Bonds Get Pre-Fed Jitters After a Mostly Flat Day

Bonds began the day in moderately weaker territory, mostly following an overnight sell-off in Europe.  Trading levels had recovered enough by 10am that we could consider the overall trend to be “mostly flat.”  In fact, MBS came fairly close to breaking even just before 2pm, but then things went sideways.  Actually, “sideways” is the wrong word.  Prices dropped somewhat abruptly and Treasury yields spiked to new super-long-term highs.  Without any clear scapegoats, we’re forced to rely on explanations filed under the category of pre-Fed jitters.  One would think traders already knew how jittery they were before 2pm today, but sometimes you don’t know until you see how jittery everyone else is.

Econ Data / Events

Housing Starts

1.283m vs 1.44m f’cast

Building Permits

1.543m vs 1.443m f’cast

Market Movement Recap

09:06 AM Selling momentum building.  10yr up 6bps at 4.361. MBS down almost an eighth.

10:44 AM Nice recovery but still weaker on the day.  10yr up 2.8bps at 4.329.  MBS down an eighth. 

01:34 PM Best levels of the day for MBS, down only 2 ticks.  10yr up 3.2bps at 4.333

02:46 PM Quickly down 7 ticks in MBS.  10yr up 5.8bps at 4.359. 

Results of the August Residential Construction report from the U.S. Census Bureau and the Department of Housing and Urban Development were decidedly mixed. While permits were issued at a rate higher than anticipated, housing starts sunk to the lowest level since June 2020, Construction was started on residential units at a seasonally adjusted annual rate of 1.283 million, an 11.3 percent decline from the July level of 1.447 million units. Further, the earlier results represent a downward revision from the original estimate of 1.452 million. Both Econoday and Trading Economics had consensus forecasts of 1.44 million units. Starts were 14.8 percent lower than in August 2022. Single-family starts were down 4.3 percent from July to an annual rate of 941,000. This, however, was 2.3 percent higher than the level a year prior. Multifamily starts plunged 26.3 percent month-over-month and 41.0 percent on an annual basis to a rate of 334,000 units. Permits for residential construction rose to a seasonally adjusted rate of 1.543 million units, a 6.9 percent increase from the 1.443 million rate in July and the highest level in ten months. Permits were, however, still down 2.7 percent on an annual basis. The number was about 100,000 units higher than consensus estimates. Construction permits were issued for 949,000 single-family homes on an annualized basis, a 2.0 percent increase from July and 7.2 percent more than a year earlier. Multifamily permits were 14.8 percent higher than the prior month but, at an annualized 535,000 units, down 17.7 percent from August 2022.

Arrr, me hearties! Last week, there be a mighty surge towards them long-term highs in the Treasury yields. But alas! With no bounty of econ data to stir the souls of treasure seekers, the market With no grand loot of data on the horizon, the market be more inclined to keep powder dry afore this Wednesday’s reckoning with the Fed. This tale nearly met an early end after the market took a cannonball shot in the mornin’, makin’ us all clutch our doubloons as 10s had a scrape with new multi-decade peaks. Alas, fairer winds prevailed and the ship stays afloat for its date with destiny.
In English now (debatable):
After last week’s surge toward long-term highs in Treasury yields, and with an absence of big ticket data on the horizon, odds favored a sideways consolidation ahead of this Wednesday’s Fed events. That remains mostly true, but only after markets recovered from a threatening morning sell-off. 10yr yields technically hit new multi-decade highs, but only by 1/100th of a percent. Bonds seem to be calming down since then. Most of the movement was attributable to positioning as opposed to organic reactions to data/events.

We’ve seen plenty of discussion and questions about the role of oil prices in the current bond drama.  The simplest way to to think about oil is that it is, of course, an input for inflation, and thus of some consequence to bonds.  Additionally, in a vacuum, it’s also an economic metric (higher = stronger economy, all other things being equal).  That said, it can also suffer from temporary headline-driven distortions as well as longer-lasting trends in currency valuations (i.e. the Euro currency deflation in 2014 led to a huge drop in oil prices as U.S. currency strengthened).  Either way, traders can’t help but consider the general suggestion in the chart since mid August.  Maybe the Fed will talk about it tomorrow…

On average since 1970, new home sales volumes decline 7% month-over-month in September and 13% quarter-over-quarter in the final months of the year, according to analysts.

Major real estate players are headed to trial next month over challenges to current commission practices, in which the seller covers fees for both themselves and the buyer.

High mortgage rates and the lack of inventory for sale pushed the sector into a downturn twice recently – between May and November last year and in May of 2023 – according to First American

Housing costs grew to an average of $2,632 during the four weeks ending Sept. 10, according to a Redfin report.

The real estate investment trust arm of two affiliate mortgage-related companies could reduce other debt and finance purchases of loans, servicing or securities.

Mortgage rates began the day in slightly higher territory, but recovered after lenders offered mid-day price improvements.  These sorts of mid-day changes are not uncommon when the bond market improves enough during any given business day. Assigning objective numbers to the changes makes them less interesting.  The average lender was only 0.02% higher than Friday this morning, and the improvement brought them 0.01% lower than Friday.  These moves are so small that many borrowers would not see any detectable different in mortgage quotes between today and Friday. Volatility stands a better chance of increasing on Wednesday afternoon when we get the next rate announcement from the Fed.  Even if the Fed doesn’t hike rates (and they probably won’t), other information released in conjunction with that decision can have a big impact.

With limited data of any particular relevance ahead of this week’s Fed announcement, the path of least resistance in the bond market would be for a broadly sideways drift between now and Wednesday.  While that’s arguably proving to be the case as the domestic session matures, it didn’t look to be the case in the overnight trading hours.  Evidence of the drama remains on the chart under the “high” reading for 10yr yields of 4.415.  The fact that such a thing could occur and not occur at the same time is due to the over-the-counter nature of the bond market.  Long story short, 4.415 wasn’t “real,” even though it technically happened.
In an over-the-counter (OTC) market, prices or yields exist as both ‘bids’ (how much a buyer will pay) and ‘asks’ (how much a seller will sell for).  When it comes to something like the 10yr Treasury yield, that means choosing one of the two in order to display a yield on a website or data terminal, etc.  In other words, it’s not common to see the 10yr quoted as 4.325 /  4.326.  
The convention is to use bid prices.  That’s usually no big deal, but sometimes some fairly ridiculous things can happen to the bid.  It can periodically drop off the face of the earth for a variety of reasons.  After all, it doesn’t much matter to the potential buyer considering they’d be getting a great deal if any seller actually dropped prices to meet the comically low bid.
This is basically what happened overnight. Tokyo was closed for a market holiday, but one dealer continued entering bid prices erroneously.  Technically, those were real bids entered by a real potential buyer.  Of course, no one would sell at those yields (i.e. super low prices), but the bids hit the screen nonetheless. 
In this particular case, it looks to be a calculation entry error as the dealer in question did indeed follow the trend of the more liquid Treasury futures market during the time that yields were in the 4.4+ territory.  You can clearly see when the error was caught as yields swooped instantly back to the prevailing range.  You can also see when European markets came online and restored the tight correlation between futures and cash.  Volume also picked up at the same time.