Update: Here’s my EconTalk with Russ Roberts, where I describe my views on macro.
A few weeks back someone suggested that I describe how my views differ from the mainstream. A few days ago I did a post describing what I thought was wrong with the standard models of monetary economics. I ended up with a call for a new paradigm and left the impression that I’m about to provide it. One commenter said my last paragraph was “remarkably ambitious,” which is a polite way I suggesting it’s crazy for a Bentley professor to be talking about new paradigms. And he’s right. So I’ll just list some of the areas where my views differ from others, provide lots of links, and then let others decide whether there is anything coherent in my approach.
1. How I think about the stance of monetary policy:
I use three ingredients here. One is Svensson’s notion that monetary policy should target the forecast. Policy should be set in such a way as to equate the forecast of NGDP or inflation, with the target. The second ingredient is Michael Woodford’s argument that AD is not so much influenced by the current setting of monetary policy, as by changes in the entire expected future path of policy. (This post applies the idea to the Great Depression.) Monetary policy actions that are expected to be temporary have almost no impact on AD. And the third ingredient is Robert King’s insight that in a forward-looking IS-LM model an expansionary monetary policy can increase inflation and growth expectations so much that both nominal and real interest rates actually move “perversely.” This means tight money can cause lower nominal and real rates and easy money can cause higher nominal and real rates. And this is not just a theoretical curiosity; this postdiscusses evidence that a contractionary surprise in December 2007 actually caused 3 month T-bill yields to decline.
Now let’s put the ingredients together. Woodford argues that in a liquidity trap you need to promise to hold rates low for a long time. The problem with that recommendation is it confuses low rates with easy money. Woodford is right that in a liquidity trap you need to promise easy money for an extended period of time. But as we saw in Japan low nominal rates may merely reflect weak NGDP growth expectations, not an easy money policy. Milton Friedman pointed out in 1997 that near zero interest rates are more likely to reflect tight money rather than easy money. So although Woodford is right to recommend a credible and sustained policy of monetary ease in a liquidity trap, low interest rates aren’t the way to implement that policy. (In fact, the tightest monetary policy imaginable would actually produce zero interest rates forever.) Instead, we need a credible policy of targeting the future path of the price level or NGDP. Later I’ll explain why level targeting is appropriate.
There are several ways to recognize the stance of monetary policy, but the best way is to try to infer the market expectation of the future path of the goal variable, and compare it to the policy goal. Svensson recommended using the central bank’s internal forecast to avoid the circularity problem, but I prefer the market forecast. Thus policy is “tight” if next year’s expected NGDP is 14.6 trillion, while the target is $15 trillion.
Now let’s think about how this approach to monetary policy forces us to rethink some of our basic concepts. The textbooks have traditionally taught demand management by considering fiscal and monetary stimulus. Under a futures targeting approach that no longer makes any sense. Monetary policy is set at the level expected to produce on-target growth in NGDP, leaving fiscal policy with no role to play.
Under this sort of approach many of the metaphors that we traditionally employ would become wildly inappropriate. For instance, how much of the “heavy lifting” should we leave to the Fed? This metaphor suggests that monetary stimulus requires more effort that not doing monetary stimulus. In fact, exactly the opposite is true. The more monetary stimulus we do, the smaller the national debt, and hence the smaller the burden on the next generation of taxpayers. In addition, under a fiat money regime we can create more money at almost zero cost. Even better, we earn seignorage from money creation, so the less “effective” is monetary stimulus, the more we should want to do it. Of course I am being a bit simplistic here. The more serious argument is that a bloated monetary base and near-zero interest rates are the result of falling NGDP, not on-target NGDP. When NGDP is on-target, interest rates and demand for base money return to more normal levels. Thus a truly effective policy of monetary stimulus would not look anything like what Woodford suggests—it would quickly raise nominal interest rates above liquidity trap levels, and also reduce the bloated demand for excess reserves.
The right metaphor isn’t “heavy lifting,” it’s turning the steering wheel to the proper setting. When piloting a ship it’s no harder to turn a steering wheel to ENE, then to set it at NNE. We need to stop thinking about how much burden we should put on the Fed, and simply think in terms of what an optimal monetary policy setting would look like.
