The Yield Curve Move That Comes From Paperwork, Not Beliefs

A yield curve is supposed to move like one line. Short rates shift, the middle follows, the long end follows. The shape can steepen or flatten, but the curve usually stays smooth.

Then a weird move shows up. One maturity jumps out. A five-year gets “rich,” with its yield dropping versus the four-year and six-year. Or a seven-year gets “cheap,” with its yield rising while its neighbors barely move. The kink can be sharp enough to notice, but not large enough to come with a headline.

People search for the usual reasons. Was there a data print? A policy hint? A sudden change in risk mood? Often, there is no clean macro story. The curve looks like it is expressing a precise opinion about one tenor, even though the economy does not operate in one-year slices.

That is the moment when the curve stops looking like a survey of beliefs and starts looking like a map of constraints.

The common story is preference. Investors “want duration” here, “avoid it” there. The market “prices” a path for growth and inflation, and one point happens to move more.

That story fits broad shifts. It struggles with narrow ones.

A real change in inflation outlook usually touches many maturities. A real change in growth risk usually touches many maturities. Even supply stories are often too wide. A single auction can matter, but it does not explain why one point stays out of line after the auction is gone.

The odd moves also repeat in familiar places. The same issues become “special.” The same sections of the curve get distorted during tight funding periods. The same shapes show up during balance sheet stress, even when the macro backdrop looks calm.

When patterns repeat without a matching story, the driver is often mechanical. In rates, one of the strongest mechanical drivers is collateral.

Collateral is an asset pledged to secure a loan or a trade. In the government bond world, collateral is not a side detail. It is the backbone of funding, clearing, and leverage.

A large share of activity sits inside systems that require collateral by rule:

These systems do not treat every bond the same. They set eligibility and haircuts.

Eligibility answers a basic question: which securities are allowed?

A haircut is the discount applied to collateral value. A bond worth $100 in the market might count as only $98 for margin purposes. The difference is not a comment on the economy. It is a safety buffer against price swings and liquidity risk.

Haircuts often vary with features that line up with maturity. Shorter bonds tend to move less in price, so they can get smaller haircuts. Longer bonds tend to move more, so haircuts can be larger. Some rules also favor “on-the-run” bonds, which are the newest and most traded issues.

Now add the key constraint: many players must meet collateral needs every day. Margin is recalculated. Funding rolls. Positions are rebalanced. This creates steady demand that is not driven by views but by requirements.

In that setting, a bond’s usefulness as collateral becomes a source of value.

When collateral usefulness changes, demand can shift across maturities even if nobody changes their macro outlook. The curve bends because the cheapest way to meet rules changes.

Consider what a haircut does in practice. If a firm must post $1 billion of collateral and the haircut is 1%, it needs about $1.01 billion in bonds. If the haircut is 3%, it needs about $1.03 billion. That extra $20 million is not a rounding error for a dealer or clearing member managing a tight balance sheet.

So the system leans toward lower-haircut collateral, even when the bonds have similar yields. This is not about liking the bond. It is about using less inventory to satisfy the same rule.

Small rule differences can map into specific points on the curve. If haircuts step up around a maturity bucket, the “cost” of posting that bucket rises. Demand drifts toward the maturities that satisfy the margin more efficiently. Prices adjust. The curve kinks where the rule boundary sits.

A second force makes the distortion sharper: delivery choice.

Many funding and clearing arrangements accept a range of eligible bonds. When a borrower can choose which bond to deliver, they naturally deliver what is cheapest for them to give up and keep what is most valuable to hold. That value can come from collateral usefulness rather than yield.

When one specific issue becomes especially valuable as collateral, it can trade “special” in repo. “Special” means it can be financed at a lower repo rate than similar bonds because lenders are willing to accept less interest to get that exact bond.

Specialness matters because it changes the economics of holding the bond. A bond that funds cheaply is easier to carry. That can pull demand into that issue, pushing its cash price up and its yield down. The curve then reflects a funding benefit.

This can create a loop that looks like conviction but is really inventory math. When the bond is special, more people prefer to keep it. When more people keep it, less of it is available to lend. When less is available, it can stay special. The “richness” sits in the bond, not in the macro story.

A third force is balance sheet pressure. Dealers and banks face constraints like leverage ratios, internal limits, and funding caps. Under those constraints, the cost is not only the price of the bond. The cost is also how much gross inventory must be held to meet margin needs.

Higher haircuts and less accepted collateral force larger positions to meet the same requirements. Larger positions consume the balance sheet. When the balance sheet is scarce, the system pushes harder toward the collateral that produces the most coverage per dollar of inventory.

That is why the curve can kink most during periods of tight funding, heavy clearing needs, quarter-end constraints, or sudden changes in margin schedules. The move does not require a new belief about inflation. It only requires a new shadow price on collateral usefulness.

The key point is simple: collateral rules turn certain maturities into tools. Tools attract demand. Demand moves yields. The curve can bend because paperwork assigns different usefulness across the line.

This explanation has clear boundaries.

It weakens when collateral is plentiful and funding is easy. When the system has slack, small haircut differences matter less. The pressure to substitute into one maturity is lower, so distortions fade.

It also weakens when rules are stable and widely understood. If eligibility lists and haircut schedules do not change, the market can adapt. Specialness can still appear, but it is less likely to create sudden, isolated kinks.

Substitution can also be blocked. Position limits, concentration caps at clearinghouses, internal risk limits, or client mandates can prevent firms from moving into the “best” collateral even if it is cheaper to use. When mobility is limited, the curve cannot bend as freely around collateral value.

And this is not a full story of the yield curve. Macro forces still drive large, broad moves in rates. Policy expectations still matter. Growth and inflation still matter. Collateral mechanics explain a different class of events: narrow dislocations, recurring “special” points, and curve shapes that appear without a matching narrative.

In those moments, the curve is not arguing about the future. It is solving a constraint in the present.