Archive for May 2012


A kindred spirit

Here’s Miles Kimball:

In calling myself a liberal, I am saying that in addition to an attachment to the liberty, limited government, constitutionalism, and rule of law emphasized by Classical Liberalism,  I hold to a view based on both classic Utilitarianism and contested elements of modern economic theory that, generally speaking, a dollar is much more valuable to a poor person than to a rich person, and that therefore, there is a serious benefit to redistribution that must be weighed against the serious distortions caused by the usual methods of redistribution.

It’s very unusual to come across someone with almost identical political views.  How rare?  Kimball continues:

Perhaps because of cognitive dissonance, it is common for people to either believe (a) that tax distortions are serious and redistribution of questionable value OR (b) redistribution is valuable and the distortions induced by taxes are small. My belief is that (c) tax distortions are serious AND redistribution is valuable.  That makes me a supply-side liberal.

I also agree with this:

Tax distortions are governed in important measure by the the consumption-constant elasticity of labor supply. The consumption-constant elasticity of labor supply measures how much less workers want to work when what they earn is taxed, but the tax revenue is recycled back to them in one form or another of government benefit they can get regardless of how little they work.  Matthew Shapiro and I argue in our paper “Labor Supply: Are the Income and Substitution Effects Both Large or Both Small?” that the consumption-constant elasticity of labor supply is large.

Karl Smith doesn’t seem to buy the consumption-constant elasticity assumption.  But it seems the most reasonable baseline to me.  In public finance the big debate is over the long run effect of expanding both taxes and expenditures.  Are higher taxes justified?  If they are recycled back to the public in some form of benefit, they might be.  But only if they overcome the efficiency cost of lower output.  It’s true that measured GDP might not fall if the money is wasted (as poorer people tend to work harder.)  But that’s not much of an argument for big government.  In addition, in recent decades the growth in government seems to be coming more in the form of social welfare expenditure, not programs like defense and space exploration.  I’d expect that trend to continue.

Put simply, there is a choice between the European model of high taxes and low work effort, the US/Japanese model of medium taxes and medium work effort, and the Singapore model of low taxes and high work effort.

Now if only he were a NGDPLT proponent . . .

HT:  Dilip

A high school student refutes the world’s most famous economist. (And then shows how to save the world economy.)

Last year I did a post criticizing this statement by Paul Krugman:

Oh, and about the exchange rate: there’s this persistent delusion that central banks can easily prevent their currencies from appreciating. As a corrective, look at Switzerland, where the central bank has intervened on a truly massive scale in an attempt to keep the franc from rising against the euro — and failed:


I argued Krugman was wrong, that the Swiss national Bank was able to depreciate the franc whenever they wanted to.  That was July 2011.  Soon after the SNB again faced an attack from speculators.  This time they decided to depreciate the franc and then cap it at 1.2 SF/euro.  And as Evan Soltas showed in this post, it worked.

The SNB made a bold surprise promise: we will not let the swiss franc appreciate any more; we will hold the exchange rate at 1.2 CHF to 1 euro. And, in short, it worked. The appreciation of the swiss franc came to an immediate halt. The CHF-EUR exchange rate stabilized at 1.2. Even during this current turmoil in the Eurozone — which one might think would lead the currency to appreciate as capital flowed into Switzerland’s financial safe haven assets — the Swiss central bank has kept the exchange rate steady, fulfilling its promise.

.   .   .

Switzerland has the ability to stabilize its currency from upward pressure because its monetary policy tool is unlimited — it could always print out more swiss franc to satisfy market demand. This is, in other words, the opposite of trying to prop up a currency, which strains the foreign currency reserves of the central bank. The result is that the Swiss intervention is entirely credible.

Its credibility is so powerful, in fact, that the SNB has stopped having to buy up foreign currencies with new swiss franc, which it did in earnest to prove its commitment in 2011, increasing its foreign exchange reserves by 177 billion from July to September. It hasn’t had to defend at all the value of its currency against appreciation since September, despite what should be enormous pressures. (See here and here for the data.) That is truly remarkable, when you zoom out for the macroeconomic big picture.

That is the power of credible monetary promises. And we can do the same thing with the price level path, of course, managing correctly the striking strength of market expectations. All it takes is the appropriate use of the expectational channel; re-establish 5 percent annual NGDP growth as did the SNB for its currency, and then the market will do the rest for you.

