Sunday, March 28, 2010

Treasury Spread Model: No Chance of Double-Dip

A few weeks ago, the New York Federal Reserve updated its "Probability of U.S. Recession Predicted by Treasury Spread" with data through February 2010, and the Fed's recession probability forecast through February 2011 (see top chart above). The NY Fed's model uses the spread between 10-year (3.69% in February) and 3-month Treasury rates (0.11% in February) to calculate the probability of a recession in the U.S. twelve months ahead (see details here).

The Fed's model (
data here) shows that the recession probability peaked during the October 2007 to April 2008 period at around 35-40%, and has been declining since then in almost every month. For February 2010, the recession probability is only 0.57% (about 1/2 of 1%) and by a year from now in February 2011 the recession probability is only .054%, the lowest reading since April 1993.

According to the NY Fed Treasury Spread model, the recession ended sometime in middle of 2009, and the chances of a double-dip recession through early 2011 are essentially zero.


At 3/28/2010 11:17 AM, Anonymous Anonymous said...

A lot has happened since February. CBO says debt is going to hit 90% of GDP. There was a weak response to treasury sales. 10s of millions of Americans just got hit with new taxes. SS just went red.

I think the east coast bankers need to go back to their calculators.

At 3/28/2010 12:06 PM, Blogger Dr Nikan Firoozye said...

The fed and most economists look to the yield curve for its predictive power based on the papers by c harvey and estrella and hardouvelis but miss some key points. The slope steepness is coincident with recessions and its flatness is coincident w overheating economies. The slope doesn't forecast. Bloody hell like traders have a frigging crystal ball.

The fact that recessions have mostly followed overheating economies and recoveries usually follow recessions have lead people to the spurious belief in the amazing predictive power. Since 1970 there have been maybe 8 recessions. 8 data points does not make a model!!!!

If the yield curve believed wholeheartedly in a recovery it would begin to flatten like mad and with the fed unwinding some of its special measures you'd better believe flattening is a real risk now. But the fact that it isn't massively flattening means that the belief in recovery is fragile.
Moreover one key ingredient that all economists tend to miss is risk premia. (Good Lord, how many look at 5Y5Y breakevens for inflation expectations. Downright dumb.)

Vol, inflation uncertainty, growth uncertainty, fiscal uncertainty, each can be attributed with the steep curve. And the massive govt spending and issuance (def/gdp although it can be very correl to the output gap is a good coincident indicator of slope)

I think its time for economists to read up on finance!!

Or start paying attention to the gifted few among their ranks like Tristani, Ang, Piazzessi, Cochrane, Hoerdahl, (and even Fischer, Bernanke's advisor) who actively study risk premia in markets and attempt to unravel the differences between a rise in breakevens and a rise in inflation outlook, or a steeper curve and the possibility of rapid recovery.
Anyhow a bit more than two cents there.

At 3/28/2010 12:18 PM, Anonymous Anonymous said...

"1970 there have been maybe 8 recessions. 8 data points does not make a model!!!!"

Some one should tell Michael Mann at Penn State.

At 3/28/2010 3:42 PM, Anonymous Benny The Man said...

Die recession, die, die, die!

And as to the predictive power of economic and investing models: I find they work until they don't.

At 3/29/2010 9:04 AM, Anonymous morganovich said...


i fear you are right.

this model is has too little data to be really useful, economics is NEVER that simple, and even if it did work, the data going in is being heavily infulenced just now by unprecedented government programs.

using this model now is like using a model of how a steel ball rolls across a table (based on 8 prior rolls) and ignoring the large magnet i just turned on.

here's some more grist for your mill:

not sure what to make of the "all the banks are upside-down on this trade" part of the argument, but the fact that 10 year USBY is higher than libor for the first time is some ominous stuff.

it's also easy to imagine how loads of prop desks piled into this swap spread trade at 40bp based on a historical tendency toward significant reversion from that level.

(look at the first chart)


Post a Comment

<< Home