Wednesday, December 17, 2014


Torsten Slok at Deutsche Bank sends the graph, along with some musings on the eternal question: When (if?) interest rates rise, will it look like 1994, or like 2004? Will rates rise quickly, leading to a bath in long-term bonds? Or will rates rise slowly and predictability?

The graph shows you actual short term rates (red) and forward curves. As this lovely graph points out, the forward curve has been predicting rises in rates for years now. And it's been wrong over and over again. Economists all over have been forecasting a robust recovery too, and that hasn't happened either.

(To non-finance people: The forward rate is the rate you can lock in today to borrow in the future. So the forward curve ought to reflect where the market expects interest rates to go. If people expect rates to rise more than the forward curve, they rush to lock in now, which drives up the forward curve. Also, the forward curve is a cutoff between making and losing on long-term bonds. If interest rates rise following the forward curve, then long bonds and short bonds give the same return. If rates rise slower, long-term bondholders make more money. If rates rise faster, long bonds make less than short or even lose money. So, should you buy long term bonds? Compare your interest rate forecast to the last dashed line and decide.)

The chart .. makes you humble when it comes to the timing of the first rate hike.
But once the Fed starts hiking, they will likely raise rates faster than the market currently is anticipating. Think about it: The Fed has basically decided that they will only start hiking rates once there are signs of inflation.. If the economy is overheating, then raising the fed funds rate to 0.5% is not going to slow the economy down....To cool the economy down, the fed funds rate needs to be above the neutral fed funds rate, which we estimate to be get inflation under control, the Fed will likely have to raise rates well above the neutral level, potentially above 5%...
So his scenario is, interest rates low and more good times for long term bonds until (if) inflation substantially exceeds 2%, then a big rout, as small rises will not do much quickly to dampen inflation. More like 1994.

An interesting view into the brains of bond traders:

The reaction I often get in client meetings when we discuss these issues is along the lines of: “Yes, I understand what you are saying but I have been positioned for higher rates for several years and it hasn’t happened. As a result I have underperformed my benchmark. Instead, I will now wait until I actually see the Fed raising rates”. The main problem from a trading perspective is that we don’t know when this will happen... In the meantime, rates will remain low, not because investors don’t believe in the recovery continuing but because investors cannot afford to be positioned for higher rates for several years.
There is so much here... "Positioned for higher rates" means "sell my long term bonds and hold short term bonds."

"Underperforming" is true. Anyone who "positioned themselves for higher rates" has watched as those willing to hold the risk of longer term bonds got higher yields, and higher prices as well. As anyone who "positioned themself for the end of the internet boom" did in 1997, or who "positioned themselves for the end of the credit boom" in 2005.

But think of the madness of "underperforming my benchmark" in bond markets. It means that long-term bond investors are hiring active managers, then rewarding the managers based on one-year returns relative to a long-term bond index, which the manager wins or loses by simply going a bit longer or shorter than the index. It would be silly enough for stocks -- rewarding managers for taking a bit more or less beta -- but it's double silly for bonds, because when bond prices go down, yields go up, and you always end up back where you started. Rewarding active bond managers for one-year duration bets is just... the standard way this nutty business works apparently.

But it echoes conversations I've had over the years. Me: "so, you really think two percent on long-term treasuries is a good deal? Inflation won't bust two percent for 20 years?" Trader: "Are you kidding? There is going to be a bloodbath. But I think it's still going up for a while before the rout." Not a recipe for long-run stability. And yet rates did not move, and trader was right, year after year.

The "main problem" from a trading perspective is that you can't sell after prices already went down!

Risk premiums

The alternative interpretation of Torsten's chart is risk premium. The market expects zero rates forever, and the upward sloping forward curve and great returns to long term bonds are just the premium for holding the risk of long-term bonds.

The deep trouble is, we really don't know that much about this premium. The models basically use regressions. And the longer we see an upward sloping forward curve and no movements in rates, the more "models" will say "it's a risk premium."

Monika Piazzesi and I did our best to get a handle on this risk premium business. The bottom line, there typically is a big risk premium early in recessions, but later in recessions the forward curve is more likely to signal rate rises that really are coming. Just how late is "late," and how sure you are about the diagnosis, is the big question separating academics from traders.

But you can't have a "risk premium" without risk. That interpretation just changes the probability of a 1994 event, not that it can't happen.


  1. How do you think the drop in oil prices will affect the economy both in terms of the robust (or bust) recovery and the impact on interest rates?

  2. Prof Cochrane,

    Could you explain what you meant by "but it's double silly for bonds, because when bond prices go down, yields go up, and you always end up back where you started. Rewarding active bond managers for one-year duration bets is just... the standard way this nutty business works apparently."

    Thank you

    1. Take a 10 year inflation protected Treasury for example. Its value in 10 years is what it is. Interest rates can change, but not the value of this security. If tomorrow its price goes down, then the rate of return from tomorrow to year 10 goes up, so that the value in year 10 is completely unaffected. The one-year return is completely meaningless to understanding the 10 year performance. And, if you're investing in long term bonds, it's the long term performance you care about.

    2. John,

      "Take a 10 year inflation protected Treasury for example. Its value in 10 years is what it is."

      Not always. Back when the U. S. government was running surpluses in the 1990's, the Rubin Treasury decided to attack the long end of the government bond market by retiring that longer dated debt early. And so there is call risk.

      When a government buys back its debt prior to maturity, it typically buys it back at par, though I am not sure that it is legally required to do so. Can a government forceably retire it's debt at a market price when that market price represents a discount to par?

  3. A nice analogy for forward rates: it's an insurance policy for people who can't afford a rise in rates. Therefore, predicting interest rates with forward rates is like predicting fires using fire insurance.

  4. Markets can remain irrational longer than you can remain solvent.

  5. Think Japan.
    Is the Fed now doing anything the BoJ didn't do 1992-2002?
    Except US federal deficits are smaller.
    And you expect what, and why?


Comments are welcome. Keep it short, polite, and on topic.

Thanks to a few abusers I am now moderating comments. I welcome thoughtful disagreement. I will block comments with insulting or abusive language. I'm also blocking totally inane comments. Try to make some sense. I am much more likely to allow critical comments if you have the honesty and courage to use your real name.