It’s easy to say “I told you so,” but what Evan is doing here is far more interesting.  He isn’t just disproving Krugman’s liquidity trap hypothesis; he’s showing that Krugman is wrong in a very interesting way.  Krugman saw the huge SNB purchases during the period of policy failure as a sign that the effort required to achieve policy success would be even more monumental.  What Soltas shows is that exactly the opposite is true.  As soon as the SNB set a firm exchange rate target, they no longer had to buy foreign exchange to defend the target.  Why invest in Switzerland at zero interest if you aren’t going to at least get some currency appreciation?  The bloated Swiss monetary base wasn’t stimulus that failed; it was stimulus that was never seriously attempted.

Then Evan makes the very astute observation that the same logic applies to US monetary policy, where NGDP level targeting would be equivalent to the pegged Swiss franc.  Krugman sees the large increase in the US monetary base during recent years as monetary stimulus that failed.  In fact, it’s just the Fed accommodating the high demand for base money when tight money and ultra-low NGDP growth drove nominal rates to zero.  Evan is right that with much faster expected NGDP growth there would actually be less demand for base money in the US.  How much QE would it take to get back to the old NGDP trend line?  Roughly negative $1.6 trillion.

Update: Evan has a new post that provides more information on just how effective the expectations channel has been in Switzerland.  BTW, I deleted the previous lame attempt at humor in the postscript.  It occurred to me that Evan is just a high school student, and doesn’t need me generating problems for him in the future.

Krugman and Hamilton on the zero rate bound

Marcus Nunes sent me an interesting old post from Paul Krugman.  Here’s Krugman in November 2008:

Nearly every forecast now says that, in the absence of strong policy action, real GDP will fall far below potential output in the near future. In normal times, that would be a reason to cut interest rates. But interest rates can’t be cut in any meaningful sense. Fiscal policy is the only game in town.

Those were the posts that had me pulling my hair out in 2008.  Why wasn’t Krugman calling for monetary stimulus?  One answer is that rates were already at zero, and hence could not be cut any lower.  But there are lots of other ways to stimulate the economy.  And even worse, rates weren’t yet at zero, the fed funds targets was 1.0% in November 2008.   So I decided to follow the link in Krugman’s article, to see why he concluded that rates could be cut no further.  And it led to this James Hamilton post:

There was yet another announcement from the Fed this week that caused my jaw to drop, though you’d think I’d be getting used to such surprises by now. The Fed announced on Tuesday that it will raise the interest rate it pays on both required reserves and excess reserves to the level of the target itself, currently 1.0%.

My first reaction was, How in the world could that work? Why would any bank lend fed funds to another bank at a rate less than 1%, exposing itself to the associated overnight counterparty risk, when it could earn 1% on those same reserves risk free from the Fed just by holding on to them?  It would seem paying 1% interest on reserves should set a floor on the fed funds rate, so that any fed funds actually lent between banks would have to offer a higher rate than the official “target.”

But I’ve always been more persuaded by facts than by theories, and the effective fed funds rate reported for Thursday– the first day of the new regime– was 0.23%. So much for that theory. But what’s going on?

The answer begins with the observation that the GSEs and some international institutions also have accounts with the Fed. But unlike regular banks, these institutions earn no interest on those reserves, so they would in principle have an incentive to lend out any unused end-of-day balances as long as they earn a positive interest rate.

But that’s not a sufficient answer by itself, because there’s an incentive for any bank that is eligible to receive interest from the Fed on reserve balances to borrow those balances from the GSE at a rate less than 1%, get credited by the Fed with 1% for holding them, and profit from the difference. Why wouldn’t arbitrage by banks happy to get these overnight funds prevent the rate paid to the GSEs from falling below 1%?

Wrightson ICAP (subscription required) proposes that part of the answer is the requirement by the FDIC that banks pay a fee to the FDIC of 75 basis points on fed funds borrowed in exchange for a guarantee from the FDIC that those unsecured loans will be repaid. If you have to pay such a fee to borrow, it’s not worth it to you to pay the GSE any more than 0.25% in an effort to arbitrage between borrowed fed funds and the interest paid by the Fed on excess reserves. Subtract a few more basis points for transactions and broker’s costs, and you get a floor for the fed funds rate somewhere below 25 basis points under the new system.

Under this regime, the effective fed funds rate– a volume-weighted average of the rate associated with fed funds traded on a given day– would come in above the target if it is dominated by actual banks borrowing fed funds, and below target when dominated by GSE lending. In the latter case, the effective fed funds rate would seem to be a particularly meaningless statistic, reflecting nothing more than the institutional peculiarities just detailed.

That means a couple of things for Fed watchers. First, fed funds futures contracts, which are based on the average effective rate rather than the target over a given month, are primarily an indicator of how these institutional factors play out– how much the effective rate differs from the target– and signal little or nothing about future prospects for the target. Second, the target itself has become largely irrelevant as an instrument of monetary policy, and discussions of “will the Fed cut further” and the “zero interest rate lower bound” are off the mark.

