Archive for the Category Monetary Theory

 
 

Neo-Fisherism, missing markets, and the identification problem

I’ve done recent posts on Neo-Fisherism, and the problem of identifying the stance of monetary policy.  I’ve also pointed out that if we can’t identify the stance of monetary policy, we can’t identify monetary shocks.

This is going to be a highly ambitious post that tries to bring together several of my ideas, in a Grand Theory of Monetary Policy.  Yes, that’s means I’m almost certainly wrong.  But I hope that the ideas in this post might trigger useful insights from people who are much smarter than me and/or better able to handle macro modeling.

NeoFisherians argue that if the Fed switches to a policy of low interest rates for the foreseeable future, this will lead to lower inflation rates.  Their critics claim this is not just wrong, but preposterous—and undergraduate level error.  I think both sides are right, and both sides are wrong.

All of mainstream macro (including the IS-LM model) is built around the assumption of a “liquidity effect”.  That is, because wages and prices are sticky in the short run, a sudden and unexpected increase in the money supply will lower short-term nominal interest rates. Interest rates must fall in the short run so that the public will be willing to hold the larger real cash balances (until the price level adjusts and the real quantity of money returns to its original equilibrium.)

One thing that makes this theory especially persuasive is that it’s not just believed by eggheads in academia.  Pragmatic real world central bankers often see nominal interest rates fall when they inject new money into the economy.  It’s pretty hard NOT to believe in the liquidity effect if you see it in action after you make a policy move.

So the NeoFisherians have both the eggheads and the real world practitioners against them.  And yet not all is lost.  Over any extended period of time the nominal interest rate does tend to track changes in trend inflation (or better yet trend NGDP.)  If I had a perfect crystal ball and saw the fed funds rate rise to three percent and then level off for twenty years, I’d expect a higher inflation rate than if my crystal ball showed rates staying close to zero for the next 20 years.  NeoFisherians are smart people, and they wouldn’t concoct a new theory without some good reason.

In previous posts I’ve found it useful to illustrate where NeoFisherism works with a thought experiment.  I’ll repeat that, and then I’ll take that thought experiment and use it to develop a general model of monetary policy.  Here’s the thought experiment:

Suppose Japan wants to raise their inflation rate to roughly 8%/year.  How do they do this?  This requires two steps.  They need a monetary regime that produces 8% steady state trend inflation, and they need to make sure that the price level doesn’t undergo a one-time jump upwards or downwards at the point the new regime is adopted.

To get 8% trend inflation, they’d need to shift the trend nominal interest rates to a much higher level.  To make things as simple as possible, suppose the US Federal Reserve targets inflation at 2%, and the BOJ has confidence that the Fed will continue to do so.  In that case the BOJ can peg the yen to the dollar, and promise to depreciate it at 6%/year, or 1/2%/month.  The interest parity theory (IPT) assures us that Japanese interest rates will immediately rise to a level 6% higher than US interest rates.  (Note, I’m using the near perfect ex ante version of the IPT, not the much less reliable ex post version.)

We’ve already gotten a very NeoFisherian result.  Currency depreciation is an expansionary monetary policy, and the IPT assures us that this particular expansionary monetary policy will also produce higher interest rates.  And not just in the long run, but right away. Although Purchasing Power Parity is widely known to not hold very well in the short run, expected inflation differentials do tend to reflect the expectation than PPP will hold.  So under this regime not only will Japanese nominal interest rates rise almost exactly 6% above US levels, Japanese expected inflation rates will also rise to roughly 6% above US levels.

Of course actual PPP does not hold all that well (although it does far better under fixed exchange rate regimes, or even crawling pegs like this system, than floating rates.)  But that doesn’t really matter in this case; just getting 6% higher expected inflation is enough to pretty much confirm the NeoFisherian result.

Unfortunately, the immediate rise in interest rates might reduce the equilibrium price level in Japan.  But even that can be prevented with a suitable one-time currency depreciation at the point the regime is adopted.  How large a currency depreciation? Let the CPI futures market tell you the answer.  Make your change in the initial exchange rate conditional on achieving a 1 year forward CPI that is 8% higher than the current spot CPI.  Thus the CPI futures market might tell you that the yen should immediately fall to say 154.5 yen/dollar, and then depreciate 1/2% a month from that level going forward.  Whatever amount of immediate currency depreciation offsets the immediate impact of higher interest rates.

