Should inflation targeting central banks hold a portfolio of assets that consists only of foreign lottery tickets? No, because if its portfolio of foreign lottery tickets became worthless, and the central bank were unable to get a bailout from the government, it would be unable to buy back the currency it has issued. So if the demand for its currency happened by sheer chance to fall at the same time, it would be unable to buy back that currency and unable to prevent inflation rising above target.
Should inflation targeting central banks hold government bonds? No. Government bonds would be even worse than foreign lottery tickets. At least the value of a portfolio of foreign lottery tickets would be uncorrelated with the central bank's need for assets. The value of a portfolio of government bonds would be negatively correlated with the central bank's need for assets. If expected inflation increased (or if the natural rate of interest increased), the (real and nominal) value of the government bonds would fall, and the (real and nominal) demand for the central bank's currency would fall at the same time. So the central bank would find itself short of assets in exactly those circumstances where it needed those assets the most.
Only an idiot would advise an inflation targeting central bank to invest 100% of its portfolio in government bonds that promise to deliver fixed amounts of the same currency that the central bank issues. It's like advising the owner of an orchard to invest 100% of his pension plan in apple futures. If the price of apples falls, the value of his orchard falls, and so does the value of his pension plan.
[Update: An inflation targeting central bank is writing butterfly CPI call and put options on the liability side of its balance sheet, so should not be holding inverse CPI futures on the asset side.(I probably said that wrong. Sod finance-speak.)]
What is a risky asset for one person can be a very safe asset for another person. Apple futures would be a very safe asset for the pension plan of a banana producer who likes to eat apples. Owning a bet that my house will burn down is a very safe asset for me, but a risky asset for anyone else. I call that bet "insurance". Similarly, what is a safe asset for a central bank to own could be a very risky asset for someone else.
What we are looking for is an asset that has exactly the right sort of risk for a central bank to own. We want an asset with uncertain returns, where those returns are positively correlated with the central bank's need for assets to buy back the currency it has issued.
And the answer to that question depends on the central bank's monetary policy target. For example, gold is a risky asset whose value fluctuates a lot. But it is a very safe asset for a central bank on the gold standard, that targets a fixed price of gold.
But targeting a fixed price of gold is not a sensible monetary policy. Gold is not a safe asset for a sensible central bank that targets inflation, or NGDP.
The best asset for an inflation/NGDP targeting central bank to hold would be a risky asset whose value is strongly positively correlated with deviations of inflation/NGDP from target. The ideal asset would becomes worthless whenever expected inflation/NGDP falls below target. So if the government committed not to bail out the central bank, the central bank would be unable to buy back currency it had issued, unless expected inflation/NGDP subsequently rose above target and raised the value of the central bank's risky asset. [Update: what I am looking for is a commitment device, like burning your boats so everyone knows you can't retreat even if you broke your promise to fight.] Because permanent changes in the stock of money have a much bigger effect on inflation and NGDP than temporary changes. But what we want are permanent changes that are conditional on whether inflation/NGDP is permanently above or below target.
Some sort of lottery ticket or bet, where the payoff is zero if inflation/NGDP falls below target, and big if inflation/NGDP rises above target, would be the ideal risky asset for central banks to own. If you made the central bank truly independent, with a no-bailout clause. (Though I wonder who would take the other side of that bet?) Update: maybe central banks should be writing options on inflation/NGDP, with the target as the strike price, so they would be forced to print and pay big money if inflation/NGDP fell below target? (Thanks louis.)
But government bonds are the worst possible risky asset for central banks that target something sensible to own. They have exactly the wrong sort of risk.
Worthwhile Canadian Initiative 
Frank: “Don’t you want the central bank buying and selling output gap futures instead of nominal GDP futures?”
No. Go and read Friedman 1968.