That seems to support Krugman’s argument.  However in the remainder of the post Hamilton goes in a very different direction from Krugman:

There’s surely no benefit whatever to trying to achieve an even lower value for the effective fed funds rate. On the contrary, what we would really like to see at the moment is an increase in the short-term T-bill rate and traded fed funds rate, the current low rates being symptomatic of a greatly depressed economy, high risk premia, and prospect for deflation.

What we need is some near-term inflation, for which the relevant instrument is not the fed funds rate but instead quantitative expansion of the Fed’s balance sheet. I continue to have concerns about implementing the latter in the form of expansion of excess reserves, which ballooned by another quarter trillion dollars in the week ended November 5. Instead, I would urge the Fed to be buying outstanding long-term U.S. Treasuries and short-term foreign securities outright in unsterilized purchases, with the goal of achieving an expansion of currency held by the public, depreciation of the currency, and arresting the commodity price declines.

But the last thing we should expect to do us any good would be further cuts in the fed funds target.

I think this is exactly right.  Once you’ve started paying IOR, what you need is not so much a bigger monetary base, but rather a bigger currency stock.  Currency doesn’t earn interest.  Hamilton was writing before I published my paper advocating negative IOR, so I wouldn’t have expected him to mention that out-of-the-box idea.  Instead he points to the need for inflation, enacted by depreciating the currency and boosting commodity prices via QE.  Krugman would counter that the Fed can’t change the price of assets like stocks, commodities and foreign exchange, because we are stuck at the zero bound.  So let’s see who was right.

On December 16th, 2008, the Fed conducted a near perfect test of Krugman’s hypothesis.  They cut the fed funds target (and the IOR) from 1.0% to 0.25%.  Krugman would predict no effect, as the actual fed funds rate would stay around 0.25%.  Recall that the Keynesian model says that what matters is the actual short term interest rate, not the target rate.  In fact, the S&P500 rose by more than 5% on December 16th.  Now I do realize that very few people share my faith in the EMH.  And stocks did resume their downward trend over the next few months as new economic data showed the recession to be worse than expected.  So let me also provide a mechanism for why the IOR cut mattered.

Unlike Keynesians, monetarists don’t believe the short term rate is the key mechanism for the transmission of monetary policy.  Rather we look at things from a supply and demand for money perspective.  Increases in the supply of money are inflationary, and increases in the demand for money are deflationary.  The IOR program provides an excellent test of this hypothesis.  A change in the IOR can change the demand for ERs, without necessarily changing the effective fed funds rate at all.  We have four observations; the initial announcement of IOR on October 6, two subsequent increases in IOR in late October and early November 2008, and the big cut on December 16th, 2008.  In all four cases stock prices moved strongly in response to the IOR announcement.  The moves ranged from 3.85% to 6.1%, with an average change of over 5%.  Those are relatively big daily changes.   More importantly, all four changes were in the right direction; stocks fell sharply after the three increases in IOR, and rose sharply when IOR was cut.   This is certainly not conclusive proof (four observations isn’t enough, and other things were going on at the time) but it’s four data points in support of the monetarist transmission mechanism.

If you pay people to hold on to ERs, there’ll be a greater demand for ERs.  And when the demand for any good rises its value rises. But when the value of the medium of account rises, its nominal price cannot change, by assumption.  Instead, the only way for the medium of account to become more valuable is for all other prices to fall.  The Fed did a beautiful experiment in late 2008 by creating IOR—pity about the economy.

PS.  Of course a much more powerful refutation of the liquidity trap view came from market reactions to QE1 and QE2.

Ram on sticky policies

Commenter Ram had a very interesting comment on why NGDP level targeting currently seems much better than flexible inflation targeting, even though in theory they shouldn’t be much different:

I support rules over discretion, too. Think about it this way: Suppose the Fed is targeting nominal variable X(t) at a value of x(t). At every instance, the Fed is adjusting its stance so as to ensure that E[X(t)] = x(t). x may vary with t, and yet this is still a rule in the only sense that matters, because it’s spelled out in advance, and adhered to without exception. Even if t (time) isn’t the dependent variable–say the output gap is, as in flexible inflation targeting, for example–it’s still a rule if the Fed implements it in a wholly non-discretionary manner (say by having some set, publicly communicated model of potential output). Nothing about the advantages of rules over discretion favors targeting one nominal variable over another.

