2s10s and the Yield Curve: Too Narrow, Too Flat

At 4.154%, a 52 bp premium to the 3.63% Effective Federal Funds Rate, the 2 year treasury yield implies roughly 2 near term rate hikes. Meanwhile, with the 10 year at 4.651%, the 2s10s spread is a mere 50 basis points. In my opinion, the curve is too flat and the spread too narrow; short run data doesn’t necessitate hikes, while structural steepeners push up the long end.

Since CME FedWatch swung to pricing hikes in late May, I have written cases against hiking. The 2 year reached a peak of 4.36% July 23rd, and though it has come in about 21 bp, it remains fairly elevated to EFFR. The belief is that the Fed will hike to fight inflation, which the FOMC has cited is a result of tariffs, Iran, and AI demand. Let us begin with the first two.

Tariffs and the Strait are both first-order supply issues. Higher interest rates do not repeal tariffs or create barrels of oil. Furthermore, they are generally one time step-ups in a price level; if there was, say, a 10% tariff on imported potato chips, they would be 10% more expensive to import in the first year of the tariff vs the prior year. But, all else equal, in year two of the tariff, prices would increase 0%; this would be disinflation by itself, and yes, that first year of inflation would be transitory. A similar thing can be said about oil prices. For simplicities sake, if the Strait were fully closed, oil supply would fall, and prices would rise, but a year later, if the strait remained closed, while supply would remain lower, and the price level higher, than prior to the closure, the year-over-year change in supply and year-over-year change in price level, or what we call inflation, would be muted. The Federal Reserve’s toolkit isn’t well-equipped to deal with these first order effects.

What the interest rate toolkit does effect is demand. When the Fed cuts rates, banks earn less risk-free interest from keeping reserves at the Fed, while money funds earn less risk-free interest from overnight reverse repurchase agreements, incentivizing both parties to take on more private sector risk. They charge lower rates than before the cut, since the risk-free bar to clear is now lower. Those lower rates increase the quantity demanded for borrowed money. Businesses borrow more to invest, consumers that had put off purchases now feel more comfortable borrowing to spend, etc. This is how lower overnight rates are conventionally thought to spur demand, and the opposite case is also true with higher overnight rates. Psychologically, higher rates also signal less marginal demand and a commitment to bringing inflation down, and can lower inflation expectations. Inflation expectations matter because they drive both wage setting, as workers negotiate future salaries based on expected future inflation to maintain purchasing power, and spending, because expected higher prices in the future pulls spending forwards to today.

The questions from these two supply shocks, then, are 1. Have they elevated inflation expectations? 2. Is demand strong enough for producers facing higher input costs to pass them through, translating to higher prices and eventually higher inflation expectations?

Starting with inflation expectations, a great measurement is the difference between a nominal treasury and an inflation protected treasury. It is what people with money on the line think inflation will be. The gold standard is called the 5 year, 5 year forward; that is, what the average annual inflation rate is expected to be over 5 years, starting 5 years out (this strips out near-term noise from today’s supply shocks). This is derived by using 10 year treasury notes and inflation protected treasuries, as well as 5 year notes and inflation protected treasuries. Today’s rate is 2.27%; lower than the Middle East conflict’s 2.35% peak earlier in May, lower than the 2.32% a year ago today, and consistent with typical levels over the last 5 years. Expectations are not at risk. Further corroborating this is wages; unit labor costs, which essentially measure wage growth in excess of productivity growth, are still just 1.4% YoY, suggesting a firm would only have to raise prices that amount to keep the same margin if labor was their only input cost. Neither of these figures shout “hike rates to reign in expectations and cool inflationary wage pressures.”

