Cross Sector P. White, L. Wash, H. Cunningham
Treasury yields and corporate spreads have largely retraced their post-payrolls move, supported by the relative strength of the data and normalization in risk sentiment. Next week we look for initial claims to remain rangebound and we think Chair Powell will provide little clarity at next week’s Jackson Hole symposium. RRP balances have declined to around $300bn and amid the tightening in liquidity conditions, we think QT has a few more months left.

Governments J. Barry, P. White, A. Borges, L. Wash
Markets are pricing in a less dovish path than our forecast, but we remain patient before adding duration given long investor positioning. Jackson Hole speeches have not traditionally led to large yield moves, except in 2020 and 2022; we do not expect clear guidance from Chair Powell next week. We contrast the upcoming easing cycle with other easing cycles: policy is more restrictive than at any point in the last 30 years, pointing toward a deeper easing cycle, but financial conditions are easy and financial stress is low, making the case for a slower pace of cuts. Hold 5s/30s steepeners as a low-beta long with a better carry profile ahead of the first cut. We review June TIC data. Initiate tactical 5yx5y inflation swap longs.

Interest Rate Derivatives S. Ramaswamy, I. Ozil, P. Michaelides, A. Parikh
Policy uncertainty is back, and with it jump risk - stay bullish on short expiry volatility despite apparently rich valuations. Maintain front end swap spread wideners, and add exposure to narrower 10Y swap spreads and a flatter 3s/7s swap spread curve. Position for a flattening of the TU inter-CTD swap spread curve. Reserves and RRP are approaching levels where liquidity is less ample - QT is likely in its end game with a few more months of headroom. An analysis of 3M expiry implieds suggests that election premia being priced currently are lower than in May, and in a smaller post-election time window.

Short-Term Fixed Income T. Ho, P. Vohra
Short-term credit investors continue to favor bank CP/CD FRNs over fixed, with this trend likely to continue in the near term. RRP balances have decreased from their local peak but may be nearing a floor. We review July MMF holdings.

MBS and CMBS J. Sim
Continue to prefer UIC conventionals. The July remit data show that serious delinquency rate for private label CMBS increased to 4.74% and we continue to see elevated levels of DLQ buyouts and loan modifications in the CRE CLO market.

ABS and CLOs A. Sze, R. Ahluwalia
BB subprime auto ABS spreads have recovered, fully erasing the previous week’s widening. We widened our base case CLO T1 AAA new issue spread forecast to 150bp (from 130bp prior) and introduced a 175-200bp risk case in the event of a hard landing.

High Yield N. Jantzen, T. Linares
High-yield bond spreads of 369bp are well inside last week’s wide of 424bp, whereas yields (7.65%) too have fallen to a low since August 2022. And tightening 48bp, 52bp, 55bp off the wide, BB, B, and CCC spreads are now wider by 1bp, 16bp, and 42bp in August, respectively.

Municipals P. DeGroot, Y. Tian, R.Gargan
Elevated tax-exempt supply is expected to persist with market reception beholden to rate volatility. We find value in certain high-quality, high-vol structures and examine sector ratings change over economic cycles.

Emerging Markets L. Oganes
In EM fixed income, we are MW GBI-EM local rates, CEMBI and EMBIGD. EM bond flows were -$1.4bn (-0.36% of weekly AUM, down from -$1.0bn).

Source: J.P. Morgan

Roundtrip

• Economics: Headline CPI rose 0.2% last month while the core index increased 0.165%, leaving the year-ago rate to moderate to 3.2%. Meanwhile total retail sales increased 1.0% m/m while the control series rose 0.3% last month. Initial claims declined to 227k and we expect filings to remain roughly constant at 230k at the next print
• Treasuries: Markets are pricing in a less dovish path than our forecast, but we remain patient before adding duration given long investor positioning. Jackson Hole speeches have not traditionally led to large yield moves, except in 2020 and 2022; we do not expect clear guidance from Chair Powell next week. We contrast the upcoming easing cycle with other easing cycles: policy is more restrictive than at any point in the last 30 years, pointing toward a deeper easing cycle, but financial conditions are easy and financial stress is low, making the case for a slower pace of cuts. Hold 5s/30s steepeners as a low-beta long with a better carry profile ahead of the first cut.
• Interest Rate Derivatives: We remain long volatility expressed in 5-year tails as elevated policy uncertainty leads to higher jump risk and more frequent jump days. Normalizing repo rates are likely to bias the front end of the swap spread curve wider while ongoing QT and elevated Treasury supply are likely to bias the long end narrower. We favor 3s/7s maturity matched swap spread curve flatteners. Liquidity conditions have tightened and we think QT is in its endgame, with reserves likely to reach $3tn and RRP balances slightly below $300bn by YE24
• Short Duration: RRP balances have trended lower, reaching $330bn on 8/16, and positive T-bill issuance over the past couple weeks has likely contributed to this trend. That said, we may be reaching a floor as net T-bill issuance stalls in early-September and as MMFs keep some amount of exposure meet liquidity needs. We expect RRP balances to hover near low to mid-$300bn range for the remainder of the month
• Securitized Products: Higher coupons outperformed lower coupons, reflecting shifting demand and relative value up the stack. We prefer UIC conventionals and are underweight 4.5s and 5s. We think mortgage rates can fall to 6% by year-end, but this will have a limited impact on house prices. In CMBS, we think upper IG mezz bonds look cheap to similar duration single-A corporates
• Corporates: HG spreads have now nearly round tripped the most recent selloff. Demand remains solid, especially from overseas investors but the 10s30s spread curve has lagged as yields nudge lower and long-end supply picks up.We revise higher our CLO T1 AAA new issue spread YE24 forecast from 130bp to 150bp
• Near-term catalysts: Aug flash manufacturing PMI (8/22), Aug flash services PMI (8/22), Jackson Hole Symposium (Aug 22-24), Aug Employment (9/6)

In the wake of the July employment report, Treasury yields initially plunged and corporate spreads widened before retracing these moves over the fortnight supported by the relative strength in the data and a normalization in risk sentiment ( Figure 1). Market liquidity conditions have also largely rebounded amid the decline in volatility. While limited, the data over the past two weeks has assuaged the worst fears surrounding an impending slowdown. The July CPI report delivered on expectations, as the core index rose 0.165% m/m last month to leave the year-ago rate to cool to 3.2% with most categories outside shelter flagging as soft, and we are currently tracking core PCE increased 0.122% in July (see US: Cooler July CPI, as expected, Michael Hanson, 8/14/24). Meanwhile the July retail sales report beat expectations, with the total index rising 1.0% m/m and the control group moving up by 0.3%, and strong consumer spending to start 3Q24 presents upside risk to our GDP tracking (see US: July retail sales suggest no slowing in 3Q real spending, Abiel Reinhart, 8/15/24). Most notably, initial jobless claims moved down from a peak 250k at the end of July to 227k, suggesting that after adjusting for one-off shocks emanating from recent weather events and auto-plant shutdowns, the level of claims are not significantly higher than in recent summers (see US: Jobless claims moved down, Murat Tasci, 8/15/24). As we look ahead, all eyes will be focused on initial claims next week, as it will overlap with the employment survey reference week, and we expect claims to remain roughly stable around 230k (see Economics). We also look forward to Powell’s comments at Jackson Hole, though it is unlikely the Chair will offer too much granularity on the magnitude or pace of easing given the importance of the upcoming August employment report.

Figure 1: Treasury yields have backed off their lows and credit spreads from their wides, as the relative strength in the data and risk sentiment has supported a normalization following the July payrolls report

Respective yields and spreads (black dots) within their 1-month range (blue section), normalized as a percentage of their 3-month histories; %

Source: J.P. Morgan

Against the backdrop of easing inflation pressures and above-trend growth, the real Fed funds rate is decidedly more restrictive than before prior easing cycles and it is notable that Fed speakers from across the spectrum have recently highlighted the need to moderate this stance ( Figure 2). Thus, with the disinflationary process resuming and labor markets significantly looser than they were a year ago, we think this makes the case for substantial Fed easing in the coming months, and we continue to project 50bp cuts in September and November before downshifting to 25bp through late-2025. However, markets have clearly moved in the opposite direction this week, pricing in just 30bp of easing by the September FOMC meeting, and 94bp for 2024. It’s tempting to add duration here, with markets pricing a slower and shallower path of cuts than our modal view, but we prefer to be patient, knowing that position technicals could be a drag over the near term. Specifically, our Treasury Client Survey index has extended significantly over the last few weeks to the longest levels since December. Should the conditions present themselves, we would look to add duration in the 3- to 5-year sector of the curve. Meanwhile, we maintain our 5s/30s steepener position as a lower-beta duration long with a better carry profile which is both supported by historical steepening pattern into Fed easing cycles and reinforced by higher term premium (see Treasuries).

Turning to inflation markets, TIPS breakevens lagged the recovery in other markets and remain cheap versus our fair value estimates. With the Fed likely to begin cutting rates next month, we note that breakevens have not exhibited a consistent performance around the start of prior easing cycles. However, a Fed that is cutting due to risks coming into better balance and a desire to return policy rates quickly to neutral should be more supportive for breakevens than a Fed that is behind the curve. Meanwhile, technicals have turned more supportive as well. Demand for TIPS-focused funds has picked up alongside the decline in real yields and improvement in realized TIPS returns, driving the 4-week moving average of inflows to its highest level in more than 2 years. Along the curve, the intermediate sector offers value, and we prefer to express longs in inflation swaps rather than breakevens, given that IOTAs remain tight. Thus, we recommend tactical longs in 5Yx5Y inflation swaps (see TIPS).

Figure 2: The Fed funds rate is considerably more restrictive than at any point over the last 30 years

Real Fed funds rate*, real longer-run median Fed dot, and Holston-Labach-Williams natural rate of interest rate; %

Source: J.P. Morgan., Federal Reserve bank of New York, Bloomberg Finance L.P.
*Fed funds target less core PCE oya

Figure 3: Policy uncertainty remains elevated and is also a feature in 2025, not just this year

Total weights on YE24 and 1H25 policy rate scenarios representing Shallow Cuts*, and Deep Cuts*, as calculated from a decomposition of the implied probability distribution associated with Dec 2024 and June 2025 SOFR futures; May 2024 – Current

Source: J.P. Morgan
* Shallow Cuts correspond to the total weights on the 0 Cut,1 Cut, 2 Cuts, and 3 Cuts scenarios for YE24, and 0 Cut,1 Cut, 2 Cuts, 3 Cuts, and 5 Cuts scenarios for 1H25 . Deep Cuts correspond to the total weights on the >3 Cuts scenarios for YE24, and 7 Cut and 9 Cut scenarios for 1H25. YE24 scenarios outlined in Figure 2 of What’s the rush?.
** We enumerate a list of scenario-specific Normal distributions with fixed standard deviations and means that are separated by 25bp, and then require the implied distribution to be a weighted combination of these individual distributions. The weights are then solved for, by fitting to the observed prices of calls and puts at various different strikes. For more details of our approach, see What’s the rush?.

While it is highly likely the Fed will deliver a cut next month, the magnitude and subsequent pace of easing still remain highly uncertain and this leaves markets in a state of limbo as there is likely to be little clarity provided over the near-term. Indeed implied probability density functions inferred from call and put premia on Z4 SOFR futures indicate comparable weights between both shallow-cut and deep cut scenarios, leaving policy uncertainty elevated ( Figure 3). Against this backdrop high jump risk combined with more frequent jump days is manifesting in high delivered volatility and we think implieds look cheap considering these factors. Thus we maintain our long volatility bias and continue to express this in 5-year tails. We also maintain our swap spread curve flattening bias, as normalization in repo rates should bias front-end spreads wider whereas ongoing QT and elevated duration supply in Treasury markets should bias the long-end narrower. We like expressing this theme through a 3s/7s maturity matched swap spread curve flattener, which additionally is near the steep end of its one year range (see Interest Rate Derivatives).

Turning to the Fed’s balance sheet, RRP balances have trended lower, reaching $330bn on 8/16, and fell as low as $287bn on 8/7, a level not seen since May 2021. Positive net T-bill issuance over the past couple of weeks has likely contributed to the lower facility balance, though we might be approaching some sort of floor as the uptick in net T-bill issuance stalls in early-September. Furthermore, MMFs maintain some amount of exposure to the Fed’s ON RRP to help meet their liquidity needs that cannot be entirely met with repo or T-bills alone. For the remainder of the month, we expect RRP balances to hover in the low to mid-$300bn range (see Short-term fixed income).

This is close to our $250-300bn guesstimate for the likely floor on O/N RRP balances that will be needed for well-functioning repo markets. This raises two important and related questions. First, is liquidity in the system (i.e., the combination of Reserves as well as RRP) still ample? Second, given the likely outlook for the Fed's major liabilities, how much further can QT continue? We argue the ampleness of liquidity is best estimated by examining the response of broader short rates (such as SOFR) to quantity shocks. Borrowing the essence of the idea in the recent note by Fed staff, we look at the empirical relationship between changes in the SOFR-versus-top-of-the-band differential and percentage changes in total liquidity (i.e., Reserves plus RRP balances). The rising magnitude of this rolling (negative) beta does indeed point to tightening liquidity conditions, and suggests that QT is in its endgame, with perhaps a few more months left. Overall, we project the Fed’s balance sheet to end the year at ~$7Tn, with Reserves flat at current levels, ON RRP balances slightly below $300bn, and a lower TGA ( Figure 4, see Interest Rate Derivatives).

High grade credit spreads widened materially in the wake of the July employment report, reaching a YTD high of 125bp on the JULI, before retracing back to pre-payrolls levels in the following fortnight. As HG spreads have made a rapid retracement, some investors have expressed skepticism as to the durability of the ongoing spread rally. Time will tell if the current selloff will mark a local high or something more, but our analysis has shown that it has been, historically speaking, advantageous to buy the dip in HG, as 70% of the time when spreads widened materially, they subsequently rallied over the next three months ( Figure 5, see JPM Daily Credit Strategy & CDS/CDX am update, Eric Beinstein, 8/15/24). Turning to relative value, we note that the 10s/30s curve looks too flat when compared to 30-year Treasury yields, and we could expect this curve to steepen as yields decline into the Fed cutting cycle, especially as the recent dearth in long-end issuance has begun to reverse ( Figure 6, see JPM Daily Credit Strategy & CDS/CDX am update, Eric Beinstein, 8/14/24). We find little evidence that Japanese investors have pared down their cross-currency exposure to US credit amid the carry unwinds, as overnight net buying has been elevated and the relative value proposition of HG credit for Japanese buyers improved as volatility drove JGB yields lower (see JPM Daily Credit Strategy & CDS/CDX am update, Eric Beinstein, 8/13/24).

Down the capital structure, high yield bond spreads declined by 15bp to 369bp and yields collapsed to 7.65%, the lowest level since August 2022, as economic data have bolstered the Goldilocks narrative of positive growth with moderate inflation while 2Q earnings are meeting elevated consensus estimates. That said, concerns over future cooling are still present, as CCC spreads are 42bp wider on the month compared to BB and B spreads wider by 1bp and 16bp, respectively. Meanwhile leverage loan spreads have tightened by 4bp over the past week, amid moderating retail outflows and light capital market activity (see High Yield). We widen our CLO T1 AAA new issue spread YE24 forecast from 130bp to 150bp given the historical precedent for negative returns for risk assets once the Fed begins easing combined with the likely reduction in floater demand. This raises the risk that refi/reset activity may not remain quite as strong, especially as some 16% of the universe is coming out of non-call period in 2024, of which the majority is currently in-the-money (see CLOs).

Figure 5: There have been 37 moderate to major HG selloffs since 2000, and spreads rallied over the next 3 months in 26 of these episodes

JULI spread versus positive and false positive flagged wide during selloffs*; bp

Source: J.P. Morgan
*Selloff is defined as a point where spreads are at their widest in 3 months and remain the widest over next 1 month, and is further restricted to points where peak spread was at least 15bp wider than tightest over past 3 months. Positive flag is where the selloff corresponds to the peak in spreads over the subsequent 3 months, and false positive when spreads are wider over the next 3 months. See JPM Daily Credit Strategy & CDS/CDX am update, Eric Beinstein, 8/15/24

Figure 6: The 10s/30s HG spread curve appears too flat with respect to 30-year Treasury yields

10s/30s spread curve versus regression* model predicted 10s/30s spread curve; bp

Source: J.P. Morgan
* Regression model: 67.65 -5.27*30-year UST yield (%) -0.11*JULI spread (bp), R^2 = 48%, SE = 9.7bp
Shaded areas are where UST 30-year yields = 4 to 5%

Turning to mortgages, higher coupon spreads continue to recover from their early August widening while lower coupons are a touch wider, perhaps on the margin reflecting concerns over bank and/or foreign portfolio reallocations or simply cheaper opportunities up the stack. Spreads remain on the snugger side of things, but the performance of mortgages in the flash rally reinforces the thesis among many mortgage-overweight money managers that mortgages offer comparable yields with less downside risk in sharp rallies to IG ( Figure 7). From this perspective, the recent retracement in corporate spreads may prove a headwind, but lower yields should be supportive of fund inflows and overall we would expect more neutral positioning going forward (see ‘Mission Accomplished’ on MBS vs. HG Corps, Nathaniel Rosenbaum and Nick Maciunas, 8/6/24). We continue to prefer UIC conventionals and are underweight 4.5s and 5s (see Agency MBS).

With a Fed easing cycle likely around the corner, we expect mortgage rates to fall marginally below 6% by the end of the year if our interest rate forecast is realized. This would improve affordability and potentially thaw a frozen housing market, but we do not expect this to be a massive driver of higher home prices as waiting sellers will likely be met by eager buyers, raising supply and demand commensurately. Indeed we look for only another 1.5% of HPI growth over the next 6months. What does this mean for relative value in the non-agency space? At the top of the capital structure, higher coupons (5.5 - 6.5s) should not face much refinancing pressure for the rest of the year and spreads should hold firm. Lower coupons meanwhile would benefit from even a modest pick-up in turnover. In credit, home prices remain supportive but spreads are tight and we look for a more sideways move in the second half. As such any spike in broader credit should be viewed as an opportunity to buy(see RMBS).

ABS spreads narrowed this week following the previous week’s volatility, with demand picking up in the higher yielding segments. Indeed the reaction function of ABS spreads to broader market volatility has been consistent in recent years with lower tier credit showing a higher beta than cash surrogate ABS versus comparable unsecured corporates. This offers a buying opportunity for ABS across the credit spectrum against the base case of no recession in the year ahead. Turning to auto loan ABS, we note that affordability appears to be returning to the market as collateral APRs have levelled off/started to decline this year, loan terms are extending, and monthly prepayments continue to fall. Delinquencies and losses have only modestly deteriorated on the prime side - likely due to the catch-up of compounding effects of inflation on household and consumer expenses among higher quality borrowers- and no worse on subprime in recent originations. Under our base unemployment forecasts, we do not expect significant deterioration in curing trends or an outsized uptick in loss ramps for prime vintages (see ABS).

In the CRE space, on-the-run conduit CMBS spreads widened 4-11bp across the stack, with AS bonds underperforming both within the stack and relative to similar duration single-A corporates. With expectations of a slowing but not breaking economy, we believe upper IG mezz bonds have sold off too much and appear cheap compared to similar duration single-A corporates. Agency CMBS spreads also widened over the past two weeks and Fannie DUS 5/4.5s appear cheap to 30-year FN 5s on both an OAS and ZV comparison. The July remit data show that serious delinquency rate for private label CMBS reached 4.74%, representing a 20bp increase from the prior month, with office loans the marginal driver of the increase ( Figure 8). Finally,we continue to see elevated levels of DLQ buyouts and loan modifications in the CRE CLO market, with $527mn in reinvestments through the entire CRE CLO market in July representing the highest volumes since October 2023 (see CMBS).

Figure 7: YTD excess returns flipped in favor of MBS during the flash rally for the first time since January

YTD Cumulative excess returns on the Bloomberg Corporate and MBS indices, %

Source: J.P. Morgan, Bloomberg L.P.

