Romer:
Of course, one can also debate the baseline and the question of whether creating or saving 3 to 4 million jobs will be enough to fully heal the economy. But it is important to acknowledge that creating or saving that many jobs would be a tremendous accomplishment...
I think by now it is fairly clear that a proper stimulus would aim at a peak employment boost of 5 to 8 million--for if things break our way and the recovery comes more rapidly than expected, then monetary policy can always cool things down. But if things come out worse than expected then we are in trouble, for monetary policy cannot heat things up.
Romer:
I feel quite confident that conventional multipliers are far more likely to be too small than too large. David Romer and I have argued that omitted variable bias is a rampant problem in estimating the effects of fiscal policy.... Robert... Barro has argued that a reasonable way to estimate the effects of increases in government spending is to look at the behavior of spending and output in wartime. But, consider one of his key observations – the Korean War. If he were using just this observation, Barro would basically divide the increase in output relative to normal by the increase in government purchases relative to normal during this episode. When one does this, one gets a number less than one. From this Barro would conclude that the multiplier for government spending is less than one. But other things were going on at this time that also affected output. Most importantly, taxes were raised dramatically; indeed, the Korean War was largely fought out of current revenues. The fact that output nevertheless rose substantially is in fact evidence that the effects of increases in government spending are very large.... David and I used narrative evidence to isolate tax changes uncorrelated with other factors affecting output.... [D]oing the same kind of narrative analysis for government spending would be very difficult: there are vastly more spending changes than tax changes, and the motivations for them are less easily classified. But the same issue of omitted variables is surely present.... [I]t is likely that conventional estimates of spending multipliers are also biased downward. In estimating the effects of the recovery package, Jared Bernstein and I used tax and spending multipliers from very conventional macroeconomic models.... [I]t is very hard to claim that those are excessively large. Indeed, if you want to know why I am more optimistic than some, it is probably because I believe my own research. I think that both the change in taxes and the change in spending may pack more bang than the official Administration estimates assume...
I agree that the evidence for large multipliers (on the order of 3) in Romer and Romer is striking. But I find it disturbing: I don't see why some of the channels they find are as strong as they are--the tremendously strong investment accelerator, for example. And thus I fear that their estimates too suffer from some opposite omitted variable bias that I am not smart enough to identify.
Romer:
I think it is possible that fiscal policy will have even more oomph in this situation. When households and businesses are liquidity-constrained by reduced lending, any money put in their pockets is more likely to be spent. More fundamentally... recovery in the real economy is salutary to the financial sector. When people are employed and buying things, loan defaults fall and asset prices are likely to rise. Both of these developments would surely be helpful.... This is, I believe, a key lesson of the Great Depression. In the Depression, the end of deflation, renewed optimism, and increased employment and output were as crucial to the recovery of the financial system as the more direct actions taken to stabilize banks. Thus real and financial recovery reinforced each other. So fiscal policy to raise employment may help to restart lending and in that way generate a more durable recovery...
Touche...
Romer:
I think almost everything in the bill is useful stimulus. The AMT patch, while desirable as public policy, is less important for the goal of net new job creation. It clearly prevented a large tax increase from occurring, but it is something that households had most likely been counting on in any event. For this reason, in our own estimates of the jobs effects, we gave it a relatively small multiplier. And there’s a small amount of spending after 2011. CBO estimates... 4%... won’t be spent out until 2012, which is probably fine in terms of having stimulus taper off, and another 5% in 2013 and beyond... having some tail is an inevitable consequence of devoting some of the spending to valuable long-term investment projects...
In other words: given that the headline stimulus package number was going to be fixed, it was a bad mistake to spend it on the AMT patch.
Romer:
I think there’s pretty broad agreement that we have been neglecting public investment in recent decades...additional investment is very likely to have substantial payoffs.... It is hard to think of a better time to make such investments than when resources are underutilized and borrowing costs are low. These investments will not only help mitigate the downturn in the short run and help spur recovery in the medium run, but will also make the economy more productive in the long run.
Indeed, given current Treasury borrowing costs there is a very big argument to be made for a much bigger public investment effort supported by borrowing: more government infrastructure financed by Treasury bonds right now looks to be a free lunch for the country even if not for the government--as long as we can borrow more without having Treasury interest rates go up.
