The following post is a sort of response to Raghuram Rajan’s recent post at Freakonomics:

If you are a college econ teacher, you’ve had this experience.  You explain how an expansionary monetary policy can boost AD.  A student raises his hand:

“But isn’t low interest rates just like the government providing a subsidy to borrowers?  And aren’t government subsidies bad?”

The student is so far off base you wonder how you are going to fix things with a short answer.  But let’s try anyway.

1.  Yes government subsidies are bad for two reasons.  They require higher future taxes, which impose deadweight costs.  And they distort relative prices.

2.  Now let’s think about monetary policy.  The first misconception is that the Fed “controls” interest rates.  In fact, the Fed controls the monetary base, and targets interest rates.  Rates are always allowed to find their free market values, given the setting of the monetary base.   So if the Fed wants to reduce rates, it might increase the monetary base until the equilibrium free market rate falls to the desired level.

3.  But doesn’t the Fed distort the bond market when they swap cash for T-bills?  Maybe a tiny bit, but that’s not really why rates fall when the Fed increases the monetary base.  The same liquidity effect used to occur in the old days when the Fed bought gold.  The effect occurs because there is more non-interest bearing money in the public’s hands.  Until consumer prices have had a chance to rise, the only way to get people to hold this extra money is for free market rates on alternatives assets to fall.  These rates are the opportunity cost of holding cash.  So monetary policy is fundamentally about the supply and demand for money.  Interest rates are just one of many variables that change as a result of changes in the money supply.  Exchange rates and consumer prices also change.  Imagine someone criticizing a reduction in the inflation target from 3% to 2% on the grounds that it would be a subsidy to consumers!

4.  Interest rates are (as a first order approximation) a zero sum game for the public.  Lower rates mean one group pays less, and the other group receives less.  But isn’t there some sort of “natural rate” and isn’t the Fed messing things up by setting rates below that natural rate?  All serious attempts to find a natural rate of interest look at the macroeconomy, especially inflation and NGDP.  Obviously credit markets (financial asset prices) can adjust to any inflation rate, but the real economy has trouble when inflation (or NGDP) rises or falls unexpectedly.  So if there is a “natural rate of interest” it would be the rate where inflation or better yet NGDP is optimal, where the macro economy is in some sort of equilibrium.  Where you don’t have mass unemployment, or overemployment, because nominal wages are temporarily stuck at the wrong level.  Even if you don’t believe in sticky wages or prices, and simply support steady 2% inflation, there is still a natural rate; it is the rate that generates that target inflation rate.  But one shouldn’t even focus on the natural interest rate, as we don’t have any way of directly estimating it (Taylor Rules notwithstanding.)  Instead, the focus should be on NGDP and inflation expectations.  Get those variables right, and then interest rates will also be at the proper level.

5.  Thus the question is never whether low rates unfairly subsidize borrowers, or whether high rates unfairly tax borrowers, because the Fed is not directly fixing interest rates, or driving any sort of tax or subsidy wedge between lenders and borrowers.  Rates are set in the market.  The question is whether the money supply is set at a level that produces the sort of interest rates, exchange rates, prices, NGDP, etc, that are consistent with optimal macroeconomic performance.

6.  Of course right now expected inflation and NGDP growth are much too low for macro equilibrium, so we need easier money.  Does that mean we need lower rates?  Here is where things get complicated.  If both long and short term rates are very low, it is generally a sign that money has been too tight, and the level of nominal spending is too low to provide optimal macroeconomic conditions.  So can you solve this problem by raising rates?  It depends what you mean by raising rates.  If you mean setting a higher fed funds rate, and implementing it through a reduction in the base, then the answer is no.  If you mean trying to raise rates by printing lots of money and promising to do more in the future, or promising higher future inflation, or promising to steadily devalue the dollar, then the answer is yes.

The problem with recent essays by people like Rajan and Kocherlakota is that they don’t seem to understand this distinction.  Or if they do, they express their ideas in a rather garbled fashion.  They both think that higher rates might be desirable for various reasons.  So far so good.  But both also strongly imply that the way to get higher rates is through the Fed raising its short term fed funds target.  But that requires tighter money–which would be a disaster right now.  I’d also like to see higher rates on long term bonds (I’d love to get back to the 4% rates we saw on the 10 year bond a few months back) but want to get there through easier money.  Since the fed funds target can’t be lowered much further, I support unconventional methods of boosting inflation and NGDP growth expectations.

