Stumbling and Mumbling

Monetary policy and Newcomb's problem

Macroblog reports that the markets are unusually uncertain about where the fed funds rate is heading. Could it be that one reason for this is that the Fed faces a version of Newcomb's problem?
To remind you, the problem is as follows.
A super-being, with amazing powers to predict your behaviour, offers you two boxes. In one, there is $1000 for certain. In the other, there is either $1 million or nothing.
The being offers you a choice: to take only the $1m or nothing box, or to take both boxes.
The catch is that he put the $1m into the box yesterday only if he has predicted that you'll take that box alone. If he predicted that you'll take both boxes, he left it empty.
What do you do? There are two theories.
Evidential expected utility: I must take only the one box. The super-being has predicted that I'll do this, and has kindly put $1m in. If I take both boxes, he'll have predicted that too, and put nothing in the box, so I'll have only $1000.
Causal expected utility: The $1m is either in the box or not by now. My choice won't affect the contents. So I must go for both boxes.  That way, I'll pick up $1000 for sure, with a chance of the $1m.
Everyone's argued the toss over this puzzle for ages. The message Robert Nozick  - who brought the puzzle to wider attention* - took from it is that there are two valid but contradictory ways of taking decisions. Rationality is not the clear-cut concept we  think it is.
What's this got to do with monetary policy?
Plenty. The Fed could think like this.
Evidential expected utility: if we hold off from raising rates, the private sector will take it as a signal that we - the experts - are worried about the economy. They'll therefore start worrying too, and cut spending and investment. We'd better raise rates rather than surprise people.
Causal expected utility: A rate pause will simply cause the economy to be stronger than it would be if we raised rates. So let's take out a little insurance against the possibility that Katrina will have serious effects on activity. That's Macro 101. 
Two decision rules conflict.
If you think this is absurd, consider this paper (pdf) by Andrew Caplin and John Leahy. They describe a model in which private agents respond to interest rate cuts by cutting spending, because they anticipate more economic weakness and rate cuts, with the result that gradualist monetary policy is counter-productive.
And if you think this is tricky, here's a further wrinkle. Causal and evidential rationality, said Nozick, are not the only forms of rationality. There's also symbolic rationality; some things (like voting?) are worth doing not because they make us materially better off, but because they symbolize who we are.
Introducing this form of rationality gives us a further dimension to the Fed's decision. Does it want to symbolize that it's keen to fight an inflation threat, or that it's sympathetic to those hurt by the effects of Katrina?
It's no wonder the market's confused.
* in this superb book. FWIW, I reckon too many people are Nozickeans because of Anarchy, State and Utopia, and not enough are Nozickeans because of this.

Are rate moves random?

It’s no surprise that the Bank of England has cut the repo rate today. What is a little surprising is that anyone should think this is the first in a series of cuts.

Logic tells us it shouldn’t be. Under inflation targeting, interest rate moves should be random. This is because rates should be set with regard to a forecast of where inflation will be in two years’ time; the consensus is that it takes this time for rates to affect inflation. An efficient forecast will embody all available information. It will therefore change only as a result of surprises. Surprises should follow a random walk; they ain’t surprises if they don’t. And if the inflation forecast changes randomly, so too should interest rates: Charles Goodhart, a former MPC member, pointed this out here (pdf).

However, since the MPC was set up in1997, rate moves seem not to have been random. Every first cut or first rise has been followed by a move in the same direction. There have been no single rises followed by cuts, or single cuts followed by rises.

Now, eight years of data isn’t sufficient to draw firm conclusions; there have been only three phases of rate rises and two phases of cuts. But what data we have suggests rate moves are positively serially correlated, rather than truly random.

This supports those forecasting more cuts.

But why are rate moves (apparently) serially correlated? There are at least three possibilities:

1. The Bank under-reacts to news about inflation, and does too little.

2. As William Brainard showed in 1967, it’s optimum for policy-makers who are uncertain about the effects of interest rates on the economy to move rates by less than pure targeting would recommend.

3. Big moves in interest rates are signals, as well as control mechanisms; Andrew Caplin and John Leahy discuss this approach here (pdf). A huge cut might signal that the Bank believes a recession is on the way. Private sector agents, believing the Bank is well-informed, might respond to this by cutting spending. In this way, big rate cuts might lead to recession. To avoid this risk, the Bank would make only gradual rate moves.

