The yield curve

Take every government security, plot its yield against its maturity, and join the dots. That line is the yield curve, and it is the most information-dense free object in Indian finance.

Why it usually slopes up

Short rates are normally lower than long rates, for two reasons that are worth separating.

Expectation. A 10-year yield is, loosely, the market's view of what short rates will average over ten years. If rates are expected to rise, the long yield has to exceed today's short rate.

Compensation for commitment. Lending for ten years means ten years of chapter 3's price risk and chapter 11's inflation risk. Lenders want paying for it, and that payment is the term premium. It exists even when nobody expects rates to change.

So an upward slope is the resting state. It does not predict anything by itself.

The three shapes

Steep. A wide gap between short and long. Typically when short rates have been cut hard, or when the market expects rates and inflation to rise. Long bonds are paying you well to take duration — and chapter 6 is why that payment is not free.

Flat. Short and long yields close together. You are being offered almost nothing extra for committing for a decade. When the curve is flat, duration is poorly paid, and that is an argument for staying short that requires no forecast at all.

Inverted. Short yields above long. Unusual, and it means the market expects short rates to be meaningfully lower in future — normally because it expects a slowdown. In several economies inversion has preceded recessions often enough to be watched closely. It is a signal with a real record and not a mechanism, and it has been wrong.

What the shape is actually telling you

Be careful here, because this is where the curve gets over-read.

The curve is the market's current opinion, priced in money. It is the same kind of object as a share price in chapter 2 of the markets subject: an aggregation of many views, backed by capital, and capable of being wrong for a long time.

What it does reliably tell you is what you are being paid for taking duration right now. That is not a forecast; it is a term sheet. And it supports a decision that needs no prediction:

If the curve is flat, I am barely being paid to lend long, so I will not.

That reasoning holds whether or not the market's implied expectations come true, which makes it far more robust than trying to trade the curve's direction.

Riding the curve, and why it is not free

A strategy that sounds clever and has a catch. Buy a 10-year bond, hold it a year, and it is now a 9-year bond. If the curve slopes up and does not move, 9-year bonds yield less than 10-year ones — so your bond's yield fell, which by chapter 3 means its price rose. You earned the coupon plus a capital gain.

The catch is in "and does not move". You have taken a full year of duration risk, and if yields rise over that year the capital loss swamps the roll. Rolling down the curve is a real effect and it is a bet on stability, not an arbitrage.

Spreads sit on top of it

The government curve is the base. Every other borrower from chapter 2 sits above it by their spread — SDLs a little, PSUs more, companies more again.

Which gives two independent questions for any bond, and keeping them apart is most of the discipline in this subject:

1. Where on the curve am I lending? That is duration, and chapter 6 prices it.
2. How far above the curve is this issuer? That is credit, and chapter 7 assesses it.

A bond offering 9% when the government pays 7% has given you one number that mixes both. Separating them tells you whether you are being paid for time or for risk — and whether the part you are being paid for is the part you wanted.

The point

The yield curve plots yield against maturity and usually slopes up, from expectation and from a term premium. Flat means duration is poorly paid; inverted means the market expects lower short rates ahead. Read it as a term sheet for what you are paid today rather than as a forecast, and keep the two questions — where on the curve, and how far above it — separate.