Now, suppose I have a rule that says at any and all times, I will pursue inflation target I(o), where o is the size of the output gap. Suppose also that I is increasing in o. In that case, there is a sense in which I’m targeting inflation, in that at any and all times my objective is to keep inflation at a particular level. It’s just that when the output gap gets really big, that level is higher then when the output gap gets really small. If pursued rigorously (as a rule), then what the Fed would do right now is say: “Listen, normally we like 2% inflation because we’re normally able to keep the output gap at a tolerably low level. Right now, however, with interest rates really low, it’s getting bigger than we’d like. So we’re going to target 4% inflation until the output gap falls back to tolerably low levels, after which we’ll go back to our normal 2% target.” This would be following a rule, it would be targeting inflation, and it would overcome the zero lower bound. In a perfect world, it would be optimal, just as optimal as the right NGDP level target.

What the crisis has shown, Scott and I believe, is that in practice it’s really hard for the Fed to follow a rule of that kind. Why? My answer is sticky policies, or sticky targets. I don’t have good microfoundations for such stickiness, but then I never really bought any of the micro-stories behind price or wage stickiness either. Stickiness in the data is enough to convince me. In light of sticky policies/targets, some policies that would be just as good as others in the flexible policy world are better than others in the stick policy world. NGDP level targeting versus flexible inflation targeting is such an example. But it’s useful to note that sticky policies are a fact of political life, not a fact of economic life per se.

So in theory the Fed should have responded very aggressively in 2008 and early 2009, as both the inflation rate (low) and the output gap (big) signalled monetary policy was much too tight.  But policy is “sticky” or slow to adjust.  I’d add that this isn’t just a zero bound problem.   During the entire  great NGDP collapse of June to December 2008, the fed funds target was consistently above the zero bound.  The Fed was slow to adjust policy.  Earlier I argued that slow policy adjustment was the reason why America doesn’t have mini-recessions; real shocks aren’t even large enough to cause mini-recessions (much less full fledged recessions) and monetary shocks end up much bigger than the Fed intended, because the Fed is slow to adjust policy when changes in the Wicksellian equilibrium rate cause policy to switch from expansionary to contractionary.

Level targeting avoids policy stickiness, as market expectations stabilize policy when the central bank is slow to react.  This is a great observation by Ram, and is precisely the sort of thing that abstract macro models tend to overlook.  Policies that work well on paper may not work well in the real world.

PS.  David Beckworth and Ramesh Ponnuru have an excellent new piece in the National Review advocating NGDPLT.

Dude, where’s your model?

Whenever I get taunted about not having a “model,” I assume the commenter is probably younger than me, highly intelligent, but not particularly wise.  I often disagree with Paul Krugman, but he’s right that simple, off-the-shelf models are all we need to evaluate aggregate demand problems.  Supply-side issues are another story—they are much more complicated.

There are basically two types of demand-side models, both of which are nearly useless:

1.  Some general equilibrium models are used to find which stabilization policy regime is optimal from a welfare perspective.  Most of these models assume some sort of wage/price stickiness.  And 100% of the models taken seriously in the real world assume wage/price stickiness.  The problem is that there are many types of wage and price stickiness, and many ways of modeling the problem.  You can get pretty much whatever policy implication you want with the right set of assumptions.  Unfortunately, macroeconomists aren’t able to prove which model is best.  I think that’s because lots of models are partly true, and the extent to which specific assumptions are true depends on which country you are looking at, as well as which time period.  And then there’s the Lucas Critique.

2.  A second type of model tries to show how to best implement a specific type of policy regime, like inflation targeting.  Thus the Taylor Rule is one way of implementing a flexible inflation target.  Unfortunately, these “implementation models” conflict with the EMH—it’s not clear why the central bank wouldn’t just peg the price of a futures contract linked to the goal variable.  This is embarrassing given that they are mostly based on New Keynesian models, which incorporate rational expectations.

To summarize, despite all the advances in modern macro, there is no model that anyone can point to that “proves” any particular policy target is superior to NGDPLT.  There might be a superior target (indeed I suspect a nominal wage target would be superior.)  But it can’t be shown with a model.  All we can do is construct a model that has that superiority built in by design.

And once we have decided on a nominal aggregate target, there is no model that can outperform a policy of having the central bank peg the price of futures contract linked to the policy goal.  And once you do all that, fiscal policy becomes a fifth wheel.

Models are toys to show our students.  When we face serious real world dilemmas it’s time to put away the toys and get real . . . er, I’m mean get nominal.

PS.  Wise readers will notice that I do have a model, indeed lots of them.  They’re just not mathematical models.

PPS.  One reason that very few models find NGDP targeting to be optimal is that very few models include a variable called “NGDP.”