Nick,
Okay, I have read Friedman 1968 here:
Click to access Friedman_AER1968.pdf
Under “Section III – How should monetary policy be conducted”, Friedman lays it out:
1. “Accordingly, I believe that a monetary total is the best currently available immediate guide or criterion for monetary policy”
2. “A second requirement for monetary policy is that the monetary authority avoid sharp swings in policy. In the past, monetary authorities have on occasion moved in the wrong direction…More frequently they have moved in the right direction, albeit often too late, but have erred by moving too far”
Not sure how that makes NGDP futures a better choice over output gap futures. Over the long term (multiple recessions / booms), it would seem that the money supply growth rate would be negative because the central bank would be buying low and selling high.
Maybe I just don’t understand the approach that the monetary authority will be taking with NGDP futures. I am only assuming the central bank’s operations – buy NGDP futures during a recession, sell them during a boom – similar to what the central bank does with government bonds and did with gold. If this is incorrect, then please let me know.
Thanks.
Here is my stupid idea of the month.
Why not do all monetary policy through a lottery and adjusting the size of the paid-out to the needs of the money supply?
In a year when the target is likely to be met – pay out exactly what is raised, when we are ahead of target pay out less (or pay in something other than money), and in years when we are below target just print up some additional prize money.
A CB that did this would never go bankrupt,could it?
“Why not do all monetary policy through a lottery and adjusting the size of the paid-out to the needs of the money supply?”
“Why not do all monetary policy through a lottery by adjusting the size of the pay-out to the needs of the money supply?”
Frank: read section 1, on what monetary policy cannot do. (Or just read a first year macro textbook, for that matter, and look at the LRAS curve.)
MF: I’ve hear worse ideas. But the only way to have a lottery that is sometimes negative is to raise taxes on everyone and vary the lottery payout. Have a lump sum tax of $100 per person per year, then vary the lottery between $0 and $200, to get a net payout of anywhere between -$100 and +$100. But those lump sum taxes are going to cause problems. Very poor people will starve, if the lottery has a $0 payout.
Nick,
Okay, from section #1:
Monetary policy cannot peg interest rates for more than very limited periods.
Monetary policy cannot peg the rate of unemployment for more than very limited periods.
I am not claiming that having the central bank buying and selling NGDP futures or output gap futures will be successful in changing either the interest rate of the rate of unemployment. I am saying that from a pure positive money supply growth rate perspective (addressed in section 3), output gap futures work better than NGDP futures because the central bank would not be operating as a profit center (buying low and selling high).
With NGDP futures (assuming a constant quantity of futures), over multiple business cycles, the central bank would be reducing the money supply on a permanent basis.
I realize that Friedman advocates a constant money growth rate, and did not explicitly say that the growth rate should be positive. But I am implying from his recommended positive 3-5% that he might have a problem with a negative growth rate.
‘But the only way to have a lottery that is sometimes negative is to raise taxes on everyone and vary the lottery payout”
I’m missing the need for taxes – couldn’t you either charge more for tickets than you pay out (and burn the difference), or pay out more than you raise (and print up the difference). All the lottery machines could be linked up to computers that calculate the payout based up-to-date NGDP info. And for most years (if you are targeting a positive rate of inflation) the returns will be greater than the stake it would be a pretty popular lottery to play. To prevent a disincentive to play in the years when you need to shrink the money supply you could (rather than offering smaller cash prizes) offer deferred prizes that will only pay out in future years when the money supply needs to expand again.
MF. Aha! I misunderstood. These are voluntary lotteries. (But who would buy a ticket if you knew the central bank wanted to reduce M?)
Well, most lotteries make a profit for those who run them and that doesn’t deter people from buying tickets – plus they could pay part of the prize as future money (bonds that had to be held until M needed to increase) if people were reluctant to play.
Market Fiscalist,
Whose claim on money printed by the central bank is senior – the lottery player or the bond holder? When the central bank decides to increase the supply of M, they now have two means – payments on bonds or payments on lottery tickets.
If lottery tickets are senior claims – winning ticket holders get paid before bond holders, then it is unclear whether people would buy tickets knowing they are going to get their winnings in bonds and future lottery winners may receive cash settlement before they do.