Here is one of the great ironies of our current predicament. A central bank that truly had a credible “we’ll do whatever it takes” attitude would never have to do very much at all; but a timid central banks that is hamstrung by inflation hawks will have to engage in endless seemingly reckless policies such as zero rates, extreme QE, bailing out banks. Australia had the world’s best monetary policy during the recent crisis. They haven’t had a recession since 1991. They keep their NGDP growing. And yet by conventional standards they look like one of the most “conservative” central banks. They never cut their interest rates to zero, and they were first out of the box raising them in the recovery. That’s what effective central banking looks like. (Of course those who follow central banking know that the Bank of Australia is actually one of the least conservative of the major central banks.)
You may find this analogy completely off the wall, but it reminds me of self-confidence. The self-confident person can have lots of fun at a party without trying very hard. The shy guy is working really hard, but just spinning his wheels. If he were to do the right thing—go to the party with an “I don’t give a s*** what other people think of me” attitude, other people would find him more interesting and he’d have a much better time. And it wouldn’t really require more effort, it would require less. It just seems like it’d take more effort, before you take the plunge. Australia is like the self-confident guy. The Fed, ECB and BOJ are the nerds.
[As an aside, this whole post might make more sense if you first read Alice in Wonderland or Through the Looking Glass.]
Recently I’ve become more interested in the way Bill Woolsey frames the issue; rather than using the “targeting” metaphor he uses the gold standard metaphor. Index futures convertibility. The reason I like this is because it helps us better understand what is going on with the money supply and interest rates. You make a NGDP futures contract into the medium of account, and make currency redeemable into that asset at a fixed price. Now both interest rates and the money supply become endogenous, just as under a gold standard. But whereas under a gold standard NGDP can move all over the place despite a fixed nominal price of gold, the ability of current NGDP to fluctuate with a fixed 12-month future expected NGDP is constrained by a complex set of macroeconomic forces. I’ll talk about those in the next section, but for now just think about how interest rates would behave in a world where expected NGDP was always on target. Expected future NGDP is the most powerful force affecting nominal interest rates; both the inflation component and the real component. Under NGDP targeting both the money supply and nominal interest rates would probably be far more stable than under current policy.
2. Recognizing the nature of the macro problem
I use a fairly mainstream AS/AD model when thinking about macro problems. I also find it convenient to define ‘aggregate demand’ as simply aggregate nominal expenditure (which makes the AD curve a hyperbola.) As with many other economists, I believe that dramatic cyclical movements in RGDP are often caused by AD shocks, as NGDP and RGDP tend to move in the same direction in the short run. Over longer periods of time AD has almost no effect on RGDP, only prices are affected. None of this is particularly controversial.
I think that one thing that separates me from other macroeconomists is that I see short run changes in NGDP as being powerfully impacted by changes in future expected NGDP. Thus if the expected level of NGDP one, two and three years out falls sharply, then current NGDP will tend to fall sharply. But the reverse is also true. If future NGDP stays on target, then it is very difficult to envision how near term NGDP could fall sharply. Most people I talk to disagree with me on this. They say that once the post-Lehman financial crisis hit it was too late to prevent the severe fall in NGDP over the next 6 months. Money affects the economy with a long lag. I disagree. I think near term NGDP can be propped up by a credible policy to make up for any deviations from the target path within a year or two. I don’t think people have clearly thought about the implications of their views, as we are so used to declines in current NGDP being associated with concurrent declines in expected future NGDP that we don’t think through what this actually means. (Perhaps someone familiar with the “unit root” literature could provide some evidence that has a bearing (either way) on my argument.)
As an analogy (and I know it’s not perfect) once the US stopped pegging the price of gold in April 1968, both the nominal and real price of gold became far more volatile, indeed several orders of magnitude more volatile. If we went back to pegging the price of gold (which I don’t recommend) both the nominal and real price of gold would become much more stable. When you target NGDP along a given trajectory, both real and NGDP will become more stable. In addition, all sorts of asset prices will also be much more stable. Why? Because changes in NGDP have a dramatic impact on asset prices, as we’ve recently seen. And since asset markets are forward-looking, the mere expectation that NGDP will return to target in a year or two means that (due to intertemporal arbitrage) asset prices will never stray far from a level consistent with on-target NGDP. Put simply, the DOW never would have crashed into the 6000s if it was understood that NGDP was going to return to its target growth trajectory in 2010. The market crashed because investors saw (correctly) that the Fed has no plans to offset the negative nominal shock. Indeed we continue to fall further and further behind; although thankfully things are getting worse at a slower rate than was expected last March, so the DOW has bounced part way back to 14,000.
