Stephen Williamson has doubts about ngdp targeting

Do read his entire post, it is full of content and difficult to excerpt.  I would make a few points:

1. A reasonable range of monetary regimes may not matter so much under good times, but we are choosing the regime for the occasional very bad time when a strong response is needed.

2. Williamson’s points about the instability of seasonal ngdp are good, but arguably considering seasonal cycles renders all or most macro theories somewhat incoherent.  Given the extreme agnosticism we should then end up in, what is a good policy rule?

3. I believe overnight financial markets could adjust to a variety of reasonable regimes, and indeed the evidence across nations appears to confirm this.

4. Scott for one would definitely admit the all-importance of eliminating interest on reserves.

5. I would add a Straussian reading of ngdp targeting: “The Fed won’t ever do it, because they don’t like to tie their hands.  But talking about it may steel their will, and give them a policy rationale, when more expansionary action would be desirable.”

Oddly Williamson does not consider what I consider to be the strongest objection to ngdp targeting, namely that in times of extreme crisis, when loan markets are collapsing, it may force the central bank into a dangerous-and-not-really-output-restoring ratio of currency/credit to meet the ngdp target.

On the off chance that Scott writes a reply to Williamson’s critique, I will link to it in due time.  Update: Here it is, I read it after writing my post.  It is interesting to compare our responses.


Eliminating the interest on reserves or putting a penalty rate on reserves is not the cure. Small negative rates make little difference to safe-haven flows - the Swiss tried enforcing negative rates on offshore deposits in the 70s and the inflows still came in. Large negative rates will lead to people hoarding bank notes and every central banker knows this. Quoting from the SNB Vice-Chairman Jean-Pierre Danthine - "With strongly negative interest rates, theory joins practice and seems to lead to a policy of holding onto bank notes (cash) rather than accounts, which destabilises the system."

If you need to really enforce large negative rates on reserves or deposits, you need to abolish or tax bank note holdings a la Gessel, Eisler, Buiter etc. In this case we've removed the zero-bound anyway and the transmission mechanism is still rate cuts.

The fundamental problem is that in the modern world of shadow money supply generated via the repoing of eligible collateral, rehypothecation etc, the monetarist concept of high-powered money is simply meaningless. And the idea that simply eliminating interest on reserves restores the old system is fantasy (as i discuss in more detail here ).

Tyler, it has taken a long time to start some serious discussion of monetary theory and policy. Unfortunately the players are just defending positions they took years or decades ago and far away from the optimal openness of mind that Bryan Caplan refers to in this post

A serious discussion should start with a functional and IO description of how each player believes that the U.S. system of payments and the U.S. system of financial intermediation work today. I bet that neither SW nor SS can provide it. Perhaps Nick Rowe who has been trying hard to do it. Can you?

Note: Just to make clear how terrible available descriptions of those systems are, please read this column posted yesterday

I used to think inflation targeting was probably best. The events, and Scott Sumner, slowly changed my mind. Now I think NGDP level path is probably best.

"Perhaps Nick Rowe who has been trying hard to do it."

Thanks. Glad someone noticed. But I don't think I can really succeed.

Don't be so hard on yourself: though you don't have a Wikipedia entry yet, you're almost famous says Google. But as a non-economist, what struck me was no mention of Fischer Black's theory that the Fed *follows* the markets in the long (year?) or intermediate (months?) term. So even if the Fed 'targets' a certain money supply growth rate, if that target is not what the market itself wants, then within a short time (weeks?) it will create arbitrage opportunities that will be exploited and the Fed policy will fail. Another thing struck me: how professional economists can be in such violent disagreement over theory. It's as if flat earth astronomers met Copernican astronomers.

Manmohan Singh's work is excellent and one of the few sources on just how far the reality of modern money creation is from the textbook description. Another excellent paper is - copying the intro over:

"Are central banks printing vast quantities of money? This column explains how money-multiplier economics (central banks create reserves that allow commercial banks to create money) no longer holds. Today, non-bank financial institutions play a pivotal role in money/liquidity creation, but hold no reserves. Their lending depends on “private reserves”, mainly highly liquid government securities. Creating more ‘public’ reserves by buying such ‘private’ reserves doesn’t trigger money creation – it just substitutes among reserve types. Open-market purchases only create money if they swap a monetary base for assets that are no longer accepted at full value as collateral in the market."

I beg to disagree with your assessment of the two columns. The authors use old concepts (money, liquidity, reserves, multipliers, monetary base) too vague to describe how payments systems and financial-intermediation systems work today. I'm quite familiar with the contributions of Peter Stella to the study of central-bank transactions and statistics (they were important in the 1980s and 1990s when we were dealing with the quasi-fiscal deficits of Latin American countries that had been hidden into CB statements) but his two columns with Mr. Singh make clear that he is still relying too much on the old, textbook description of central banking.