Figure 8: Within CMBS, office loans were the marginal driver of the increase in serious delinquencies last month

60d+ delinquency rate (including NP matured and FC/REO) by property type, as of July 2024; 5

Source: J.P. Morgan, Trepp

Figure 9: Cross sector monitor

Current levels, change since 8/02/24, 1-year average, minimum, maximum, and current z-score for various market variables; units as indicated

Source: J.P. Morgan, Bloomberg Finance L.P., ICE, IHS Markit
* 8/15/24 levels for 30Y FNCL, AAA CMBS, JULI, US HY, EMBIG, CEMBI, iBoxx Euro HG, US financials, AA taxable munis, GBI-EM Global FX, gold, and brent oil; 8/16/24 levels for all others

Figure 10: YTD returns on various fixed income indices; %

Source: J.P. Morgan

• The July CPI report was tame, as the core measure increased only 0.17%
• Firm retail sales trend led us to revise up 3Q GDP growth to 1.25%
• Initial jobless claims move lower as automaker-related distortions fade
• Powell’s Jackson Hole remarks next week unlikely to tip hand for September FOMC

The immaculate disinflation thesis received more support this week, as inflation continues to moderate against a backdrop of healthy growth in economic activity. The CPI report for July was benign as the ex-food and energy core measure increased only 0.17%, and over the last year is up 3.2%. After a few unwelcome upside surprises at the beginning of the year, core inflation has been well behaved more recently, increasing at only a 1.6% annual rate over the last three months. The lagging rental measures are still slow to moderate, but excluding shelter costs core inflation was up 1.7% in the 12 months through July.

Meanwhile, most indicators of growth were generally favorable this week. Most notably, retail sales rose 1.0% in July and the core “control” measure increased 0.3%. These figures suggest real consumer spending is off to a good start in 3Q and accordingly we revised up our estimate of current-quarter GDP growth to 1.25%. We also separately rejiggered the contour of real GDP growth for 2025, effectively swapping the first-half and second-half growth profiles to reflect more modest growth in 1H25 followed by a moderately stronger rebound in 2H25 as the impact of the front-loaded Fed easing cycle starting next month percolates through the economy. This revision leaves 4Q/4Q and year-on-year growth for 2025 unchanged.

The latest jobless claims data also peeled back some of the recent increase, suggesting the deterioration in labor market conditions seen in the July employment report is not building on itself. Some of the prior rise in claims may owe to extended automaker shutdowns, a factor that also weighed on the July industrial production report released this week. State-level data suggest some drag on payrolls in Texas that might be attributed in part to Hurricane Beryl; impacted sectors showed an earlier rise in jobless claims as well, so we could see a modest bounce-back in August nonfarm payrolls. Beryl does not appear to have distorted the national unemployment rate, however, although the automaker shutdowns might have led to temporary layoffs in the Midwest.

The data flow next week slows to a trickle. Instead, the most notable development will likely be Fed chair Powell’s remarks at the annual Jackson Hole symposium. There has been only limited Fed-speak since the disappointing July jobs report. Of those who have spoken most have asserted that risks to the employment and inflation mandates are close to balanced. They generally have also judged that policy is currently restrictive. In this setting we think it makes sense to get policy back to neutral quickly and believe a 50bp reduction in the funds rate at the September FOMC meeting is warranted. That said, we doubt Chair Powell will pre-judge the outcome of that meeting, given almost a full month of data releases between now and then.

Inflation genie gets back in the bottle

This week saw the release of the full suite of July inflation data—CPI, PPI, and import prices—and the net result is that we expect only a modest 0.12% increase in the core PCE price gauge last month. Both headline and core CPI rose 0.2% on the month, with core very close to our expectations (a realized 0.165% unrounded print versus our forecast of 0.170%) but headline slightly softer. As a result, the year-ago pace of headline CPI inflation eased a tenth to 2.9%, while core slid a tenth to 3.2% as expected (Figure 1).

Both of these readings mark the slowest annual pace of inflation since 2021. The unexpected softness in headline can largely be traced to a flat contribution from energy on the month, while food prices rose 0.2% as expected. Away from food and energy, the details on the core were more varied but broadly aligned with our expectations. Core services firmed 0.3% while core goods declined 0.3% for July. The former was the strongest in three months while the latter was the softest reading for core goods prices in a year.

As we forecasted, shelter prices firmed on the month—albeit a bit more than anticipated, with owners’ equivalent rent (OER) up 0.4% and tenants’ rent up 0.5% on the month (Figure 2).

New and used vehicle prices eased broadly as forecast as well, with the latter accelerating their recent monthly pace of decline to 2.3%. Industry data point to a reversal of this drag in coming months. Airfares (down 1.6%) and postage (up 0.8%) also provided offsetting but broadly anticipated swings last month. Motor vehicle insurance costs appear to have resumed their upward trend after a brief interruption in May; they rose 1.2% in July. More surprising was the 0.3% drop in medical care services, which appears to have been nearly entirely due to an unexpected 1.9% decline in outpatient hospital services. The rest of the subcomponents for medical services rose last month, and several accelerated.

The July producer price index (PPI) report was generally softer than expected, re-establishing a disinflationary trend for producers’ prices after some firming earlier this year. The headline final demand measure rose 0.1% on the month, while the core PPI index (excluding food and energy) was unchanged. With revisions to prior months, year-ago headline PPI inflation has cooled to 2.2% from an upwardly-revised 2.7% in June, while core eased to 2.4%oya from 3.0% in June. Although the pass-through from producer into consumer prices is incomplete and of variable length, the July reading is well within the range that allows the Fed to continue to place its primary focus on the labor market at upcoming policy decisions.

Consumer spending stays firm

The July retail sales report came in strong and points to another solid gain in real consumer spending in 3Q. Based on the retail data as well as a reported decline in utilities in the IP report, our expectation is that real PCE rose 0.2%-0.3%m/m in July. Even if there were no further gains in spending in August and September, real PCE for the quarter would still increase a solid 2.1%q/q, saar.

Total retail sales increased 1.0%m/m in July, supported by a 3.6%m/m increase in sales at motor vehicle and parts dealers. These data are consistent with a 4.2%m/m rebound in unit vehicle sales reported for July. Excluding autos and gas, total sales rose 0.4%. The control category, which further excludes food services and building materials, rose a high-side 0.3%, which comes after a large 0.9% increase the prior month. The control category rose 4.9%3m/3m, saar, reversing a slowdown earlier in the year (Figure 3).

We had expected food services to be a drag on sales given the signal from Chase card data, but in fact that category rose 0.3%m/m. That said, the 2% annualized increase over the last three months is still a slow pace. Results for housing-related components were also decent, as sales at building material stores rose 0.9% while furniture, electronic, and appliance store sales increased 1.0% in July. Sales at both of these store types have trended upward from earlier in the year after being weak last year. Sales in some of the other larger categories, including grocery stores, general merchandise stores, and e-commerce, were all up on the month as well.

Stronger than expected retail sales data for July paint a relatively upbeat picture for consumers, but the preliminary August print of the University of Michigan’s consumer sentiment index showed only a mild improvement that was driven by the expectations component while current conditions index slid a bit. Inflation expectations remained unchanged both at the one-year (2.9%) and five-year (3.0%) horizons.

Housing starts slide

Housing starts and permits declined substantially in July, with starts down 7%m/m to 1.238mn saar, and permits down 4% to 1.396mn saar. Moreover, there were net downward revisions to starts in prior months. The decline in total starts was driven by a 14% plunge in the single-family segment, which left them near their cycle lows from early 2022. Despite modest declines in mortgage interest rates that are likely to continue as the Fed easing cycle commences, we think residential investment growth in the GDP accounts should be soft in 2H given the past declines and given that units under construction are still falling from elevated levels. In addition, builders have been giving a lot of rate incentives to buyers, which they may withdraw as mortgage rates decline.

Excerpted from, United States Data Watch , Michael Feroli, August 16, 2024

Just keep starin’ at the sun, pray for summer’s end

• With the disinflationary process resuming and labor markets looser, the path for substantial Fed easing in the coming months is becoming clearer. We continue to forecast 50bp cuts in September and November, followed by 25bp cuts through summer 2025. Though markets are pricing in a less dovish path than our own forecasts, we think we can be patient before adding duration...
• ...We think investor positioning has had a role in the recent move to higher yields: our Treasury Client Survey Index has reached its highest level since December. This level of the survey index has historically preceded rising yields in subsequent weeks, indicating potential near-term bearish risks
• Chair Powell’s Jackson Hole speech will be closely watched. Historically, these events have not driven significant volatility in Treasuries, except in 2020 and 2022. We do not expect the Chair to give granular policy guidance, as the magnitude of easing will depend on labor market data, particularly the August employment report
• The current environment does not closely resemble the periods prior to the last 4 easing cycles. Policy is more restrictive than at any other point in in the last quarter-century, suggesting more significant easing than in 1995 and 2019, but financial conditions are easier and displaying less stress than prior to the 2001 and 2007 cycles
• We continue to favor 5s/30s steepeners as a representation of our strategic view, as the curve tends to steepen ahead of the first rate cut. This trade behaves like a lower-beta duration long with a better carry profile. The rapid growth of the Treasury market and reliance on price-sensitive investors warrant higher term premiums and steeper curves
• 2s/5s steepeners are becoming more attractive, but a weak employment report is needed to shift steepening exposure further in the curve
• Foreign investors purchased $17bn of long-term Treasuries in June, the least since December 2023, with demand concentrated among Euro-area investors while net selling pressure emanated from Japan

Market views

It’s been a wild two weeks since our last US-FIMS publication, with yields plunging first in the aftermath of the weak July employment report and later following weakness in risk assets due to the unwind of the Yen carry trade. However, yields have risen off their lows, supported by a rebound in risk assets and relative strength in data: with these moves, markets have normalized strongly from their most extreme levels on August 5 ( Figure 11). Initial claims have fallen more than 20k from their local peak just two weeks ago and are firmly back in the range they’ve held since early summer, and the 4-week average has retreated back as well. With the rises in July concentrated in Texas and Michigan now fading, it appears these were indeed technical, related to Hurricane Beryl and auto plant shutdowns, respectively (see US: Jobless claims moved down, Murat Tasci, 8/15/14). Separately, the July retail sales report came in strong, pointing to upside risk to growth early in 3Q24: control group sales rose 0.3% over the month (consensus: 0.1%), and early in the quarter, this indicates consumption has not stepped down materially from the 2.3% pace in 2Q24 (see US: July retail sales suggest no slowing in 3Q real spending, Abiel Reinhart, 8/15/24). Since our last weekly publication on August 2, 2-, 5-, 10- and 30-year yields have risen by 20bp, 15bp, 10bp, and 4bp, respectively.

Though there was normalization in markets this week, it can’t be ignored that inflation data was benign once again. Headline CPI rose 0.2%, allowing the year-ago rate to tick lower to 2.9%, while core CPI also rose 0.2% (0.165% unrounded), leaving the year-ago rate at 3.2%. With CPI and PPI data in hand, we estimate a 0.12% rise in core PCE in July, with the year-ago rate holding steady at 2.6% (see US: Cooler July CPI, as expected, Michael Hanson, 8/14/24). Given three consecutive months of cooler inflation prints and the ongoing cooling in labor markets, this ratifies the market’s expectation the Fed will ease next month. It’s notable that the CPI data seemed to contribute to a tone shift from Fed speakers this week: Bostic’s (Atlanta, nv) comments shifted markedly in two days, hinting at support for a September ease. Separately, Goolsbee (Chicago, nv) and Musalem (St. Louis, nv) both discussed the restrictive stance of monetary policy and seemed comfortable with easing in the near future ( Figure 12). While the Fed is focused on the totality of the data in determining when to lower rates, it seems that this week’s CPI data was the proverbial “nail in the coffin,” though markets have clearly been in this camp for weeks.

From a high-level perspective, it may seem strange that yields rose this week in the face of inflation data that gives the FOMC confidence inflation is trending back toward 2%, but fleeting expectations of an intermeeting or larger-sized cuts were being driven by fears of a more severe slowing in growth and weakening in labor markets, and this week’s data helped to ease those fears for now. What’s more, we think investor positioning has played a role as well : our Treasury Client Survey Index extended significantly over the last few weeks, to the longest levels since December ( Figure 13). Thus, it could be that this week’s moves were somewhat exaggerated by investor positioning as well, as the index had risen to 1.27, nearly 1.5 standard deviations above its 1-year average. Historically, when the Treasury Client Survey Index has moved to this level, yields have consistently risen over the ensuing weeks (see Survey says Using the Treasury Client Survey to predict rates moves, 7/21/23). We will be interested to see next week’s survey, as this technical factor leaves the risks skewed more bearishly over the very near term.

Figure 13: Our Treasury Client Survey Index rose to its highest levels since December, a move that has been traditionally consistent with yields rising over the following weeks

J.P. Morgan Treasury Client Survey Index*;

Source: J.P. Morgan*(Longs +Neutrals)/(Shorts+Neutrals)

As we look ahead, policymakers and academics will descend on Jackson Hole late next week for the Kansas City Fed’s annual monetary policy symposium, and Chair Powell will speak on the economy at 10am ET next Friday morning. Given this backdrop, we think it’s worthwhile to go back and look at the behavior of the Treasury market in the day and week after the Chair’s appearance at Jackson Hole. These events have not frequently corresponded with large moves in Treasury yields, either on the day of the speech or in the week following the Chair’s remarks, as Jackson Hole has not typically been a forum for discussing the near-term path of Fed policy ( Figure 14). 2020 and 2022 were exceptions: in 2022, Chair Powell used this forum to push back on expectations for easing in the following year, and this event represented a turning point for policy expectations, with long-term yields rising more than 100bp over the next two months (see Short but not sweet, Michael Feroli, 8/26/22). Separately, in 2020, Chair Powell formally announced the adoption of flexible average inflation targeting (see We know what they want but we just don’t know how they go about getting it, Michael Feroli, 8/27/20). However, we think it’s unlikely Chair Powell offers too much granular policy guidance, as the expected magnitude of eases in September and beyond will probably depend highly on the evolution of labor market data, and the August employment report will be key in that regard.

Away from the near-term event risks, we’ve often debated whether the coming easing cycle will look more like the shallow maintenance cutting cycles of 1995 or 2019, or the full-blown easing cycles of 2001 or 2007. With the data we have in hand now, we do not think the current cycle resembles any of these cycles in the modern monetary policymaking era since the Fed began releasing policy statements in 1994. On one hand, the easing in inflation pressure against the backdrop of above-trend growth has this cycle somewhat reminiscent of the mini 75bp cycles of 1995 or 2019, but the similarities end here.

The current stance of monetary policy is decidedly more restrictive than it was in these other episodes: the real Fed funds rate is approximately 2.75%, a full 200bp above the Fed’s real longer-run dot. Using either the Fed’s longer-run dot (with data back to 2012) or the Holston-Laubach-Williams r* estimate, policy rates were negligibly restrictive in 2019, and only about 100bp restrictive in mid-1995 before the first cut ( Figure 15). To be fair, in 1995, the Fed had the luxury of shallow cuts because the neutral rate of interest was rising thanks to the IT-driven productivity boom, and it’s hard to say whether we are at the early stages of another productivity boom now. However, even if we adjust these measures of neutral policy rates higher by 50bp, there would still be more than 150bp of easing to bring policy rates back to more neutral levels, or twice the magnitude of shallower cycles.

Figure 15: The Fed funds rate is considerably more restrictive than at any point over the last 30 years...

Real Fed funds rate*, real longer-run median Fed dot, and Holston-Labach-Williams natural rate of interest rate; %

Source: Federal Reserve, J.P. Morgan

Source: J.P. Morgan., Federal Reserve bank of New York, Bloomberg Finance L.P.* Fed funds target less core PCE oya

Thus, could this cycle more closely resemble the 2001 and 2007 easing cycles, when the Fed eased aggressively for more than a year in each instance? We think it’s hard to make these comparisons, as financial conditions are clearly easier and showing fewer signs of stress than they did prior to these other major easing cycles. Figure 16 shows the OFR’s Financial Stress Index, and this series began in 1999. The index indicates financial stress is well below its longer-term averages, while this index demonstrated financial stress was more acute before the Fed began easing in 2001 or 2007. Certainly, were financial conditions to tighten more aggressively, the analog could tip in favor of these more aggressive easing cycles, but this is not apparent right now.

Figure 16: ...but as of now there are few signs of financial stress to motivate a deeper easing cycle like 2001 or 2007

OFR US Financial Stress Index (lhs) versus Fed funds target rate (rhs; %)

Source: OFR, J.P. Morgan

Overall, with the disinflationary process resuming and labor markets substantially looser than they were a year ago, we think this makes the case for substantial Fed easing in the coming months, and we continue to project 50bp cuts in September and November before downshifting to 25bp through late 2025. However, markets have clearly moved in the opposite direction this week, pricing in just 30bp of easing by the September FOMC meeting and 94bp for 2024 ( Figure 17). It’s tempting to add duration here, with markets pricing a slower and shallower path of cuts than our modal view, but we prefer to be patient, knowing that the position technicals we described above could be a drag over the near term. Should the conditions present themselves, we would look to add duration in the 3- to 5-year sector of the curve.

Figure 17: Markets are pricing in a less dovish path, but we do not yet recommend fading this

Easing priced by September and December 2024 (lhs), and December 2025 FOMC meetings (rhs); bp for both axes

Source: J.P. Morgan

Turning to the curve our views are unchanged, and we continue to favor 5s/30s steepeners for multiple reasons. First, as we’ve written in this piece multiple times over the last 8 to 9 months, the curve tends to steepen outside in, with the long end steepening ahead of the first cut, as medium-term monetary policy expectations ease. This steepening tends to occur consistently as the time to the first cut draws near, suggesting this is no time to pare back on steepening exposure. Second, this trade tends to behave like a lower-beta duration long with a better carry profile than an outright long position. Third, as the rapid growth of the Treasury market outstrips demand from its traditional base of less price-sensitive investors like the Fed, foreign official investors, and US commercial banks, and relies more heavily on price-sensitive investors, this warrants higher term premium and steeper curves (see In the eye of the beholder, 9/12/23). Now it’s clear our choice of steepener biases us toward a more normalizing Fed cutting cycle: our recent work has found that the behavior of the front end varies greatly, traditionally trading in a range around shallower easing cycles, and steepening sharply in more aggressive easing cycles. In this vein, 2s/5s is beginning to look more attractive, as the risks point toward quicker and deeper cuts, but we think it will require another weak employment report to have us shift our steepening exposure further in the curve.

June TIC update

Turning to the monthly TIC data, foreign investors purchased $17bn long-term Treasuries over the month of June, below the six-month average, and the weakest month of net purchases since December 2023. Private investors added $23bn, a step down from the $41bn of purchases in May, which includes $7bn of net selling from International and Regional Organizations (IROs), the first month of net selling from this community since January. Meanwhile foreign official investors sold $6bn for the second consecutive month ( Figure 18).

Figure 18: Foreign investors on net purchased $17bn of long-term Treasuries in June, the least since December 2023...

Total net sales of long-term Treasuries by US residents by sector and 6-mo. moving average; $bn

Source: Treasury International Capital System

Figure 19: ...and geographically, demand was concentrated in Europe while net selling flows emanated from Japan

Net sales of long-term Treasuries by US residents from April- June 2024 by region as well as cumulative net sales year-to-date; $bn

Region	Jun-24	May-24	Apr-24	YTD 2024
Euro Area†	20.6	10.0	8.1	83.1
China	5.9	-23.9	23.9	-38.8
Ireland	0.2	5.7	0.3	13.3
Oil Exporters*	-2.2	2.1	0.3	10.4
EM ex-China**	-2.3	13.1	15.5	51.6
UK	-5.1	15.2	-3.7	25.9
Cayman Islands	-10.7	3.7	18.0	2.0
Japan	-30.1	-5.8	-4.6	8.8

Source: Treasury International Capital System
*Oil exporters include all OPEC members (excluding Equatorial Guinea & Angola) and also include Bahrain, Indonesia & Oman
**EM ex-China includes Russia, Brazil, African countries, Taiwan (China), South Korea and Mexico
† Euro area include Austria, Belgium, Cyprus, Estonia, Finland, France, Germany, Greece, Ireland, Italy, Latvia, Luxembourg, Malta, Netherlands, Portugal, Slovakia, Slovenia, Spain

Turning to the geographical composition, Figure 19 shows that flows were rather muted across the domiciles we monitor. That said, demand in June was concentrated among European investors, who net purchased $20.6bn of long-term Treasuries over the period. Indeed, European investors have comprised the largest source of demand for US Treasuries this year, as buying US bonds on both a rolling and matched-maturity FX hedged basis remains attractive (see Global Fixed Income Markets Weekly, 8/2/24). Within this group, French investors purchased $18bn over the month, accounting for the lion’s share of this demand. Meanwhile Japan investors sold $30bn of long-term Treasuries, the largest monthly sale since September 2022. This should not come as a surprise as more granular data from the MoF showed that Japanese investors net sold $4.26bn JPY of foreign bonds in June, of which roughly 60% were USD-denominated, with the majority of this selling pressure emanating from banks. Selling continued at a more moderate $2.11bn JPY pace in July (see Japan Flows in Pictures, Takafumi Yamawaki, 8/8/24).