Why would it help to have more stable asset prices? Here’s another area where I differ from my colleagues. Imagine a forward-looking model like Woodford’s, but make it monetarist. Unlike traditional monetarism, we use Woodford’s emphasis on changes in the expected future path of monetary policy. As with monetarism, I assume an excess cash balance transmission mechanism, but only in the long run. This means an expansion in the money supply that is expected to be permanent will cause future expected NGDP to rise via the excess cash balance effect. This will immediately increase the prices of assets like stocks, commodities and real estate. And this will increase current AD. Because wages are sticky, the higher prices of these assets will boost output of corporate assets, commodities and real estate. And since employment and wealth rise, so does consumption. So policies that are expected to have a major effect on future NGDP also have a powerful effect on current NGDP, just as in Woodford. And that brings us to the issue of policy lags.
3. Policy lags aren’t what they seem.
The standard view of macro is that there are a set of monetary and fiscal policy tools that affect the broader macroeconomy with “long and variable lags.” The transmission mechanism lies in some sort of “black box” that is poorly understood. So policy actions go into this mysterious black box, and a year later out pop some macroeconomic outcomes. And we can use VAR models to estimate the effects of policy. I don’t buy this view of things, as I think it implicitly assumes that asset markets are inefficient. Any expected long run effect on inflation should be reflected immediately in TIPS spreads and CPI futures prices. That’s how we should evaluate policy.
In my research on monetary history I discovered again and again that the policies that economists focused on often had little or no impact on asset prices, even though economic theory suggests that if they were expected to have important macro effects, asset prices should have been strongly impacted. And we know this because when there is a sudden and unexpected change in prices or output, the effect on the stock market is often very dramatic.
So I’d like to see us reconstruct macro in a way that eliminates lags. The basic units of analysis would no longer be inflation and real growth, but rather expected inflation and expected real growth. No more “depression economics,” it would be replaced with “expected depression economics,” aka bad economic policy economics. Now analysis could operate in “real time.” There would be no lags between changes in policy and changes in expected NGDP or inflation. You might ask “but isn’t it actual growth and actual inflation that we really care about?” Actually, I’m not the only one who thinks expectations may be more important. Many new Keynesian-type models emphasize the importance of keeping expected inflation under control, in order to prevent a change in expected inflation that would feed into wages and core inflation. And I think this view is correct. Where I differ from my fellow economists is that I think that with the right approach to macro policy, with a policy of targeting the forecast, this is surprisingly easy to do. Indeed I think it is so easy that if and when we do start targeting the forecast in a serious way, VAR models of policy will almost instantly become just as obsolete as old money demand models from the 1970s. The market has the only structural model that matters. It doesn’t matter whether some model at MIT tells us NGDP is likely to rise too fast, it only matters whether bond investors, and commodity markets, and labor unions think inflation will rise, as NGDP growth expectations are only destabilizing to the extent that they effect financial markets, goods markets, and labor markets. What if an “accident” happens where MIT is right and all the normal people are wrong? Very little, because if we assume a policy where next period’s NGDP will always be expected to return to target, then any short run deviations won’t cause significant macroeconomic instability—for the simple reason that they won’t have been reflected in wage and debt contracts.
Sure there are some lags. The economy is full of sticky wages and prices. But that just means we need to choose a medium of account which has properties consistent with macro stability, given that price stickiness. And I think expected NGDP comes closest. A low and stable expected NGDP growth path will definitely anchor long run inflation expectations; it will also minimize demand-side business cycles and will reduce the impact of supply shocks (when compared to a strict price level target.) And targeting expected future NGDP will do this far more effectively than a backward-looking regime that responds to past deviations of NGDP from target.
4. Markets are smarter than great economists (and also lesser ones like me.)
You’ve heard about how if a tree falls in the forest and nobody hears it then it doesn’t really make any sound. (Perhaps it would be more accurate to say it doesn’t make sound as people think of the term ‘sound’ but it still does make vibrations in the air.) I’d say the same about major policy errors. If a catastrophic policy error is alleged to have occurred, and the markets don’t care, then I say it didn’t occur. Was Greenspan’s decision to lower rate to 1% the root cause of our current crisis? Who’s to say? But for all practical purposes the answer is no. We should not go around acting like a bunch of Sherlock Holmes’s; searching for the secret key to our predicament. It is right in front of us. The policy errors occurred when the asset markets crashed. We don’t have an NGDP futures market, but if you look at CPI futures, as well as indirect measures of real growth expectations such as commodity and stock prices, there can’t be much doubt that NGDP growth expectations plunged precipitously between July and November 2008. Case closed. There’s your smoking gun. No need for an Oliver Stone movie.
