Scott Sumner  

The new "voodoo"

Working Around FDA Regulation... Open Borders Philanthropy Bleg...

Here is Arnold Kling:

John Cochrane writes,
Economics is a work in progress. But it is certainly brand-new, made-up-on-the spot economics, designed to buttress policies decided on for other reasons.

He is describing the economic analysis that claims that policies to distort labor markets to try to increase wages will increase aggregate demand, so that instead of reducing employment these policies will raise employment.

I am reminded of the made-up-on-the spot economics of the Laffer Curve, which claimed that cutting taxes would reduce budget deficits. That became known as "voodoo economics."

I'd say this is the just the tip of the iceberg. Old hydraulic Keynesianism from the 1960s was already a pretty implausible model. But what's happened since 2009 involves not just one, but at least five new types of voodoo:

1. The claim that artificial attempts to force wages higher will boost employment, by boosting AD.

2. The claim that extended unemployment benefits---paying people not to work---will lead to more employment, by boosting AD.

3. The claim that more government spending can actually reduce the budget deficit, by boosting AD and growth. Note that in the simple Keynesian model, even with no crowding out, monetary offset, etc., this is impossible.

4. More aggregate demand will lead to higher productivity. In the old Keynesian model, more AD boosted growth by increasing employment, not productivity.

5. Fiscal stimulus can boost AD when not at the zero bound, because . . . ?

In all five cases there is almost no theoretical or empirical support for the new voodoo claims, and lots of evidence against. There were 5 attempts to push wages higher in the 1930s, and all 5 failed to spur recovery. Job creation sped up when the extended UI benefits ended at the beginning of 2014, contrary to the prediction of Keynesians. The austerity of 2013 failed to slow growth, contrary to the predictions of Keynesians. Britain had perhaps the biggest budget deficits of any major economy during the Great Recession, job growth has been robust, and yet productivity is now actually lower than in the 4th quarter of 2007.

Arnold is right about the similarity between this and the old supply-side voodoo (especially item #3), but I'd say this is even worse. Supply-side economics was taken to extremes by some of its more fanatical proponents, but was ultimately based on sound economic principles---incentives matter. The new demand-side voodoo represents (in part) a denial of many of the most basic tenets of economics. As recently as 10 years ago, New Keynesians would have scoffed at the list above. Their current willingness to adopt these heterodox views represents a triumph of wishful thinking over hardheaded reasoning.

Comments and Sharing

COMMENTS (19 to date)
TravisV writes:

Hmm, I'm not at all sure that case #4 is "voodoo," at least the way it's written. Below is a good new post by Benjamin Cole citing Ed Yardeni on that topic:

TravisV writes:

Whoops, apologies, I just saw you already addressed this in the comments section at TheMoneyIllusion. You wrote "Travis, I don’t quite see how excess capacity could lower productivity growth, if the employment rate is rising."

That's an excellent point, thank you very much.

Rajat writes:

Kocherlakota seems to have made the 4th argument recently:

Kevin Erdmann writes:

On #4, isn't productivity held back by lack of demand because quits have been low, reducing matching and risk taking in the labor market?

Matthew Waters writes:

If there is monetary offset, then no argument is really good for fiscal stimulus. Monetary policy will already increase AD. If some government spending should be done, then it should be done through taxes rather than debt. So all 5 arguments are invalid with monetary offset.

If you DO have a simple Keynesian model at the zero bound, then #4 is true and #2 could be true.

For #4, there are productivity losses from lack of accruing experience with slack in the labor market. A high unemployment environment also probably does a worse job matching employees to where they would produce most.

For #2, unemployment benefits increase supply-side unemployment, but it's counter-weighted by higher AD. IF deficit spending increases AD at the zero-bound, then it could increase employment. #2 is much more marginal than #4, I agree.

#1, 3, and 5 are all trash. No disagreement there.

Matthew Waters writes:

Also, the real voodoo of supply-side arguments is that tax cuts increase AD. Most laymen arguments for supply-side economics either use increased AD ("job creators") or fantastically overstate elasticity with respect to overall tax rates.

Tax rates in subsectors, such as health insurance, have very high elasticity, with ready substitutes. But most people don't change their behavior much for 35% or 39.6% top marginal rate.

Scott Sumner writes:

Everyone, I don't doubt that one can invent stories where AD affects productivity, but that's certainly not the standard model. One can also invent stories where tax cuts boost revenue. Arthur Laffer did invent such a story.

As I recall, productivity did well during the 1930s. Why? If falling AD hurts productivity, then shouldn't productivity have done very poorly
during the 1930s? Or is my memory off on the productivity numbers?

B Cole writes:

My anecdotal evidence from running a small factory was that productivity went up when output went up. Some of this was reduced unit costs, or overhead labor costs that get spread among a greater number of produced units. Also, it made sense to buy equipment when there was a good backlog of orders.

Yardini also raises the point that productivity is slowing down globally. Thus slumping productivity is not something structural in some national economies but rather something ubiquitous and global and that would be demand.

Trevor Adcock writes:

You do realize that the Great Depression had some of the fastest productivity growth in US history, B Cole?