For the purposes of forecasting rate moves, it doesn’t matter which of these three is most important; this speech (pdf) by Paul Tucker, an MPC member, shows that Brainard uncertainty does feature in their thinking.

From the point of view of judging the MPC’s effectiveness, however, we should hope that 2 and 3 are more relevant than 1.

Dancing on the grave of monetarism

The Bank of England's monetary policy committee seems to be dancing on the grave of monetarism. Today's minutes (pdf) show that two members of the committee voted to cut interest rates earlier this month. This is despite the fact that figures at the time showed that the M4 money stock had grown by 10.8 per cent in the previous 12 months; subsequent figures show that growth has accelerated to 11.6 per cent, the fastest real increase  since 1996.
However, this warranted just a single line in the minutes; it "raised questions about the extent of the weakness in domestic demand." There's no reference to this in the discussion about the immediate policy decision.
But should the committee pay more attention to the boom in bank deposits? Eyeball econometrics suggests so; accelerations in real M4 in 1987, 1995 and 2000 all led to accelerations in consumer spending. This paper has even found that real  M4 "provides the best single leading indicator of recessions." And common sense says we should at least think about why money demand has risen. To argue that it has no implications for future demand and hence inflation, one must show that there has been a permanent increase in the demand to hold bank deposits. Why might this have happened at a time when real interest rates are relatively low and when stock markets are rising?
Perhaps, therefore, the MPC is tempting fate by dancing on the grave of monetarism. As one prominent economist has said (pdf):

The absence of money in the standard models which economists use will cause problems in future, and...there will be profitable developments from future research into the way in which money affects risk premia and economic behaviour more generally.

Who said that? Bank of England Governor Mervyn King, that's who.

Limits of transparency

The Bank of England has failed. That's the gist of the FT's report of this working paper (pdf) by Ana Lasaosa, a Bank economist.
Ms Lasaosa has found that since the Bank became independent in 1997 interest rate decisions have had just as big an impact on asset prices as they did before independence. This suggests Mervyn King's ambition that monetary policy will become boring - because markets should know the Bank's decision in advance - has not been  achieved.
But perhaps there's another interpretation here. Maybe Mr King's ambition was never realistic in the first place. Maybe central bank transparency is consistent with big reactions to interest rate decisions. There are two reasons for this.
1. A central bank will always know its policy reaction function better than the markets do, so a rate decision will always convey some information. Think of it this way. For much of the post-97 period, inflation has been below target. The Bank has therefore often had a choice: to raise rates to pre-empt rising inflation, or to wait for more data. Its decision here tells the market something about its assessment of the strength of inflationary pressures relative to the merits of collecting more data. Markets should react to that.
2. Markets might not be rational (you don't say). A rate change, even if economists have anticipated it, might move markets simply because it draws investors' attention towards inflationary or recessionary pressures.
Decisions by a transparent central bank, therefore, can affect asset prices. And the converse is true - decisions by an opaque or capricious central bank might not do so. Imagine such a Bank were to raise rates. Half investors might think: the Bank is determined to reduce inflation. I'll buy bonds." The other half might think: "the Bank thinks inflationary pressures are greater than I thought. I'll sell bonds." These two reactions might net out, so bond prices don't change.
There is, therefore, no necessary link between central bank transparency and market reactions to interest rate announcements. Maybe Mr King was wrong. Maybe interest rate decisions can never be wholly boring.
There are lessons here for the US. If British experience is anything to go by, the Fed's recent move towards greater transparency - publishing minutes of FOMC meetings earlier - might not make interest rate decisions any less important for markets.

