Two extreme models of the macroeconomy have recently been discussed on the internet.  On the extreme right you have Casey Mulligan, arguing that demand shocks can’t explain high unemployment, and that our economy is faced with a supply problem.  If workers cut wage demands we could return to full employment.  On the extreme left people like Paul Krugman argue that wage cuts won’t boost employment at all; instead they will simply depress AD even further.

In the middle is the sensible AS/AD model, with downward-sloping AD curves and upward sloping AS curves.  The textbook workhorse.  This is the model that accurately describes the US economy.  If AD declines, output will also decline (due to the upward sloping SRAS curve.)  If workers accept pay cuts, then SRAS will shift right and output will increase.

Neither the Mulligan nor the Krugman models are consistent with the empirical evidence.  We know that autonomous increases in wages do reduce AS and employment, as with the NIRA.  So we can’t assume that the effect of wages on AD (if any), will offset the effects of wages on AS.  Indeed the opposite is likely to be true.  Imagine a central bank targets inflation.  Now lower wages and shift AS to the right.  What happens?  Prices would fall if the central bank did nothing, so it will increase the money supply enough to shift AD to the right by enough to maintain stable prices.  Of course real world central banks do not succeed in hitting their inflation targets precisely (as I showed in my previous post.)  But they certainly pay enough attention to inflation to prevent an increase in AS from actually decreasing output.

Mulligan’s approach is also inconsistent with the empirical evidence.  When there is a large exogenous negative monetary shock, which reduces NGDP, you will observe a rise in the rate of unemployment, rather than simply a fall in the price level.  If labor markets are flexible, output will recover within a year or two (as in 1921-22.)  If wages are sticky, output will recover much more slowly (2009 – ???)

Both Krugman and Mulligan are partly right.  More monetary stimulus really would reduce unemployment (at least if it succeeded in raising NGDP.)  Krugman’s right about that.  And cutting the minimum wage and reducing maximum UI benefits really would reduce unemployment.

A competent government would do all of those things.  An incompetent government would do none of them.  Which one are we?

PS.  I haven’t actually read enough of Mulligan to know that he thinks demand stimulus wouldn’t help.  He says Keynesian theory should be discarded, which I took as meaning demand-side theory should be discarded.  But perhaps he’s just attacking the more extreme Krugmanian version of Keynesian theory, as Tyler Cowen argues.  If someone finds a link where he supports demand stimulus I’ll add an update.

Karl Smith has a new post on sticky wages that makes the following claim:

So growing up in the New Keynesian paradigm I learned that sticky wages don’t make sense because the real wage is pro-cyclical. That is, in contrast to Keynes’s suggest labor is actually cheaper during recessions than during booms.

This moved the story to sticky-prices.

Not so fast.  It’s true that Keynes’s original model implied real wages should be countercyclical.  But that’s not the implication of the standard AS/AD model, with sticky wages.  Instead, the AS/AD model predicts that wages will be countercyclical when the economy is hit by demand shocks, and procyclical when the economy is hit by productivity shocks.  And a paper I wrote with Steve Silver (Journal of Political Economy, 1989) showed that this is the case.  Real wages are strongly countercyclical in the face of demand shocks, and strongly procyclical in the face of supply shocks.  I’d add that real wages rose especially sharply in some of the most easily identifiable adverse demand shocks (1920-21, 1929-32, 1937-38.)  Most researchers simply look for “the” cyclicality or real wages, just like they look for “the” fiscal multiplier, and lots of other unicorn-like parameters.

Since that time I’ve come to believe that “inflation” is pretty much a meaningless concept, just a number pulled out of the air by those who calculate price indices in Washington.  But I’m in the minority, most economists watch core inflation very closely, even though it is 39% housing, and even though the CPI showed housing prices rising during the greatest housing price crash in American history–even in relative terms.  To each their own.  In any case, I look at the ratio of the nominal hourly wage rate to per capita NGDP, which seems like a better indicator of how nominal shocks effect the labor market.

In the comment section to the previous post, many people don’t quite understand how strong the evidence is for nominal wage stickiness, and how big a problem it is for New Classical models.  Yes, one can explain some wage stickiness in a New Classical model.  And yes, one can explain why workers might be reluctant to accept real wage cuts.  But the real problem is the distribution of NOMINAL wage adjustments.  The sharp discontinuity at zero percent is extremely hard to explain.  The basic problem is that in New Classical models nominal variables shouldn’t matter, except perhaps to the extent that there might be menu costs to adjusting wages and prices.  But the problem is that given that wages are being adjusted, there is no difference between the menu cost of raising someone’s wage 1% and cutting someone’s wage by 1%.  The number zero doesn’t have magical, talisman-like qualities in New Classical models.  But in the real world it apparently does.

Could it be money illusion?

This is a response to Tyler Cowen.  Take a simple example.  Nominal wages are fixed for the employed.  NGDP falls 5%, and 5% of workers are laid off.  Now the unemployed workers lower their wage demands by 20%.  Why not by even more?  Because of minimum wage laws, unemployment insurance, fear of loss of prestige, etc.