Notice that monetary policy has two components, a level shift and a growth rate shift. This idea will underpin my grand theory of monetary policy.  In this case the level shift was depreciating the yen from 120 yen/dollar to 154.5 yen/dollar.  The growth rate shift was going to a regime where the yen is expected to gradually appreciate against the dollar, to one where it is credibly expected to depreciate at 1/2%/month.

Now think about the expected money supply path that is associated with that new policy regime for the yen.  My claim is that if the BOJ simply adopted that expected money supply path, all the other variables (exchange rates, interest rates, inflation, etc.) would behave just as they did under my crawling peg proposal.

So far I’ve been supportive of the NeoFisherians, but of course I don’t really agree with them.  Their fatal flaw is similar to the fatal flaw of Keynesian and Austrian macro, using the interest rate to identify the stance of monetary policy.  But in some ways it’s even worse than the Keynesian/Austrian view.  Those two groups correctly understand that a Fed rate cut is a signal for easier money ahead.  The NeoFisherians implicitly assume the opposite.

People say that the longest journey begins with a single step.  But what the NeoFisherians don’t seem to realize is that central banks generally signal an intention to go 1000 miles to the west by taking their very first step to the EAST.  (Nick Rowe has much better analogies.)  Thus when Paul Volcker decided that he wanted the 1980s to be a decade of much lower inflation and much lower nominal interest rates, the very first thing he did was to raise the short term interest rate by reducing the growth rate of the money supply.

As soon as you realize that central banks usually use changes in short term nominal interest rates achieved via the liquidity effect as a signaling device, then the NeoFisherian result no longer makes any sense.

But now I’m being too hard on the NeoFisherians.  Look at my crawling peg for the yen thought experiment.  And what about the fact that low rates for an extended period do seem associated with really low inflation, in places like Japan.  That suggests there’s at least some truth to the NeoFisherian claim, and at least some problem with the Keynesian/Austrian view.

In my view the two views can only be reconciled if we stop viewing easy and tight money as points along a line, but rather as multidimensional variables:

Monetary policy stance = S(level, rate)

A change in monetary policy reflects a change in one or both of these components of the S function.  You can have a rise in the price level, but no change in the trend rate of inflation.  You can have a rise in the trend rate of inflation, with no change in the current (flexible price) equilibrium price level.  The beauty of the thought experiment with the yen is that it makes it much easier to see this distinction.  You can imagine once and for all change in the exchange rate, as when the dollar went from $20.67 an ounce to $35.00 an ounce in 1933, and then stayed there for decades.  Or you can imagine a change in the trend rate of the exchange rate, as in my crawling peg example.  Or you can imagine both occurring at once.

When NeoFisherians are talking about higher interest rates leading to higher inflation, they are (implicitly) changing the “rate” component of my monetary policy stance function.  Keynesian and Austrians tend to (implicitly) think in terms of changes in levels, once and for all increases in the money supply that depress short-term interest rates and have relatively little effect on long-term interest rates or long run inflation.

In previous posts I’ve expressed puzzlement as to why easy money surprises lower longer-term interest rates on some occasions, and at other times they raise long-term interest rates.  The “perverse” latter result occurred in January 2001, September 2007, and (in the opposite direction) December 2007.

Indeed the December 2007 FOMC meeting produced the most NeoFisherian (i.e. “rate”) shock that I have ever seen in my entire life.  A smaller than expected rate cut (contractionary shock) led to a huge stock market sell-off (no surprise) but also lower bond yields for 3 months to 30 years T-securities (a big surprise.)  This policy announcement had an unusually large “rate shift” component, whereas most US policy announcements are primarily “level shifts”.  The markets (correctly) saw the Fed’s passivity in December 2007 as indicating that inflation and NGDP growth might be lower than normal going forward.  And not just in 2008, but indefinitely.

Unfortunately, while exchange rates nicely capture the rate/level distinction, they are not ideal for developing a general monetary policy theory, as the real exchange rate is too volatile.  And that brings me to the missing market, NGDP futures.  If this market existed we would have all the tools required to describe the stance of monetary policy in the multidimensional way necessary to sort through this messy NeoFisherian/conventional macro debate.  Suddenly we could clearly see the distinction between level shifts and rate shifts.

Unlike exchange rates, NGDP responds to monetary policy with a lag.  But we could use one-year forward NGDP as a proxy for levels.  Thus if one year forward NGDP rose and longer-term expected NGDP growth rates were unchanged then you’d have a level shift.  If the opposite occurred, you’d have a rate shift.  Or you might have both.  The response of interest rates to the monetary policy shock would largely depend on the response of NGDP futures to the monetary policy shock.