Trade recommendations

• Maintain 5s/30s steepeners

-   Stay long 100% risk, or $112mn notional of T 4.875% Oct-28s
-   100% risk, or $29.9mn notional of T 4.75% Nov-53s
- (US Treasury Market Daily, 11/22/23: P/L since inception: 10.3bp)

• Maintain 75%/6% weighted 5s/10s/30s belly-cheapening butterflies

-   Stay long 75% risk, or $43mn notional of T 4.625% Sep-28s
- Stay short 100% risk, or $33.3mn notional of T 3.875% Aug-33s
- Stay long 6% risk, or $1mn notional of T 4.125% Aug-53s
- (US Fixed Income Markets Weekly, 9/29/23: P/L since inception: -6.5bp)

• The 2-year note backup from 3.65% approaches the first zone of potential support at 4.115-4.20%. We expect the market to find its footing near that support or the 4.30-4.35% second confluence of levels over the near term. Bigger picture, what appears to be a multi-year yield top pattern developing behind the 3.55% Mar 2022 yield low and other signals across markets that point to the potential transition to an end-of-cycle dynamic leave us looking for late-summer consolidation to give way to an impulsive break to lower yield levels into 2025 as a bull market enters its acceleration phase.
• The 5-year note consolidates after the mid-summer acceleration to lower yields. We expect the 3.91-4.02% support zone to cap yields through late summer. A break through 4.15-4.20% is needed to derail the bullish pattern structure at this point. The rally stalled near the 3.50% Oct-Apr equal swings objective and the associated channel trend line. We suspect that area will act as resistance for a few weeks.
• The 10-year note yield bullishly consolidates behind the 3.625% Apr 2023 78.6% retrace and in front of key support at 4.02-4.14%. We expect those two zones to define the trading range into the early fall. The next zone of chart resistance sits at 3.22-3.248%, targets for later this year or early 2025.
• The 30-year bond bullishly consolidates richer than the 4.33-4.345% Jun-Jul pattern riches and in front of the 4.40-4.50% confluence of moving averages. Look for further coiling around the 4.00% area in the weeks ahead.
• Look for continued 5s/30s curve range trading below the 59bp Mar 2021 50% retrace and above the 28-32bp support zone in the weeks ahead. We think that range will consolidate the breakout from the 2022-2024 base pattern and give way to a more aggressive steepening trend in the months ahead.
• The 10-year TIPS breakevens rebound from the 203bp Aug low stalled at the key 215-219bp breakdown resistance zone. We used the test of that resistance as an opportunity to initiate a tightening trade. While the market can see more backing and filling over the near term, we expect additional tightening pressure to build later this year and into 2025. We see the 180bp area as a base-case target for that period.

US

While we believe the Apr-Aug moves kicked off powerful trend dynamics that will lead to further bond market rally impulses into 2025, we suspect the rest of the month will give market participants the opportunity to enjoy what are usually heavy vacation weeks. We are looking for further consolidation after the early-Aug cascade to lower global bond yields. That same consolidation is likely to play out on curve and TIPS breakevens charts. The speed at which some equity indexes retook ground over the past week lowers our conviction that trends have entered a more acute phase of an end-of-cycle transition to some degree. However, the majority of those markets are still trading below Jun-Jul breakdown levels and thus still look vulnerable into risk-off seasonals that take hold into Sep provided they respect that resistance over the near term. For now, we suggest carrying a full 5s/30s curve steepening trade and a newly added 10-year TIPS breakevens tightening trade that we entered as the market tested expected resistance levels. We will look for opportunities to enter outright duration longs if the markets respect support as we anticipate in late Aug.

At the front end, the 2-year note has nearby support at the 4.115% Jan yield low, 4.18-4.20% Apr-Jul Fibonacci retracement confluence and payrolls bull gap ( Figure 21). We are looking for material buying interest to contain Aug-Sep backups near there, or at the 4.30-4.35% second zone of levels. As we get increased confidence in which of those two it will be, we will look to start entering a bullish trade. The medium-term bullish chart structure stays in gear as long as the market does not cheapen back thorugh the 4.595-4.66% pattern breakout levels and moving averages.

Figure 21: The 2-year note backup from 3.65% approaches the first zone of potential support at 4.115-4.20%. We expect the market to find its footing near that support or the 4.30-4.35% second confluence of levels over the near-term...

2-year note yield, daily bars with momentum divergence signals; %

Source: J.P. Morgan, CQG

Bigger picture, we view the late-2022 through 2024 price action that has developed cheaper than 3.50% as a large bullish reversal pattern ( Figure 22). We suspect bullish consolidation to continue behind the 3.55% Mar 2022 yield low through the fall, but see the prospects for a release richer growing as the months roll by. Next resistance rests at the 3.285% Jul 2022 38.2% retrace and then 2.98% 2018 bear cycle cheap. We think the market will challenge those levels later this year or into early 2025.

Figure 22: … Bigger picture, what appears to be a multi-year yield top pattern developing behind the 3.55% Mar 2022 yield low and other signals across markets that point to the potential transition to an end-of-cycle dynamic leave us looking for late-summer consolidation to give way to an impulsive break to lower yield levels into 2025 as a bull market enters its acceleration phase.

2-year note yield, weekly bars with momentum divergence signals; %

Source: J.P. Morgan, CQG

The 5-year note has a similar short-, medium-, and longer-term pattern setup. Nearby support rests at the 3.915% Jul 22 61.8% retrace and 3.935% May internal trend line ( Figure 23). Medium-term bulls keep the agenda as long as backups respect the 4.02% Aug equal swings objective, 4.15% Aug 2022 trend line, and 4.195% Jun 14 yield low. While the recent data have reduced some angst regarding US growth, we do not envision a retest of those cheaper levels going forward. The Aug rally satalled near the 3.50% Oct-Apr equal swings objective and 3.485% Oct 2023 channel. Look for that area to keep a floor under yields for a few weeks. Next resistance rests at the 3.205% May 2023 yield low.

Figure 23: The 5-year note consolidates after the mid-summer acceleration to lower yields. We expect the 3.91-4.02% support zone to cap yields through late summer. A break through 4.15-4.20% is needed to derail the bullish pattern structure at this point. The rally stalled near the 3.50% Oct-Apr equal swings objective and the associated channel trend line. We suspect that area will act as resistance for a few weeks.

5-year note, daily bars with momentum divergence signals; %

Source: J.P. Morgan, CQG

The 10-year note backup from near the 3.625% Apr 2023 78.6% retrace has stalled in front of key support at the 4.02% Apr internal trend line, 4.05% Jul 24 61.8% retrace, and 4.14% Jul 17 yield low ( Figure 24). The market is currently working off the most extreme overbought reading on our TY premium-weighted Put/Call ratio z-score indicator since 2021, but the absence of the momentum divergence signaling that normally accompanies a more lasting trend reversal points to bull market consolidation and not a protracted retracement to higher yield levels. We suspect the intermediate sector will spend a few weeks in a 3.65% to 4.15% range and pressure to lower yields in the months ahead. Next resistance rests at the 3.50% Oct-Apr equal swings objective, with longer-term levels at the 3.248% Apr 2023 yield low and 3.22% Mar 2020 38.2% retrace.

Figure 24: The 10-year note yield bullishly consolidates behind the 3.625% Apr 2023 78.6% retrace and in front of key support at 4.02-4.14%. We expect those two zones to define the trading range into the early fall. The next zone of chart resistance sits at 3.22-3.248%, targets for later this year or early 2025.

10-year note yield, daily bars with momentum divergence and TY premium-weighted Put/Call z-score signals; %

Source: J.P. Morgan, CQG, CME

Figure 25: The 30-year bond bullishly consolidates richer than the 4.33-4.345% Jun-Jul pattern riches and in front of the 4.40-4.50% confluence of moving averages. Look for further coiling around the 4.00% area in the weeks ahead.

30-year bond yield, daily bars with momentum divergence signals; %

Source: J.P. Morgan, CQG

Figure 26: Look for continued 5s/30s curve range trading below the 59bp Mar 2021 50% retrace and above the 28-32bp support zone in the weeks ahead. We think that range will consolidate the breakout from the 2022-2024 base pattern and give way to a more aggressive steepening trend in the months ahead.

5s/30s curve, daily closes; bp

Source: J.P. Morgan, CQG

Figure 27: The 10-year TIPS breakevens rebound from the 203bp Aug low stalled at the key 215-219bp breakdown resistance zone. We used the test of that resistance as an opportunity to initiate a tightening trade. While the market can see more backing and filling over the near term, we expect additional tightening pressure to build later this year and into 2025. We see the 180bp area as a base-case target for that period.

10-year TIPS breakevens, daily closes; bp

Source: J.P. Morgan, CQG

This report was excerpted from Global Fixed Income Technical, Jason Hunter, August 16, 2024

Initiate 5Yx5Y inflation swap longs

• Core CPI came in close to expectations, with some offsetting surprises within the details, and does little to change our medium -term inflation outlook. The fixings continue to imply a softer core CPI inflation over the balance of the year relative to our forecast
• Despite a partial rebound in breakevens over the last two weeks, the product has lagged the recovery in other markets and breakevens remain cheap versus our fair value estimates. Given the recent improvement in economic data, we now recommend initiating long exposure
• With the Fed likely to begin cutting rates next month, we note that breakevens have not exhibited a consistent performance around the start of prior easing cycles. However, a Fed that is cutting due to risks coming into better balance and a desire to return policy rates quickly to neutral should be more supportive for breakevens than a Fed that is behind the curve
• Demand for TIPS-focused funds has picked up alongside the decline in real yields and improvement in realized TIPS returns, driving the 4-week moving average of inflows to its highest level in more than 2 years
• Along the curve, the intermediate sector offers value, and we prefer to express longs in inflation swaps rather than breakevens, given that IOTAs remain tight. 5Yx5Y inflation swaps also avoid the seasonally negative carry of spot breakeven positions

Market views

Over the last two weeks, 5-, 10-, and 30-year breakevens widened 8bp, 8bp, and 6bp, respectively, alongside an improvement in economic data. Specifically, the ISM employment index rose from 46.1 to 51.1, and initial jobless claims moved back down to 227K, after rising to 250K two weeks ago. The July retail sales data came in stronger than expected, leading our economists to revise up their 3Q real consumption growth estimate to 2.0% from 1.1% previously (see US: July retail sales suggest no slowing in 3Q real spending, Abiel Reinhart, 8/15/24). Inflation news was relatively benign. Though core PPI was softer than expected, with the index unchanged in July, core CPI came in line with expectations, rising 0.17% m/m and leaving the year-ago rate unchanged at 3.2%. With CPI and PPI data in hand, we estimate a 0.12% rise in core PCE in July, with the year-ago rate holding steady at 2.6% (see US: Cooler July CPI, as expected, Michael Hanson, 8/14/24).

Turning to the details of the CPI report, the headline CPI-U NSA printed at 314.54 in July, below the market fixing of 314.63. Energy prices were flat on the month, while food prices rose 0.2% as expected. Meanwhile, core goods prices fell 0.32%, the softest reading since January, held down by a 2.3% decline in used vehicle prices ( Figure 28). Though industry data imply another decline in used car prices in August, more recent data have shown an increase in prices, and we generally look for greater stability in vehicle prices over the remainder of the year relative to what we’ve observed YTD (see Focus: Vehicle Prices, Michael Feroli, 7/19/24). Core services also came in line with our forecast, rising 0.31%, though with some offsetting surprises within the details. Rent and OER firmed somewhat more than anticipated, rising 0.49% and 0.36%, respectively, with both components rising faster in larger cities. Overall, this does not derail our view that we should see softer rent inflation readings over the balance of the year, particularly on a seasonally-adjusted basis. However, we note that fundamentals in the rental market remain strong, which could make it difficult for rent inflation to return to its pre-pandemic pace, absent more significant economic weakening (see CMBS Weekly, Chong Sin, 8/2/24). Away from rents, medical care services came in softer than expected, driven by an outsized 1.9% decline in outpatient hospital services, which is unlikely to be repeated going forward. Other components were generally in line with expectations.

Over the past month or so, we have recommended a neutral stance on breakevens and had noted in late July that we would look for cheaper valuations to add bullish exposure. Though that cheapening came in the wake of the July employment report, we were patient to initiate wideners given that the details of the report, combined with the large move up in initial jobless claims, suggested a greater risk of a more significant deterioration in labor markets than we had expected (see TIPS Strategy, 8/2/24). However, the improvement in data over the last two weeks, including the full reversal lower in claims, leaves us more comfortable with fading the cheapening. Indeed, we think there’s room for breakevens and inflation swaps to richen in the near term for a number of reasons.

First, the very front end of the curve continues to imply a very soft trajectory for core CPI inflation over coming months, well below our own forecast. Specifically, we estimate the fixings imply roughly a 0.15% m/m pace on average through October, below our own forecasts near a 0.21% m/m pace ( Figure 29). As discussed above, though we look for some softening in rent inflation through the fall, the weakness in some of the other categories are unlikely to be repeated, which should drive a modest rebound in the pace of sequential core CPI inflation.

Second, though breakevens across the curve have already partially rebounded from their narrowest levels alongside better data, the recovery in inflation has lagged other markets, and intermediate breakevens in particular remain near their narrowest levels since the failure of SVB ( Figure 30). Meanwhile 10-year breakevens still appear roughly 11bp too narrow versus our fair value framework, more than the standard error of the model ( Figure 31).

Figure 30: Despite rebounding partially wider alongside better data, intermediate breakevens remain near their narrowest levels since the failure of SVB...

1-month forward, seasonally-adjusted breakevens; bp

Source: J.P. Morgan

Figure 31: ...and remain 11bp narrow versus our fair value framework

Residual on J.P. Morgan 10-year breakeven fair value model*; bp

Source: J.P. Morgan
*1m-forward, seasonally-adjusted breakevens are regressed on the J.P. Morgan Commodity CurveIndex (JPMCCI) as well as its square and the 3mx3m/15mx3m OIS curve; regression over the last 7years; R2=93%, SE=9bp

Third, with the Fed very likely to begin cutting interest rates next month, we think the context in which the Fed is easing and the evolution of the economic data is likely to be important in determining the path forward for breakevens. A Fed that is cutting due to risks coming into better balance and a desire to return policy rates quickly to neutral to sustain a soft landing should be far more supportive for breakevens than a Fed that is behind the curve and responding to a deterioration in labor markets. Notably, even as nominal yields have consistently declined in the months around the first ease in a cutting cycle, the behavior of breakevens has not been uniform. Figure 32 shows the cumulative change in 10-year breakevens from 1 month prior to the first ease through 3 months after for each of the last three easing cycles. While breakevens narrowed at the start of the 2001 easing cycle, breakevens widened in 2007 and were virtually unchanged in 2019.

Fourth, given the significant decline in real yields over recent months and the improvement in realized TIPS returns, inflows into TIPS-focused funds have picked up, despite the continued moderation in inflation: Figure 33 shows that the 4-week moving average of weekly net inflows has risen to $215mn, the highest in over 2 years. Meanwhile end-user takedown at TIPS auctions has continued to come near record levels, with takedown at the July 10-year auction rising to 92.1%. Against this backdrop, we think it’s unlikely that the recent cheapening in breakevens can be sustained.

Figure 32: TIPS breakevens have not displayed a consistent pattern in the months after the first ease in a cycle

Cumulative change in 10-year TIPS breakevens from 1 month prior to the first ease in a cycle through 3 months after; bp

Source: J.P. Morgan

Figure 33: As expected, the improvement in realized TIPS returns has driven a pickup in inflows to TIPS-focused funds

Weekly net inflows into TIPS-focused ETFs and 4-week moving average; $mn

Source: EPFR

Along the curve, we think the intermediate sector offers value, having underperformed versus the wings in recent weeks. Notably, the 5Yx5Y/10Yx20Y curve has steepened sharply, trading 1.3 standard deviations above its 1-year average ( Figure 34), and the 2Yx3Y/5Yx5Y/10Yx20Y PCA-weighted butterfly is trading at its cheapest levels of the past year. Moreover, consistent with the valuation signal from our fair value framework, we can see that 5Yx5Y inflation swaps had been tightly correlated with 5Yx5Y Treasury yields for much of the last 9 months before deviating sharply in recent weeks ( Figure 35). 5Yx5Y inflation swaps offer the benefit of relative value along the curve while also avoiding the seasonally negative carry of spot breakevens. Lastly, we’ve discussed in the past, we prefer to express longs in inflation swaps rather than cash breakevens, given that IOTAs continue to trade near multiyear tights and likely have room to widen further from here (see TIPS Strategy, 7/26/24). Putting the pieces together, we recommend adding tactical longs in 5Yx5Y inflation swaps at current levels (see Trade recommendations).

Figure 34: Along the curve, we think the intermediate sector offers value, having underperformed versus the wings in recent weeks

5Yx5Y/10Yx20Y inflation swap curve; bp

Source: J.P. Morgan

Figure 35: 5Yx5Y inflation swaps had been tightly correlated with 5Yx5Y Treasury yields for much of the last 9 months before deviating sharply in recent weeks

5Yx5Y inflation swaps (LHS) versus 5Yx5Y Treasury yields (RHS) derived from our par curve; % both axes

Source: J.P. Morgan

Trade recommendations

• Initiate 5yx5y inflation swap longs

-    Initiate long 100% risk, or $50mn notional of 5yx5y inflation swap (swap start: 8/20/29, swap end: 8/20/34) at 243.0bp.

Hopscotch

• Policy uncertainty is back in the picture. Although markets appear to agree on September with respect to timing, the pace and extent of easing remains uncertain for this year and the next. Such policy uncertainty is an important driver of jump risk, which is already evident. All of this supports a bullish view on short expiry volatility, despite the apparent richness of implieds - although implied volatility in most sectors appears rich relative to our fair value estimates, our analysis suggests that policy uncertainty is currently significant enough to be a strong offset. Therefore, we maintain our bullish stance on short expiry vol
• The zero duration swap spread appears too narrow to fair value and we believe this is likely to correct in the near term as dealer capacity returns in repo markets. With front end spreads appearing to find a bottom, we maintain our widening bias on swap spreads at the front end
• At the same time, term funding premium is likely biased wider in coming weeks and months, which should pressure longer maturity swap spreads narrower and the spread curve flatter - initiate 10Y swap spread narrowers and initiate 3s/7s spread curve flatteners
• One a relative basis, the swap spread curve between the 2-year note front and back contract CTDs appears too steep, likely because of futures market technicals, and should correct as the roll progresses - initiate 0.875% Jun 2026 / 0.875% Sep 2026 maturity matched swap spread curve flatteners
• The ampleness of Reserves received some attention this week, thanks to recent work by Fed staff that outlines an elegant approach to measuring this by examining the slope of the demand curve in the fed funds market. We borrow the spirit of this work, but believe that it is important to (i) expand the concept of liquidity to include RRP as well as Reserves, and (ii) focus on SOFR rather than the effective funds rate when examining the sensitivity to quantity shocks. Our expanded analysis suggests that liquidity conditions are clearly tightening and QT is likely in its end game. But thanks to a tapered pace, a little more room for RRP declines, and an expected reduction in the TGA later this year, there will likely be a few more months of headroom for QT before it ends. We look for QT to reach an end later this year with the Fed balance sheet near ~$7tn, RRP balances slightly below ~$300bn and Reserves near ~$3.3tn
• Earlier this month, the expiry of benchmark 3M expiry swaptions first crossed the November election date. By examining the relative performance of election-impacted expiries (3M in this case) versus a control expiry such as the 6M, we can infer election premia being priced into the vol markets. The current experience suggests that options markets view election risk as extending to ~4 business days after election, for a cumulative election premium of ~6 additional vol days. This is lower than the experience in May when 6M expiries first crossed the election - back then, cumulative premia peaked near 15 additional vol days, spanning a period ending 7 business days after the election. This is a reminder that shifts in election sentiment could impact short expiry implied volatility in coming weeks
• We include a short summary of our outlook for Treasury futures calendar spreads as we head towards the September - December roll cycle; we recommend selling the FV and buying the US weighted calendar spreads
  
  

  ---

Hopscotch

August started with a bang, as the soft Payrolls report caused markets to pivot sharply towards pricing in aggressive easing, and yields reached their recent trough on that same day (the 2nd of August). But even right then, it appeared to be an over-reaction. After all, earlier that same week, Fed Chair Powell had noted during the press conference that the Fed would watch the data, not data points. Given that, the post-Payrolls appeared to be a case of "too-much, too-soon" even then. The past two weeks appear to be proving that point. Forward OIS rates have risen rather steadily from their post-NFP lows, and are now near the higher end of their 2-week range (Figure 1). This normalization in Fed expectations has been helped by Fed-speak in recent weeks. As seen in Figure 2, it is apparent that Fed speakers have increased their focus on labor markets relative to inflation, but they also appeared to dispel or downplay recession fears (Figure 2). All in all, markets appear engaged in a game of hopscotch, collectively trying to land on the correct square.