Policymakers can’t outsmart markets, and shouldn’t be expected to. If investors didn’t see a crash coming 5 years down the road, then there is no reason to expect policymakers to have seen the crash. (And of course they didn’t.) Experts aren’t superman, although they often claim to be as Einzig suggested in this brilliantly devastating skewering of Cassel from 1937 (a crisis with some interesting similarities to our recent crisis):
“On June 9, 1937, this veteran monetary expert [Cassel] published a blood-curdling article in the Daily Mail painting in the darkest colours the situation caused by the superabundance of gold and suggesting a cut in the price of gold to half-way between its present price and its old price as the only possible remedy. He took President Roosevelt sharply to task for having failed to foresee in January 1934 that the devaluation of the dollar by 41 per cent would lead to such a superabundance of gold. If, however, we look at Professor Cassel’s earlier writings, we find that he himself failed to foresee such developments, even at much later dates. We read in the July 1936 issue of the Quarterly Review of the Skandinaviska Kreditaktiebolagetthe following remarks by Professor Cassel: ‘There seems to be a general idea that the recent rise in the output of gold has been on such a scale that we are now on the way towards a period of immense abundance of gold. This view can scarcely be correct.’ . . . Thus the learned Professor expected a mere politician to foresee something in January 1934 which he himself was incapable of foreseeing two and a half years later. In fact, it is doubtful whether he would have been capable of foreseeing it at all but for the advent of the gold scare, which, rightly or wrongly, made him see things he had not seen before. It was not the discovery of any new facts, nor even the weight of new scientific argument that converted him and his fellow-economists. It was the subconscious influence of the panic among gold hoarders, speculators, and other sub-men that suddenly opened the eyes of these supermen. This fact must have contributed in no slight degree towards lowering the prestige of economists and of economic science in the eyes of the lay public.” (1937, pp. 26-27.)
So I don’t agree with the way my fellow economists go about their Monday morning quarterbacking. I look for the market crash, and infer that that is probably where the policy error occurred. And at least in the Great Depression, I found that it often did occur at that point. BTW, I don’t mean to suggest that Monday morning quarterbacking is always wrong; Smoot-Hawley did cause a stock market crash in mid-1930, as did the National Industrial Recovery Act in mid-1933. My point is that it isn’t fair to blame policymakers for mistakes that were invisible to the markets. Hoover and FDR are certainly fair game.
5. Expected NGDP growth is more meaningful than expected inflation:
a. Inflation is an arbitrary concept that has never been clearly defined, or at least defined in a way that relates to how it is actually measured.
b Where expected inflation is alleged to be important (i.e. nominal interest rates, nominal wage negotiations, etc.) then expected NGDP growth is more useful than expected inflation.
c. When there is a sudden nominal shock hitting the economy then the size of the shock is much more accurately proxied by NGDP than by the CPI.
d. NGDP targeting will provide greater macro stability that inflation targeting.
Some of these issues are discussed here.
6. Recessions should be unpredictable, if we are doing our jobs.
As James Hamilton once argued:
“You could argue that if the Fed is doing its job properly, any recession should have been impossible to predict ahead of time.”
Unfortunately, he then went on to talk about how we might better forecast recessions in the future. To the extent that economists cannot forecast recessions, we should wear that failure like a badge of honor. Would you want to fly on an airline that could forecast its own fatal accidents? The search for a recession forecasting formula is like the search for the Holy Grail. We’ll never find it. If we could forecast recessions we could (and would) prevent them. But we will eventually develop formulas that are capable of “retrocasting” past recessions, if that makes anyone feel better.
