Friday, August 8, 2014

Interest Rates Should Have Risen More (updated)

I have been disappointed in the movement of interest rates so far this year.  I have been positioning myself to trade volatility around a range of rates that corresponds to a first rise in the short term rate in the first half of 2015.

The market has basically come to agree with me.  Here is a graph of the changing expectations of the market.  When rates shot up in the summer of 2013, the initial rise expectations shot all the way back to the fall of 2014, which was too optimistic.  Then, they fell all the way back to the winter of 2015, which was too pessimistic.  And, as employment has shown improvement this year, the expected date of the first increase has crept back to spring 2015 - currently around April or May.  This is precisely where I wanted it to go.

But, I expected this to correspond to forward rates for June 2016 Eurodollars of just over 2% and for June 2017 of just over 3%.  Instead, rates have trended around a mean of about 1 5/8% and 2 5/8%.  If that had been my target mean, it would have been perfect for this trade.  (Basically, buying when the price declines and selling when it increases, profiting from random movements over time.)  But, the farther the price moves from my target, the harder it is to profit from the position.  Why haven't the prices followed the market expectation?

The reason is that the slope of the yield curve has declined while the expected date of the rise has moved back.  Late in 2013, the slope was up to around 33 bp.  (Rates would be expected to rise 33 basis points per quarter after the initial rise.)  For reference, in the past two cycles, during the rate recovery period, short term rates rose at a pace of 50 to 75 bp per quarter.

So, what's going on?  I think that the market has been surprised by how healthy the economy has remained in the face of the tapering of QE3.  This diminishes inflation fears and also signals a hawkish intention from the Fed.

But, I think this reflects the Wizard of Oz view of the Fed's interest rate policy.  I think that there is a systematic underestimation of how much Fed policy chases the Wickesellian interest rate.  I think the Wickesellian rate is probably already above zero.  This is part of the reason that we have seen an acceleration in economic activity and employment this year.  QE3 appears to have been only slightly accommodative, for reasons I don't completely understand, and so its taper has probably not changed the objective stance of monetary policy that much.  But, before QE3, a non-QE zero rate policy was probably disinflationary.  The increase in the Wickesellian rate over the past 2 years means that at the end of QE3, a non-QE zero rate policy will probably be inflationary.  (As a small caveat, I do think that home prices will start rising again with the expansion of real estate credit at commercial banks, and that this will be deflationary, with regard to the measured CPI.)

Since the Wickesellian rate is so opaque, there is a tendency to vastly underestimate its movement.  So, by the end of 2014, with no QE, Fed policy will be increasingly inflationary.  It seems to me that the Fed is signaling that they intend to raise interest on reserves along with the Fed Funds rate in order to raise rates without sucking up all those excess reserves.  If that's the case, then rates will be rising sooner and at a faster pace than currently expected.

But, even if the Fed starts buying up treasuries, there will be a decent amount of institutional inertia involved, and there simply is no way that they will start buying them quickly enough to chase down the rising Wickesellian rate.  If QE3 was only very slightly inflationary, then reverse QE3 will probably be only very slightly deflationary.  And if there was some sort of inverse causal relationship between QE and the expansion of bank credit, then reverse QE might even accelerate commercial bank real estate credit.  This will probably cause home prices to rise and the Wickesellian rate to rise, rents to moderate and measured inflation to moderate, with the net effect of keeping the Fed looser than it appears, at least as long as there isn't a freak out if home prices start rising too quickly.

So, I think the expectation of a slowly rising rate is backwards.  Rates should be right where I expected them to be.  And I blame all of you for messing it up.  ;-)

Update:  Scott Sumner offers an alternative explanation.  He thinks the flat yield curve reflects the disappointing GDP numbers.  He has a good point.  I have been putting more weight on the employment numbers than the GDP numbers, to a certain extent.  Here are some graphs which show the historical relationship between employment and real GDP relative to recent movements.  GDP growth was outpacing employment growth in the early part of the recovery, then abruptly moved below trend in 2011, and has bounced around with the long term trend as the upper limit since then.

In the end, it will depend on whether GDP rebounds to reflect the labor market or the labor market moderates to reflect the GDP performance.  I expect the former, but that's only one man's forecast.

However, I do think it is important to distinguish between the eventual level of interest rates, which I agree with Scott, will probably be low, and the speed at which they get from here to there.  My speculative position is that they will end up lower than the current forward market price, but that they will get there more quickly than the market currently has priced it.  I go into that topic here.

Update #2:  In response to input from Benjamin Cole, I adjusted the scatterplots for hours worked.  I thought he made a good point, so I was surprised to see that it weakened the long term correlation between labor growth and GDP growth.  It does pull the relationship closer to the long term trend back in 2010, but hours worked has recovered from the cyclical drop, so it doesn't change the recent relationship between real GDP growth and employment growth, during which real GDP growth has been low, but not outside the historical range.


  1. TravisV from TheMoneyIllusion comments section here.

    "Jeff Gundlach Warned Us Rates Could Fall, And They Did — Here's What He's Saying Now"

  2. Interest rates are tricky right now because the yield curve is so distorted by the zero lower bound. Normally most rate changes are from level changes, and almost all the rest are from changes in slope, with a small occasional effect from curvature.

    Now, there are two distinct curves in the yield curve that are both changing frequently, both in their location on the curve and their convexity. So there are three different yield curve segments all changing based on different inputs.

    The decrease in the long bond yields has been entirely due to the limit on the long term bond rates coming down. But, look at 5 year yields. They have held pretty steady at last summer's high.

    I prefer to look at interest rates through forward rates at specific dates in order to isolate these effects. I expect rates over the next couple of years to surprise to the high side, but the experienced rates over the remaining years of the 10 year window will probably justify the low 10 year rates we have today, even if the rates in the early years come in a little higher.

  3. That's why one has to look at the 30 year yield versus the 5 year yield. That give a much better picture of what the markets are thinking, how much risk appetite the average investor wants to take.

    1. Do you have a method for separating out inflation expectations, risk premiums, NGDP expectations, etc.? Couldn't a falling 30 - 5 spread signal either an increased appetite for risk or pessimism about future NGDP growth, depending on what factors are moving it?