Monetary Offset in a World with Automatic Stabilizers
Monetary offset has been on my mind for a while now. Scott Sumner (among other market monetarists) are running victory laps (and to an extent, rightfully so) considering the relatively healthy growth last year despite significant fiscal drag. There’s no doubt, as stock and bond market reactions prove, that monetary policy has been helpful. But both the theory and empirics behind a strong and automatic offset – as favored by market monetarists – is weak. I should preface this by noting I’m largely in agreement with the market monetarist argument for nominal income targeting.
Let’s imagine (for now) that the zero bound did not bind – whether that be through the efficacy of unconventional monetary tools, a higher inflation target, or a Herculean ability for the Fed to handcuff its own hands years into the future and convince the market that it threw away the key. Standard economic theory supposes that in this world fiscal policy does not determine the price level given an inflation-targeting central bank arguing that if the government increases its budget and hence aggregate demand, the central bank will increase rates to maintain credibility. Hence government spending cannot decrease unemployment.
If we’re talking about totally discretionary stimulus this may be true. But consider a government that offers generous unemployment insurance (UI) with reemployment credits or guarantees employment (either generally or in a recession). Soon after recession, the government institutes very long UI and, in doing so, increases its primary deficit from 2% to 10%. Let’s say hysteresis effects are minimal and expansionary policies don’t simultaneously increase aggregate supply. Expansionary spending then, by Law of the Excluded Middle, either increases the price level or it does not. Given an upward-sloping supply curve (depressed as the economy might be), the former case is more likely. Monetarists argue that an independent central bank offsets policy in one of the following ways:
- By force of expectation, given its credibility to an inflation target.
- By being more cautious with its stimulus programs (or halting them altogether, depending on relevant magnitudes) than the counterfactual without stimulus or deeper austerity.
It feels like the first point used to be more popular than it is now, given that the Fed has zilch credibility on its inflation target (by definition, if it had any credibility, long term expectations wouldn’t be as low as they are). The second point is pretty fragile given behavioral features, decentralization of central banking decisions, and the need to have a precise ability to estimate price level elasticity of aggregate supply if it is low (which it is in a weak economy).
So after the government promises insurance to layoffs and credits to employers who hire said layoffs the central bank estimates the effect this has on the price level and accordingly decreases the rate at which it purchases assets. This creates a new wave of unemployed workers – that, after all, is the core of monetary offset models – which would require even more deficit spending to finance the promised unemployed benefits. This would require an even greater offset, requiring even more stabilization.
In this case two things can happen. Either the value of a credit default swap on Treasuries increases, as the market starts loosing confidence that we can service future deficits, or prices rise as markets expect the Fed to monetize deficits in an effort to prevent default. In a world where bond yields and CDS values aren’t soaring, the only possible conclusion is that the Fed stops offsetting government spending.
In fact, to the extent the market knows the Fed would never let the government default, the Fed’s offset would be offset by expectations of its future relaxation of its offset. This sounds a lot like the fiscal theory of the price level, and in some sense it is, but the distinction is that there must be some mechanism in place that requires the government to increase its deficit in response to monetary contraction. If there was no such mechanism – i.e. fiscal policy was only a one time, discretionary cash hand out – monetary policy could offset austerity perfectly well. (A helicopter drop of money and cash hand out financed by bond buying is actually the same thing, so offset could be surgically precise, as both Keynesians and monetarists agree). The only way fiscal theories could work in this environment is a government that engages in discretionary policy every time the central bank tightens policy which is unrealistic and, by definition, not rules-based. So the possibility of hyperinflation from ARRA was well, nonexistent.
There are second order effects too. If the interest elasticity of government spending is higher than the interest elasticity of investment (and studies suggest that this is probably the case), much of the benefit from easier money comes from cheap finance to beneficiary governments, reducing net outlays. Therefore tighter policy would decrease both the government’s primary and non-primary balance. This, by the way, is not negligible – the United States may face $75 billion in increased debt servicing to finance the same level of operations.
If the political situation is such that the government may only engage in a certain level of deficit spending (either by law as in Europe or institutional arrangement as in the US) offset would require the government itself to tighten its budget.
The point of the post so far is that monetary offset cannot be as theoretically sound as its proponents make it seem. There are multiple sources of positive and negative feedback, and actual results depend on the precise role of each which itself depends on the complex slew of automatic stabilizers, central bank learning mechanisms, and so forth. However, as outlined above, that economic conditions today resemble that setup seem unlikely given the preponderance of automatic stabilizers.
The empirical case for full monetary offset is stronger, but still wanting. Yes growth was a lot stronger than some Keynesian models suggested. That itself doesn’t mean anything, especially for anyone that (like me) believes in an at least approximate efficient market hypothesis. No model that can predict growth can exist. The question is whether growth today violates the Keynesian story. Perhaps a macroeconometrician will answer this better than I, but frankly the magnitudes don’t justify that explanation either. While fiscal drag was unfortunate, the United States certainly didn’t succumb to the same austerity as Europe and within the margin that it did plenty of other factors, including an improving supply side, can explain strong growth beyond monetary offset. As for Europe, where’s the offset?
Let me end this post with a final example which captures the point of the above reasoning. Imagine the government guaranteed employment at below market wage rates as a primary automatic stabilizer. In a recession, as deficits increase, monetary offset would force a growing number into government employment. The logical conclusion would be a huge deficit and huge government work force, but not unemployment by virtue of the government’s promise to employ. The only way total GDP would be affected would be a decreased output per worker, a supply-side phenomenon because the government makes for a bad employer. But supply-side concerns are not market monetarists’ concern. Is there any model with guaranteed employment monetary offset decreases total employment?
Of course, deficits would never get so out of hand before the central bank stopped offsetting. But even monetarists agree that monetary offset would not increase employment, only government deficit. By virtue of that transfer of liabilities, the private sector is allowed to deleverage which itself increases aggregate demand.
The feedback loops here are just way too complicated for the simple monetary offset story to be true.
Late Addendum: Scott Sumner comments on his blog (in response to another):
I’ve always argued that zero is a sort of benchmark, a starting point in the analysis. If the fiscal stimulus is large enough to bankrupt a country, then for fiscal theory of the price level reasons I’d expect a positive multiplier. In not (i.e. in the US) I expect the multiplier is zero on average, but may be above or below zero for the reasons you indicate. What matters is the expected multiplier, not the actual multiplier, and I see no reason to expect a multiplier that is significantly different from zero. In 2013 we saw about what I expected.
That’s fair enough. But the point here is bankruptcy conditions are non-negligible with automatic stabilizers. Not in general, but certainly if the offset is persistent (that is if the “expected multiplier” remains at or near zero).