In a richer model there would be more than two dimensions, indeed the expected future NGDP at each future date would tell us something distinct about the stance of monetary policy, and thus more fully characterize any monetary shocks that occurred. But I see great benefits of using the simpler two variable description, levels and growth rates.  This simpler version is enough to address many of the perplexing features of monetary policy, such as why economists can’t seem to come up with a coherent metric for the stance of monetary policy, and why the NeoFisherians reach radically different conclusions than the Keynesians.

Ideally we’d create highly liquid NGDP and RGDP futures market, and then we could easily identify the stance of monetary policy, in both dimensions.  This would also allow us to ascertain the impact of policy shocks on both NGDP and RGDP. NeoFisherians could say, “A monetary policy that raises expected inflation rates and/or expected NGDP growth rates will usually raise nominal interest rates.”  That’s a much more sensible way of making their claim than “Raising interest rates will raise expected inflation and/or expected NGDP growth.

PS.  I’ve benefited greatly from discussions with Daniel Reeves (a Bentley student), and Thomas Powers (a Harvard student).  They understand NK models better than I do, and bouncing ideas off them has helped me to clarify my thinking.  Needless to say they should not be blamed for any mistakes in this post.

Stance, shock, cause

OK, the title’s not as dynamic as Camille Paglia’s Break, Blow, Burn, but I’m only an economist.  Recently I’ve been trying to figure out several questions; including what do we mean by the stance of monetary policy, and what do we mean by a monetary shock?  I suspect that at some deep level these are actually the same question, much as I suspect, “Does free will exist” and, “Is there such a thing as personal identity?” are the same question.  This post is a reaction to some good comments I’ve received, and also an excellent new post by Nick Rowe.  Don’t expect any final answers here, go to Nick’s post for specifics on VAR models.

Let’s work backwards from “cause.”  Perhaps a monetary shock is a change in monetary policy that causes something or many things to happen.  But that forces us to examine the thorny issue of what do we mean by “cause?”  In a sense, monetary policy could be said to cause all nominal changes in the economy, and many real changes.  After all, under a fiat money regime there is always some alternative monetary policy that would have prevented a nominal variable from changing.  Thus if the price of zinc rises from $1.30 to $1.35 a pound, you could say the cause of the increase was the Fed’s refusal to use OMOs to peg the price of zinc at $1.30/oz.  I think we can all agree that this is not a very useful view of causation.  But this problem will creep in to some extent no matter how hard we try to pin down ’cause’.

Now let’s look at ‘stance’ and ‘shock’.  By now you are sick of me telling you that interest rates don’t measure the stance of monetary policy.  But why not?  And why can’t I provide a definitive definition, if I’m so sure interest rates are wrong? Let’s consider three groups of possible indicators:

1.  Interest rates and the monetary base

2.  Exchange rates and the monetary aggregates

3.  Inflation, NGDP growth and zinc prices

I’d like an indicator that always moves in one direction in response to a given change in the stance of monetary policy. That’s why I hate interest rates; tight money sometimes makes them rise, and sometimes makes them fall.  So we can’t look at interest rates and identify the stance of policy.  Ditto for the base, which might rise because we are monetizing the debt Zimbabwe style, or it might rise because we are accommodating a high demand for reserves at the zero bound, a la Japan since the late 1990s.  That’s not to say that we can’t assume a given nudge in rates or the base leaves policy predictably tighter than a few minutes earlier.  The problem is that we can’t look at rates or the base and tell whether policy is looser or tighter than 3 months ago, which makes it useless for projects like VAR studies.  Nor does it help to look at rates relative to the natural rate, as the natural rate is highly unstable, and hard to estimate.

The second group is better.  In general, tight money will appreciate the exchange rate and reduce M2.  But I’m not 100% sure that’s always true.  Is it possible that tight money could lead to expectations of depression, and that expectations of depression could lead to lower future expected exchange rates, and that this would reduce the current value of the currency?  Is it possible that tight money could lead to such uncertainty that people want to hold more M2, relative to other assets?

