Arguing with the Best, or Why You Should Lock RIGHT NOW

Dan Green of Mortgage Reports is about as good as there is in the business.  He’s been at this a long time, he blogs and he tweets (@mortgagereports) and following him is a worthwhile enterprise.

And today, as insane as this is, I’m going to pick a fight with him.  Okay, not really.  But he did post today on a mortgage “myth” that he takes apart point-by-point.  Specifically, the “myth” that mortgage rates take the stairs down and the elevator up, i.e. that rates rise much faster than they fall.  He argues that this is false.  And I think his argument is all wet.

To prove his point, he calls out bond rates from the last 12 months, and shows – convincingly and correctly – that bond rates have been much more likely to fall than to rise, and more likely to fall dramatically than to rise dramatically. Here are the data:

So, if we compare these groupings to the last year’s actual, daily price changes, and we see an interesting pattern emerge for rate shoppers.

  1. Days of no change in rates : 65 days trending worse, 63 days trending better
  2. Days of 0.125% change : 36 days of higher rates, 50 days of lower rates
  3. Days of 0.250% change : 21 days of higher rates, 50 days of lower rates
  4. Days of 0.375% change : 6 days of higher rates, 3 days of lower rates

In other words, mortgage rates [sic] were unchanged for nearly half of the last 12 months. For the other half, they overwhelmingly moved toward “improvement”.

Now, we should throw out 0.125% changes because they’re somewhat “ordinary”; it’s just one tick higher or lower in rates. Instead, let’s focus on big shifts in pricing which, in turn, lead to big shifts in rates.  That’s where we see the myth debunked, specifically.

On days with big rate changes — 1/4 percent or more — decreases in rates outnumber increases by a 2:1 margin. That’s a huge difference.

His data is unassailable.  So his argument appears sound.

Except for one little bait-and-switch that he probably didn’t notice himself (see the [sic]).  These data are for bond rates.  But Dan says “mortgage rates” in his analysis.  If they were the same thing, then the data would support his premise exactly.  But they aren’t.  Everyone knows they aren’t.  There is not and never has been anything like a one-to-one correlation between bond rates and mortgage rates.

Worse than that, I think Dan’s entire premise is flawed.  Here’s the premise of his analysis:

  1. Changes of less than 25 basis points often lead to no change in mortgage rates
  2. Changes of 25-37.5 basis points often lead to a 0.125% change in mortgage rates
  3. Changes of 37.5-75 basis points often lead to a 0.250% change in mortgage rates
  4. Changes of 75+ basis points often lead to a 0.375% change in mortgage rates

But this isn’t true, or at least, it’s only true in one direction – higher. There’s a reason that it’s proverbial among mortgage people that rates are sticky down, that they don’t drop nearly as fast as they rise. That is because there is a disconnect between bond movement and mortgage rate movement. When bonds move higher, rates should fall, and they do, but not very fast.  And when bonds move lower, rates should rise, and they do, but far out of proportion to the movement in bonds.

Dan has his data, and he points out that bonds have moved higher, taking rates down, far more often than they have moved lower, taking rates up, over the last year.  I don’t mean to be snide here, Dan, but duh.  Anyone can see that.  The bond market has been very robust this past year. Your data show exactly what we ought to see in a hot bond market.

But my data show how banks actually react to bond market movement, and what we see backs up the conventional wisdom and contradicts Dan’s argument.  To demonstrate this, I have the intraday reprices from three different mortgage lenders for the past six months.  And the story THOSE data tell is very different than the one the bond market tells.

To wit:

Intraday reprices for the better: All lenders – 29 (14, 11, and 4)

Intraday reprices for the worse: All lenders – 54 (20, 15, and 19)

There were, on top of this, five second reprices to the worse, making a total of 59, or more than two negative reprices for every one positive reprice.  Even if you throw out the most conservative lender, the total is still 39 to 25 in favor of reprices to the worse.  And yet, over this period, the FNMA 4.0 bond rose dramatically, from a low of 98.74 to a high of 104 at one point in mid-October (currently trading at 101.68).  How can this be, unless lender reaction really is skewed to raise rates instead of dropping them?  Dan’s data show that there are twice as many days of robust positive movement than negative movement in the bond market over the past year.  Taking that into account, what we see here is that lenders are four times as likely to raise rates than to drop them, when the market moves.

It’s actually even worse than that.  Lenders improved prices, in the 29 improvements, an average of .188 bps.  But the average deterioration was .262.  Not only are there significantly more deteriorations than improvements, when lenders move pricing, the moves to raise rates are far larger than the moves to lower them.  There were five days in the last six months that saw two negative reprices in the same trading day, but no days when there were multiple price improvements, this in a market where the bond rose over 600 bps in four months.

Bottom line, Dan, you have convincingly demonstrated that when bond demand is high, bonds rise.  I’m not sure this needed demonstrating, but there certainly isn’t any doubt about it after your analysis.  But you either forgot to establish the correlation between bond rates and mortgage rates, and how the movements correspond, or else the data support no such correlation.  My data are far less scientific than yours, but mine show very stiff circumstantial evidence that lenders behave exactly as the conventional wisdom says they do: they price higher faster, and more aggressively, and they drop rates reluctantly and as slowly as they can get away with.

All of which is unfortunate for rate shoppers.  Dan does conclude his post with some very sage advice: if you’re more upset with a 1/8 rise in rate than you’d be happy with a 1/8 drop – and most people are – you should be locking as fast as you can.  I’ll even extend that.  If you can live with the rate you have right now, take it, because there’s a 4x greater chance that you’ll get something worse than something better if you wait.

Leave a Reply