Second, demand is unlikely to bring passthrough. Real consumption did grow 3.2% in Q2, but with strong tailwinds of tax refunds, the world cup, and an earlier prime day. The prior two quarters (Q4 ’25, Q1 ’26) were just 1.3% and .5%, respectively. Bank of America account and card data suggests spend fell .2% in July vs. June, with tax refunds spent down, and waning world cup + prime day boosts. Cleveland Fed President Beth Hammack, perhaps the biggest advocate for raising interest rates, shared on Monday that business owners were telling her how much trouble they had passing through higher input costs to consumers. These signs suggest underlying demand is not causing inflation or even at a point where supply issues manifest into expectation issues.

The third inflation risk is the AI buildout. In my last post, I argued that much of its disinflationary effects are being overlooked. I’d also add a growth narrative to AI. Recent employment reports show a continued contraction in the labor force participation rate. Just this year, it has fallen from 62.1% to 61.4%, and the level itself has genuinely contracted from 170.412 million laborers in July 2025 to 169.094 million in July 2026. Economic growth is driven by labor force growth, capital deepening, and productivity gains; the US is facing a shrinking labor force, and AI investment is driving both capital deepening and productivity. Even if AI investment wasn’t a disinflationary wage pressure and if it proved to be very rate elastic, raising rates to quell it would prove very punitive to long run economic growth. Chairman Warsh seems to appreciate this, claiming that AI will bring “the most productivity-enhancing wave of our lifetimes.”

Altogether, anchored expectations, muted labor-inflationary pressures, softer consumption demand, and a capital investment cycle carrying potential growth that no one should want to slow, I believe it is unlikely the two year gets its prescribed near-term rate hikes, bringing its yield closer to EEFR. Now, let’s move over to the long end.

On the 10 year/duration side, things are very different. The same AI buildout that pushes out hikes in the immediacy will eventually mean that the marginal product of capital will be higher; this will drive more investment, more borrowing, and eventually raise the neutral rate. We are in the early stages; over 40% of investment grade debt issuance this year has a maturity of 15+ years. The insurance firms and pension funds previously yearning for long dated assets to match against their liabilities suddenly have an overwhelming amount of supply; hyperscalers’ like Alphabet have even issued 100 year bonds.

Adding to the supply of duration is the Fed’s balance sheet; while Warsh has stated that a departure from ample-reserves is unlikely, and quite frankly he likely wouldn’t have the FOMC support to do so, he has stated a preference for switching the asset composition from long-dated mortgage backed securities and treasury bonds to short term treasury bills. There are a couple great reasons for this; the Fed’s liabilities are shorter-term, with reserves and the TGA as deposits at the Fed while ON RRP literally stands for Overnight Reverse Repurchase Agreements, and it makes most sense to match duration of assets and liabilities. Additionally, the Fed as a demand source for duration artificially has raised price and lowered yield; lower yields on low risk assets like T-bonds and mortgage backed securities are accommodative in all sorts of manners. Cash flows from equities are discounted less, boosting stock multiples; non-yielding assets like gold, art, crypto, and Bitcoin become relatively more attractive; corporates and households who borrow at risk-free + a spread can now finance projects, automobiles, homes, etc. for cheaper. Holding a lot of duration distorts risk-free rates, makes policy more accommodative and disproportionately benefits asset holders; it made much more sense when interest rates could physically not go below zero, but given that is no longer a consideration, the portfolio composition should be altered. A T-bill portfolio, on the other hand, doesn’t compress rates in the same way as bonds, since bills are much more reflective of short term policy expectations; if the fed compresses the t-bill yield below SOFR/EFFR or even ON RRP, institutional holders of treasuries will sell their bills, raise cash and lend in more profitable overnight markets, pushing the yield right back up. Switching the portfolio from bonds to bills, then, will have a notable effect on bond yields, with marginal demand flipping to marginal supply, leading to falling prices and rising yields, without changing bill yields much.