The last group seems safest, and the first two items in group #3 are also the indicators chosen by Ben Bernanke back in 2003.  I can’t imagine a case where tight money raises either inflation, NGDP, or zinc prices.  So at least in terms of direction of change, they seem completely reliable.  One can think of the CPI as measuring the (inverse of the) value of money.  That’s a nice definition of easy or tight money—changes in its purchasing power.  NGDP is slightly more ungainly, the (inverse of the) share of national income that can be bought with a dollar bill.

Unfortunately I’ve engaged in circular reasoning.  I’ve defined easy and tight money in terms of inflation and NGDP growth, because I believe they are reliably related to the stance of monetary policy.  But how do I know that the thing that causes NGDP to rise is easy money?  Here I don’t think we can escape the necessity of relying on theory.  Theory says that an unexpected injection of new money will have all the effects associated with easy money, such as temporarily lower interest rates, a depreciated currency, more inflation and NGDP growth.

If we define the stance of monetary policy in terms of inflation or NGDP growth, then it seems to me that it makes sense to think of “shocks” as policy actions that change the stance of monetary policy.  Thus if Fed actions moved expected GDP growth from 4% to 6%, you could say that monetary policy eased and a positive monetary “shock” occurred.  Vice versa if expected NGDP growth fell from 4% to 2%.

But lots of people aren’t going to like the implications of all this, or any of this, as there are deep cognitive illusions about distinctions between “errors of omission” and “errors of commission” that seem important (but in my view are not.) For instance, VAR studies could no longer disentangle monetary and demand-side fiscal shocks—-all changes in NGDP would be monetary shocks, by assumption.  There would be no difference between a change in M and a change in V, both would be monetary shocks.

If you try to flee back to your comforting notions of causality, they will fall apart under close inspections.  Suppose Russians hoard 5% of the US monetary base in 1991, and the Fed does not accommodate that increase, even though they easily could have done so.  Interest rates soar, V falls, and the US goes into recession.  What do the “concrete steppes” people say? Unfortunately they’d start arguing with each other.  The monetary base concrete steppes people would insist the Fed did not cause the recession, it was caused by Russian currency hoarding.  The base didn’t change.  The interest rate concrete steppes people would insist the Fed did an error of commission; they increased interest rates sharply and caused the recession.

Nor is it possible to fall back on “unexpected changes in interest rates.”  Suppose that prior to the Sept. 2008 FOMC meeting, markets had expected a huge negative monetary shock, which would occur because the Fed foolishly kept interest rates at 2%.  But instead (suppose) the Fed surprised us and avoided a negative monetary shock by cutting rates to zero and switching to NGDP level targeting.  There would be a huge negative surprise to interest rates, but no change in the stance of monetary policy, by the NGDP criterion.

Macroeconomics is riddled with unexamined assumptions about stances of policy, shocks, and causation.  It’s ironic that we use the term ‘stance’ as there’s no stable ground here to stand upon.  It’s like we’ve moved from a classical Newtonian world to a relativistic universe.  What’s is the stance of policy?  It depends where you are standing, how fast you are moving, what variables you are interested in, etc., etc.

People are constantly telling me that my “tight money” theory of the 2008 recession is loony.  But I am never provided with any good reasons for this criticism.  I have no doubt that there are hundreds of macroeconomists who are much smarter than I am, but I do occasionally wonder if my profession is somewhat lacking in imagination.

PS.  Going back to the opening paragraph, the answers are no and no.

Monetary policy is not about banking

When I advocate something like QE or negative interest on reserves, I often get people complaining that this will not boost bank lending, or that we shouldn’t even be trying to boost bank lending.  It almost makes me want to tear out my hair. What in the world does banking have to do with monetary policy?  Yes, it may or may not boost bank lending, but it doesn’t matter, as monetary policy is about the hot potato effect.  And yes, the Fed should not be trying to boost lending, any more than it should try to boost sales of microwave ovens.  NGDP is what matters.

Jim Glass directed me to a new study by the Bank of England, which confirms that monetary policy is about the hot potato effect (aka portfolio rebalancing channels) not bank lending.  Here is the abstract.

We test whether quantitative easing (QE) provided a boost to bank lending in the United Kingdom, in addition to the effects on asset prices, demand and inflation focused on in most other studies. Using a data set available to researchers at the Bank, we use two alternative approaches to identify the effects of variation in deposits on individual banks’ balance sheets and test whether this variation in deposits boosted lending. We find no evidence to suggest that QE operated via a traditional bank lending channel (BLC) in the spirit of the model due to Kashyap and Stein. We show in a simple BLC framework that if QE gives rise to deposits that are likely to be short-lived in a given bank (‘flighty’ deposits), then the traditional BLC is diminished. Our analysis suggests that QE operating through a portfolio rebalancing channel gave rise to such flighty deposits and that this is a potential reason that we find no evidence of a BLC. Our evidence is consistent with other studies which suggest that QE boosted aggregate demand and inflation via portfolio rebalancing channels.