Suppose companies are not worried about workers making invidious comparisons (a big if, but I’ll grant this point to my opponents.)  In the best case scenario firms lay off 4% percent of their workers and hire back the 5% who are unemployed at the same total wage bill.  The excess unemployment is now 4% instead of 5%.  The total unemployment rate falls from 10% to 9% (assuming 5% is the natural rate.)  No big deal, we are still deep in recession.  Thus wage flexibility among the unemployed doesn’t really help very much.  If all employed workers accepted a 5% pay cut (or if the government ordered such a cut) and the Fed kept targeting inflation, we’d experience rapid economic growth.  BTW, I’m not advocating an incomes policy, I favor monetary stimulus.

Now you might claim that this process would keep repeating, and eventually we’d reach full employment.  But that would violate the assumption that wages are sticky for the employed.  I.e. could firms really say “You are all fired; now we’ll hire you back for 20% less?”  I suppose so, but then there’d be no wage stickiness for the employed, and we have lots of evidence that there is wage stickiness for the employed.  Indeed one piece of evidence (this pay increase dispersion graph from a Paul Krugman post, plus quotation) is so overwhelmingly persuasive that it should be considered, by itself, a 100% conclusive refutation of New Classical economics.  There is no possible New Classical explanation for this graph.  None.

So as I see it, we’re in a state of censored wage deflation; underlying forces are trying to make wages fall, but thanks to the combination of dispersion and rigidity actual average wages are still rising slowly.

And that’s all we need, empirical evidence that wages are sticky for the employed.  It is a bit puzzling that wages are so inflexible.  But remember that 2008-09 wasn’t the only adverse nominal shock.  We had another in 2009-10, and then another in 2010-11.  It’s possible that wages have now adjusted to the 2009 shock (although I doubt they’ve fully adjusted) but haven’t adjusted to the more recent adverse shocks.  And of course there are also lots of adverse supply shocks that are slowing the adjustment.  Adverse supply shocks partly caused by the adverse demand shock.  That’s why the sticky wage hypothesis seems implausible when looking at the world from a micro perspective—it is implausible from a micro perspective.  Lots of the problem is supply side, when viewed from a micro perspective.   But macro can’t be understood at a micro level.  For instance there is no way I can explain to someone at the micro level why if the Fed buys a $1,000,000 T-bill from me for $1,000,000 in cash it will create more AD.  But it will.

When NGDP growth falls 8% below trend, revenues also fall 8% below trend.  Or much more than 8%, if you are a state like California with a fiscal regime all leveraged-up like a hedge fund.

But public employee wages and pensions do not fall 8% below trend, creating massive fiscal deficits.  One solution is to have the Fed bring NGDP all the way back to trend.  That won’t happen, and indeed at this point I’m not sure it should.   What did happen is that after mid-2009 NGDP growth continued to fall further and further below trend, worsening the fiscal situation of state governments.  Inevitably, you eventually end up with the following:

Still, discussions about something as far-reaching as bankruptcy could give governors and others more leverage in bargaining with unionized public workers.

“They are readying a massive assault on us,” said Charles M. Loveless, legislative director of the American Federation of State, County and Municipal Employees. “We’re taking this very seriously.”  [New York Times]

That’s right, 2 1/2 years after NGDP started plunging, states are “readying” an attempt to bring their costs in line with falling NGDP.  Still think wages aren’t sticky?

The Times also had this interesting tidbit:

Meet the new banking system, same as the old banking system

And then there is this article:

Oregon was one of the few to buck that trend. Last January, voters approved a temporary increase in taxes for individuals making more than $125,000 a year and on couples whose income exceeded $250,000. An editorial in The Wall Street Journal later stated that these rates caused thousands of upper-income residents to flee the state, but studies revealed that a large majority simply made less money, and so fell beneath the income cutoff for the higher rates.

If I was a twenty-something high tech worker in Portland making 60k to 80k, I’d certainly expect to make over 125k at some point in my career.  Thus I’d move up to Washington where there is no state income tax.    And I’d do it now, not when in my 40s with kids in the public schools.

Another example of how income distribution data is misleading.

Here’s Tyler Cowen:

No one — and I mean no one — has a coherent story about how nominal stickiness of wages lies at the heart of our current dilemma.

Good news:  I’m now a celebrity with an instantly recognizable face.

Bad news:  About that face . . . maybe a Bernanke/Krugman/Stiglitz beard would give it more gravitas.

Good news:  Yes, I’m incoherent, but the phrase “and I mean” suggests I’m the least incoherent of a bad lot.

Bad news:  He’s right, it is hard to make a sticky wage theory plausible—real wages aren’t particularly countercyclical.

Good news:  But not impossible.  Forget inflation and real wages; the key is NGDP.  If NGDP growth falls faster than wage growth, mass unemployment is almost inevitable.

I’ll do a more serious response to Tyler’s post later today.