Steve Williamson left this comment over at Nick Rowe’s blog, in support of Kocherlakota:


Let’s be more explicit. I know you don’t like math and symbols, but they force some discipline on what you do. Suppose this is an infinite horizon, discrete-time model, t = 0,1,2,… . Now, in the first case, the initial money stock is M, the growth rate of the money stock is m. Suppose the environment is stationary (population, technology, preferences constant over time). Suppose in the first case the equilibrium nominal interest rate is R. What do we know about the first equilibrium? (i) M does not matter, i.e. money is neutral, and R is an equilibrium interest rate for initial money stock kM for any k > 0. (ii) As you said, the Fisher relation holds, if I increase m, then R increases one-for-one. Now, consider the second case. Suppose the central bank pegs the interest rate at R. Then there is a continuum of equilibrium money stock paths (and maybe more besides, but there are at least these) with initial money stock kM and money growth rate m that support that interest rate peg. The central bank gets to choose the one it wants. If the central bank choose a nominal interest rate R+x, it can support that with a money stock path kM and a money growth rate m + x. Done.

In a flexible price world Williamson would be correct.  If the Fed suddenly increases the fed funds rate by 200 basis points as Rajan suggests, then the money supply growth rate must also increase by 200 basis points.  Since the real rate is unaffected (money is neutral) the expected inflation rate also increases by 200 basis points.  And that means right away, inflation rises literally overnight.

In the real world prices are sticky.  How do we know?  Because all the various markets respond to fed funds rate surprises as if prices are sticky.  If there is a sudden and unexpected rise in the fed funds target at 2:15 pm, then here is what typically happens at 2:16 PM:

1.  Nominal stock prices fall sharply.

2.  Nominal commodity prices fall sharply.

3.  Foreign currency prices fall sharply.

4.  Inflation expectations (TIPS spreads) fall sharply.

So Williamson’s model is not applicable to the fed funds tool that most people refer to when discussing interest rates.  Increases in the fed funds target cause short term real rates to rise by at least as much as nominal rates.  It is only true as a long run proposition.

It’s funny how right-wingers who supposedly believe in markets go out of their way to lecture markets about how they don’t understand the true model of the economy, which it seems has only been revealed to freshwater economists.  Here is Rajan:

Indeed, the Fed is now trapped because of the expectations it has set “” because the market “expects” ultra-low rates, the Fed cannot even return to normal low rates without the market taking fright. And it is hard to find a Wall Street economist who is not urging the Fed to undertake stronger, unorthodox actions.

I’ve spent most of my life studying the 1930s.  And I have to agree with Rajan.  Just as in the 1930s, the markets are very “frightened” of monetary tightening during a recession.

Here is Kocherlakota:

The FOMC’s decision has had a larger impact on financial markets than I would have anticipated. My own interpretation is that the FOMC action led investors to believe that the economic situation in the United States was worse than they, the investors, had imagined. In my view, this reaction is unwarranted. The FOMC’s decisions were largely predicated on publicly available data about real GDP, its various components, unemployment, and inflation. I would say that there is no new information about the current state of the economy to be learned from the FOMC’s actions or its statement.

Just the opposite is true.  The market isn’t falling because they think the Fed knows more than they do, in fact the problem is just the opposite.  The market understands that there is a serious shortfall in NGDP growth–nowhere near enough to generate economic recovery.  It is the Fed that seems clueless, and that is what has markets very frightened.  If the Fed really did do something aggressive, more than expected, markets would not fall because they saw the Fed was worried, they’d soar in relief that the Fed was finally doing something.  Ironically it is people like Krugman and I that are doing ratex modeling.  We have models where the various markets’ expectations are consistent with the predictions of the model.  The difference between Krugman and I is that I always make this assumption.

Apologies to commenters–I will be way behind for a while.  This is an important moment.

HT:  Daniel Carpenter, Marcus

A few thoughts on nominal wage stickiness

Tyler Cowen recently responded to some wage stickiness comments by Bryan Caplan and myself:

But if people who work on commission and tips are out of work in large numbers, or if truly flex-wage workers are being laid off, why see wage stickiness as the #1 culprit?  (Scott isn’t following through the logical implications of his cyclicality point.)  In economies with truly flexible wages, people are forced to retreat into household production in down times and that is perhaps a better parable for America today.  No one will hire them, flexibility or not.  Plus if workers are irrational by focusing on the nominal rather than the real values, it’s easy enough to trick them by cutting real benefits and working conditions, thereby saving the employer money.  Real wage flexibility should be enough to keep them at work, yet it isn’t.

I have 5 comments on this passage.

1.  I don’t follow the logic of the last sentence.  First of all, nominal wage rigidity may be due to contracts, not money illusion.  But even if it is money illusion, Tyler’s argument doesn’t follow.  Just because a factory worker with limited skill at math doesn’t understand the distinction between a cut in nominal wages and a cut in real wages, does not mean that he wouldn’t realize if the assembly line sped up from 60 to 80 cars an hour.  And if he truly wouldn’t realize this, why hadn’t the car company already sped up the line?