Figure 2: Fed speakers are now focused on both inflation and labor markets, but voiced caution in terms of cutting expectations and downplayed recession risks

Selected Fed-speak excerpts, 8/2/2024 - 8/16/2024

Source: J.P. Morgan., Bloomberg Finance L.P.

Going forward, given the extent of retracement that has already happened, front end yields and Fed expectations will likely remain rangebound in the near term, until more clarity emerges from next month's employment report (which remains 3 weeks away), or perhaps at the Jackson Hole symposium next week. Thus, until then, markets will likely remain somewhat in limbo, but in a state of heightened sensitivity to data. We have often noted that policy uncertainty is an important driver of jump risk, and policy uncertainty is now back after several months of relative clarity. To be sure, it now appears highly likely that rate cuts will begin in September, but the pace and extent of easing remain uncertain in the eyes of the market. By carefully de-constructing the implied probability density function inferred from call and put premia on Z4 SOFR futures, we obtain combination weights (which must sum to one) that are associated with policy outcome scenarios (see Powell sees the data, markets see one data point more details). Right now, the sum total of weights associated with shallow-cut scenarios (defined as under 75bp of easing by year end, or under 150bp by mid-2025) is quite comparable with the total weight on deep cut scenarios (defined as 125bp or more of rate cuts by year end, or 175bp or more easing by mid-2025). This means that policy uncertainty is once again elevated, after a brief 2-month period where markets had converged on shallow-cut expectations. Moreover, such policy uncertainty is also a feature further forward, which can be seen by examining implied distributions on June 2025 SOFR futures as well (Figure 3).

Figure 3: Policy uncertainty remains elevated, not just in the near term but also at a mid-2025 horizon

Total weights on YE24 and 1H25 policy rate scenarios representing Shallow Cuts*, and Deep Cuts*, as calculated from a decomposition of the implied probability distribution associated with Dec 2024 and June 2025 SOFR futures**; May 2024 – Current

Source: J.P. Morgan.
* Scenarios involving 75bp or less in rate cuts by YE24, or 150bp or less of rate cuts by mid-2025, are defined as shallow cut scenarios. Deep cut scenarios are those with even more easing by those respective horizons. For more details, see Figure 2 of What’s the rush?.
** We enumerate a list of scenario-specific Normal distributions with fixed standard deviations and means that are separated by 25bp, and then require the implied distribution to be a weighted combination of these individual distributions. The weights are then solved for, by fitting to the observed prices of calls and puts at various different strikes. For more details of our approach, see What’s the rush?.

Figure 4: The frequency of jumps is already higher, and will likely remain so thanks to policy uncertainty

Rolling 2-week jump frequency (%), and policy certainty metric (right, inverted)**, 6/20 - current

Source: J.P. Morgan.
*A jump is defined as any daily change in UST 10Y yields over 10bp, on an absolute basis. Jump frequency is defined as the number of times a jump has occurred in the past two weeks, divided by the number of observations in that period** Policy certainty metric is the weight on the "shallow" cuts scenario (for YE24) as defined in Figure 3. A high weight on this scenario implies policy certainty, and a weight closer to 0.5 implies policy uncertainty

But why do we dissect implied distributions to measure policy uncertainty in this manner? Because policy uncertainty is an important driver of jump risk - as new information causes markets to reassess these weights, that can result in large moves in forward OIS rates since the cone of policy outcomes is quite wide. Indeed, this is already evident - in the two weeks since Payrolls, there have been three days when 10-year yields changed by 10bp or more from the previous days close (Figure 4). Thus, jump risk is already high, resulting in high delivered volatility. This is a strong argument in favor of maintaining a bulllish stance on short expiry volatility going forward. But one factor that stands in opposition is valuations. Our longer term fair value framework, which uses 5 years of history and is based on intuitive observables (see Interest Rate Derivatives 2024 Mid-Year Outlook) suggests that implieds currently are rich with respect to fair value (Figure 5). This is especially true in shorter expiry / shorter tail structures, where the residuals are above 0.5 bp/day. Is this sufficient to erode the case for bullish vol exposure?

We think it is not, and we can see this in two independent ways. One, given the current observed frequency of jumps, and assuming this persists (which we think is likely), and given the typical size of moves on jump and non-jump days, delivered volatility will likely be above current levels of implied volatility. This is seen in Figure 6, where we project future delivered using a simple calculation that assumes current jump frequencies and jump and non-jump move magnitudes in each sector. Of course, this relies heavily on jump frequency assumptions.

Figure 5: Implied volatility appears rich compared to our longer term fair value framework...

Residual from our implied volatility fair value framework*, 8/15/2024; bp/day

Source: J.P. Morgan.
* For details please see Interest Rate Derivatives 2024 Mid-Year Outlook

Figure 6: … but appears cheap given the observed frequency of jumps and size of typical moves in the past two weeks

Observed jump frequency* (%), average jump size and non-jump size (bp), as well as projected realized volatility** and current implied volatility (bp/day), 8/15/2024

Source: J.P. Morgan.
*A jump is defined as any daily change in yields over 10bp, on an absolute basis. Jump frequency is defined as the number of times a jump has occurred in the past two weeks** Calculated as square root of jump size squared times jump frequency plus non-jump size squared times non-jump frequency

A more robust approach to answering this question also points to the same conclusion. We turn to our empirical model for daily delta hedged short gamma returns, which we modify to include the current implied volatility residual (to capture the impact of current valuations) as well as our metric for policy clarity (as a factor that drives jump risk). Figure 7 shows regression statistics from our empirical model. As one might expect, policy clarity and a high residual benefit short gamma positions, and the corresponding coefficients all have the expected signs (except in 2Y tails, where the ex ante residual turns out to be statistically less significant.

As Figure 7 shows, returns from selling volatility over the next month are projected to be mostly flat, despite the richness in implieds. Thus the carry cost of holding long volatility positions in short expiries is likely to be fairly low. Therefore, given the potential for significant moves and high jump risk in the next few weeks from Jackson Hole symposium, PCE and August payrolls, we maintain our long volatility bias in shorter expiries, and continue to recommend expressing it in 5-year tails.

Figure 7: Returns from selling volatility are likely to be flat in the near-term

Statistics from regressing* monthly short delta hedged straddle returns (bp/notional) against ex-ante implied vol (bp/day), ex-ante metric of policy expectations stability** (unitless) ,ex-ante ATMF (%), and ex-ante residual† (bp/day) as well as current (8/15) values of drivers and projection and normalized projection***

Source: J.P. Morgan.
*Regression over past 6 months** Policy expectations stability is defined as the weight on "shallow cuts" scenario shown in Figure 3***Projection is calculated by using the current drivers and the coefficients. Normalized projection is calculated by dividing the projection by standard error
†Residual from fair value framework as shown in Figure 5

Swap spreads

In a period where yield levels and the curve have been fairly volatile, swap spreads have been rather stable. After trading in relatively tight ranges of 1.5-2.5bp, swap spreads are roughly flat across the curve - maturity matched swap spreads are about 1bp wider in the front end, modestly narrower in the belly of the curve, and flat at the long end (Figure 8).

Figure 8: Swap spreads have remained stable over the past two weeks

Selected statistics for maturity matched SOFR swap spreads, 8/2 - 8/16; bp

Source: J.P. Morgan.

We continue to view swap spreads predominantly through the lens of the macro-drivers that shape the term structure of swap spreads - term funding premium, which specifies the slope of the term structure, and zero-duration spreads, which specifies the intercept and thus the level of front end spreads more broadly (for more details on our framework, see Term Funding Premium and the Term Structure of SOFR Swap Spreads). Currently, term funding premium stands near its lows of the past 5 months or so, and is now about 0.25bp/year below fair value (Figure 9). Should this correct itself, which we expect since this has been reasonably mean-reverting around estimates of fair value, that would imply a narrowing bias of 2bp and 4bp respectively in 10- and 30-year swap spreads, respectively. But at the front end, perhaps the bigger driver of swap spreads in recent weeks has been zero duration spreads, which have narrowed sharply on an outright basis as well as relative to its fair value - currently, we estimate that zero-duration spreads are as much as 7bp too narrow relative to fair value (Figure 10). We believe this will correct in the near term. As we have noted recently (see Joie de Louvre), the cheapness of zero-duration spreads was likely due to upward pressure on GC repo rates stemming from repo market plumbing issues as foreign dealers pulled back in late June and July. But this appears to be normalizing and money market funds' allocations to repo has climbed recently, as evidenced by the sharp drop in O/N RRP balances (see July MMF holdings update: MMFs and their use of Fed ON RRP, T. Ho, 8/15/2024). In other words, the RRP appears to be finally playing its role in moderating stresses in repo markets. This is healthy, and we expect GC repo rates to drift back lower in coming weeks, which should help support a normalization in zero duration spreads.

Figure 9: Term funding premium stands near its recent lows, and is now about 0.25bp/year below fair value

Term funding premium*, actual versus current (8/14) fair value**, 3/4/2024 - 8/14/2024; (bp/year)

Source: J.P. Morgan.
* For details on our methodology for calculating Term funding premium, see Term Funding Premium and the Term Structure of SOFR Swap Spreads
** For details on our fair value framework for term funding premium, see Interest Rate Derivatives 2024 Mid-Year Outlook

Figure 10: We estimate that zero-duration spreads are as much as 7bp too narrow relative to fair value

Residual from regression of zero-duration swap spread* versus its drivers**, past 6 months; bp

Source: J.P. Morgan.
* For details on our methodology for calculating Zero-duration swap spreads, see Term Funding Premium and the Term Structure of SOFR Swap Spreads
** For details on our fair value framework for zero-duration swap spreads, see Interest Rate Derivatives 2024 Mid-Year Outlook

These two observations together underpin our views on swap spreads across the curve. First, our anticipation of wider zero-duration spreads is a factor that strongly supports a widening bias in front end spreads, since the front end will be much less negatively impacted by any increase in term funding premium. The case for front end wideners is further strengthened by the fact that we would expect 2- and 3-year swap spreads to trade ~2bp wide to the fitted term structure. In other words, as seen in Figure 11  , 2- and 3-year maturity matched swap spreads are currently trading pretty close to the fitted term structure of spreads (that is specified by term funding premium and zero-duration spreads), but empirical models of historical deviations suggest that they should trade slightly wide relative to the term structure (to the tune of 1.5-2bp). Thus, all things considered, we maintain our bullish bias on front end swap spreads.

In the 10- and 30-year sectors, the upward impact on spreads from a widening in zero-duration spreads will likely be counteracted by three other considerations. First, a near term correction in term funding premium, should it occur, would pressure 10- and 30-year spreads narrower by ~2bp and ~4bp respectively. Second, and perhaps more important, we also expect term funding premium to drift higher over the medium term on the back of continuing QT and elevated duration supply in the Treasury market. Should term funding premium climb to 5bp/year by year end as we projected (see Interest Rate Derivatives 2024 Mid-Year Outlook), that would imply an additional 4-5bp and 8-10bp of narrowing pressure respectively. Even if this is only partially realized, the narrowing pressures at the long end are significant. Lastly, Figure 11 also suggests that 10Y and 30Y spreads are currently trading above the fitted term structure by ~2bp more than we would expect based on empirical models of historical deviations. Thus, given this confluence of factors, we are biased towards narrower swap spreads in the long end of the curve (see Trade recommendations).

Figure 11: 10Y and 30Y spreads are currently trading above the fitted term structure by ~2bp more than we would expect based on empirical models of historical deviations

Swap spread deviation* and fair value (expected)** swap spread deviation relative to the term structure of swap spreads for 2Y, 3Y, 5Y, 7Y, 10Y, 20Y, and 30Y swap spreads, 8/16/2024; bp

Source: J.P. Morgan.
* Swap spread deviation relative to the term structure of swap spreads is calculated for any given day as the actual maturity matched swap spread minus the fitted value as of that day. The fitted value is calculated from a cross sectional regression of maturity matched swap spreads at benchmark tenors (2s, 3s, 5s, 7s, 10s, 20s, 30s) versus their modified durations, and evaluated at the OTR note’s modified duration.
** Expected deviation is detailed in our note (Term Funding Premium and the Term Structure of SOFR Swap Spreads)

The above discussion also implies a straightforward flattening bias on the broader swap spread curve, and we recommend maintaining such positions. One such trade that currently appears attractive is a 3s/7s maturity matched swap spread curve flattener. The fundamental case for this is covered in the discussion above - an upward bias in zero duration spread as well as term funding premium, and an upward bias in 3Y spreads relative to the term structure, and a downward bias in 7Y spreads relative to the term structure, all support a 3s/7s swap spread curve flattener. But even when viewed much more simply, the 3s/7s maturity matched swap spread curve is at the steep end of its range over the past year, and appears poised to correct (Figure 12). Therefore, we now recommend positioning for a flatter 3s/7s swap spread curve (see Trade recommendations).  

Finally, on a relative basis, we recommend positioning for a flatter swap spread curve between the 0.875% June 2026 and 0.875% Sep 2026 issues. These two issues are CTD into Sep and Dec 2-year note contracts, and the steepening of this spread curve is likely the result of futures market technicals (Figure 13). This means that once the roll happens and open interest migrates from the Sep to Dec contracts, this idiosyncratic steepness of the spread curve will likely correct itself. Therefore, we recommend this trade (see Trade recommendations).

Figure 12: The 3s/7s maturity matched swap spread curve is at the steep end of its range over the past year, and appears poised to correct

3s/7s maturity matched OTR swap spread curve, past 1 year; bp

Source: J.P. Morgan.

Figure 13: The spread curve between the front and back CTDs into the 2-year note contract has steepened idiosyncratically, and should correct itself as the Treasury futures roll progresses

Maturity matched swap spread curve between TU Sep CTD (0.875% Jun 2026) and TU Dec CTD (0.875% Sep 2026) candidates, past 1 year; bp

Source: J.P. Morgan.

Ampleness of liquidity and the outlook for the Fed’s balance sheet

As we noted above when discussing repo market developments, money market funds have recently grown their dealer repo exposure at the expense of the O/N RRP facility. Locally, this is a positive - indeed, such reallocations between the O/N RRP and the private repo market will likely help GC repo rates decline from their recent highs relative to the RRP rate. But one consequence of this has also been that O/N RRP balances have now fallen to just over $300bn. This is close to our $250-300bn guesstimate for the likely floor on O/N RRP balances that will be needed for well-functioning repo markets.

This raises two important and related questions. First, is liquidity in the system (i.e., the combination of Reserves as well as RRP) still ample? Second, given the likely outlook for the Fed's major liabilities, how much further can QT continue?

The first question is a variation on a closely related question recently addressed by Fed staff in a research note and blog post on Liberty Street Economics (see Scarce, Abundant, or Ample? A Time-Varying Model of the Reserve Demand Curve, Federal Reserve Bank of New York, May 2022 (Revised April 2024) and When Are Central Bank Reserves Ample?, Liberty Street Economics, 8/13/2024). Specifically, they explore the question of whether Reserves are ample. To answer this, they describe an elegant approach to measuring the ampleness of Reserves that is based on estimating the slope of the Reserve demand curve. In other words, in a world of scarce Reserves, the effective funds rate ought to exhibit greater sensitivity to quantity shocks, while such sensitivity would likely be small in an abundant Reserves regime.

Here, we rephrase their question and adapt their approach in two important ways. First, we note that volumes in the fed funds market are quite small relative to risk free short term markets as a whole - the volume of repo transactions used in estimating SOFR, for instance, are upwards of $2tn, versus a much smaller $100bn in the fed funds market. Since the Fed likely does focus on a broader complex of short rates rather than just the effective funds rate when assessing its monetary policy’s operational effectiveness, we believe it is preferable to measure ampleness by looking at the response of broader rates such as SOFR. Second, we also believe that the question itself must be recast in a manner that also includes RRP balances. This is true for at least two reasons. First, Reserves appear to have become quite sticky at ~$3.3tn. As seen in Figure 14, Reserves have basically remained flat all year in a +/- $100bn range, while RRP balances have fallen by nearly $400bn.The current level of Reserves is a high enough level that would previously have been thought of as quite abundant, but the remarkable stickiness of Reserves around this level suggests that RRP balances are more likely to be the source of marginal liquidity right now.

Second, the recent behavior of GC repo rates is additional evidence in support of the importance of RRP balances as a source of marginal liquidity. After all, the steady upward rise in GC repo rates since late June likely happened because of issues such as counterparty limits facing money market funds. The fact that it happened at a time when Reserves were elevated but not deployed into repo markets suggests that they are less important as a source of marginal liquidity.

Therefore, to summarize, we believe that (i) liquidity is better defined as the sum of Reserves plus RRP balances, and (ii) the ampleness of liquidity is best estimated by examining the response of broader short rates (such as SOFR) to quantity shocks. Borrowing the essence of the idea in the recent note by Fed staff (but devoid of its statistical rigor in the interest intuition and simplicity), we look at the empirical relationship between changes in the SOFR-versus-top-of-the-band differential and percentage changes in total liquidity (i.e., Reserves plus RRP balances). This rolling empirical beta (rolling 1Y history, using weekly changes) is shown in Figure 15 - as can be seen, the rising magnitude of the (negative) beta in recent months does indeed point to tightening liquidity conditions, and suggests that QT is in its endgame. A different bottoms-up analysis by our Short Term market strategists also concludes that we are not too far from a floor on RRP balances (see July MMF holdings update: MMFs and their use of Fed ON RRP, T. Ho, 8/15/2024).

Figure 14: Reserves have remained mostly flat all year in, while RRP balances have fallen by nearly $400bn

Cumulative change in Fed balance sheet Reserves, RRP, and TGA, 1/5/2024 - 8/14/2024; $bn

Source: J.P. Morgan., FRED, Federal Reserve H.4.1, Federal Reserve H.8

Figure 15: Liquidity conditions appear to be tightening as QT is now in its endgame

Rolling 1Y beta of weekly changes in the SOFR minus FF target differential (bp) versus weekly percent change in total liquidity (defined as Reserves plus RRP), past 3 years

Source: J.P. Morgan.

Fortunately, the fact that (i) QT is now happening at a tapered pace, (ii) there is perhaps a little more room for RRP balances to decline before it becomes critically low based on our Short Term strategists' $200-325bn estimate, and (iii) the anticipated decline in the TGA once we get past 3Q, collectively create some limited headroom for QT to continue. Thus, we see QT as being in its end game, but with perhaps a few more months left. Our projections for the Fed's balance sheet are shown in Figure 16 - we now estimate that the Fed's balance sheet will end the year at ~$7tn, with Reserves essentially flat to current levels, O/N RRP balances dipping in the near term but recovering to slightly below $300bn, and with TGA lower from current levels.

Election Premium in Swaptions

About 10 days ago, the benchmark 3M expiry sector of the swaptions market started to expire after the US Presidential election in November. The same thing happened with 6M expiries back in May and we noted back then that a comparison of the impacted expiry (6M back then) versus a "control" expiry (we used 3M back in May) can shed light on the election premium being priced into vol markets (see election enters). Now that the 3M benchmark expiry has crossed the election, we can repeat the exercise in reverse (3M expiries being the election-impacted expiry, and 6M being the control expiry) to infer the election premium that is now being priced into swaptions. 

To quickly recap our approach - we start by estimating the recent beta between changes in the impacted expiry (3M) implied vols and the control expiry (6M) implieds. Using this beta and the change in the control expiry implied volatility, we calculate an expected change in 3M expiry implied volatility on each day. By comparing this to the actual change, we can estimate the election premium on a per-day basis as well as the cumulative impact being priced in.

There are two questions one could ask with respect to election premia. First, how many post-election days does the swaptions market deem to be impacted and worthy of some incremental election premium. Second, what is the market's assessment of the total election premium once all the impacted days are included within the benchmark expiry window.