7. I also differ from most economists in the way I interpret the stylized facts of macro history:
a. In 1929 a policy of tight money among all the big central banks led to expectations of sharply falling NGDP, and this caused the stock market crash. Stock crashes of equal size that are not associated with falling NGDP expectations (i.e. 1987) don’t result in recessions, much less depressions.
b. The was no liquidity trap in 1932. The aggressive open market operations did fail (as Keynes argued.) So Friedman and Schwartz were wrong about that. But Keynes was wrong about the reason why. They failed because of the constraints of the gold standard prevented the OMOs from affecting the expected future path of monetary policy.
c. In 1933 neither liquidity trap-type conditions nor an almost completely dysfunctional banking system prevented extremely fast NGDP (and RGDP) growth. FDR did what Krugman says can’t be done now.
d. In 1933 inflation rose sharply, despite the most “slack” that we’ve ever had in our entire history. So much for the NAIRU model. (Finally a point my Austrian readers will like.)
e. After the BOJ adopted QE, it hit its zero percent CPI target almost perfectly. Mission accomplished (contrary to what you read in 99% of the discussions of Japan.) Policy was “effective.”
f. And of course my claim to (minor) fame, it was tight money that caused NGDP expectations to fall sharply between July through November 2008. All the other things we observed (discussed here, here and here) were symptoms of that tight money. That includes the second and much more severe part of the financial crisis, as well as the second half of the housing collapse, as well as the entire collapse in the commercial property market. Not to mention a severe world-wide recession, high real interest rates, a soaring dollar, and a good portion of the fall in commodity prices.
g. One area where I do agree with many economists is that I think most recessions are due to demand shortfalls, and hence monetary policy errors. But even here my views are slightly different. I think most economists only understand this about earlier recessions, and they continually misdiagnose each new recession as being due to special factors. When they are “in the moment” they are too mesmerized by the froth on top of the waves, and are missing the powerful currents below the surface that are creating all this turbulence.
8. Escaping from a liquidity trap. (Arnold Kling said I needed this section.)
First adopt the attitude of Peter Pan. (If I think I can, then I can.)
Seriously, there are several options, none ideal:
1. Target the price of NGDP future contracts. Too radical in today’s policy environment.
2. FDR’s strategy—currency depreciation. It worked, but is politically problematic (although it isn’t the beggar-thy-neighbor policy most assume it is.)
3. Unfortunately we are left with the least attractive option. First create an explicit NGDP target. Use level targeting, which means you promise to make up for under- or overshooting. If excess reserves are a problem, get rid of most of them with a penalty rate. Commit to doing QE until various asset prices show (in the view of Fed officials) that NGDP is expected to hit the announced target one or two years out. If necessary buy up all of Planet Earth. In practice, if you forced banks to disgorge their bloated excess reserves, then the monetary base would probably fall from $1.5 trillion to under $1 trillion, even as NGDP growth expectations rose sharply. So you’d being doing open market sales. That’s because banks would hold very few excess reserves if penalized (they’d substitute T-securities) and the public’s demand for cash to put under beds would not be very high if asset prices showed expected NGDP to be at the policy target.
9. What would macro texts look like if I had my way?
By “my way” I don’t just mean if I wrote the textbooks, I mean if I was monetary dictator of the world. OK, so let’s say expected future NGDP is pegged along a 3% or 5% growth trajectory with index futures. What’s left to talk about? Certainly not fiscal stimulus, or multipliers for investment spending. (Indeed, even the current multiplier debate is incoherent.) No more paradox of thrift or liquidity traps. Banking instability no longer affects the broader economy. Fiscal policy becomes a long run topic, focused on issues like savings, investment and the current account. Interest rates would no longer be very interesting, nor would the money supply. There would be increasing emphasis on supply-side problems.
Notice that with the exception of interest rates my dream scenario pretty well describes what was already beginning to occur in higher level new Keynesian textbooks during the Great Moderation. So let’s finish the job.
Of course there is that one glaring exception; nominal interest rates continued to be the center of attention during the Great Moderation, and this proved to be the Achilles heel of the whole system. It turns out that interest rates are about the worst possible “instrument” of policy. Why? Because they fail you just when you most need them. When the economy is in danger of falling into a depression the nominal interest rate will hit zero, and interest rate targeting becomes ineffective. And so here we are, spinning our wheels while the nominal economy falls far short of policy goals. Meanwhile our libraries are full of academic papers with various “foolproof” escapes from liquidity traps, all of which sit on the shelf collecting dust:
Currency depreciation in Japan? The BOJ let the yen soar in value last year.
Inflation or NGDP targets in the US? The Fed has refused to make any, and instead keeps insisting that their goal is low inflation.
In Europe they do have an explicit target, but when they undershot it the ECB refused to take any actions to remedy the situation.
If this doesn’t call for a new paradigm, I don’t know what does.