It seems my colleague Stephen Grenville is somewhat sceptical of quantitative easing (QE). He says 'QE might have been an admirable second-best policy, but it was still "beggar-thy-neighbour".'
I agree that QE (where the Federal Reserve purchased long term government bonds along with securities backed by mortgages) was second-best policy, but for different reasons. I think negative interest rates would have been better. But I strongly disagree that QE was 'beggar-thy-neighbour'.
In fact I think QE was quite effective.
My thinking on the topic has been strongly influenced by a paper released by some of my old colleagues at the Federal Reserve. In the paper they evaluate the effects of quantitative easing and the forward guidance provided by the Fed (forward guidance is when the Fed signals it will keep interest rates at zero). The authors of the paper estimate that these policies subtract 1.25 percentage points from the unemployment rate and add 0.5 of a percentage point to inflation. That's quite impressive! [fold]
These estimates come from a large macro-econometric model of the economy called FRB/US. I worked closely with the model when I was at the Fed. The way the authors have estimated the effects of forward guidance is by changing the interest rate expectations of the private sector, keeping them lower than otherwise, which is expansionary. The effects of QE are modeled as pushing down the long term Treasury bond interest rate, which in the model positively affects stock market valuations and pushes down the corporate bond yield and the cost of borrowing for households. And yes, it also lowers the exchange rate, but it is only one channel of many that QE operates through.
It is important to emphasise that the strength of the relative channels of monetary policy are based on estimated relationships from past experience. These things are not just made up.
Now, the authors do not break out how much work is done by the exchange rate, but it beggars belief to think that the exchange rate response would account for much of that 1.25 percentage point decrease in unemployment in a relatively closed economy like the US (trade as a percentage of GDP is around 30%, which is low). Most of the work is being done by other channels, such as easier borrowing conditions for households and firms. QE and forward guidance is not beggar-thy-neighbour because most of the effect does not work through the exchange rate. QE likely has positive spillovers, as the IMF found. Quiet, sceptics!
Should we be cynical about the apparently large effects of QE?
I say 'no'. Sure, the effects are uncertain, and there should be largish confidence bands associated with these guesses. The model is not perfect after all, but I certainly think it is informative. These models are, in fact, invaluable for policy making. Without them, we would have no idea about the effects of many policy changes (economists call these policy changes 'multipliers').
Moreover, I'm confident that the guesses represent the best guesses of the authors. There would have been no political interference here to get answers that would make the Fed look good. Indeed, there are some messages in here that imply QE was not implemented optimally. For example, they state that the Federal Reserve's actions 'apparently provided only a small boost to the real economy during the recession and the initial recovery period'. Rather they find that the largest effect is happening this year, and next, well after the worst of the recession has passed.
When I worked at the Fed, there was never any trace of interference in the independent research of the staff – if there had been it would have been fiercely and stridently resisted. No, the large effects were not found in order to make QE look good. Rather, the large effects in these and other calculations likely convinced people in the Fed to apply QE with the force they did.
Photo courtesy of Flickr user International Monetary Fund.