When I discuss the effect of monetary stimulus on aggregate demand with other economists, I notice that they often want an explanation couched in terms of the major components of GDP. I find this very frustrating, as this approach does more to conceal than illuminate. Suppose you were policy czar in a liquidity trap (such as right now), and you were asked to increase nominal GDP by 3-fold (i.e. 200%) in the next five years. If you were given a choice of only one tool, which would it be–monetary or fiscal policy? Any economist with an ounce of common sense would take monetary policy. OK, so how would you explain its effect in terms of the 4 components of GDP?
One might object that this isn’t a fair question, such a rapid increase in GDP can only occur through inflation, and in that case the classical model applies–money (and velocity) determines the price level in the long run. OK, so lets go as far from the classical model as possible. You are at the lowest level of output (relative to trend) in American history, i.e. March 1933. What policy tool do you choose? FDR used both monetary and fiscal policy, although initially it was monetary policy that had the greatest effect.
So let’s look at FDR’s most effective stimulus–dollar devaluation. How does one explain its effect using the famous four components of GDP? One might have expected a sharp increase in the NX component, as a lower dollar boosted exports and discouraged imports. Output did soar after the dollar was devalued, but it wasn’t because of a rise in NX, rather it was despite a sharp fall in NX. Imports rose much faster than exports, as the “income effect” outweighed the “terms of trade effect.”
There are many ways that one could model the transmission mechanism between dollar devaluation and rising output. The devaluation certainly raised future expected prices. This probably reduced real interest rates, which is a very new Keynesian way of looking at things. But it also directly impacted commodity prices in two ways, the obvious PPP effect, but also the fact that even in a closed economy that was expected to eventually return to a gold-dollar peg, a higher nominal price of gold will increase the expected future price of commodities, and thus their current price as well. Furthermore, the resulting inflation will reduce the “debt-deflation” problem that was plaguing the economy.
In my view the most important transmission mechanism was real wages, which fell very sharply in the first four months of dollar devaluation. One piece of evidence that real wages were important shows up in July 1933, when a policy of sharply higher wages (the NIRA) aborted a promising recovery in monthly industrial production after only 4 months.
Of course this is just one example, but it is also one of the few monetary shocks that we can identify with any confidence. Now let’s look at another well-identified shock, the biggest year-over-year fall in the monetary base in modern American history, 1920-21. The sharp fall in the base caused the sharpest 12 month deflation in modern American history between 1920-21. And it also caused the sharpest one year increase in real wages in modern American history between 1920-21. And real output plummeted. What does the C+I+G+NX approach add to this story? Nothing. Of course investment usually falls more sharply than consumption in a depression, but that would be true almost regardless of what caused the depression.
Between October 1929 and October 1930 the same thing happened again, to a lesser extent. The monetary base fell significantly, the price level fell, and real wages rose sharply. Higher real wages made it less profitable to produce all sorts of goods–both consumption goods and investment goods. Economists often flounder around seeking the mysterious cause of the drop in AD after late 1929. Did consumers suddenly stop spending? Or did a change in animal spirits hold back investment? The answer is much simpler, as with any decline in nominal spending either the monetary base declined, base velocity declined, or both. In 1930 it was both. The various components of GDP will respond in different ways to the lower nominal spending under different conditions, but they don’t add any explanatory value.
Macroeconomics should be about aggregates, not components of spending. Yes, changes occurring in the various components of GDP can impact interest rates, and thus velocity. And if monetary policy is inept (i.e. doesn’t offset changes in velocity) that can impact nominal spending, but it certainly isn’t the most illuminating way of looking at the issue. It’s like trying to explain changes in the overall price level by modelling changes in the nominal price of each good—theoretically possible, but a waste of time.