Figure 17 seeks to answer these questions, by calculating the cumulative election premia priced into the impacted expiry for each day after the election was first crossed, using 10Y tails. As can be seen, cumulative election premia in 3M expiries quickly ramped up to their peak of ~6 additional vol days, within a mere 3-4 days after 3M expiries first crossed the election. This suggests that markets view the event risk surrounding the election as relatively brief, impacting only 3-4 days after the election. Moreover, the cumulative election premium is about 6 additional vol days, as also seen in the figure. This is quite different with the election premia that were being priced in back in May - back then, cumulative election premia on 10Y tails reached a peak of nearly 15 additional vol days, and did so 6 business days after first crossing the election. In other words, election premia currently being priced into the swaptions market is less than half as much in total as it was back in May, and is being priced into an impacted time window that is only 3-4 business days after the election (as opposed to 6 business days back in May). Lastly, Figure 18 shows the cumulative premia being priced into different tails. As seen here, the message is fairly similar across tails, and 3M expiry implied volatility across different tails are all pricing in a cumulative 5-6 additional vol days. This is likely a reflection of shifts in the election landscape, and is a reminder that shifting election sentiment can have very real impacts on short expiry implieds going forward, in sectors where the expiry falls after the election. 

Figure 17: Markets appear to view the event risk surrounding the election as relatively brief, impacting only 3-4 days after the election

Number of election premium days as priced in to 10-year tails * calculated for 2024 election-impacted business days**, versus # business days after the election

Source: J.P. Morgan.
* Election premia is calculated by using 6Mx10Y swaptions for the May period, and 3Mx10Y swaptions for the August period. See footnote under Figure 18 for details on our method for calculating election premia
** Election-impacted business days are defined as the number of business days where 6M (May) and 3M (August) expiry swaptions expire after the Presidential election

Treasury Futures September - December Rollover Outlook Summary

As we approach the period where the majority of Treasury futures positions will need to be rolled forward from September to December expiries, we discuss our outlook for the various Treasury futures calendar spreads. A detailed discussion may be found in our U.S. Bond Futures Rollover Outlook September 2024 / December 2024, and we include a short summary here.

As we noted earlier, policy uncertainty is back in the picture, and it can also have an impact on calendar spreads. First, near term Fed expectations get priced into forward repo rates between the Sep and Dec contract delivery horizons. Lower (higher) forward repo rates imply higher (lower) carry on the back contract CTD, and therefore lower (higher) forward prices, which in turns widens (narrows) the calendar spread. Thus, volatile Fed expectations can beget volatile calendar spreads. Second, shifts in Fed expectations also have the potential to shift optimal delivery in December to last delivery day, which could bring about additional cheapening in the back contract.  In all sectors but the 2-year note, back contract CTDs have coupons in the mid-4% range. Given our current Fed outlook (we look for two 50bp cuts in Sep and Nov, and a 25bp cut in Dec), this implies a forward repo rate outlook of ~4.3-4.4% in the month of December (Figure 19). This means that back contract CTDs will likely experience carry that is cuspy - i.e., either modestly positive or modestly negative carry in the month of December. But if Fed easing expectations turn more aggressive, December carry could turn positive and shift optimal delivery to late in the month. All else equal, this implies additional cheapening in the back contract (to account for one more month of then-positive carry) and additional widening pressure on calendar spreads.

Our best guess is that risks to Fed expectations continue to be two-sided from here, and will likely remain stable over the horizon of this roll especially with the passing of recent economic data. Recent Fed-speak suggests that the Fed will likely not rush to ease aggressively in response to one data point after having exercised considerable patience thus far. Continued retracement towards higher yields on the back of strong data implies some upside pressure on forward financing rates. But there is an offset to this, stemming from a likely retracement in GC repo rates relative to RRP rates / SOFR, which seems to be already underway (see Joie de Louvre). Thus, all told, we expect financing rates to be rangebound over the roll horizon, but the risk to calendar spreads from this factor is now higher than it has been in recent quarters.

Policy uncertainty also has an impact on market liquidity and volatility. As might be expected, liquidity provision is quite dependent on policy clarity and therefore it is unsurprising that market depth was quick to tumble last week (and recovered since). But, this backdrop of policy uncertainty, elevated volatility and weaker broad-market liquidity is likely to exacerbate the effect of any technical imbalances. Commercial accounts are either moderately or significantly net long in all sectors except for the classic bond contract, and since these investors prefer to avoid delivery and roll their positions early, this preponderance of commercial longs can lead to selling pressure on calendar spreads going into first notice day. Given that net commercial futures open interest is significantly elevated in the five-year note (FV) and ultra-long bond (WN) contracts, as well as somewhat elevated in the two-year note (TU) and ten-year note (TY) sectors, we expect technical narrowing pressures to be significant in these sectors. In the ultra ten-year note (UXY) contract, net commercial longs are closer to recent averages and we expect modest narrowing pressure in this sector. In the classic bond contract, commercial accounts are net short as well as well below historical average levels - technicals (all else equal) could exert widening pressure in this sector (see Figure 20).

Figure 19: Optimal delivery in the December contract can shift to last delivery day or become cuspy if more aggressive easing gets priced in

Term repo and forward repo for different policy paths*, and optimal delivery date for the December contract CTD under each scenario

Source: J.P. Morgan.
* Projections assume that the IOER/GC basis is zero.
** Policy path assumes the eases shown take effect on September, November and December meeting dates and repo rates are calculated as a weighted average of policy rates, weighted by the number of days between each meeting

Figure 20: Commercial accounts are net long in all sectors but the US and UXY contracts

3-year statistics* on net commercial longs for the various treasury futures contracts; ‘000s of contracts, past 3Y history

Source: J.P. Morgan., CFTC
* Current as of 8/9/2024

Delivery optionality and basis convergence is likely to be a factor in the ultra-long bond contract and (to a much lesser extent) the classic bond contract. In the ultra-long bond contract, the front basis is underpricing the wildcard, and could widen as we head into the roll. Moreover, in the back month, the CTD’s much higher coupon and conversion factor make for a much smaller wildcard option value. Thus, a widening of the front WN basis will likely help pressure the WN calendar narrower. In the classic bond contract, the back net basis is underpricing delivery optionality to the tune of ~2.5/32nds. Although back bases do not usually exhibit basis convergence during the roll to the same degree that front bases do, we think in this case back basis convergence could exert some widening pressure on the calendar spread. In all the other contracts, we don’t expect basis convergence to be a driving factor of calendar spreads. 

As of this writing all contracts have different CTDs into front and back contracts, which creates the potential for the curve between CTDs to impact the calendar spread in each sector. As we discussed earlier in the piece, in the TU contract, the maturity matched swap spread curve between the front and back CTDs appears too steep relative to recent history. A correction in this could richen the front CTD relative to the back, which would exert widening pressure on the calendar spread. In the FV contract, the yield curve between the front and back contract CTDs is near the steep end of its range in recent months, mostly as a result of fairly recent moves. A flattening would exert narrowing pressure on the calendar spread . Similarly, in the TY sector, the sharp recent steepening of the spread curve suggests that the back CTD has significantly richened relative to the front - a correction would (all else equal) pressure the calendar spread wider. Additionally, given the different CTDs into the front and back contracts in all sectors, we recommend BPV-weighted hedge ratios when rolling positions forward to mitigate the directional exposure of calendar spreads. 

In summary (Figure 21), we are bearish on the weighted ultra-long bond contract calendar spread as the wildcard option is underpriced in the front contract and commercial accounts are net long; we are bullish on the classic bond contract calendar spread on the back of back-basis convergence towards fair value and investor positioning; we are mildly bearish on the weighted ultra ten-year note contract calendar spread on the back of investor positioning; we are bearish on the weighted five-year note calendar spread as commercial accounts are significantly net long and the CTD curve between the front and back contract appears too steep; and we are neutral on the ten-year note and two-year note weighted calendar spreads due to offsetting influences from investor positioning and CTD relative value considerations.

Figure 21: We are bearish on the weighted calendar spreads in the WN and FV sectors, mildly bearish in the UXY sector, bullish on the weighted calendar spreads in the US sector, and neutral on the weighted calendar spreads in the TY and TU sectors

Details for various Treasury future contracts including front price, calendar spread (/32nds), hedge ratios (as of 8/13), CTD characteristics (as of 8/13), view and drivers on each of the calendar spreads.

Source: J.P. Morgan.* Hedge Ratio: recommended number of Back contracts per 1000 Front contracts. Contract prices and calendar spread levels are based on live levels on 8/16/2024. Calendar spread levels are quoted at theoretical mids and front contract prices are quoted at the bid side.

Trading Recommendations

• Initiate 10Y swap spread narrowers
  Term funding premium has decreased and is near its recent lows, and a near term correction would pressure 10-year spreads narrower by ~2bp. We also expect term funding premium to drift higher over the medium term on the back of continuing QT and elevated duration supply in the Treasury market. Should term funding premium reach our YE24 projections, that would imply an additional 4-5bp of narrowing pressure on 10-year spreads. Lastly, fair value considerations suggest that 10-year spreads are currently trading above the fitted term structure by ~2bp more than we would expect based on empirical models of historical deviations.
  -Receive fixed in 4.375% May 2034 maturity matched SOFR swap spreads. Sell $100mn notional of the 4.375% May 2034 (yield: 3.892%, PVBP: $822.6/bp per mn notional), and receive fixed in $100mn notional of a maturity matched SOFR swap (coupon: 3.444%, PVBP: $822.3/bp per mn notional) at a swap spread of -44.8bp.
• Initiate 3s/7s swap spread curve flatteners
  The 3s/7s maturity matched swap spread curve is at the steep end of its range over the past year, and appears poised to correct. Also, an upward bias in zero duration spreads as well as term funding premium, an upward bias in 3Y spreads relative to the term structure, and a downward bias in 7Y spreads relative to the term structure, all support a 3s/7s swap spread curve flattener.
  -Pay fixed in 3.75% August 2027 maturity matched SOFR swap spreads. Buy $100mn notional of the 3.75% August 2027 (yield: 3.868%, PVBP: $279.2/bp per mn notional), and pay fixed in $95.9mn notional of a maturity matched SOFR swap (coupon: 3.652%, PVBP: $291.0/bp per mn notional) at a swap spread of -21.6bp.
  -Receive fixed in 4.875% October 2030 maturity matched SOFR swap spreads. Sell $49.3mn notional of the 4.875% October 2030 (yield: 3.799%, PVBP: $566.5/bp per mn notional), and receive fixed in $50.5mn notional of a maturity matched SOFR swap (coupon: 3.447%, PVBP: $553.2/bp per mn notional) at a swap spread of -35.2bp.
• Initiate 0.875% June 2026 / 0.875% September 2026 swap spread curve flatteners
  The swap spread curve between the 0.875% June 2026 and 0.875% Sep 2026 issues has steepened idiosyncratically, as the result of futures market technicals. This means that once the roll happens and open interest migrates from the Sep to Dec contracts, this steepening will likely correct itself.
  -Pay fixed in 0.875% June 2026 maturity matched SOFR swap spreads. Buy $100mn notional of the 0.875% June 2026 (yield: 4.09%, PVBP: $171.2/bp per mn notional), and pay fixed in $92.9mn notional of a maturity matched SOFR swap (coupon: 3.916%, PVBP: $184.3/bp per mn notional) at a swap spread of -17.4bp.
  -Receive fixed in 0.875% September 2026 maturity matched SOFR swap spreads. Sell $88.7mn notional of the 0.875% September 2026 (yield: 3.998%, PVBP: $192.9/bp per mn notional), and receive fixed in $82.1mn notional of a maturity matched SOFR swap (coupon: 3.831%, PVBP: $208.4/bp per mn notional) at a swap spread of -16.7bp.
• Initiate calendar spread wideners in US Futures
  We are bullish on the US calendar spread on the back of basis convergence and the net open interest in commercial accounts being significantly short.
  -Buy 1000 US calendar spreads at -5.25/32nds. This trade uses a 1000:1000 hedge ratio between front and back contracts, as per the recommendation in Figure 21.
• Initiate calendar spread narrowers in FV Futures
  We are bearish on the weighted FV calendar spread on the back of elevated net commercial longs and relative value in the yield curve between the front and back CTDs.
  -Sell 933 FV calendar spreads at -17.5/32nds and sell an additional 67 FVU4 contracts at 108-23.5. This trade uses a 1000:933 hedge ratio between front and back contracts, as per the recommendation in Figure 21.
• Maintain longs in 6Mx5Y swaption implied volatility on an outright basis, delta hedged daily
   P/L on this trade is currently -0.7abp. For original trade write up, see Fixed Income Markets Weekly 2024-08-02.
• Maintain conditional exposure to a flatter 1s/7s swap yield curve in a selloff using 6M expiry payer swaptions
  P/L on this trade is currently -4bp. For original trade write up, see Fixed Income Markets Weekly 2024-07-12.
• Continue to overweight 6Mx5Y swaption straddles versus 150% of the vega risk in 6Mx30Y straddles
   P/L on this trade is currently -1.8abp. For original trade write up, see Fixed Income Markets Weekly 2024-07-12.
• Maintain TU/TY invoice spread curve flatteners (1:0.35 weighted)
  P/L on this trade is currently -6.6bp. For original trade write up, see Fixed Income Markets Weekly 2024-06-07.
• Continue to Pay-fixed in 4.625% Feb ‘26 maturity matched swap spreads
  P/L on this trade is currently -1.2bp. For original trade write up, see Fixed Income Markets Weekly 2024-05-31.
• Maintain 1:0.75 risk weighted 7s/10s maturity matched swap spread curve steepeners
  P/L on this trade is currently -2.8bp. For original trade write up, see Fixed Income Markets Weekly 2024-05-31.
• Continue to Pay-fixed in 4.375% Aug ‘28 maturity matched swap spreads
  P/L on this trade is currently -1.2bp. For original trade write up, see Fixed Income Markets Weekly 2024-05-31.
• Continue to overweight 1Yx10Y straddles versus a gamma-neutral amount of 1Yx15Y straddles
   P/L on this trade is currently -2.2abp. For original trade write up, see Fixed Income Markets Weekly 2024-05-03.
• Stay long A+100 1Yx5Y payer swaptions versus selling A-100 1Yx5Y receiver swaptions, delta-hedged daily, to position for a correction in skew
  P/L on this trade is currently -10.6abp. For original trade write up, see Fixed Income Markets Weekly 2024-04-19.
   
• Continue to overweight 6Mx5Y and 6Mx30Y swaption volatility (vega weights of 0.32 and 0.76, respectively) versus selling 6Mx10Y swaption volatility
  P/L on this trade is currently 0abp. For original trade write up, see Fixed Income Markets Weekly 2024-04-05.

Closed trades over the past 12 months

P/L reported in bp of yield for swap spread, yield curve and misc. trades, and in annualized bp of volatility for option trades, unless otherwise specified

Note: trades reflect Thursday COB levels, and unwinds reflect Friday COB levels

• With still a considerable amount of uncertainty around monetary policy, short-term credit investors continue to favor bank CP/CD FRNs over fixed, with volumes now at their highest point since late January
• The recent rise in T-bill issuance over the past couple of weeks has likely contributed to the lower facility balance, though we might be approaching a floor as the uptick in net T-bill issuance stalls in early September and MMFs maintain some minimal amount of exposure to ON RRP for liquidity reasons
• We see QT as being in its endgame, with perhaps a few more months left, and project the Fed’s balance sheet to end the year at ~$7Tn, with Reserves flat at current levels, ON RRP balances slightly below $300bn, and a lower TGA
• In a complete reversal from June month-end, MMFs increased their dealer repo exposure by $205bn, with significant growth in repo exposure to FICC, UK banks, and French banks, and reduced their Fed ON RRP balances
• Current MMFs’ usage of the Fed’s ON RRP facility is highly concentrated. The top 5 and 10 individual money funds account for 44% and 62% of total ON RRP balances respectively as of July month-end. More notably, the top 2 fund families account for 66% of total ON RRP balances
• Assuming funds park 3-5% of their AUMs at the ON RRP for operational ease to meet unexpected liquidity needs, demand for ON RRP should range between $200bn and $325bn, suggesting that we are not far from the floor in terms of ON RRP balances
• Near-term catalysts: Aug flash manufacturing PMI (8/22), Aug flash services PMI (8/22), Jackson Hole Symposium (Aug 22-24), Aug Employment (9/6)

Market commentary

Treasury yields experienced heavy price action over the past two weeks, as the narrative shifted once again back towards a soft landing. In the front-end, 2-year yields witnessed a sharp increase of 19bp to 4.07%, while 10-year yields rose 10bp to 3.89%, retracing nearly all the rally witnessed so far this month. At these levels, yields have moved closer to pre-NFP. To be sure, weekly initial jobless claims fell for the second consecutive week, with the 4-week moving average dropping below this year’s peak. A closer look at the claims by state show normalization in the data for Texas and Michigan, suggesting that the the rises in July might have been technical, related to Hurricane Beryl and auto plant shutdowns respectively. Overall, the high frequency report does not suggest a major weakening in the labor market but more a gradual slowing (see US: Jobless claims moved down, M. Tasci, 8/15/24). July retail sales also exceeded expectations, with total sales up 1.0% MoM and the control category rising by 0.3%, indicating similar consumption spending as we saw in 2Q. Notably, the three-month annualized rate for the control category was 4.9%, reversing the slowdown earlier in the year (see US: July retail sales suggest no slowing in 3Q real spending, A. Reinhart, 8/15/24). Separately,the latest inflation data indicates a continued downward trend. In July, both headline and core CPI increased by 0.2% on the month, with the year-ago levels at 2.9% and 3.2%, respectively. These figures represent the slowest annual pace of inflation since 2021. With both July CPI and PPI data in hand, we foresee July core PCE to show a MoM increase of 0.122% (see US: Cooler July CPI, as expected, M. Hanson, 8/14/24).

Given the continued disinflationary process and the labor markets being looser than a year-ago, we believe there is a case for substantial Fed easing in the coming months. Our economists anticipate 125bp of cuts this year, with 50bp of easing in both September and November, followed by 25bp of reductions thereafter. However, markets have pared back on policy expectations as reflected in OIS forwards, which now indicate 95bp of cuts by year-end compared to 121bp just two weeks ago ( Figure 37).

Figure 37: Markets have pared back on policy expectations as reflected in OIS forwards, which now indicate 95bp of cuts by year-end compared to 121bp just two weeks ago

OIS-implied change in fed funds effective rate by FOMC meeting, as of 8/16/24, 8/2/24, and 1/12/24 (bp)

Source: J.P. Morgan

In the money markets, not much has changed over the past couple of weeks. Short term credit investors still favor bank CP/CD FRNs over fixed, with volumes now at their highest point since late January ( Figure 38). This trend makes sense given the substantial inversion of the money marekts curve (e.g., 1m1y T-bill yield spread is around -75bp), resulting in a relative all-in yield advantage in FRNs vs. fixed. Not to mention, there is still considerable uncertainty as to how the easing cycle will evolve as evidenced by the dramatic shift in Fed expectations over the past two weeks. In the near term, investors will likely continue to favor FRNs.

Elsewhere, RRP balances have trended lower, reaching $330bn on 8/16, down from the local peak at the end of July. Interestingly, RRP balances hit a low point of $287bn on 8/7, a level not seen since May 2021. The recent T-bill issuance over the past couple of weeks has likely contributed to the lower facility balance, though we might be approaching some sort of floor as the uptick in net T-bill issuance stalls in early September. Furthermore, as we discussed below, MMFs generally maintain some amount of exposure to the Fed’s ON RRP to help meet their liquidity needs that cannot be entirely met with repo or T-bills alone.

With ON RRP balances now hovering around $300bn, bank reserves sticky at $3.3tn, and repo rates steadily drifting higher in the fed funds corridor, it naturally raises the question of whether we have transitioned from an abundant reserves regime to an ample reserves regime. Two blog posts from Liberty Street Economics this week evaluate four indicators on the ampleness of reserves, including the use of an elasticity measure of the fed funds rate to reserves shock, and generally conclude that we have not hit that transition point (see here).

Taking a different approach in terms of measuring ampleness of liquidity in the market, our Interest Rate Derivative strategists adapted to the Fed’s analytical framework which is generally based on reserves and fed funds to be one based on reserves and ON RRP and broader short-term rates. That is, the ampleness of liquidity, defined as the sum of reserves and ON RRP, is best estimated by examining the response of broader short rates (such as SOFR) to quantity shocks. To that end, a look at the empirical relationship between changes in the SOFR-versus-top-of-the-band differential and percentage changes in total liquidity shows that indeed liquidity conditions have tightened.