Unemployment is below 5% in the US right now. So, I'm not sure why you would bring up demand at all. Do you think demand was permanently too low during the classical gold standard, because they had no inflation then?

Brian Donohue writes:

Great stuff, Scott!

Benjamin Cole writes:

Trevor and Scott Sumner:

This seems like a sensible article. The 1930s saw okay productivity growth.

Not sure what that means. There may have been in the 1930s the wider introduction mass manufacturing--big factories--and related technologies that boosted productivity, even into limp demand. Maybe productivity would have exploded in the 1930s, except for the Fed suffocating AD.

So instead productivity exploded in WWII--the amount of aircraft and Liberty ships etc pumped out was amazing.

"Permanente Metals Corporation (Kaiser) No.2 Yard in Richmond, California won the competition. The keel for the SS Robert E. Peary (Liberty Ship) was laid at 12:01 AM on November 8, 1942 and 250,000 parts weighing about 14,000,000 pounds were assembled in 4 days, 15 hours and 29 minutes. On November 12, 1942, she was launched."

Imagine building a cargo ship in four days! The first such ship took 150 days to build.

But we are talking about the world now. Take the auto industry, it is operating about 70% capacity. That reduces productivity, if only due to overhead labor.

People invest in plant and equipment when sales are robust. Monetary suffocation is not the way to higher productivity or prosperity.

Thomas B writes:

Matthew Waters writes, "most people don't change their behavior much for 35% or 39.6% top marginal rate".

I guess it's a question of how much is "not much". If a 35% marginal rate causes people to change their behavior by some amount (call it A) then a 39.6% rate will cause them to change it by 25-30% more than A. If A is small, then 1.25-1.30 x A may be "not much". But does anyone really think that the response to a 35% tax rate (i.e., A) is small?

How do we know it's 25-30%? Recall high school math. If you have a curve that starts on the x axis, goes above the x axis, and later again touches the x axis, the lowest term in a polynomial approximation to the curve will be the squared term. Here, we have such a curve: if the tax rate is zero, there is no economic activity, and if it is 100% there is no economic activity, and economic activity is positive in between. So, to a first approximation, economic activity varies as the square of the tax rate. Now, this is too simple, but only in the sense that economic activity must vary at least as the square of the tax rate. (.396/.35)^2=1.28, so the 25-30% range is pretty safe as a lower bound.

Now, it's true that people in mid-career can't cut their work hours by that much: they either stay in their job, or they don't. So, Matthew may expect that behavior doesn't change at all, but that's not likely to be true. Mid-career people have already invested in their specializations, but young people haven't.

Career development is like capital investment: once you've made the investment, there's not much you can do about an increase in the tax rate - but if you haven't made the investment yet, you certainly take the tax rate into consideration. Young people don't have to take up relatively unpleasant, but socially important jobs; they can pursue personal fulfillment on other, untaxed, dimensions instead. Those decisions have large, long-lasting consequences for them - and for everyone. Remember, once having raised the tax rate and undermined people's incentives for career investment, you can lower it again but many of those people will never be able to fully reverse course either. You'll have to wait for a new crop of young people to respond to the improved incentives - and even then, they'll know that taxes were high in the past and may be again. Mistreatment is not quickly forgotten.

ThaomasH writes:

I think the error made is not to specify the conditions under which each of these propositions could be true. For example with #5, if the additional investment is in projects with NPV greater than zero and some project inputs are not fully employed (their marginal cost is less than their market price. [I do not take "not at ZLB" as synonymous with "optimal monetary policy. For example today we are not at the ZLB, but the Fed has still not returned the price level to its pre-crisis trend, not to mention the NGDP level.] Granted that with optimal monetary policy the NPV rule of investment increase AS, not AD.

baconbacon writes:
So instead productivity exploded in WWII--the amount of aircraft and Liberty ships etc pumped out was amazing.

Productivity isn't "number of things built" in economics, it is the value of things built.

Benjamin Cole writes:


Agreed, mostly.

Although, in the time and place, the Liberty ships were worth their weight in gold 1000 times over. The value was incalculable.

I think it is safe to say productivity exploded in WWII.

Consider how many bombs it would take to flatten Hiroshima at the start of the war, vs. the end.

By any measure, American workers produced more during WWII than during the Great Depression.

I am not advocating government make-work. I am advocating central banks stop asphyxiating global economies.

José Romeu Robazzi writes:

In a simple model where capital is fixed cost and labor is variable cost, boosting AD leads to no labor productivity increases, but leads to capital productivity increases.

This is another way to put that boosting AD actually benefits vested interests (capitalists with capital structures in place) and raises profits, but since capital structure does not change (improve), probably no productivity increases happen...

LK Beland writes:

About productivity during the great depression: TFP fell by nearly 20% from 1929 to 1933.

This was largely the case all across the world (see page 51):

Stefano Fiore writes:

The Laffer curve is not voodoo and it is empirically proven MANY times during the history in diffrent countries... especially in Eastern Europe in recent years.

Jason writes:

[Comment removed pending confirmation of email address. Email the to request restoring this comment. A valid email address is required to post comments on EconLog and EconTalk.--Econlib Ed.]

Comments for this entry have been closed
Return to top