Chapter 2 · Indian Markets and Instruments

2.3 Bonds and Fixed Income: Yield, Duration and Debt Funds

Lending to the government and to companies: how T-bills and G-secs are priced, how to solve for a bond's yield to maturity, why prices fall when yields rise, and how duration (a derivative) and convexity measure the rate risk of a bond or a debt fund.

📍 The situation

Meera, 52, a school principal in Nagpur, has ₹10,00,000 from a matured insurance policy and wants something safer than shares. Her bank's app shows four choices (all numbers illustrative): a 364-day Treasury bill at ₹93.35 for every ₹100 of face value; a 10-year Government of India security paying a 7.20% coupon, quoted at ₹98.60; a five-year AAA-rated bond of "Beta Finance Ltd" offering a yield of 8.10%; and two debt mutual funds — a short-duration fund and a gilt fund — whose factsheets list a "YTM", a "Macaulay duration" and a "modified duration". Her nephew, who follows the bond market, adds: "If the RBI cuts rates, the gilt fund will jump. If rates rise, it can fall — even though it only holds government bonds."

Meera's questions are the questions of this section. What does a bill bought at ₹93.35 really earn? What is the true yield of a 7.20% bond bought for less than ₹100 — and why is it not 7.20%? Why do bond prices fall when interest rates rise, and by how much? How can a fund full of "risk-free" government bonds lose money in a year? And what extra yield should a company's bond pay over a government bond? The answers combine the present value and annuity formulas of §1.1–§1.2 with one new idea from calculus: duration is a derivative — the slope of the price–yield curve — and convexity is its curvature.

⚠️ About the numbers on this page

All prices, yields, coupons, credit spreads and fund figures on this page are illustrative, chosen to look realistic; they are not current market levels, forecasts or recommendations. The bonds, companies and funds are hypothetical. Market conventions (day counts, quotation rules), the ways retail investors can buy government securities, SEBI's debt-fund categories and the taxation of bonds and debt funds can change — check the RBI, SEBI, AMFI, the offer document of the instrument and the Income Tax Department for the current position. Nothing here is investment advice.

The Language of Bonds: G-secs, T-bills and Corporate Bonds

A bond is a loan cut into tradable pieces. The borrower (the issuer) promises fixed payments on fixed dates; whoever holds the bond on those dates receives them. In India the largest borrower is the Government of India, whose securities are issued through auctions conducted by the Reserve Bank of India; state governments issue State Development Loans (SDLs); and companies, banks and financial institutions issue corporate bonds and non-convertible debentures (NCDs). Retail investors can buy government securities through the RBI's Retail Direct platform and through bidding facilities offered via stock exchanges and brokers; corporate bonds trade on the exchanges' debt segments. Most of the trading, however, is done by banks, insurers, provident and pension funds and mutual funds.

📘 Definition 2.13 — Bond, face value, coupon and maturity

A (fixed-coupon) bond is described by

  • its face value $F$ (par value, the amount repaid at the end; ₹100 for G-secs, often ₹1,000 or more for corporate bonds);
  • its coupon rate $c$, the annual interest as a fraction of face value, paid $m$ times a year in equal coupons $C=\dfrac{cF}{m}$ (dated G-secs pay half-yearly, $m=2$; many corporate bonds pay annually, $m=1$);
  • its maturity date, when the last coupon and the face value are paid; the number of coupons still to come is $n$.

A G-sec (government security) is a bond issued by the Government of India: short-term ones are Treasury bills (Definition 2.14), longer ones are dated securities with coupons. A zero-coupon bond pays no coupons, only $F$ at maturity; government "STRIPS" are zero-coupon pieces created by separating the coupons and principal of a dated G-sec. A corporate bond is issued by a company and carries credit risk — the risk that the issuer pays late, pays less, or does not pay at all.

Bond prices in India are quoted per ₹100 of face value: "₹98.60" means you pay ₹98.60 for each ₹100 you will be repaid.

Government bonds of the issuing country carry essentially no credit risk in their own currency — the government can always raise the rupees to pay. That is why they are the benchmark "risk-free" asset. But "risk-free" refers only to the promised rupees being paid. As we shall see, the price of a G-sec before maturity moves every day, and the rupees it pays may buy less than expected if inflation runs ahead (§1.1).

