A bond's expected return decomposes into premiums that can each be named, sourced and argued with:
E(R_bond) = real risk-free rate
+ inflation premium
+ maturity premium
+ credit premium
+ liquidity premium
+ tax effects
− expected credit loss
The first three build the government curve. The next two build the spread, and the tax line adjusts what the investor keeps rather than what the bond pays. The last line is the step most often skipped, and skipping it is the single most common error in a credit assumption.
What are the building blocks of a bond's expected return?
The real risk-free rate is the short real rate an economy can sustain. It is not observable, it is estimated either from inflation-linked bonds where they trade or from a structural model, and a plus or minus 0.75 point range around the estimate is honest for a developing economy. The inflation premium is expected inflation over the bond's life: for India, the Reserve Bank's 4% target with its plus or minus 2 point band anchors the long end while breakeven inflation or a professional forecasters' survey supplies the level; for the United States, the breakeven spread between nominal Treasuries and TIPS is directly observable, which makes this the easiest block in the stack. The maturity premium compensates for holding a longer bond rather than rolling short ones, and it is the term premium the curve module owns. Empirically it is positive on average, small and highly variable, and estimating it as the current curve slope is wrong because the slope also contains expected rate changes.
The credit and liquidity premiums arrive together inside a quoted spread and have to be separated, because they behave differently. A corporate spread pays you for expected loss, for bearing the uncertainty around that loss, and for the fact that you cannot sell the bond quickly at a fair price. Only the second and third are premiums you expect to earn. The first is a cost you expect to pay.
Advertisement
How is expected credit loss calculated?
Expected credit loss is probability of default multiplied by loss given default, and it is subtracted rather than added. A bond yielding 7.40% with a 0.15% annual default probability and a 55% loss given default has an expected loss of 0.0825 percentage points, so its expected return is 7.3175%, not 7.40%. The gap looks trivial for investment grade and is not: it is 9.17% of the entire spread on that Indian AAA bond and 30.00% of the spread on a comparable US BBB. Build a credit assumption from yield and you have systematically overstated it by whatever the market thinks defaults will cost.
Tax effects matter where the instrument is tax-advantaged: interest versus capital gains across holding periods in India, municipal bonds in the United States. State the investor's tax status before the table is built rather than adjusting afterwards.
How do you check the blocks against the quoted yield?
The reconciliation is what makes the method verifiable: build the blocks, sum them, and compare against the quoted yield on the actual bond. Worked example 3 does this for a US Treasury and a US corporate. The Indian government leg reconciles the same way, with a 1.20% real short rate, a 4.50% inflation premium and a 0.80% maturity premium summing to exactly the 6.50% quoted 10-year G-sec yield (illustrative September 2026 level, check against the Reserve Bank's Database on the Indian Economy or CCIL). A residual above about 15 basis points means a block is wrong, and the discipline is to find out which rather than to plug it.