fed acquired a good thermostat:
The contrast between the periods before and after the middle of the 1980s is remarkable.
Before, it is like a chart of the temperature in a room without a thermostat in a location with very variable climate; after, it is like the temperature in the same room but with a reasonably good though not perfect thermostat, and one that is set to a gradually declining temperature.
Sometime around 1985, the Fed appears to have acquired the thermostat that it had been seeking the whole of its life…
Prior to the 1980s, the Fed got into trouble because it generated wide fluctuations in monetary growth per unit of output. Far from promoting price stability, it was itself a major source of instability as Chart 1 illustrates.
Yet since the mid ’80s, it has managed to control the money supply in such a way as to offset changes not only in output but also in velocity.
Nick Rowe explained the difficulty of causation and correlation under different policy regimes and Milton Friedman’s thermostat superbly as an econometric problem Nick Rowe:
If a house has a good thermostat, we should observe a strong negative correlation between the amount of oil burned in the furnace (M), and the outside temperature (V).
But we should observe no correlation between the amount of oil burned in the furnace (M) and the inside temperature (P). And we should observe no correlation between the outside temperature (V) and the inside temperature (P).
An econometrician, observing the data, concludes that the amount of oil burned had no effect on the inside temperature. Neither did the outside temperature. The only effect of burning oil seemed to be that it reduced the outside temperature. An increase in M will cause a decline in V, and have no effect on P.
A second econometrician, observing the same data, concludes that causality runs in the opposite direction. The only effect of an increase in outside temperature is to reduce the amount of oil burned. An increase in V will cause a decline in M, and have no effect on P.
But both agree that M and V are irrelevant for P. They switch off the furnace, and stop wasting their money on oil.
Subsequent work of Levis Kochin showed that if the effects of fluctuations in monetary aggregates were not precisely known then the optimal policy would produce negative correlations between monetary aggregates and inflation:
The negative correlation results from coefficient uncertainty because the less certain we are about the size of a multiplier, the more cautious we should be in the application of the associated policy instrument.
Therefore, although optimal policy leads to lack of correlation between the goal and control variables if the coefficient is known, it will lead to a negative relationship if there is coefficient uncertainty. The higher the uncertainty, the more cautious will be the optimal policy response. Also, if the control variable can’t be controlled perfectly then the correlation between the goal and the control variable becomes positive i.e., the control errors are random…
Uncertainty about the impact of a policy will stay the hand of any bureaucrat , much less a central banker, as Kochin and his co-author explain:
Uncertainty should lead to less policy action by the policymakers. The less policymakers are informed about the relevant parameters, the less activist the policy should be. With poor information about the effects of policy, very active policy runs a higher danger of introducing unnecessary fluctuations in the economy.
Jan 30, 2015 @ 09:46:49
Jim,
I am unsure why monetarism is/was a success.
LikeLike