📘 Definition 2.14 — Treasury bill, discount and its yields

A Treasury bill (T-bill) is a zero-coupon G-sec with a maturity of less than a year; in India they are issued with maturities of 91, 182 and 364 days. A T-bill is bought at a price $P\lt F$ and redeemed at face value $F$; the difference $F-P$ is the discount, and it is the investor's entire return. For a bill with $d$ days to maturity:

  • the yield (money-market yield) relates the gain to the price paid: $\ y=\dfrac{F-P}{P}\times\dfrac{365}{d}$;
  • the discount rate (discount yield) relates the gain to the face value: $\ \delta=\dfrac{F-P}{F}\times\dfrac{365}{d}$.

The yield convention used for Indian T-bills is the first one — the gain on the price paid, with a 365-day year. Some other markets quote the discount rate, often with a 360-day year, so always check which number a quote is.

📐 Theorem 2.12 (Formula) — T-bill price and yields

For a T-bill with face value $F$, $d$ days to maturity and yield $y$ (price basis, 365-day year):

  1. the price is $$P=\frac{F}{1+y\,\frac{d}{365}};$$
  2. the discount rate is always below the yield: $$y=\frac{\delta}{1-\delta\,\frac{d}{365}}\gt\delta\quad(0\lt\delta\lt\tfrac{365}{d});$$
  3. the effective annual yield (if the same return could be earned again and again) is $\left(\dfrac FP\right)^{365/d}-1$, which exceeds $y$ for bills shorter than a year.

Why. (1) Solve $y=\frac{F-P}{P}\cdot\frac{365}{d}$ for $P$: $\frac FP=1+y\frac{d}{365}$. This is simple interest (§1.1) on the price paid. (2) From $\delta$, $P=F\left(1-\delta\frac{d}{365}\right)$, so $\frac{F-P}{P}=\frac{\delta d/365}{1-\delta d/365}$; multiplying by $\frac{365}{d}$ gives the formula, and the denominator is less than 1, so $y\gt\delta$. The discount rate understates the return because it divides the gain by the larger number $F$ instead of the amount actually invested. (3) Reinvesting $365/d$ times a year at the growth factor $F/P$ is compounding; by Bernoulli's inequality $(1+x)^k\gt1+kx$ for $k\gt1$, $x\gt0$, the effective rate is above the simple rate when $d\lt365$. ∎

✏️ Example 1 — What does a T-bill at ₹93.35 earn?
Meera can buy the 364-day T-bill at ₹93.35 per ₹100, or a 91-day T-bill at ₹98.40. For each, find the yield, the discount rate and the effective annual yield. How many rupees does ₹10 lakh of face value of the 364-day bill cost, and what does it earn?
364-day Gain ₹6.65 per ₹100. Yield $y=\dfrac{6.65}{93.35}\times\dfrac{365}{364}\approx7.143\%$; discount rate $\delta=\dfrac{6.65}{100}\times\dfrac{365}{364}\approx6.668\%$; effective $(100/93.35)^{365/364}-1\approx7.144\%$ (almost the same as $y$, because the bill is almost a year long).
91-day Gain ₹1.60. $y=\dfrac{1.60}{98.40}\times\dfrac{365}{91}\approx6.522\%$; $\delta=1.60\%\times\dfrac{365}{91}\approx6.418\%$; effective $(100/98.40)^{365/91}-1\approx6.683\%$ — noticeably above $y$, because rolling over a 91-day bill four times compounds.
Rupees ₹10,00,000 of face value is 10,000 bills of ₹100; it costs $10{,}000\times93.35=$ ₹9,33,500 and returns ₹10,00,000 after 364 days — a gain of ₹66,500 (before tax).
Check Theorem 2.12(2): $\dfrac{0.06668}{1-0.06668\times364/365}\approx0.07143$. ✓ The discount rate, computed on ₹100 rather than ₹93.35, understates the return by almost half a percentage point.

Pricing a Coupon Bond and Solving for its Yield

A coupon bond is two things glued together: an annuity of $n$ coupons $C$ and a zero-coupon payment of $F$ at the end. If the market's yield for such a bond is $y$ a year, compounded $m$ times a year (for G-secs, yields are quoted as annualised rates with half-yearly compounding), the periodic rate is $i=y/m$, and the price is the present value of everything the bond will pay — exactly the discounting of §1.1 and the annuity formula of §1.2.