2.  Just because aggregate nominal wage stickiness causes aggregate unemployment to rise, doesn’t mean wage flexibility in a single profession can prevent unemployment in that profession from rising during recessions.  There is a coordination problem here.  Suppose the money supply falls 10%, and all economic agents get in a room and immediately decide to cut all wages and prices by 10%.  Will it prevent recession?  We can’t be sure, but later I’ll argue that it will.  But if they don’t do that, then we will have a recession.  In that case most nurses will not lose their jobs even if they refuse to take pay cuts, and many factory workers will lose their jobs even if they take 10% pay cuts.  Nominal shocks cause recessions that have very different effects on cyclical and non-cyclical industries.  I believe that is a prediction of every important business cycle model.

3.  I agree that nominal wage flexibility plays no role in explaining the chronically high unemployment in certain economies.  The argument is that nominal rigidity explains why unemployment rises in the face of nominal shocks, and that’s all it explains.  I have no idea why Haitian businesses don’t cut wages when there are Haitians who are willing and able to do the same work for less.  Perhaps there is some sort of efficiency wage explanation.  In any case, the chronically high unemployment that one observes in certain developing countries, or even in some eurozone members, has nothing to do with nominal wage rigidity.

4.  I don’t see “wage stickiness as the #1 culprit”.  Here is an analogy.  Suppose we observe engine failure once a month on jetliners.  Each time the plane crashes.  What’s the fundamental problem here, bad engines, or gravity?  Most people would say bad engines.  Now assume that every few years the Fed creates a negative nominal shock.  Because nominal wages are sticky, it creates a temporary recession (until wages adjust.)  What’s the fundamental problem here, sticky wages or monetary policy?  I’d say monetary policy.  To me, nominal stickiness is just a part of nature, like gravity.  It is not something you’d think about altering with government policies.  (BTW, it certainly isn’t the fault of “workers” most of whom don’t even set their nominal wage.)  Just as gravity is something airplane engineers must take into account, nominal stickiness is something that the Fed must take into account.  Of course there are government policies that increase real wage stickiness (minimum wages, extended IU benefits, etc) and those can make the problem worse.  At the risk of making my airliner analogy even more ludicrous, these rigidities are analogous to installing giant magnets on the ground, which try to suck airplanes out of the sky.  (Yes, I know that airplanes are aluminum.)

5.  Here’s why I can’t shake the idea that nominal wage (and perhaps price) stickiness is the key problem when AD falls.  Consider the 1920-21 deflation, which was quite rapid.  It was caused by a big drop in the monetary base (I believe around a 17% decline.)  Unemployment rose sharply, until wages had adjusted.  Now it seems to me that if these two facts are not related, if the rapid deflation (more than 20% depending on the index) did not cause the high unemployment in 1921, then I should just quit economics.  It would mean that everything I think I know is wrong–that I have nothing useful to say.  (Some people might wish I’d quit.)  But let’s assume I and the other 95% of economists who believe deflationary monetary policies increase unemployment are correct.  What then?  Well then explain this:  A few years back Mexico experienced 99.9% deflation almost overnight.  Prices plunged far more sharply than they did in the US between 1920-21.  And yet there was no sudden spike in the unemployment rate.  How could this be?  My theory is that Mexico avoided high unemployment during this severe deflation because the government ordered all nominal wage rates to be immediately cut by 99.9%.  Prices fell by a roughly similar amount (I don’t know if the government ordered them to fall.)  The point is that even severe deflation doesn’t cause additional unemployment if there is no nominal wage stickiness in the system.

Until someone can find a plausible alternative explanation for why America experienced high unemployment in 1921, but Mexico did not during a much more severe deflation during 1993, then I’m sticking with the sticky wage (and price) explanation of why nominal shocks have real effects.

One final point.  Once a major deflationary shock creates a severe recession, all sorts of real supply-side factors enter the equation.   I’ve already mentioned the cyclical effects of nominal shocks.  There is also the decision to raise IU from 26 weeks to 99 weeks—something that never would have happened without the initial nominal shock that created the recession.  But once those real factors kick in, the outcome can look very much like a real shock to a casual observer.  Don’t underrate the importance of unemployment benefits and other labor market rigidities.  The eurozone experienced long periods of near 10% unemployment (for reasons unrelated to nominal wage rigidities), there is no reason that it could not happen here—if we follow similar policies.

If the Fed boosts NGDP sharply, unemployment will fall sharply, Congress will let the extended IU benefits expire, and SRAS will shift right as AD is shifting right.  A win-win scenario.  Oh, and the budget deficit will automatically get much smaller.