Central banker pay for performance

Nick Rowe had a recent post suggesting that we might want to consider tying the pay of employees at the Bank of Canada to their performance.  He noted that the BOC’s most important objective is 2% inflation.  But how do we compensate each employee on their contribution to meeting that goal?  After all, there is only one monetary policy, toward which each employee contributes.  Here’s how:

1.  Each member of the Canadian monetary board votes on a policy setting for the monetary instrument (short term rate or monetary base.)  They are told to vote for the instrument setting that they believe is most likely to lead to 2% inflation.   The BOC counts the votes sets the monetary instrument at the median vote.  A year later all those who were “right” get paid a $1000 bonus.  All those who were wrong get $1000 deducted from their paycheck.  The votes occur once a month.  Being “right” means voting for a more contractionary than average instrument setting if inflation overshot the target, and vice versa.  Thus if you voted for a 4.5% policy rate, but the median vote was 3.5%, your vote was more contractionary that the median.  In that case are considered right if actual inflation exceeds 2%, and wrong if actual inflation falls short of 2%.  (At the zero bound you’d vote on a monetary base setting.)

2.  Why stop here?  Let’s open up the committee to all 6.7 billion humans.  One man, one vote.  Make participation voluntary.  And wouldn’t one dollar, one vote be more effective at getting the optimal instrument setting?  If you are still with me we have arrived at my 2006 Contributions to Macroeconomics paper.

3.  Why not adopt the policy in the US as well?  And how about switching from a 2% inflation target to a 4% NGDP growth target?  Now we have arrived at my 1989 Bulletin of Economic Research paper.

That’s right; Nick is proposing CPI futures targeting.  At least I hope he is.


Tags:

 
 
 

10 Responses to “Central banker pay for performance”

  1. Gravatar of Indy Indy
    6. June 2010 at 17:20

    Is there any paper which has tried to argue against NGDP targeting? What, in your view, is the most powerful criticism or counterargument available?

  2. Gravatar of Doc Merlin Doc Merlin
    6. June 2010 at 18:02

    @Indy:
    There was an argument a while back that went thus: what if you had 10% CPI inflation but per capita NGDP growth was still on track (lets say 4% growth).
    Are you willing to cut per capita NGDP growth to fight the inflation?

    I’m not exactly sure how to answer this critique.

  3. Gravatar of Doc Merlin Doc Merlin
    6. June 2010 at 18:04

    Markets in money creation. I love it. Now if only Bill and I could only persuade you to support competitive money issuance.

  4. Gravatar of Bret Bret
    6. June 2010 at 19:37

    Why 4%? I prefer 5%. Was there some empirical discussion I missed that support the 4% target?

  5. Gravatar of Nick Rowe Nick Rowe
    7. June 2010 at 04:15

    That wasn’t *quite* what I was proposing Scott 😉

    Funnily enough, while working at the Bank of Canada about 10 years ago, I did propose something rather similar. Only it would be *models*, not people, who would get “rewarded” for having given the “right” advice, as you define “right”. And the models would be “rewarded” by being listened to more.

    (The horizon was the Bank of Canada’s preferred 18 to 24 months ahead for annual CPI, rather than your proposed 1 year, but that’s a detail.)

    I found the top people at the Bank surprisingly (to me) receptive to new ideas, even “left-field” ones. They listened.

  6. Gravatar of scott sumner scott sumner
    7. June 2010 at 06:41

    Indy, I am sure there are, but don’t know of any.

    Doc Merlin, I have heard that argument, but it makes no sense. If you switched to inflation targeting under that sort of supply shock, the depression would be even deeper.

    I am OK with competitive money, as long as there is a single medium of account.

    Bret, I’ve also been pushing for 5%. But with futures targeting you don’t need to worry about liquidity traps, so in the long run 4% would be better. Right now I am sticking with 5%, because we lack futures targeting.