📐 Theorem 2.13 (Formula) — Price of a coupon bond

On a coupon date (just after a coupon has been paid), a bond with face value $F$, coupon $C$ per period and $n$ periods to maturity, valued at periodic yield $i=y/m\gt0$, has price $$P=\sum_{k=1}^{n}\frac{C}{(1+i)^k}+\frac{F}{(1+i)^n}=C\,a_{n}+\frac{F}{(1+i)^{n}},$$ where $a_n=\dfrac{1-(1+i)^{-n}}{i}$ is the annuity factor of §1.2. Equivalently, $$P-F=(C-iF)\,a_n.$$ Hence the bond trades at par ($P=F$) when the coupon rate equals the yield ($c=y$), at a premium ($P\gt F$) when $c\gt y$, and at a discount ($P\lt F$) when $c\lt y$.

Derivation. The coupons form an ordinary annuity, whose present value is $C\,a_n$ (§1.2); the face value is a single amount discounted $n$ periods. For the second form, note that $a_n=\frac{1-(1+i)^{-n}}{i}$ gives $(1+i)^{-n}=1-i\,a_n$, so $F(1+i)^{-n}=F-iF\,a_n$ and $P=C a_n+F-iFa_n=F+(C-iF)a_n$. Since $a_n\gt0$, the sign of $P-F$ is the sign of $C-iF=F\,(c-y)/m$. ∎

The second form has a clean meaning: a bond whose coupon is $C-iF$ more per period than the market currently pays is worth exactly the present value of that extra annuity above par. If market yields rise above the coupon, the bond's coupon is now "too small", and its price must fall until a buyer earns the market yield anyway — partly from coupons and partly from the climb back to $F$.

✏️ Example 2 — One bond, three yields
A five-year G-sec with a 7.40% coupon (half-yearly) has just paid a coupon. Find its price per ₹100 if the market yield for five-year government bonds is (a) 7.40%, (b) 6.80%, (c) 8.00%. Check (b) and (c) with $P-F=(C-iF)a_n$.
Set up $F=100$, $m=2$, $C=3.70$ every half-year, $n=10$ half-years.
(a) $i=3.70\%=C/F$, so the price is exactly ₹100.00 (par).
(b) $i=0.034$, $a_{10}=\dfrac{1-1.034^{-10}}{0.034}\approx8.3587$. $P=3.70\times8.3587+100\times1.034^{-10}\approx$ ₹102.51. Check: $(3.70-3.40)\times8.3587\approx2.51$ above par. ✓ (premium)
(c) $i=0.04$, $a_{10}\approx8.1109$, $P\approx3.70\times8.1109+100\times1.04^{-10}\approx$ ₹97.57. Check: $(3.70-4.00)\times8.1109\approx-2.43$. ✓ (discount)
Notice A fall of 0.60 percentage points in yield raises the price by ₹2.51, but a rise of 0.60 points lowers it by only ₹2.43. The price–yield relation is not a straight line — it curves upward. This asymmetry is convexity, studied below.

In practice the price is what the market tells you, and the question runs the other way: what yield does this price imply?

📘 Definition 2.15 — Yield to maturity and current yield

The yield to maturity (YTM) of a bond bought at price $P$ is the annual rate $y$ (compounded $m$ times a year) for which the present value of the bond's remaining cash flows equals $P$: $$\sum_{k=1}^{n}\frac{CF_k}{(1+y/m)^k}=P,$$ where $CF_k=C$ for $k\lt n$ and $CF_n=C+F$. It is the internal rate of return (§1.3) of buying the bond and holding it to maturity, if every coupon can be reinvested at the same rate $y$.

The current yield is the simpler ratio $\dfrac{\text{annual coupon}}{\text{price}}=\dfrac{cF}{P}$. It ignores the gain or loss from the price moving to $F$ at maturity, so it is not a return measure on its own.

📐 Theorem 2.14 — The price–yield curve, and why the YTM is unique

For a bond with positive cash flows $CF_1,\dots,CF_n$, the price $P(y)=\sum_k CF_k\,(1+y/m)^{-k}$ is, for $y\gt-m$,

  1. strictly decreasing in $y$: prices and yields move in opposite directions;
  2. strictly convex in $y$: the curve bends upward, so a fall in yield raises the price more than an equal rise lowers it;
  3. $P(0)=\sum_k CF_k$ (the undiscounted total) and $P(y)\to0$ as $y\to\infty$.

