Bond Index Mechanics: What Makes an Index Investable
Fixed Income Trading and Risk Management by Alexander Düring (ISBN 9781119756354)
Previous: Principal Component Analysis | Next: Portfolio Risk Management
Bond indices look objective. Chapter 34 is short and technical about why they are actually two maps: who is in the index, and at what price.
Three principles
Every index needs:
- Ex-ante definition: Universe known before performance, not “top ten performers this month.”
- Replicability: You can actually buy it. Broad indices include old illiquid corporates fine for legacy portfolios, impossible for new cash.
- Measurability: Valuable often enough (usually daily) to compare portfolios.
Replicability has a nuance: replication within the stated universe. Deutsche Börse REX uses theoretical coupons that do not exist; investors replicate with Pfandbriefe and take off-index spread risk.
Two maps
Universe map: bond i → outstanding notional Nᵢ (zero if excluded).
Price map: bond i → price pᵢ (executable, conservative).
Returns add coupons, accrued, inflation uplift mechanically once prices are set.
Pricing and frictions baked in
No universal price exists; indices average contributors and trades. Long positions valued on bids (liquidation view). New index entrants priced on ask because trackers must buy. That embeds bid-offer in reported index levels.
Global indices use T+0 settlement everywhere even though JGB is T+2 and US Treasuries T+0/T+1. Cross-market reallocations are timing-wrong on paper; funds negotiate non-standard settlement in practice (repo the US leg until JGB cash arrives).
Rebalancing
Bonds issue and mature daily. Daily full replication would crush managers in costs. Most indices adjust constituents and amounts monthly (quarterly for illiquid markets). Daily versions exist for comparison at some providers.
Why corporates in broad indices matter
Düring stresses a tension in replicability. A global aggregate index that includes every investment-grade corporate older than one year is economically representative of what legacy funds hold. It is also a fiction for a new mandate: those lines do not trade in size on a normal Tuesday. Index providers still include them because the index describes the asset class, not your next ticket.
That split explains a persistent gap between published index return and achievable return for inflows. The index assumes you could have bought at index prices on rebalance day. You often cannot. Tracking error from illiquidity is not manager skill; it is structure.
Settlement fiction and why funds get away with it
The T+0 global settlement assumption in cross-market indices is the clearest example of “consistent wrong over inconsistently right.” If you modeled JGB sale proceeds landing T+2 against US Treasury purchase T+1, every global rotation would show a financing hole. Indices skip the hole. Real managers negotiate delayed US settlement or repo the bonds. The index level never sees that financing cost unless the provider adds a explicit cash reinvestment assumption.
For performance attribution, that means comparing your fund to the index is partly comparing your repo desk to a spreadsheet zero.
What to read next in the book
Chapter 34 sets up Chapter 35’s index tracking discussion. If you do not know how the index prices bonds and when it changes weights, you cannot explain tracking error or cash drag. Lagrange optimizers and spanning sets in Chapter 35 only make sense once you accept that the liability side of a passive manager is literally “the index definition,” not an abstract market.
Düring keeps it factual: indices are products with rules, not passive truth from the market gods. Treat the rulebook like a term sheet. It tells you where tracking error is allowed to hide.