Another structural steepener is the removal of forward guidance. I want to start off by saying that I am a large fan of this. In Q3 2021, Core PCE breached 4%, but Fed governor commentary suggests an over reliance on forward guidance post-covid caused them to delay raising rates When it finally reached 5.5% in Q1 2022, they jumped in to hike, but now they were in an inflation expectations war; the 5 year 5 year forward reached an expected inflation rate of 2.59%, a rate not seen since 2014. This required a much more aggressive tightening cycle, with four 75 bip hikes in a row, than had they responded earlier. Institutions like Silicon Valley Bank, who took on duration to chase yield under the belief that rates would stay low for some time found themselves caught offsides with low coupon assets plummeting in value. This case study shows the dangers of forward guidance; unnecessary inertia in setting policy that distorts market risk setting. Getting rid of it will healthily force markets to price risk.

However, this does mean long end yields will be elevated relative to shorter term yields. Any yield can be decomposed to the average expected short term rate over the period + a term premium. Baked into the term premium is interest rate risk. In finance, a critical concept is the time value of money; that is, a dollar in the future is worth less than a dollar today. To find a future dollars value, that dollar is discounted by a certain interest rate over a certain number of periods. The conventional formula is Cash Flow / ( 1 + interest rate) ^ holding period. That exponent in the denominator is critical; longer holding period cash flows are worth less. What can also be understood in this formula is that changes in the interest rate affect further out cash flows more. When rates fall, further out cash flows benefit the most; this is why growth stocks often perform best in falling rate regimes. When rates rise, closer cash flows perform better relative to further out cash flows. We can draw the same conclusions to interest rate risk. When there is more uncertainty regarding the levels of interest rates, there is more uncertainty regarding the value of further out cash flows, so long duration assets fare worse. For bonds, that means lower prices and higher yields. The removal of forward guidance intentionally tells the market to start pricing interest rate risk for themselves; the result is a structurally higher term premium.

That said, treasury won’t let long end rates rise without a fight, nor should they; At 15%, interest is the second largest government expense, and there is about $12 trillion in debt to be refinanced over the next year. Strategies like new issuance being 85% short term bills, coordinated FX intervention with Japan to sell euros instead of treasuries to defend the yen, and encouraging FIMA repo for future FX engagements are all efforts to prevent increased supply of duration that would cause yields to spike. These are intelligent treatments to rising yields. They don’t cure the underlying drivers; robust expected future productivity & marginal product of capital, corporate issuance of duration, the Fed returning interest rate risk to the private sector, and record peace-time deficits as a % of GDP. This week’s 10 and 30 year auctions are the proof; despite coming in at 10.3 bips higher than last auction (4.683% vs 4.580%), the 10 year’s bid to cover fell from 2.59 to 2.53, while indirect bidders fell from 81.5% to 76.7%. The 30 year stopped at its highest yield since 2001.

In summary, I believe today’s yield 2s10s spread to be too narrow, and the yield curve too flat. Rate hike expectations are likely to be walked back thanks to contained inflation expectations, muted labor inflationary pressures, and the necessary role capital investment is playing in long run economic growth, bringing the 2 year yield down with it. The long end ought to stay elevated with a higher neutral rate brought on by AI productivity gains, increased duration supply from both corporates and the Fed’s balance sheet, and higher interest rate risk overwhelming short-term yield suppressors like bill heavy issuance and coordinated fx intervention.

There may be periods of volatility, even as soon as the next PCE print, which may come in firmer than expected due to July’s 6.5% MoM increase in Portfolio Management Fees in PPI and the .5% MoM increase in Computer Software and Accessories (core PCE weights these categories much higher than CPI does while weighting categories like shelter lower), causing short lived spikes in the 2 year, I believe the structural set up is for a steeper curve. I am currently eyeing the 10Y-2Y spread at the end of 2026 Contract on Kalshi; above 30 bps offers a 1.32x payout, and above 40 bps a 1.57x payout. I will likely wait for Core PCE in August to see if the statistical weighting mechanics result in a firmer print and immediate 2 year action, providing more favorable odds, before placing a trade.