If bonds are senior claims – bond holders get paid before winning ticket holders, then why not just have the central bank sell bonds and skip the rigamarole of lottery tickets?
The lottery ticket discussion reminds me of a hypothetical Nick once brought up – ‘what if the only new money is interest paid on old money’. Here’s a hybrid approach.
Give everyone with a social security number a perpetual bond that entitles the bearer to a discretionary quarterly interest payment made by the CB. The interest payment would be equal to the amount of money that the CB wants to add to the money supply in the given period. No need to buy assets to increase the money supply.
From an accounting perspective every interest payment is a loss to the CB, but functionally this poses no limitation on the CB since cash is not a true liability.
To reduce the money supply, the bank can issue and sell new bearer bonds. This doesn’t commit the bank to grow the money supply in the future, only commits it to pay less to each existing bond holder than it would do otherwise.
Nick Rowe: “But who would buy a ticket if you knew the central bank wanted to reduce M?”
Well, they might not know. The CB might not know. Also, as Market Fiscalist points out, actual lotteries are bad bets in terms of expected monetary payoffs. That does not necessarily mean that for some people they are bad bets in terms of utilities. Lower income people in the US buy state lottery tickets, because winning is a game changer — or so they believe. As the lady said, “My pension is the lottery.” If she does not win, she will not have much in her old age, but she is used to that.
Assets then that are positively correlated with inflation.
Inflation linkers (TIPS), Oil, Gold, Foreign Government bonds.
I think that most central banks do own some sort of mix along these lines.
Update: maybe central banks should be writing options on inflation/NGDP, with the target as the strike price, so they would be forced to print and pay big money if inflation/NGDP fell below target?
Excellent post on why NGDPLT does not work, at least not as envisioned by Sumners. And if the above proposal is adopted, given, as history has shown, that money is largely neutral, you can bet that a future George Soros will take a central bank ‘to the cleaners’ and cost some taxpayers a lot of money. Recall on Black Wedneday the Major government had the option of ‘unlimited funding’ for the Bank of England to defeat Soros, yet it blinked. Do you think the US Fed could take on all of Wall Street, when there’s a concerted push betting the Fed will fail to make the strike price? The lesson would be: ‘don’t fight Wall Street’. It might be a rare event, but the history of markets shows it will happen. So ask yourself: when the US taxpayer is asked to make a payment of, say, $2 trillion USD because the Fed lost a bet vs Wall Street, do you think the Tea Party will get a surge of support? I think so. Audit then abolish the Fed indeed.
Ray: Google “Neutrality of Money”, then remember that NGDP is a nominal variable.
Nick, you are spot on with the critique — bonds are the stupidest of all CB assets. You and others totally miss the point about gold, however — it is quite stable.
Click to access WGC%20purchasing%20power%20of%20gold%20copy.pdf
Many currencies have appeared- and disappeared- over the centuries, gold might not be such a bad holding for a long lived central bank — who knows, gold holdings might enhance the longevity of monetary regimes.
Looking narrowly at the US during the relatively brief postwar window just possibly could lead one to incorrect conclusions over gold.
The point is not what is perfect — pegging in any form to these bizarrely reified abstractions of CPI or NGDP is entirely implausible — but what works. Gold works infinitely better than bonds.
In the quest to find the perfect mechanistic provision of base money, why assume reliance on dubious PhD star chambers in the first place? We have an incredible innovation called a marketplace. Float (now interest bearing) reserves to anybody, including outside the banking system. Have banks compete harder for reserves, and let the marketplace demand drive the creation of base money across the board — all the CB has to do is meet demand, nothing more. If the CB and Treasury are the same, how would this be different from simply selling bills?
If floating reserves does not float your boat, float the Fed — allow people to buy and sell equity rights to receive base money.
Use gold as the automatic asset (assets are mostly irrelevant in a no-delivery world), allow the broad public to freely buy and sell both IOR reserves and currency cash base money in any form, in absolutely any volume. Problem solved.