Does that mean the end of QT? Fortunately, the tapered pace of QT, the potential for RRP balances to decline further within the $200-325bn range, and the anticipated decline in the TGA after 3Q collectively create some limited headroom for QT to continue. Thus, we see QT as being in its endgame, with perhaps a few more months left, and project the Fed’s balance sheet to end the year at ~$7Tn, with Reserves flat at current levels, ON RRP balances slightly below $300bn, and a lower TGA ( Figure 39) (see Interest Rate Derivatives).

Figure 38: Short term credit investors still favor bank CP/CD FRNs over fixed, with volumes now at their highest point since late January

Bank CP/CD volumes for 6m FRN vs. 6m fixed ($bn)

Source: DTCC, J.P. Morgan

Figure 39: We project the Fed’s balance sheet to end the year at ~$7Tn, with Reserves flat at current levels, ON RRP balances slightly below $300bn, and a lower TGA

Current* and projected values for Fed balance sheet assets, RRP, TGA, Reserves, and Commercial bank deposits through 2024, $bn

Source: J.P. Morgan., FRED, Federal Reserve H.4.1, Federal Reserve H.8
* Current as of 8/8/2024 Fed H.4. release
 

July MMF holdings update: MMFs and their use of Fed ON RRP

In a complete reversal from June month-end, MMFs grew their dealer repo exposure and lowered their Fed ON RRP balances at the end of July. MoM, MMFs’ dealer repo holdings increased by $205bn, driven by a surge in exposure to FICC (+$94bn), followed by UK banks (+$45bn) and French banks (+$42bn) ( Figure 40). The return in dealer activity from the latter banks is not surprising. As we previously noted, the timing of the French and UK elections likely prompted some dealers to de-risk heading into the event, and upon conclusion of those elections, they were able to loosen their balance sheets. Separately, MMFs’ dealer repo exposure to Canadian banks fell $25bn in July as it was Canadian banks’ fiscal quarter-end. We would expect this to be reversed next month.

Figure 40: MoM, MMFs’ dealer repo holdings increased by $205bn, driven by a surge in exposure to FICC (+$94bn), followed by UK banks (+$45bn) and French banks (+$42bn)

J.P. Morgan estimate of MMF repo counterparties by jurisdiction ($bn)

Source: Crane Data, J.P. Morgan

In light of the above, usage at the Fed’s ON RRP facility experienced a sharp MoM decline, with balances dropping to $413bn, the lowest level of usage at month-end since April 2021. Government MMFs accounted for the bulk of the decline, with their participation falling to $273bn, or just 66% of total RRP balances, while prime MMFs accounted for 26% ( Figure 41). Additionally, the number of MMF counterparties utilizing the RRP facility at the end of July was relatively low, with only 56 MMFs participating.

Figure 41: Government MMFs accounted for the bulk of the decline at the RRP, with their participation falling to $273bn, or just 66% of total balances

MMF and non-MMF RRP balances (LHS, $bn) vs. Govt MMF RRP % of total (RHS)

Source: Crane Data, J.P. Morgan

Even so, it is worth noting MMFs’ usage of the Fed’s ON RRP facility is currently highly concentrated. Among the MMFs that utilized the facility on July month-end, the top 5 and 10 individual money funds accounted for 44% and 62% of total ON RRP balances, respectively, with the largest placing over $70bn ( Figure 42). From a fund family perspective, the composition is even more concentrated, with the top 2 fund families comprising 66% of total ON RRP balances, which is to say portfolio allocation decisions across those two fund families could result in meaningful swings at the Fed’s ON RRP facility.

To that end, with Fed ON RRP balances hovering around $300bn, questions have emerged about whether the market is getting close to a floor. A closer look at individual MMF RRP counterparties reveals a mix of MMF types (e.g., government institutional, government retail, prime institutional, prime retail) accessing the facility, with usage ranging from <1% to 89% of fund AUMs at the end of July. To be sure, there are only two funds that dedicate more than 75% of their portfolios to ON RRP, and interestingly, those funds are prime institutional funds ( Figure 43). By and large, the vast majority of funds dedicate <25% of the portfolios to ON RRP, and within that, most dedicate <5% of the portfolios to ON RRP. If we assume that funds park 3-5% of their AUMs at the ON RRP for operational ease to meet unexpected liquidity needs, this would imply the demand for ON RRP to range between $200bn and $325bn, which suggests we are not too far off from the floor in terms of ON RRP balances.

Figure 42: On July month-end, the top 5 and 10 individual money funds accounted for 44% and 62% of total ON RRP balances respectively, with the largest placing over $70bn

MMF counterparty usage at the ON RRP ($bn)

Source: Crane Data, J.P. Morgan

Figure 43: Only two funds dedicate more than 75% of their portfolios to ON RRP, and interestingly, those funds are prime institutional funds

MMF counterparty usage at the ON RRP shown as a % of fund AUMs (# of funds)

Source: Crane Data, J.P. Morgan

In total, MMF AUMs increased by $93bn to $6.5tn in July. Government MMFs experienced the bulk of the rise, increasing by $87bn MoM. From a portfolio perspective, they continued to allocate more towards T-bills, which rose by $113bn MoM ( Figure 44). Government MMFs continued to allocation more towards T-bills, which rose by $113bn MoM. This is not surprising, particularly as T-bill outstandings were in positive territory during the month, up by $150bn MoM. Meanwhile, prime funds allocated more of their holdings towards credit, with the largest increases in Eurozone and Other Yankee banks. On net, prime funds reduced their exposure to rates and agencies ( Figure 45).

Figure 44: Government MMFs continued to allocation more towards T-bills, which rose by $113bn MoM

Asset allocation breakdown of government MMFs ($bn)

Source: Crane Data, J.P. Morgan

Figure 45: Prime funds increased their credit holdings, while reducing exposure to rates and agencies at July-end

Asset allocation breakdown of prime MMFs ($bn)

Source: Crane Data, J.P. Morgan

Nonetheless, most institutional prime funds that announced their conversion to government MMFs or liquidation of their prime funds have completed the transition, leaving only about $24bn of balances as of the end of July. As shown in Figure 46, total credit exposure among the funds that will soon convert or liquidate was just about $7bn. This is relatively small compared to the credit exposure held by the remaining universe of prime institutional and retail funds. Even with over $200bn in balances that already transitioned away from prime funds, CP/CD credit spreads have remained remarkably rangebound ( Figure 47). With the October prime MMF reform deadline rapidly approaching, it is possible that spreads could widen. However, as we have witnessed throughout this year, any spread widening will likely be limited as demand continues to outweigh supply, particularly given the amount of liquidity in the front end and a much more diversified CP/CD buyer base than in prior years.

Figure 46: Most institutional prime funds that announced their consolidation into government MMFs or liquidation of their prime funds have completed the transition, leaving only about $24bn of balances as of the end of July

Asset allocation breakdown of retail prime MMFs, institutional prime MMFs that are converting or liquidating, and remaining institutional prime MMFs, as of 7/31/24 ($bn)

Source: Crane Data, J.P. Morgan

Figure 47: Even with over $200bn in balances that already transitioned away from prime funds, CP/CD credit spreads have remained remarkably rangebound

Bank CP/CD 6m and 1y FRN spreads (bp)

Source: J.P. Morgan

Excerpted from Short-Term Market Outlook and Strategy, Teresa Ho, August 16, 2024

Watch the LTVs on your LLBs

• This week, higher coupon mortgage spreads continued to recover from their early August widening, while lower coupons edged a bit wider
• The stability of mortgages into the rally at the beginning of the month was notable and in line with the thesis held by many money managers—that if mortgages offer similar outright yields to corporates, and much less downside risk in sharp rallies, it can make sense to remain overweight. This should be a supportive of near term technicals
• Refi applications have spiked, but it remains difficult to know exactly how those applications will be apportioned on the coupon stack (and across seasoning); VA loans are already showing signs of acceleration in the Black Knight sample, and will likely pick up at locks from the first week of August flow through to closes near the end of the month
• It’s too early to glean much from the GSE daily prepay data
• LLBs and MLBs from 2023 have paid surprisingly fast over the past few months; we dig into the details and find that a greater cashout share is pulling speeds up as borrowers cashout a second time into a higher loan balance
• While this ramp effect should only be temporary, we are wary of the lower LTVs that have made their way into LLB pools, primarily from cashouts
• In the latest prepayment experience, LLBs with the lowest LTVs were more reactive to rate incentive, eroding call protection on the story
• Option cost, vega, and convexity all are tied to underlying prepayment risk, but are fundamentally different ways of attributing it; we explore how the three relate, and why the absolute maximums for each are currently occurring at different parts of the stack
• As the curve re-steepens, we should return to more ‘normal’ conditions with respect to which moneyness produces the peak values for these metrics
• We review trends in public originator Q2 2024 earnings; though many non-banks are focused on maintaining and growing capacity in anticipation of a rally, many servicers are still focused on growing their MSR holdings
• TIC data implies that foreign holdings of agency MBS rose $44.8bn in June (implying that YTD net growth was +48.8bn); we are not entirely sure what drove this large increase, but it was led by Canada followed by domiciled countries

Views

• Continue to prefer UIC conventionals, u/w 4.5s&5s
• Prefer seasoning in discounts
• Within loan bal, stick to relatively higher cuts to avoid low LTV cashouts

This week, higher coupon mortgage spreads continued to recover from their early August widening, while lower coupons edged a bit wider. In particular, lower coupon Ginnies and 15yrs were 3-5bp wider, while conventionals were just 1-2bp wider. On the margin, further concerns over bank and/or foreign portfolio reallocations may have played into some of the underperformance, though fundamentally cheaper opportunities up the stack may have just prompted some rotation as well.

We’re back on the snugger side of things, but the performance of mortgages into the flash rally was notable and in line with the thesis ascribed to by many money managers—that if mortgages offer similar outright yields to corporates, and much less downside risk in sharp rallies, it makes sense to remain overweight them. As IG has continued to retrace the 15bp widening at the start of the month, tight corporate spreads may remain a tailwind for MBS demand. Fund flows slowed a bit after the early August dislocations, but lower yields should be generally supportive of new dollars shifting to money managers.

Figure 48: Higher coupons are starting to recover from their early August widening, and lower coupons are a touch wider

Current, 1m, and 6m Treasury OAS ranges across the TBA stack in our research beta model. The black dots represent the current OAS, the blue boxes represent the 1m range, and the black lines represent the 6m range (as of 4:30PM, 8/15/2024)

Source: J.P. Morgan

Daily prepays: This week, we didn’t produce a daily prepayment flash note because the 5 days of data we got from the GSEs weren’t enough to produce a strong signal. As Figure 49 and Figure 50 show, speeds are running perhaps a bit below last month’s pace through a very limited observation window – but not by enough that we have a high degree of confidence in the readings. In addition, it will be the daily readings from later in the month that will tell us whether August prepays (September print) might incorporate some short-lag closings that were locked at the start of the month.

Figure 49: It’s too early to read much into the GSE daily prepay data...

Daily Fannie 30yr speeds vs. business day, prepaid

Source: J.P. Morgan, Fannie Mae

Figure 50: … since we have too few days and the latter half of the month is when refis might start to appear

Daily Freddie 30yr speeds vs. business day, prepaid

Source: J.P. Morgan, Freddie Mac

On the Ginnie side, the rapid 3-4 week lag length on many VA IRRRLs means that the Black Knight data for the last week or two of the month should start to get interesting. Those closes haven’t yet factored into the Black Knight closing data (or to our rate-agnostic projection methodologies) and already there are signs of a significant pickup in VA prepays in the higher coupons. 6s are generally tracking in line with our forecasts, though 2023 6.5s appears quite a bit faster. We’d expect fast peaks, and perhaps a re-emergence of the Ginnie mesa profile. Some of the mid-to-late 2023 issued 6.5s didn’t see low enough rates to sharply peak (i.e. mid 50s instead of 75c), and thus may still be somewhat sensitive to incentive.

Whenever there are such significant spikes in application volume data (see Figure 51 for the context of the magnitude of the spike), there is the question of how to apportion the single application total into actual prepayments. There is a loose correlation between the UPB of ramped and in-the-money VA loans and the application volume in a given week ( Figure 52). Clearly, a lot more high coupon loans have been created over the past two years, and as they ramp up and meet the Ginnie seasoning tests, for the same level of rates there should just be more applications. But the precision on this relationship is hardly airtight (see the regression in Figure 53), with a notable deviation in the first week of August. Moreover, you have to know how to apportion the speeds by coupon, and with Ginnies it gets even more complicated because the seasoning ramp is so sharp. As a result, even though the locks from the first week of August seem to have outstripped the eligible refi population, it’s hard to make a definitive statement about whether that means the peaks or post peaks (or the 6.5s or 7s) will perform worse than model expectations. There are just too many degrees of freedom. Still – signs point to fast prints ahead.

Figure 51: Application volume spike in the first week in August, though it has cooled somewhat this week

Optimal Blue VA Rate locks (indexed, left) and MBA VA Refi Index (right)

Source: J.P. Morgan, MBA, Optimal Blue (https://www2.optimalblue.com/)

Figure 52: Locks and eligible population roughly track…

Optimal Blue VA refi locks (left, indexed) and VA eligible UPB (right, $bn)

Source: J.P. Morgan, Ginnie Mae, Optimal Blue (https://www2.optimalblue.com/)

Figure 53: …but there can be significant deviations, like the high print we saw in the first week of August…

Optimal Blue VA Refi Locks (indexed) vs. VA eligible UPB ($bn)

Source: J.P. Morgan, Ginnie Mae, Optimal Blue (https://www2.optimalblue.com/)

This week, we look at how LLBs and MLBs have been polluted by low LTV cashout loans; in our view, this worsens the relative appeal of the collateral, both in terms of near terms speeds and its convexity into a rally. We also examine the bond math of how option cost, convexity, and vega relate (i.e. why doesn’t the same coupon have the absolute maximum for all three across the stack?). Finally, we review mortgage originator earnings; there was a general focus on adding MSR and being ready for potential refi events.

What’s up with high speeds in low loan balance?

For many years, the stability of LLB/MLB prepayments were up there in life’s certainties, but that assumption has been challenged in recent months by some odd behavior out of the 2023 vintage. In this piece, we’ll take a close look at the story and how it has evolved during the run up in home prices since 2020. These small loans have for a while now been a combination of buyers with large downpayments plus those who own outright using a cashout for liquidity. This latter group has shot up as a share of originations in the current logjammed purchase market. In our view, these faster speeds are a temporary shift upward in the baseline, but the growing cashout share is worrisome since they carry such low LTVs. In the latest prepayment episode, LLBs with low LTVs were more reactive to rate incentive. Investors should prefer LLBs/MLBs with lower cashout shares, particularly up in coupon, to avoid a near term hit to carry and generally underweight the collateral.

Recent speeds (and what’s causing them)

Figure 54 shows the 2024 prepayment prints for 2023 vintage 6.5s by spec story along with non-spec. After ramping up for the first few months, LLBs printed as high as 14c in May, and roughly tied with MLBs and nonspec in the July report. Now, while its possible to imagine nonspec having some incentive in that coupon, the lower WAC and origination costs on the LLB/MLBs mean that these borrowers are unlikely to be refinancing to get into a lower rate. Also note we’ll be referring to these stories rather interchangeably in this write-up; we did argue a year ago that they might as well be combined given their similarity in profile and small origination volumes.

Figure 54: LLBs/MLBs are paying surprisingly fast in 2023 6.5s

1m CPR for UMBS 30yr 2023 6.5s by spec story

Source: J.P. Morgan, Fannie Mae , Freddie Mac

Expanding the scope of the speeds we’re looking at for the 2023 vintage helps to explain it. Even below 6.5s, as you look at lower size tiers, speeds incrementally go up ( Figure 55) and are quite flat across coupons for LLBs/MLBs. There’s also a noticeable increase in cashout share along with smaller loan sizes.

Returning to just the 2023 6.5s, we can split the latest month of speeds by the loan’s purpose. Most are cashout/purchase, and as Figure 56 shows, the faster speeds for LLB/MLBs are being driven by the cashout loans prepaying at nearly 20c. This, along with their greater presence in pools these days, is leading to the mid-teen prints that look so unusual for out-of-the-money LLBs. We also show curtailments, since they are typically stronger in lower loan size, but they aren’t contributing as much to the recent discrepancy. Figure 57 displays how LLB/MLB cashout loans have prepaid equal to or faster than nonspec for coupons lower down the stack. One should keep in mind that very few cashouts are making it into nonspec pools, to the order of 5-7% on 5s-6.5s against 20-35% for LLBs and MLBs.

Figure 56: An increased cashout share and fast cashout prepays is driving the increase in LLB/MLB overall speeds…

2023 6.5 July 1m CPR, curtailments, and purpose share, split by original loan purpose of the loan and spec type

2023 6.5s	Spec	All	Cashout	Purchase	Rate/Term
1m CPR	LLB	13.7	17.1	10.9	22.7
	MLB	14.3	20.4	11.3	10.4
	Nonspec	13.8	20.6	13.2	16.8
1m Curtailments	LLB	1.7	1.6	1.9	1.4
	MLB	2.4	1.4	2.8	2.7
	Nonspec	1.7	1.4	1.8	0.7
Share %	LLB		35%	60%	5%
	MLB		32%	64%	4%
	Nonspec		7%	90%	3%

Source: J.P. Morgan, Fannie Mae, Freddie Mac

Figure 57: …and that trend can be seen across coupons issued in 2023

July CPR for UMBS 30yr 2023 6.5s cashout loans by spec type and coupon

Source: J.P. Morgan, Fannie Mae, Freddie Mac

So where are all these small loans rushing off to? We can use the Black Knight loan-to-loan transition data to get some answers. Focusing on just cashout loans, and 6-7 WAC, we can see that the main differentiation between MLBs and non-specs has been a much greater share of cashouts for the smaller loans ( Figure 58 and Figure 59). Turnover and curtailments show little difference between MLBs and nonspecs.

Figure 58: MLB cashout loan prepays have been driven by additional cashouts…

Prepayments by reason (excluding rate-term) for 2023 originated MLBcashout loans with 6-7% WACs

Source: J.P. Morgan, Fannie Mae, Freddie Mac

Figure 59: …which are not nearly as large a factor in >$250k cashout loans

Prepayments by reason (excluding rate-term) for 2023 originated >=$250kcashout loans with 6-7% WACs

Source: J.P. Morgan, Fannie Mae, Freddie Mac

Phrased another way, “serial cash-outs” are more likely for LLBs/MLBs. A borrower might have taken out $100k of equity in 2023 for a home remodel or college tuition, and then turns around and takes out some more in 2024. This can be a cagey approach for handling large lumpy payments over a period of years; the borrower avoids paying interest on the full balance right away which hopefully makes up for origination costs. It could also be tempting at a time when rates are historically high and the borrower could potentially get a lower overall rate on the next cashout (even if it hasn’t panned out so far in 2024).

Looking at the 2023 loans that were originally purchase, there is a slight amount of cashout activity (~1c) for MLBs versus almost nothing for >250k loans ( Figure 60 and Figure 61). This also makes sense from the context of the borrower’s decision at the time of the first loan. Either they could have put less down (and not needed to cash-out a year later) or they were already fully levered on a cheaper house, leaving little room for cashouts a year later. It’s possible that the cashouts that do happen come from borrower’s unexpectedly needing more liquidity within this short time window.

Figure 60: When restricted to just purchase loans, MLBs show only a slight amount of prepayments for the purpose of cashing out in 2024…

Prepayments by reason (excluding rate-term) for 2023 originated MLBpurchase loans with 6-7% WACs

Source: J.P. Morgan, Fannie Mae, Freddie Mac

Figure 61: …though turnover and curtailments are even lower for larger loans

Prepayments by reason (excluding rate-term) for 2023 originated >=$250kpurchase loans with 6-7% WACs

Source: J.P. Morgan, Fannie Mae, Freddie Mac

What has changed with LLB borrower after the sell-off?

As hinted at above, composition is a major part of the story. Taking a stroll down memory lane, cashout share was roughly equal between LLBs and nonspec from 2014 to 2017 ( Figure 62). At that point, the gap started to open up, with LLBs more likely to be the result of cashouts as home values (and loan sizes) crept upward. They re-aligned during the 2020-2021 refi wave as plenty of borrowers opportunistically cashed out while grabbing lower rates across the loan size spectrum. The subsequent sell-off caused the largest rift yet, with nonspec cashouts nearly nonexistent while they make up a third of LLBs. This can be seen as a migration of the purpose behind LLBs as the loan limit cap of $85k dwindles in comparison to the median home price. A greater share of them overtime should come from borrowers accessing equity on a previously paid off home rather than for purchase.