All talk

You might be expecting a blog about fiscal policy. So, here's one on monetary policy.
William Polley points out that, despite its apparent success in the UK, some economists doubt whether inflation targeting works.
Why don't we push this argument further, and ask whether any monetary policy works at all? Ingemar Bengtsson does just this. He says: "the central bank's control of inflation - or interest rates - is true as long as it is believed."  And no longer.
This is because price-setting is a coordination game; if one company expects others to set high prices, it will set high prices, so we'll get inflation.
Now, the solution to coordination games requires what Thomas Schelling called a focal point - some point, perhaps arbitrary, on which everyone's expectations converge. And, says Bengtsson, an inflation target is just such a point.
Whether inflation stays on target , he says, depends solely on the power of this focal point - that is, on the success of the Bank's rhetoric. Central banks, he says, have no other proper tools to control inflation.
What Bengtsson omits - though it corroborates his argument beautifully - is the "price puzzle" identified years ago by Chris Sims. This is the paradox that rises in interest rates lead to rises in inflation.
If all this is right, it destroys the managerialist view of  central bankers as near-deities with hydraulic control of the economy - whereby they pull monetary policy levers with predictable consequences.
Instead, central banks perform not open market operations, but open-mouth operations. They are all talk.

Does thinking help?

Ruth Lea tries to resurrect Norman Lamont's reputation. In this pdf from the Centre for Policy Studies she says he gave Gordon Brown a "golden economic legacy" of low inflation and stable growth. It was Lamont, she reminds us, who introduced the inflation-targeting which contributed to the UK's fine macroeconomic performance.
In strict consequentialist terms, this seems right. And it is not merely a party political point. In this paper (pdf) published by the Bank of England last year, Luca Benati uses some statistical tests to estimate that there was a structural improvement in inflation and GDP volatility in late 1992 - when inflation targeting was introduced. he says:

It is implausible  that the inflation targeting regime did not play any role in the marked increase in overall UK macroeconomic stability [after 1992]. A reasonable interpretation of the evidence is probably that the introduction of inflation-targeting, in 1992, was one of the key factors behind what the Governor of the Bank of England recently described as the NICE [non-inflationary consistently expansionary] decade.

For me, though, this highlights one of the most curious and over-looked paradoxes in the recent history of UK monetary policy.
We can all agree that the UK's membership of  the exchange rate mechanism in 1990-92 was a disaster, whilst the inflation-targeting that followed it was a success.
Herein, though, lies the interesting contrast. The UK joined the ERM after years of public debate and consideration; "should we join?" was a staple question in exams, job interviews and City lunchrooms throughout the mid-late 80s. Sometimes, I got the impression that I couldn't leave the house without someone asking for my tuppenceworth on the question.
By contrast, inflation-targeting was introduced with barely any prior public debate and during a time of panic and disarry in the Treasury. It was chosen because the government needed some montary anchor and the obvious alternatives - money stock or exchange rate targets - had been tried and failed.
So, my questions are: does this show the validity of George Orwell's famous gibe, that the British ruling class will always do the right thing, once it has exhausted all the alternatives? And does it show that consultation, debate and deliberation are not always necessary to produce good policies?

Transparency at the Fed

William Polley draws attention to an interesting discussion at the last FOMC meeting. In discussing whether to release minutes of the meeting earlier than in the past, the committee:

debated the possibility that the markets would misinterpret the minutes and that the prospect of early release would lead to either less productive discussions at the meetings or to less comprehensive, and therefore less useful, minutes.

To English eyes, this seems a peculiar concern. The Bank of England has been releasing minutes two weeks after meetings for years, and no-one worries much about this problem. Indeed, almost everyone believes monetary policy is better run today than it has been in our lifetimes (not that this is a demanding benchmark.)

So, why did the FOMC decide to follow the English (and the Swedes) and start releasing minutes earlier?

One justification lies in research by Petra Geraats. She’s found that “improvements in central bank transparency are typically associated with significantly lower interest rates than economic conditions would predict.”

The theory is straightforward. If everyone knows a central bank is determined to hold inflation down, they won’t need high interest rates to compensate for the risk of unforeseen inflation.

There is, though, a quirk here. Dr Geraats has also found that the most transparent central banks also have clear inflation targets.

By contrast, the Fed has no such clear goal. Its mandate is to “promote effectively the goals of maximum employment, stable prices, and moderate long-term interest rates.” As every undergraduate knows, this is confused; the Fed has (in effect) just one tool, so how can it do two jobs?

So, here are some questions. Could it be that what FOMC members were worried about is not that market participants would stupidly misinterpret the minutes, but rather that they would more reasonably misinterpret the FOMC’s precise goals? Does this small increase in transparency at the FOMC therefore strengthen the case, at the margin, for it adopting an inflation target?