Monday, January 12, 2009

Worst-case economic scenarios...

Mark Thoma (via Paul Krugman) says:
I think the stimulus package is like driving up an icy hill. If you don't have enough momentum from the start and fail to provide enough "stimulus" to get the car over the crest of the hill, you can slide all the way back to the bottom, crashing into things along the way and ending up worse off than when you started.
Krugman and Thoma have been trying to figure out just how big a stimulus will be sufficient for the economy to pull itself up by its bootstraps, and both are worried that a trillion dollars isn't enough.

Here's a best guess on what unemployment will look like over the next few years with and without the Obama stimulus (from Krugman's blog):

I have very little faith in macroeconomic models of the economy, so naturally I'm skeptical that the data on this graph will have any relationship to reality. I think that we understand the economy about as well as we understand the weather-- we can make reasonably precise predictions about what conditions will be like over the next few days. And we can make some very broad, imprecise guesses about what will happen over the next decade or two. But I think both the weather and the economy are impossible to predict with any certainty in between.

But I'll swallow my doubts, and pretend that the experts will be right. So: should we do nothing, suffer from 9% unemployment for three years, and let the economy recover on its own?

Should we spend a trillion dollars, suffer from 8% unemployment for two years, and risk "sliding down the icy hill" and ending up worse than we were before?

Or should go all-in: spend three (or four or five?) trillion dollars to get to the top of the hill? What if the hill is bigger than THAT? Or what if the Keynesian theory of economic stimulus is wrong?

My brain kind of understands numbers as large as a couple of million dollars. Billions and Trillions are just way too big for me.

So, lets break it down: Lets say we spend an extra trillion dollars to get up the icy hill. And lets be wildly optimistic and assume the unemployment rate immediately drops from the current 7% to the normal, "full employment" rate of around 5%.

Two percent of the US workforce is 3.2 million people.

One trillion dollars divided by 3.2 million people is about $300,000 per person.

Somebody please check my math; are we really talking seriously about spending THREE HUNDRED THOUSAND DOLLARS PER JOB SAVED?!?

I don't have a PhD in economics, but common sense tells me that just cannot be a good idea.
Update: I checked my own math, and I goofed on the first version of this post. It's "only" $300,000 per job saved, not $600,000 as I originally thought.


Joseph said...

Gavin. Good points. Seems wacko. I twittered this link out to the world. Perhaps there will be some response from that crowd.

Jonathan said...

So I don't have a PhD in Economics either, but I think that to look at just the cost of the job losses is to oversimplify things a bit. I think the real measurement of the cost of a recession is the lost GDP growth, which (measured per capita) is one of the best quantitatative measurements of overall standard of living.

US GDP growth in good times is typically in the ballpark of +2%. In the last two bad recessions (1973 and 1981) it has been around -3%. With our GDP around $14 trillion annually, the difference between +2% growth and -3% recession is about $700 billion. This represents the net loss to the US economy in goods and services produced, per year.

Of course, the stimulus package is not going to magically transport us from -3% GDP to +2%, just as Krugman's chart doesn't reduce unemployment to 5% overnight. We're talking about the difference between doing something and doing nothing, not about the difference between the current state of affairs and the ideal state of affairs. But we're also talking about a long period of time. Notice that the chart shows an unemployment advantage for the stimulus plan going out to 2013 or 2014. If you assume a +1% annual GDP increase from the stimulus package ("assume" == "numbers coming out of my butt" - I'm not a Nobel laureate like Paul) over a five-year period, you end up with a net gain for the US economy that is roughly similar to the cost of the plan. Tinker with my 1% estimate and you end up either slightly ahead or slightly behind, but you're definitely in the ballpark, in terms of an overall cost-benefit analysis.

Gavin Andresen said...

You're right Jonathan, I am ignoring standard of living increases for employed people; macro economists like to argue about what the "multiplier" is for government spending; Arnold Kling (my favorite economist these days) posted about that today at EconLog.

So: worst case scenario: we spend a trillion dollars, don't get up the "icy hill" (the "stimulus multiplier" turns out to be less than 1), and we're back where we started, but with a trillion more dollars in debt.

Best case scenario: those of us with jobs are a little bit richer, and 3 million more people are working over the next few years. Oh, and we've got however many more bureaucrats needed to make sure that the mumble-hundred-billion-dollar stimulus isn't being wasted.

It seems to me the just extending and expanding unemployment benefits for those 3-4 million people would be a whole lot less risky. A trillion dollars is enough to pay 6 million people $30,000 per year (the median individual wage) for 5 years straight...

(I'm ignoring structural changes needed to the financial system, because that's separate from the stimulus-- that's what TARP and all the stuff the Treasury/Fed have been doing is all about... not that they've been particularly successful, as far as I can see...)

Anonymous said...

Hi Gavin,

the latest from Paul Krugman addresses this


Gavin Andresen said...

Even if the numbers are "only" $60,000 per job per year (I'll let me anti-government bias show through: when have you EVER heard of a government underestimating how much something will cost...), I still don't see how that's wise.

If you knew nothing about the stimulus and I ask myself:

"Is it a good idea to spend $60,000 per year to put an unemployed person to work?"

... I have to answer No-- because that is more than the average EMPLOYED person makes in a year. You're taxing that average employed person (who makes $30,000 per year) and giving the money to somebody so they can make $60,000 per year?

(or, even worse, the process is so inefficient half or less of the money actually flows to the people you're trying to help)