Sunday, November 30, 2008

The Second Coming of J. Maynard Keynes

by the Sandwichman

It's official. Big fiscal stimulus package is the flavor of the day! Let's get fiscal, fiscal. I wanna get fiscal. Let's get into fiscal. Let me hear your money talk, your money talk, Let me hear your money talk.

Sandwichman was wondering, though, "how do they get from $X billion of fiscal stimulus to X million jobs created?"

Dean Baker must have been reading my mind. "Assuming that employment is roughly proportional to output..."

Is that all there is?

Percent of GDP in additional government spending times multiplier equals percent increase in GDP and -- assuming an employment effect roughly proportional to output -- equals percent increase in employment?

Is that actually how economists estimate the employment effect of a fiscal stimulus package? Because if it is, the old Sandwichman has news for you economist folks:


But you already knew that, didn't you?


Barkley Rosser said...

Well, I suppose if one starts monkeying around a whole bunch with how many hours people work, but it is not all that bad of a rough correlation, although with some lags (firms are slow to lay off and then slow to rehire).

Sandwichman said...

O.K., Barkley, you and I both know that "roughly proportional" and "rough correlation" don't mean the same thing. Also I don't need to recite to you the bit about correlation not implying causation (which I just did).

But, obviously, if economists assume that $x billion of fiscal stimulus will create y million jobs there must be scads of robust studies confirming such an effect. Scads of 'em. Am I right? So, setting aside the suppositions and the rough correlations -- WHERE'S THE BEEF?

The Congressional Budget Office (Peter Orszag, director) issued a paper last January assessing "options for responding to short-term economic weakness" and in it they included a table describing the cost-effectiveness, length of lag from enactment to stimulus and uncertainty about policy effects for various policies. Investing in public works projects was described as having small cost-effectiveness, a long lag from enactment to stimulus and small uncertainty about effects. Providing aid to state and local governments: medium cost-effectiveness, medium lag, large uncertainty. Extending or expanding unemployment benefits: large cost effectiveness, short lag, small uncertainty, etc. etc.

Would the employment effect of any and each of those various stimulus policies be roughly proportionate to its effect on output, regardless of its other characteristics?

I think I can anticipate a short-cut answer to my question. Since "output" is measured by income in the GDP, somebody's got to be receiving that extra income and we're just assuming that they are doing so through the agency of a new job. So a 5% increase in aggregate income implies a "roughly proportional" 5% increase in employment.

But aren't there other, unstated, assumptions underlying that roughly proportional notion? Doesn't it assume that the qualifications of the unemployed approximately fit the skills being sought by the new program spending? Doesn't it assume that the average wage and intensity of employment in those new jobs are roughly equivalent to the average in the current employed labor force?

But surely economists have asked these questions and they've done the studies and there are some canonical econometric studies measuring the employment effects of fiscal stimulus packages. I'm just asking for somebody to point me to 'em. said...

Actually I have not seen any projections of what the job impact of the proposed stimulus will be, aside from the desired increase of 2.5 million jobs that Obama has requested. But that is not the same thing as a projection.