PS.  I have a new post over at Econlog that is far more important than this post.  Read the whole thing.

Apart from boosting NGDP and RGDP, euro depreciation will not help Italy

That’s the conclusion of a new paper that Tyler Cowen has linked to.  And I think that’s right.  Many stimulus advocates (including me and Lars Svensson) have pointed out that currency depreciation caused by monetary stimulus would not be expected to boost net exports, as the substitution effect will often be dominated by the income effect (a booming economy sucking in more imports.)

Unfortunately, the paper (by Alberto Bagnai and Christian Alexander Mongeau-Ospina) starts off with a misleading summary of the results:

It is frequently claimed that the current EUR/USD exchange rate is too high and that a depreciation of the EUR against the USD would contribute to relieve the Eurozone economy from the current state of persistent crisis. Evidence provided by the a/simmetrie annual econometric model suggests that this claim is unsupported by the data, at least as far as the Italian economy is concerned. In fact, the size and sign of the trade elasticities show that the increases in net exports towards non-Eurozone countries, brought about by the depreciation of the euro, would be offset by an increase in net imports towards Eurozone countries, brought about by the increase in Italian domestic demand.

And indeed Tyler also assumed that this pessimistic conclusion was their key finding, in his quick summary of the results.  But in fact that’s not at all what the paper says. Here’s the key paragraph:

Before presenting the results, it is worth noting that the simulations proposed were performed using only the foreign trade block of the model, supplemented with the national income identity and the price deflators equations. As a consequence, the results presented have only a partial equilibrium meaning and are still preliminary. In particular, they take into account the feedback on imports following from the expansion of aggregate demand caused by the increase in exports, as well as the inflationary effects following from the increase in import prices determined by the nominal exchange rate devaluation, but they do not take into account the “second round” inflationary effects determined via Phillips curve by the decrease in unemployment, which could possibly offset in the longer run the effect of a nominal realignment.

So if you ignore the fact that monetary stimulus that depreciates the euro will also boost NGDP, and that this will boost RGDP and employment via the “Phillips curve” mechanism, and only focus on the fact that a faster growing Italian economy will suck in more imports and hence stimulus will not improve the trade balance, then it appears a weaker euro will not help Italy.  And I certainly can’t disagree with that!

However I certainly don’t agree with this:

These results have important policy implications.

The euro is still far too strong

The title of this post does not refer to the exchange rate, the importance of which is overrated.

Tyler Cowen has a new post on the euro.  Here’s his conclusion:

All in all, the weaker euro is likely to prove a net benefit to the eurozone, all the more so if monetary policy can drum up some expansionary domestic benefits above and beyond the exchange rate effect.  Still, if you deliberately engineer a depreciation of your currency out of weakness and desperation, the long-run benefits usually don’t match up to that immediate feeling of short-run juice.

This is correct.  There are many examples of Latin American or Mediterranean countries devaluing their currency, and merely ending up with higher inflation in the long run.  But it’s also important to point out that the euro is still far too strong. Unfortunately there is no single measure of the strength of a currency, but surely it is more meaningful to talk about it’s ability to purchase a basket of all goods and services, as compared to its ability to purchase a pound of zinc, a share of Apple stock, a US dollar, an Australian dollar, or a Zimbabwean dollar.  I’d argue that a still better measure would be the fraction of a year’s eurozone NGDP that can be bought with a single euro.

In any case, whether you use the price level or NGDP as your metric, the euro is far too strong.  So while there are many examples throughout history of countries debasing their currency, the eurozone is not currently one of those examples.

PS.  If in fact the ECB has engineered a weaker euro in the forex markets, then ipso facto it has engineered “some expansionary domestic benefits above and beyond the exchange rate effect” relative to a tighter monetary policy stance. When using monetary policy, you can’t have one without the other.

Update:  Marcus Nunes has a related post on the yen.  Yes, there are actually people claiming the yen is too weak (not Marcus!)  Isn’t the yen right up there with the Swiss franc, vying for the title of the strongest fiat currency in the history of the world?