We would also mention that on a coupon level, the overall greater amount of primary rate dispersion plus the additional cash-out LLPAs leads them to be extra concentrated on the high-end of the coupon range. We illustrate this in Figure 63 by looking at the cashout% of the highest (and lowest) coupon of LLB issued in each month with at least 10% of the overall balance. Lately that has meant 6.5s and 7s in the mid-forties range against the average of a third and the lowest coupon (5.5s) in the mid-twenties.

Figure 62: The cashout share of LLBs/MLBs has fundamentally disconnected from nonspec pools…

Share of cashout loans by spec type

Source: J.P. Morgan, Freddie Mac

Figure 63: …and the greater degree of dispersion in primary rates, combined with cashout LLPAs, has increased the cashout% of high coupons

Cashout loans % of LLBs based on coupon (lowest/highest with >10% bal)

Source: J.P. Morgan, Freddie Mac

Looking across the most relevant loan characteristics, the other main change that stands out over the years is the drop in LTV ( Figure 64). This is the expected side-effect of a static loan balance cap and rising home values; there just aren’t many places anymore where an LLB-sized loan is going to be more than 70LTV. This is important to speeds later and we would note that even while the average has gone from 70 to 56 (‘18 to ‘24), the purchase loans have only gone from 76 to 66 while the cashouts have dropped from 56 to 40. The investor % within LLBs has remained a steady 20%. Some of that is just pooling dynamics; investor loans are more likely to be cut out into separate pools at higher loan sizes.

On the originator side, Rocket has been the most prolific seller within <=$110k cashouts, and their overall share has risen markedly since sell-off ( Figure 65). The share of the 2^nd/3^rd sellers has remained mostly in the 3-5% range with a rotating cast of sellers and a general migration towards nonbanks. Beyond mirroring the general change in the originator landscape, this trend could hint towards Rocket doing an effective job of reaching owners with paid-off houses and steering them towards a small cash-out rather than another product for tapping home-equity. That would also establish a relationship for the “next” cash-out as the borrower has more liquidity needs come up a year later.

We can glean one final bit of information on the changing LLB/MLB borrower from the HMDA dataset. It includes granular income data, which isn’t available in the agency disclosures, providing a more detailed look at the borrower. Figure 66 shows income by loan purpose from 2018 to 2023. While purchase income was steady before the sell-off and has by necessity risen to match the higher interest costs today, the most remarkable change has happened in cashouts. LLB cashout borrowers had incomes in the $50k range in 2018, down to almost $40k during the 2021 refi frenzy, and that level has now soared to almost $80k. Figure 67 shows how this surge is notable in absolute terms for both LLBs and MLBs and exceeds higher loan size buckets. This income data helps highlight how there are some fundamental differences with the LLB borrower today versus before the 2022 selloff.

Figure 66: For LLBs, cashout incomes soared in 2022/2023, well above the change in purchase loans

Borrower income ($k) for GSE loans <$85k in size by loan purpose, yearly

Source: J.P. Morgan, HMDA

Figure 67: The surge in LLB/MLB incomes on cashouts is impressive even in absolute terms relative to other loan size categories

For cashout loans, change in income ($k) since 2018 by size of loan

Source: J.P. Morgan, HMDA

What does it all mean for LLB speeds (and reactivity) going forward?

Figure 68: The unusual nature of the 2024 LLB cashout ramp stands out against previous years…

1m CPR on LLBs by loan age, for owner-occupied cashouts, observed yearly since 2015

Source: J.P. Morgan, Freddie Mac

Figure 69: …and for more seasoned LLBs, the greatest differentiator seems to be LTV, which matters with the low LTV cashouts being made currently

1m CPR versus incentive for LLB loans by current LTV, 18-36 WALA, observed 2019-2022

Source: J.P. Morgan, Freddie Mac

On the convexity side, we would first highlight that regardless of loan purpose or borrower, the cold hard economics behind the call protection of LLBs/MLBs remain true. As we wrote about recently, origination costs have enough of a fixed component that the payback period is significantly longer for these smaller loans at an equal rate incentive to a larger loan. When looking at s-curves, we didn’t notice that cashouts were much reactive than purchase loans after this ramp period. Figure 68

However, there does seem to be a significant connection between rate sensitivity and current LTV. Figure 69 shows how lower LTV LLBs were noticeably faster than higher LTV, particularly when moving to 50 and below. This matters today because of the bifurcation in LTVs on new issuance; cashouts as a whole are now starting at 40LTV while purchase (which have also dropped) are closer to 65LTV. So into a rally, It is possible that those lower LTV cashouts will drop out of the pool at a quick rate, and moreso than a model looking at just the blended LTV might expect.

The remaining wildcard is the nature of the borrower. Today’s LLBs have a much greater share of cashouts from borrowers with incomes relatively high to the size of the loan. On one hand, these borrowers could be more financial savvy, and more ready to pounce on a refi opportunity when presented a reasonable payback period. On the other hand, they could view the potential refi savings as a less meaningful share of their overall cashflow.

As a result of this analysis, our first recommendation would be to seek out new issue LLBs/MLBs with relatively lower cashout components. With even 5s in these stories trading at a premium, it is desirable to avoid the quicker than expected ramp from sequential cashouts. We also are wary of the greater share of low LTV loans in these pools generally and what that has meant for prepays into rallies.

Option cost, vega, and convexity—understanding why the metrics peak at different points on the stack

Mortgage brainteaser: why is option cost highest for 5.5s and 6s, convexity worst for 5s and 5.5s, and vega most negative for 4.5s and 5s? On any given day, the coupons may shift a bit, but we get the same general question from time to time. The metrics seem like they should all be tied to the prepayment option that the investor is short to the borrower, and thus the bonds that are most exposed to prepayment risk should fare ‘worst’ in all three categories. Vega and option cost are actually quite closely linked (it just takes a spread duration transformation), while OAC is related more loosely to the other two.

Option cost and vega should be closely linked, given that implied vols are a key driver of option cost; they don’t need to be precisely on top of one another, given that curve slope also is a key factor in estimating option cost (in addition to the level of vol), but they should be close. However, option cost is quoted in spread terms, while vega is quoted in price change per change in vol. We need to translate that vega into spread terms to make the comparison more direct, as shown in Figure 70. Since spread durations currently increase in magnitude as one moves down the stack, applying this transformation effectively shifts the ‘peak’ vol impact up the stack, at least as quoted in spread terms. As a result, the relationship between option cost and vega becomes much tighter ( Figure 71).

Figure 70: Vega needs to be divided by spread duration to be more comparable to option cost …

FN 30yr option cost, OAC, vega and spread duration as of 8/7/2024

FN 30yr	Opt Cost	OAC	Vega	Sprd dur	Vega / Sprd dur
2.0	14	0.29	-0.033	7.1	-0.005
2.5	20	0.02	-0.041	6.8	-0.006
3.0	27	-0.16	-0.048	6.5	-0.007
3.5	38	-0.63	-0.056	6.1	-0.009
4.0	51	-0.85	-0.060	5.6	-0.011
4.5	65	-1.33	-0.062	5.1	-0.012
5.0	86	-2.46	-0.064	4.5	-0.014
5.5	107	-2.35	-0.057	3.6	-0.016
6.0	105	-1.67	-0.046	2.9	-0.016
6.5	76	-0.92	-0.035	2.4	-0.015
7.0	47	-0.21	-0.023	2.0	-0.011

Source: J.P. Morgan

Figure 71: … after this transformation, the relationship between vega and option cost is much clearer

FN 30yr vega and vega/spread duration vs. option cost, 8/7/2024

Source: J.P. Morgan

Convexity, on the other hand, isn’t going to lend itself to such a simple conversion. The change in duration is typically largest in absolute magnitude (equivalently, most negative) for MBS at the steepest part of the s-curve (usually those around 50bp in-the-money). When forward primary mortgage rates are very different from today’s, the bonds that are currently at the steepest part of the s-curve may not be the same as those on that part of the curve in the future. For example, with today’s inverted s-curve, bonds that are somewhat out-of-the-money will become more at-the-money in the future, and thus may have a worse convexity ( Figure 72). In contrast, with a steeper curve (and for most of the last three decades, the curve has been positively sloped), bonds that are at-the-money today become out-of-the-money in the future.

Figure 72: The most convex point on the stack was typically 50bp above the current coupon when the curve was steep – but that’s not been the case for the past two years

OAC by relative moneyness (vs. the current coupon)

Source: J.P. Morgan

In addition, the magnitude of convexity is partly tied to the magnitude of duration; a peak convexity bond with a longer OAD, all else equal, typically models with an OAC of larger absolute magnitude. The inverted curve also shortens durations up the stack, which means that the expected change in duration for a move in rates is smaller. In Figure 73 and Figure 74, we show how the relationship between option cost and OAC is improved with the introduction of OAD as another explanatory variable (the second chart is shown as a rearrangement of the regression: OAC = a*option cost +b*OAD + c, or OAC-b*OAD=a*option cost +c). This helps to explain why the level of OAC is not at its most negative (vs. a long time series), but option cost is near the historical maximum.

Figure 73: OAC and option cost haven’t correlated particularly well over time …

Current coupon OAC vs option cost, 2010-now

Source: J.P. Morgan

Figure 74: … but after adjusting for the level of OAD, the relationship is tighter

Rearranged regression of OAC vs OAD & option cost, 2010-now

Source: J.P. Morgan

Hopefully this helps clear up some of these differences. As the curve resteepens, we should return to a more ‘normal’ relationship between the metrics. Until then, suffice it to say that there’s still plenty of exposure to prepayment risk in mortgages, but one’s choice of measurement of that risk is important when making the comparison to other market environments.

Public non-bank originator roundup: 2024Q2

With most public originators having reported for Q2, we review trends in the non-bank origination space this quarter, and how these trends have evolved from Q1 (see our discussion of Q1 non-bank earnings here).

Unlike the previous quarter, originator gain-on-sale margins were generally flat to slightly narrower in Q2 ( Figure 75). In earnings call transcripts, many originators are focusing on maintaining or increasing capacity in anticipation of a rally and a more robust origination market ( Figure 76).

While UWM is still biased to shedding their MSR portfolio, servicers such as Rocket and Pennymac are focused on maintaining the size of their MSR portfolios and taking advantage of higher recapture rates into a rally ( Figure 77). Other servicers, such as Mr. Cooper and Rithm, are less focused on a potential origination market and are still primarily focused on growing their MSR portfolios; in fact, Mr. Cooper is so focused on growing their MSR portfolio that they are slated to acquire all of Flagstar’s MSR holdings in the fourth quarter.

Even without the Flagstar portfolio, Mr. Cooper has acquired the most servicing of any servicer this year, with a significant portion actually coming from UWM ( Figure 78 & Figure 79).

Figure 78: UWM has still been selling a sizeable portion of its MSR holdings in Q2 …

2024 YTD cumulative transfers from the top 5 “transfer-from” servicers this year by factor date, $bn

Source: J.P. Morgan, Fannie Mae, Freddie Mac, Ginnie Mae

Figure 79: … while Mr. Cooper and Lakeview have been growing their MSR portfolios

2024 YTD cumulative transfers to the top 5 “transfer-to” servicers this year by factor date, $bn

Source: J.P. Morgan, Fannie Mae, Freddie Mac, Ginnie Mae

Week in Review

• MBA Weekly Survey: For the week ending August 9, the purchase application index rose 1.69% w/w and was 7.6% lower than year ago levels, while the refinance index rose 34.5% w/w and was 55.3% higher than the 3-month trailing level (daycount-adjusted, not seasonally-adjusted) ( Figure 63 & Figure 64).
• Freddie Enhanced Primary Survey: For the week prior to August 15, 2024, 30-year conventional conforming fixed-rate mortgages averaged 6.49%, up 2bp from the previous week ( Figure 65)
• Primary dealer specified pool positions rose to $437.8bn (+$4.8bn w/w) as-of close trading July 31. Including TBA positions of -$378.0bn, dealers were long $59.8bn (-$0.3bn w/w) pass-throughs. Other agency MBS holdings fell $0.6bn to $30.4bn.
• Fixed-rate agency gross and net issuances were $95.9bn and $19.0bn, respectively, in July. August gross supply currently stands at $67.8bn ( Figure 66).
• ICI Total Bond Long-term Mutual Fund and ETF Weekly Flows: Inflows were +$5.0bn for the week of August 7, and +$49.4bn for the month leading up to August 7 ( Figure 67).
• Our OAS optimized portfolio favors FNCL 1.5s & 5.5s, FNCI 1.5-3s, and G2 2s & 5.5-6.5s, while reducing exposure to FNCL 2-3s & 4-5s & 6.5s, and G2 2.5-3.5s & 4.5-5s. Our optimized portfolio achieves an OAS of 36 with an OAD of 5.00 versus the index’s OAS of 33 with OAD of 5.05. Cohort weights are limited to 40% redistributions in our optimized portfolio, while duration cannot change by more than 1 year and convexity can be no worse than the index ( Figure 68 & Figure 69).

Figure 80: MBA Purchase Index, calendar year overlay with daycount adjustments

Source: J.P. Morgan, MBA

Figure 81: MBA Refi Indices, seasonally adjusted

Source: J.P. Morgan, MBA

Figure 82: Primary mortgage rates, %

Source: J.P. Morgan, Optimal Blue, Freddie Mac, Mortgage News Daily, MBA

Figure 83: Gross and net fixed-rate MBS monthly issuance, $bn

Source: J.P. Morgan

Figure 84: ICI Total Bond Long-term Mutual Fund and ETF Weekly Flows, $bn

Source: J.P. Morgan, ICI

Rising CES prepayment speeds

• Our rates strategists expect the 10yr to be 3.5% by 24Q4. Along with declines in the 2yr (3.2% 24Q4) and 5yr (3.1% 24Q4) rates, mortgage rates could be at, or marginally below, 6% by the end of the year
• If realized, this will improve affordability and potentially thaw a frozen housing market. However, we do not think this becomes a massive driver of higher home prices as supply and demand potentially rise commensurately
• Our home price forecast for the remainder of the year is benign, only expecting another 1.5% of growth over the next 6 months
• Closed-end second (CES) lien prepayment speeds have been picking up. Speeds have increased in seasoned 2023 vintage deals from low to high 10c
• The recent pickup in CES speeds is likely due to borrowers either paying off the loan in cash or refinancing just the second lien
• The CES loan balance is small but paying off the loan in cash is likely still challenging for a typical GSE borrower. Additionally, partial curtailment rates in CES loans are low at less than 2c
• Borrowers refinancing just the second lien makes more sense as CES rates have come down since last year. In 2023, CES loans were originated at 10-11% note rates. CES rates are closer to mid-9% this year
• Rates are lower due to significantly better private label execution. Loan bids continue to stay competitive as pools trade at steep premiums. However, given recent higher prepayment prints, market expectations of CES speeds will have to recalibrate

Figure 85: RMBS credit issuance to date...

Source: J.P. Morgan, Bloomberg Finance L.P.

Figure 86: ...and spreads

Source: J.P. Morgan
Note: Includes our on-the-run indices. Jumbo is TOAS, non-QM and SFR are spread to treasuries. CRT is SOFR DM@10CPR. HG/HY are spread to treasuries.

Market Commentary - Are we there yet? Long anticipated FED cuts are pending...

The latest CPI print gave the market what it wanted, albeit some may have expected an even softer number. Regardless, our economists continue to believe that we will see a 50bp Fed cut in September. However, the market is assigning a lower probability to that happening. For example, on August 8^th, the market was pricing in a 45bp cut versus a 32bp cut as of August 15^th. As the short end of the curve declines, clearly short, floating rate financing and leverage benefit. However, what about mortgage rates? Currently, our rates strategists expect the 10yr to be 3.5% by 24Q4. Along with declines in the 2yr (3.2% 24Q4) and 5yr (3.1% 24Q4), this would mean mortgage rates could be at, or marginally below, 6% by the end of 2024. If realized, this will improve affordability and potentially thaw a frozen housing market. However, we do not think this becomes a massive driver of higher home prices as waiting sellers will be met with eager buyers. In other words, supply and demand potentially rise commensurately. Our home price forecast for the remainder of the year is benign, as we only expect another 1.5% of growth over the next 6 months (year-to-date HPI is +4%).

What does this all mean for relative value in the non-agency space? At the top of the capital structure, higher coupons (5.5-6.5s) do not look like they will have a ton of refinancing pressures for the rest of the year. Spreads should hold pretty firm. Lower coupons should benefit from even a modest pick-up in turnover. In credit, home prices remain supportive, but spreads are very tight. We look for more sideways moves in the second half. Any VIX-spike-induced widening in broader credit should be viewed as an opportunity in mortgage credit.

CES speeds pick up

Closed-end second (CES) lien prepayment speeds have been picking up. Speeds have increased in seasoned 2023 vintage deals from low to high 10c ( Figure 87). Some deals paid even faster. For example, the first two Rocket deals issued in 2023 paid 31c and 22c in June, respectively. Speeds saw a decline in July but were still faster compared to a couple of months ago. Since Freddie’s announcement earlier this year, prepayment speeds of CES loans have been a topic du jour, particularly as loan pools continue to trade at steep premiums.

Figure 87: Closed-end second (CES) lien prepayment speeds have been picking up

1m CPR for CES securitizations

Source: J.P. Morgan, CoreLogic

Rocket and Spring EQ are the two largest originators of securitized CES loans ( Figure 88). Rocket has securitized $3.8bn, while Spring EQ has done $1.4bn of loans. Rocket has its own private-label shelf (RCKT), but its loans have also appeared across other issuers. Spring EQ loans have been primarily securitized by Saluda Grade, but the originator was acquired by Cerberus at the end of 2023. Spring EQ tends to have lower FICO, higher CLTV, and marginally higher loan balance compared to the Rocket production ( Figure 89). Recent Rocket transactions also have a higher share of QM loans. The share of QM loans has been increasing in securitizations. As a reminder, CES loans are QM if the rate is within APOR + 3.5% for $78k-plus loans and within APOR + 6.5% for loans less than $78k. Freddie will purchase only QM CES loans.

Figure 88: Rocket and Spring EQ are the two largest originators of securitized CES loans

Original balance ($bn) of CES loans in securitizations by originator and loan vintage

Source: J.P. Morgan, CoreLogic

Figure 89: Spring EQ tends to have lower FICO, higher CLTV, and marginally higher loan balance compared to the Rocket production

Characteristics of CES loans in securitizations by originator

Source: J.P. Morgan, CoreLogic

Prepayment speeds have been increasing across all CES originators. Figure 90 and Figure 91show prepayment speeds for 2022 and 2023 vintage securitized CES loans broken out by the originator. In the 2022 vintage, Rocket speeds picked up versus Spring EQ over the last two months. The pickup in the 2022 vintage drove faster speeds in a couple of RCKT deals issued in early 2023. In the 2023 vintage, Rocket was faster in June, but it converged to other originators in July.

Figure 90: In the 2022 vintage, Rocket speeds picked up versus Spring EQ over the last two months

1m CPR for 2022vintage CES loans broken out by originator

Source: J.P. Morgan, CoreLogic

Figure 91: In the 2023 vintage, Rocket was faster in June but it converged to other originators in July

1m CPR for 2023vintage CES loans broken out by originator

Source: J.P. Morgan, CoreLogic

We think that most of the recent pickup in CES speeds is driven by originators refinancing just CES loans, and it is unlikely that borrowers are consolidating into a cash-out loan. CES speeds have been in the mid to high teens. Meanwhile, first lien borrowers who got a second lien in 2023 have been prepaying in the mid-single digits ( Figure 92). Prepayments for first lien loans with a second lien have been driven by turnover rates. Compare this to first lien borrowers who got a second lien in 2019. Most of the peak speeds in 2021 were due to borrowers refinancing both liens into a cash-out loan. This, along with the absolute level of prepays, shows that borrowers today are not consolidating two liens into a cash-out loan.

Figure 92: First lien borrowers who got a second lien in 2023 have been prepaying in the mid-single digits

1m CPR broken into turnover, cash-out and rate/term refi components for first lien GSE loans with a second lien originated in 2023

Source: J.P. Morgan, Black Knight

Figure 93: Compare this to first lien borrowers who got a second lien in 2019. Most of peak speeds in 2021 were due to borrowers refinancing both liens into a cash-out loan

1m CPR broken into turnover, cash-out and rate/term refi components for first lien GSE loans with a second lien originated in 2019

Source: J.P. Morgan, Black Knight

Hence, the recent pickup in CES speeds is due to borrowers either paying off the loan in cash or refinancing just the second lien. It is difficult to come up with an argument for the former. The CES loan balance is small, but paying off the loan in cash is likely still challenging for a typical GSE borrower. Additionally, partial curtailment rates in CES loans are low at less than 2c ( Figure 94). Higher curtailment rates would have been an indicator of borrowers’ financial strength.