I'm not sure how the terminology they use matters. The institutional understanding of the importance of collateral chains and rehypothecation in Manmohan Singh's work is accurate and important. We can use different words - say collateral, safety, haircuts, length of chain etc as he does in many of his papers such as .

Guillermo Calvo ( )
starts a recent draft paper with this statement "Future generations will likely remember the turn of the 21st century as the time when mainstream macroeconomics was about to completely remove money and finance from its models, and perished in the attempt." I think Guillermo is right and his recent papers are small steps in the right direction. At best Singh's paper on collateral is another small step to understand the role of collateral in the new systems of financial intermediation. We are, however, far away from understanding how these systems have evolved and how they work today, and in particular how they relate to payments systems.

My biggest worry about NGDP targeting is there is not a single microfounded model (afaik) of the thousand permutations of RBC, New Keynesian, OLG, Menu Costs, DSGE, etc, etc where it is optimal to target NGDP. This might be ok except that these models are super flexible and can be used to show about anything you want if you're clever enough about the framework and the parameters. I have to see an NGDP proponent seriously grapple with this fact and the logic behind it, which David Andolfatto recently summarized quite nicely. Scott Sumner (and Nick Rowe to an extent) blame this deficiency on their inability to do math, which seems like a strange defense imo. Surely a grad student would have written down a model that generated this optimal targeting rule if it were just an issue of technical skill.

NGDP proponents also like picking apart the details on any one particular model that says that NGDP targeting is suboptimal but that misses the point: the robustness of the 'NGDP targeting is not optimal' result to broad sets of models is a huge strike against it.

Hmmm, I don't think they're accepting T-bills at the checkout at Walmart yet...

All you creditists (including Tyler) - how do you explain 1933?

Ashwin, you really think the public wants to hold 3 trillion dollars in cash???

Saturos - here's how it works. The Fed announces a -0.50% negative interest rate on excess reserves, banks reduce their deposit rate to around -0.50%. Most people swallow this safety premium and keep money in deposits. Fed tries a -5% negative IOER - banks reduce deposit rates to -5% and people like you and me decide that it's more profitable to take all their money out in banknotes and put it in a safety deposit box somewhere which costs them say 1% per annum to insure and protect against theft.

On the use of T-bills at Walmart - if you're an institutional client of a bank and hold some T-bills and you need some cash, you can simply borrow from the bank using the T-bills as collateral and get cash. You don't even need to sell the T-bills. And what's more, the bank you borrowed from can usually borrow from some other bank using the same T-bills as collateral (a process known as rehypothecation).

So you don't think the public might decide at some point that they have, I dunno, excess cash balances?

But the guy at the checkout will in the end want to see some cash or my credit card before he hands over that new microwave, won't he?

re #1: Was this not the rational for the creation of the Reserve in the first place? Is there any basis, other than theory, for believing that the Fed will ever be capable of performing its intended task? Will they prove better at hitting NGDP in a future crisis than they were at producing inflation in the last one? Insofar as faith in the Fed's toolkit led people to buy into the housing bubble, wouldn't even stronger tools - like NGDP targeting - have made the situation worse? If this theory has nothing to say about what caused the crisis, what good will it be in addressing (or preventing) a situation that everyone understands is not simply monetary?

The real problem for Sumner's faction is that, for twenty years, inflation-targeting did work. It has not been falsified by theoretical advances, but by empirical setbacks. There is nothing at this point to differentiate monetary theory from Catholic theology. It is an attempt to construct faith out of knowledge... it is intended to be useful rather than true.

So we are discussing theories of action. This speaks to the inability of economists to come up with a workable theory of knowledge. We don't really understand the business cycle or the operation of money - not least of all because the rules appear to change - however one time we did a funny dance and the next day it rained, so now the study of markets begins with the study of theater. Is this really the proper course for economics?

Stephenson hits upon something with his graphs. Assumptions have been made about periodic vs. unpredictable fluctuations, catch-alls like "sticky prices", and the desirability of certain trends in certain aggregate figures. If economics were reduced to what might be proven, would it cease to be of any use as a science? Would it become politically inert like history, musty like archaelogy?

The two dictats of our time seem to be 'grow at all costs' and 'apres mois le deluge'. I worry less about the adventures of the Federal Reserve and the pharmacological exploits of modern economists, than I do about the role of economics. Pumping out theories from assumption and dubious-but-useful policy recommendations has proven, for the moment, to be expansionary for the field; but if economics is merely a branch of political science, it won't long survive the politicians that employ it.

"Given the extreme agnosticism we should then end up in, what is a good policy rule?"

Don't have any policy. Seriously, none. No monetary authority, no SPEND SPEND SPEND fiscal policy. Let the market sort itself out as it always has. Any other approach is destructive.

Comments for this post are closed