    Nick, You said;

    “Funnily enough, while working at the Bank of Canada about 10 years ago, I did propose something rather similar. Only it would be *models*, not people, who would get “rewarded” for having given the “right” advice, as you define “right”. And the models would be “rewarded” by being listened to more.

    (The horizon was the Bank of Canada’s preferred 18 to 24 months ahead for annual CPI, rather than your proposed 1 year, but that’s a detail.)”

    I am OK with a longer lead time, as long as we use level targeting. The entire academic process is supposed to weed out the weaker models, but we seem to keep revisiting the debates of the interwar period, so I’m not sure that we are making much progress.

    Of course I was kidding about you supporting this idea. 🙂

  7. Gravatar of fmb fmb
    7. June 2010 at 14:06

    This seems like a reasonable place to ask something I’ve been wondering about for a while. Say you use futures to come up with the policy setting to hit your NGDP target in 1 year. What if that setting undermines your ability to hit your target in 2 years? We don’t just want our path to intersect with the trend at a particular point in time, we also want it to be reasonably parallel. It seems like if you correct for a shock too quickly that the volatility of NGDP would get unreasonably high. If we’re 10% below trend, it might be better *not* to make that up in 1 year and instead expect to be 5% below trend a year from now, rather than be on trend in a year but having to turn sharply to stay there.

    Perhaps my proposed answer will also help explain what I’m getting at. It seems like the central bank might want some discretion to adjust the horizon (or, maybe they need options targeting to set the horizon of today’s futures contract). Perhaps normally they use futures to adjust the policy setting to hit their NGDP target 12 months out. But, after a shock it might make sense to allow more than 12 months to get back on track — they could suspend the current contract (it should still settle, of course) and start trading a new one with a longer horizon. This would enforce level targeting (always trade some contract with some horizon where you target having the level back on trend) but allows for some discretion about the pace of adjustment after a shock.

  8. Gravatar of scott sumner scott sumner
    8. June 2010 at 13:00

    fmb, I think 18 months or 2 years are defensible, but i also think a promise of a fairly quick return to trend would reduce deviations in the first place. But I’d be thrilled with a two year time frame as well, so I have an open mind on the issue. My hunch is that if it were done, it would be a two year time frame.

  9. Gravatar of fmb fmb
    8. June 2010 at 14:49

    To clarify: I did not mean to quibble over the “normal” horizon, but rather to raise some issues that come up when considering that we’re trying to hit a target over time, not just one target at one point in time. As you say, “a promise of a fairly quick return to trend would reduce deviations in the first place”, so perhaps my concerns are overblown. But, presumably there are plenty of skeptics who still believe that a shock could put us far enough off course that we’ll struggle to get back, and a futures scheme should hopefully be expected to perform well under such circumstances.

    As I understand your proposals, the Fed sets and NGDP goal for some future date (let’s say 12 months out), and stands willing to both buy & sell futures that settle based on differences in NGDP from target on that date. Every future they trade generates a corresponding open market purchase or sale.

    1. How long do they stand willing to buy/sell? Presumably they need to *stop* trading a given contract well before expiration or they’ll just get picked off. (Probably each contract should only trade once, in a call auction.)

    2. Presumably if the fed stops trading a given contract, they create a new one. Maybe this happens once a quarter, say with an initial maturity of 12 months. At any given time, the fed trades (and uses those trades to dictate its policy setting) the “on-the-run” that expires in 9-12 months. But, the “off-the-run” that expires in 6-9 months will still be affected by that policy setting.

    I think having an on-the-run and an off-the-run would be required, but that it creates some potential problems that are missed when the scheme is imagined as setting one policy to hit one target (perhaps there are details I’m unaware of, of course). I attempted to describe one of these in my previous comment. I’ll try to come back later and elaborate a bit more on that.

  10. Gravatar of ssumner ssumner
    9. June 2010 at 04:34

    fmb, I only have them target on the run contracts. You can get fine-grained estimates of daily NGDP by taking weighted averages of quarterly data. So they could target a different contract every day. Of course the market value of two adjacent days would be very highly correlated.

Leave a Reply