Borrowers refinancing just the second lien makes more sense as CES rates have come down since last year. In 2023, CES loans were originated at 10-11% note rates. CES rates are closer to mid-9% this year; although, there still remains a noticeable spread among different originators. Rates are lower due to significantly better private label execution. Loan bids continue to stay competitive as pools trade at steep premiums. However, given recent higher prepayment prints, market expectations of CES speeds will have to recalibrate. Historical data, of course, is limited as we rely on only the past 2 years of performance. Furthermore, consolidating both liens is unlikely to be economical anytime soon, even with the first Fed cut expected in a month’s time. However, as Freddie enters the CES market and more originators offer the product, CES rates will decline, leading to faster prints for the early borrowers of second liens ( Figure 95).

Figure 94: Partial curtailment rates in CES loans are low at less than 2c

1m CPR and CCR for 2023 vintage CES loans in securitizations

Source: J.P. Morgan, CoreLogic

Figure 95: As Freddie enters the CES market and more originators offer the product, CES rates will likely decline, leading to faster prints for the early borrowers of second liens

S-curves for 6-24 WALA CES loans in Black Knight’s home equity database broken out by the observation year

Source: J.P. Morgan, Black Knight

Higher office delinquencies in July and more CRE CLO mods and buyouts

• CMBS spread movements were mixed over the past week but, over the past two weeks, on-the-run conduit CMBS spreads widened by 4-11bp across the stack, with AS bonds underperforming both within the stack and relative to similar-duration single-A corporates
• Our outlook on the economy remains one of slowing growth that avoids a recession. In light of this view, we believe that upper IG mezz bonds have sold off too much and now appear cheap compared to similar duration single-A corporates
• Agency CMBS spreads have also widened over the past two weeks. Relative to current coupon mortgage OAS and ZVs, 5-year benchmark Agency CMBS is leaning cheap
• The refi success rate for conduit CMBS loans scheduled to mature in 2024 (excluding defeased loans) was 64% through the July remit period. This is aligned with our view coming into the year where we projected that the refi success rate for 2024 would be biased towards the higher end of the 50%-66% range
• During the July 2024 remit period, the serious delinquency rate (60d+ including NP matured and FC/REO) for private label CMBS reached 4.74%, representing a 20bp increase from the prior month. Office loans were the marginal driver of the increase in the private label CMBS serious delinquency rate. In July, the serious delinquency rate for office loans increased by 100bp to 7.6%
• We have continued to see elevated levels of DLQ buyouts and loan modifications in the CRE CLO market through the first seven months of the year. At the current pace, the total volume of DLQ buyouts in 2024 will far surpass levels seen in 2023. CRE CLO loan modifications have already exceeded the total for 2023
• In July, we estimate that there were about $527mn in reinvestments throughout the entire CRE CLO market, representing the highest volume since October 2023. This increase was driven by two Arbor deals

Weekly market snapshot

Market commentary - after a modest selloff, some better relative value emerges in CMBS

CMBS spread movements were mixed over the past week but, over the past two weeks, on-the-run conduit CMBS spreads widened by 4-11 basis points across the stack, with AS bonds underperforming both within the stack and relative to similar-duration single-A corporates ( Figure 98). Our outlook on the economy remains one of slowing growth that avoids a recession. In light of this view, we believe that upper IG mezz bonds have sold off too much and now appear cheap compared to similar duration single-A corporates ( Figure 99).

Figure 98: CMBS spread movements were mixed over the past week but are modestly wider over the past two weeks

Spreads to Treasuries (bp)

	8/14/2024	w/w chg.	2w chg.
5yr Conduit CMBS LCF AAA	115	0	4
10yr Conduit CMBS LCF AAA	105	0	4
10yr Conduit CMBS AS	169	4	11
10yr Conduit CMBS AA	197	-2	9
10yr Conduit CMBS A	245	-7	7
10yr Conduit CMBS BBB-	628	-11	5
7-10yr Single-A Corporates (JULI ex-EM)	105	-6	4
7-10yr REIT Single-A	107	-2	5
7-10yr REIT BBB	146	-3	3
High Yield BB Corporates	239	-4	16
High Yield Single-B Corporates	393	-7	34
High Yield CCC Corporates	880	-14	46
HY CRE REITs	422	-1	32
10yr Freddie K A2	53	-1	3
10/9.5 Fannie DUS	57	-1	4
FRESB A10F (5CPR)	83	0	2
7-10yr AA Corporates (JULI ex-EM)	70	-6	2
FN 30yr 2% Tsy OAS	22	0	-2
5yr Freddie K A2	50	0	5
5/4.5 Fannie DUS	56	-1	4
FN 30yr 5% Tsy OAS	19	-2	2
FN 30yr 5% Tsy ZV	113	7	14

Source: J.P. Morgan

Figure 99: Conduit AS bonds look attractive to similar duration single-A corporates

10yr conduit CMBS AS spreads (actual) versus modeled spread and difference between actual and modeled spreads (residual), bp

Source: J.P. MorganNote: Modeled spreads linearly regresses 10yr conduit CMBS AS spreads against 7-10yr single-A corporate spreads over the last 10 years. R² = 73.5%, SE = 23bp.

Agency CMBS spreads have also widened over the past two weeks. Two weeks ago, we highlighted that Fannie DUS 5/4.5s appeared cheap compared to 30-year FN 5s OAS. However, this comparison may not be entirely fair, as some mortgage relative value investors might also consider ZVs when evaluating the relative value of mortgages to Agency CMBS ( Figure 100). To address this, we can perform a simple two-variable linear regression of Agency CMBS against mortgage OAS and ZVs. This regression indicates that 5-year benchmark Agency CMBS is leaning cheap ( Figure 101).

Figure 100: Mortgage relative value investors may consider ZVs as well when considering Agency CMBS

Spreads to Treasuries (bp)

Source: J.P. Morgan

Figure 101: A two-variable regression comparing Agency CMBS to current coupon mortgage OAS and ZVs indicate Agency CMBS is leaning cheap

5yr Freddie K A2 (actual) versus modeled spread and difference between actual and modeled spreads (residual), bp

Source: J.P. MorganNote: Modeled spreads linearly regresses 5yr Freddie K A2 spreads against current coupon FNMA 30yr Treasury OAS and current coupon FNMA 30yr Treasury ZV spreads over the last 5 years. R² = 72.4%, SE = 8.7bp.

Ratings Tracker

Figure 103: CMBS ratings downgrades to upgrades ratio since March 2021

Source: J.P. Morgan, Bloomberg Finance L.P.

Primary Markets

July 2024 remit review - office loans continue to push delinquencies higher

Refi success rates

The refi success rate for conduit CMBS loans scheduled to mature in 2024 (excluding defeased loans) was 64% through the July remit period ( Figure 105). This is aligned with our view coming into the year where we projected that the refi success rate for 2024 would be biased towards the higher end of the 50%-66% range. The refi success rate for conduit CMBS has been trending upwards in recent months. The low refi success rate is primarily due to office loans, with conduit office loans at a 34% refi success rate ( Figure 106). Similarly, the SASB and CRE CLO sectors also show historically low refi success rates for loans maturing this year, although this is largely due to high levels of extensions.

Figure 105: The refi success rate for conduit CMBS loans scheduled to mature in 2024 is 64% in July

Refinance success rates by CMBS product type by original maturity year and monthly breakout of refi success rates for 2024. as of July 2024

Source: J.P. Morgan, Trepp

Figure 106: Office loans continue to be the main laggard with regard to refi success rates

Refinance success rates for conduit CMBS loans by product type, as of July 2024

Source: J.P. Morgan, Trepp

Delinquencies, special servicing, and modifications

During the July 2024 remit period, the serious delinquency rate (60d+ including NP matured and FC/REO) for private label CMBS reached 4.74%, representing a 20bp increase from the prior month ( Figure 107). Additionally, the loan modification and foreclosure rate experienced slight increases reaching 10.5% and 2.4%, respectively. Meanwhile, the special servicing rate for private label CMBS remained flat at 7.4%. By private label CMBS product type, conduit CMBS experienced the largest month-over-month increase in serious delinquencies. In July, the serious delinquency rate for conduit CMBS increased by 30bp to 5.3% ( Figure 108). The serious delinquency rate for SASB and CRE CLOs increased slightly to 4% and 4.3%, respectively. As for agency CMBS, the serious delinquency rates for both Freddie K and FRESB remained relatively flat over the month at 0.4% and 2.7%.

Figure 107: The serious delinquency, loan modification, and foreclosure rate all increased in July for private label CMBS loans

60d+ delinquency rate (including NP matured and FC/REO), modification rate, modification rate excluding temporary forbearance, special servicing, and foreclosure rate for private label CMBS, as of July 2024

Source: J.P. Morgan, Trepp

Figure 108: Conduit CMBS loans experienced the largest month-over-month increase in July, reaching to 5.3%

60d+ delinquency rate (including NP matured and FC/REO) by CMBS product type, as of July 2024

Source: J.P. Morgan, Trepp

When reviewing July’s remit data by property type, it is apparent that office loans were the marginal driver of the increase in the private label CMBS serious delinquency rate. In July, the serious delinquency rate for office loans increased by 100bp to 7.6%, while the serious delinquency rate for all other major property types was either flat or fell slightly ( Figure 109). This increase was largely the result of maturity defaults as the current rate and credit environment have continued to challenge office property owners' ability to successfully refinance. In the conduit CMBS space, an additional $1.3bn of outstanding balance across 36 office loans became seriously delinquent in July. On the SASB side, the largest office loan to become delinquent was 230 Park Avenue, which is securitized in MSC 2021-230P ($670mn). Despite being in a prime area of New York City, across the street from Grand Central Station, the property entered special servicing in November 2023, claiming that a maturity default was imminent.  Contributing to the property’s underperformance have been increased operating expenses and higher borrowing costs. As of the most recent special servicer commentary, the loan has been granted a 90-day forbearance period to provide the property owner with time to consider converting some portion of the property, possibly to residential apartments. We will continue to follow the eventual outcome of the loan, especially considering that the property owner has seemingly opted against or is unable to exercise the first of three embedded 1-year extension options.

Figure 109: The marginal driver for the increase in serious delinquencies last month were office loans

60d+ delinquency rate (including NP matured and FC/REO) by property type, as of July 2024

Source: J.P. Morgan, Trepp

More CRE CLO mods and buyouts

The CRE CLO market has faced heightened stress since the second half of last year as increased interest rates have resulted in significantly higher borrowing costs and stifled new loan origination. The serious delinquency rate in the CRE CLO market has now reached 4.3%, discussed above, and is likely to continue increasing through the rest of the year. Earlier this year, we released a 3-part series (here, here, and here) discussing the different tools currently being used by CRE CLO managers to deal with this heightened stress and to manage the deleveraging of their CRE CLO vehicles.  As discussed in the series, DLQ buyouts and loan modifications are two of the most widely used tools by CRE CLO managers. As highlighted in Figure 110, we have continued to see elevated levels of DLQ buyouts and loan modifications in the CRE CLO market through the first seven months of the year. As of July 2024, the 3-month average volume of DLQ buyouts was about $120mn, which is well above historical norms. The total outstanding balance of modified loans reached $15.6bn in July, representing the highest level on record.

Figure 110: DLQ buyouts and loan modifications have remained elevated through the July remit period. Additionally, we noticed a spike in reinvestments in July

3-month average of managed CRE CLO loan payoffs, reinvestments, total cash, cumulative modified loan balance, and delinquent payoffs ($bn), as of July 2024

Source: J.P. Morgan, Trepp

In total, we estimate that CRE CLO managers have executed about $1bn of DLQ buyouts through the first seven months of the year ( Figure 111). Our methodology for estimating total DLQ buyouts can be found here. To put in context the volume of DLQ buyouts we’ve already seen this year, we estimated that $1.3bn of delinquent loans were bought out of CRE CLOs in all of 2023. At the current pace, the total volume of DLQ buyouts in 2024 will far surpass levels seen in 2023. Nearly 90% of all buyouts to occur this year have been from the 2021 and 2022 vintages as some of these deals are still in their reinvestment periods and these two vintages account for 80% of the total CRE CLO market. Similarly, $6bn of CRE CLO loans have been modified in 2024, already exceeding the total for 2023 ( Figure 112). These modifications indicate a commitment from both lenders and borrowers to the continued performance of the loans, often providing additional time for asset performance improvement and a better interest rate environment. Our economists forecast that the Fed will cut rates by 50bp in both September and November, then 25bp in every meeting thereafter. If this materializes, next year’s rate environment should allow for more refinancings and new loan originations. Consequently, we expect DLQ buyouts and loan modification volumes to decrease in 2025.

Figure 111: DLQ buyouts in 2024 will likely exceed 2023 levels

Balance of CRE CLO DLQ buyouts split by vintage ($bn), as of July 2024

Source: J.P. Morgan, Trepp

Figure 112: Loan modifications in 2024 have already surpassed 2023 levels

Balance of modified CRE CLO loans excluding benchmark transition modifications by vintage ($bn), as of July 2024

Source: J.P. Morgan, Trepp

Each manager implements its own strategy to maximize the performance of their CRE CLOs, depending on investment objectives and available resources. In Figure 113, we summarize the different strategies used by the top 15 CRE CLO managers by lifetime issuance (excluding reinvestment loans). There is clear differentiation in preferences: some managers prefer DLQ buyouts, while others favor loan modifications. Since the start of 2024, Arbor has engaged more in DLQ buyouts and loan modifications than any other CRE CLO manager. As highlighted in Figure 113, we estimate that Arbor has executed about $327mn in DLQ buyouts and over $2.6bn in loan modifications. Arbor’s increased activity in both buyouts and modifications demonstrates their commitment to maintaining the leverage in their outstanding CRE CLO vehicles.

The aggregate level of reinvestments in the CRE CLO market had been relatively low in recent months due to the lack of new origination volumes and most outstanding managed CRE CLOs being outside their reinvestment period. However, in July, we estimate that there were about $527mn in reinvestments throughout the entire CRE CLO market, representing the highest volume since October 2023. This increase was driven by two Arbor deals, ARCLO 2021-FL4 and ARCLO 2022-FL1, which accounted for 90% of all reinvestment volumes. Arbor called its ARCLO 2021-FL2 deal back in June 2024, which had about $700mn outstanding across 32 performing loans. In July, 17 of these loans, totaling about $470mn in outstanding balance, were reinvested into ARCLO 2021-FL4 and ARCLO 2022-FL1. Based on the limited data available, minor changes were made to the loan balance and gross margin of just 9 of these loans before reinvestment ( Figure 114). The bottom half of Figure 114 shows the 15 loans that have not been reinvested as of the July remit period. It will be interesting to see if any or all of these loans are eventually reinvested into Arbor’s outstanding CRE CLO vehicles.

Weekly Tracker

Figure 115: Delinquency rate

Conduit CMBS 30-day+ delinquency rate including FC/REO and NP matured (%)

Source: J.P. Morgan, Trepp

Figure 116: Delinquency cure rates

Conduit CMBS 30 day+ delinquency to performing transition rates (%)

Source: J.P. Morgan, Trepp

Figure 117: Specially serviced rate

Conduit CMBS percentage of loans in special servicing (%)

Source: J.P. Morgan, Trepp

Figure 118: Office RTTO indexed to pre-pandemic levels

Kastle System Back to Work Barometer, weekly

Source: Kastle Systems, Bloomberg Finance L.P.

Cross Sector Spreads

Cross Sector Spreads (continued)

August 16, 2024

• BB subprime auto ABS spreads have recovered, fully erasing the previous week's widening
• We review ABS auto loan level data with a focus on new/used origination trends by credit score bands, as well as delinquency/loss metrics for super-prime, prime and non-prime borrowers
• The data show early signs of improved affordability on this year’s originations with APRs starting to plateau and decrease, loan terms extending once again, and monthly payment amounts continuing to fall
• We continue to see stabilization of credit trends for nonprime and modest deterioration for recent prime auto loan originations
• Recent recovery rates on auto loan ABS pools have normalized to long term historical levels as used car values are expected to stabilize in the year ahead with limited supply

ABS spreads recovering this week from market hiccup

The tone in the ABS market improved this week with spreads settling back down following the previous week’s volatility. Demand picked back up after the credit spread widening, particularly in higher yielding segments. Our indicative BBB subprime auto ABS spreads (3-year), for example, tightened by 10bp to Treasury +165bp this week after widening by 25bp last week, while the BB spread narrowed 30bp this week, fully erasing last week’s widening. Spreads on cash surrogate ABS sectors stood unchanged to slightly tighter from last week. This reactive function of ABS spreads to broader market volatility has been consistent in recent years, wherein lower tier credits show higher beta versus comparable unsecured corporates ( Figure 119). While that habit may be hard to break, particularly if economic data turns weaker-than-expected for consumer, it offers buying opportunity for ABS across the credit spectrum against the base case of no recession in the year ahead. Consumer ABS collateral performance trends remain positive with minimal deterioration, as expected (more on auto loan and credit card metrics in the next sections). Meanwhile, the rest of August seems to lean towards the conventional summer doldrums, but vacation breaks will not last long, as the September calendar is already filling up and will be further punctuated by the widely expected first rate cut of the policy cycle.

Figure 119: Indicative ABS spreads versus comparable corporate credits

bp

Source: J.P. Morgan

Auto ABS pools showing early signs of improving loan originations

We review auto loan ABS pool origination and performance trends using ABS-EE loan tapes (for SEC-registered transactions), focusing on APR, term and monthly payment metrics for new and used vehicle loans, as well as delinquencies and losses across FICO credit bands. Overall for these securitized auto loans in aggregate, we note that affordability seems to be returning in 2024 as seen in the collateral APRs that have levelled off and/or started to decline this year, loan terms that are extending once again after shortening previously in 2022 and 2023, and monthly payments that continue to fall. Furthermore, delinquencies and losses across auto loan ABS credit bands have only deteriorated very modestly on the prime side and is no worse on subprime for recent originations. In addition, the recent recovery rates on auto loan ABS pools have normalized to long term historical trends as used car values are expected to stabilize in the year ahead with limited supply. We note that our “Auto ABS Loan Tape Rewind” Excel data supplement report to this commentary was published earlier today.

Figure 120: 600-619 auto loan 30+ delinquencies by loan age (months) and origination vintage (year) in nonprime auto ABS

% current balance

Source: J.P. Morgan, form ABS-EE via 1010data

Figure 121: 600-619 auto loan cumulative net losses by loan age (months) and origination vintage (year) in nonprime auto ABS

% original balance

Source: J.P. Morgan, form ABS-EE via 1010data

Starting with credit trends, we continue to see stabilization for nonprime auto loan ABS pools, with delinquencies and losses for 2023 originations tracking lower than 2022 ( Figure 120 and Figure 121), even as prime pools continue to see an uptick in arrears. We attribute the recent underperformance to compounded effects of inflation on household and consumer expenses, along with FICO drift/inflation, catching up with even higher quality borrowers. Nonprime borrowers experienced significantly more immediate impact from inflation. Lenders also pulled back notably on lower credit originations for their ABS pools. Taking a closer look at the prime auto loan ABS pool by finer credit score bands, we note that the super prime cohort (>720), which accounts for bulk of captive and bank auto loan ABS pools, is still largely insulated, while the lower prime bands saw the modest uptick in delinquency. For example, on the 660-719 credit score band, 30+ delinquency tracked 1.1% at loan age 6 months for 2023 originations, slightly higher compared to 0.9% for 2022 ( Figure 122). However, on the 720-779 band there is no discernible difference for delinquency ramps across 2022 and 2023 ( Figure 123). We note that there is still limited seasoning across 2023 vintages, as loans originated in 2H23 are currently only 6-12 months seasoned, and the trends could evolve in subsequent months. Given the more default conscious nature of the prime segments with the borrower juggling through payments to avoid default, these arrears have not translated to elevated losses ( Figure 124 and Figure 125). In our base case scenario, factoring decent labor market conditions and unemployment forecasts from our economists (pointing to 4.5%-4.6% through 2025), we do not expect a significant deterioration in curing trends and/or outsized uptick in loss ramps for prime vintages.  

Figure 122: 30+ delinquencies on 700-719 loans in prime auto loan ABS by loan age and vintage

% of current balance

Source: J.P. Morgan, form ABS-EE via 1010data

Figure 123: 30+ delinquencies on 720-779 loans in prime auto loan ABS by loan age and vintage

% of current balance

Source: J.P. Morgan, form ABS-EE via 1010data