Consequently, for every market price $P$ with $0\lt P\lt\sum_k CF_k$ there is exactly one positive yield to maturity.

Proof. Each term $CF_k(1+y/m)^{-k}$ has derivative $-\frac{k}{m}CF_k(1+y/m)^{-k-1}\lt0$ and second derivative $\frac{k(k+1)}{m^2}CF_k(1+y/m)^{-k-2}\gt0$; adding the terms gives $P'(y)\lt0$ and $P''(y)\gt0$, which is (1) and (2). Part (3) is direct: at $y=0$ nothing is discounted, and each term tends to $0$ as $y\to\infty$. Since $P$ is continuous and strictly decreasing from $\sum CF_k$ down towards $0$ on $(0,\infty)$, the intermediate value theorem gives a yield with $P(y)=P$ and strict monotonicity makes it unique. ∎

The YTM equation is a polynomial of degree $n$ in $v=1/(1+i)$, so for $n\gt4$ there is no algebraic formula — it must be solved numerically, exactly like the XIRR of §1.3. Because $P(y)$ is decreasing and convex, Newton's method started at a sensible guess converges fast.

🧭 How to compute a yield to maturity
  1. List the cash flows: $n$ coupons $C=cF/m$ and the face value $F$ with the last coupon. (Off a coupon date, price with the dirty price and fractional periods — see accrued interest below; spreadsheets do this for you.)
  2. Bracket the answer. If $P\lt F$ the YTM is above the coupon rate; if $P\gt F$, below it (Theorem 2.13). The current yield $cF/P$ is a good first guess.
  3. Iterate. Newton: $y_{\text{new}}=y-\dfrac{P(y)-P_{\text{mkt}}}{P'(y)}$, with $P'(y)=-\sum_k\frac{k}{m}CF_k(1+y/m)^{-k-1}$. Or bisection: halve the bracket, keeping the half where $P(y)-P_{\text{mkt}}$ changes sign.
  4. Stop when the price error is below ₹0.0001 per ₹100, and report $y$ as an annual rate with the bond's compounding ($m=2$ for G-secs). In a spreadsheet, =RATE(n, C, -P, F)*m on a coupon date, or =YIELD(settle, maturity, c, P, 100, m) on any date.
✏️ Example 3 — The YTM of Meera's 7.20% G-sec at ₹98.60
The 10-year 7.20% G-sec (half-yearly coupons) is quoted at ₹98.60 on a coupon date, with exactly 20 coupons to come. Find its current yield and its yield to maturity by Newton's method.
Current yield $7.20/98.60\approx7.302\%$. The price is below par, so the YTM is above 7.20% and (since there is also a gain of ₹1.40 at maturity) above the current yield too.
Iteration 1 Start at $y_0=7.302\%$: $P(y_0)\approx99.285$, $P'(y_0)\approx-697.8$ (rupees per unit of yield). $y_1=0.07302-\dfrac{99.285-98.60}{-697.8}\approx7.4002\%$.
Iteration 2 $P(y_1)\approx98.6030$, $P'(y_1)\approx-691.7$, so $y_2\approx7.4002\%+\dfrac{0.0030}{691.7}\approx7.4006\%$. The next step changes nothing at this precision.
Answer YTM $\approx7.40\%$ a year (half-yearly compounding). Check: $P(7.40\%)=3.60\,a_{20}+100\times1.037^{-20}\approx98.60$ ✓. Of the 7.40%, about 7.30 points is coupon income relative to the price, and the rest is the "pull to par" from ₹98.60 to ₹100.

Pull to par. If the yield stays constant, a premium bond's price drifts down to $F$ and a discount bond's price drifts up to $F$ as maturity approaches — Theorem 2.13 with a shrinking $a_n$. (Exercise II.5 traces a premium bond.) A bond held to maturity therefore delivers its promised cash flows whatever happens to prices in between; the price moves matter to anyone who sells early, and to a fund that must value its bonds at market prices every day.

📘 Definition 2.16 — Accrued interest, clean price and dirty price

Between coupon dates, the seller of a bond has "earned" part of the next coupon but will not receive it — the buyer will. The buyer therefore compensates the seller with the accrued interest $$AI=C\times\frac{\text{days since the last coupon}}{\text{days in the coupon period}}.$$ Bond prices are quoted clean (without accrued interest); the amount actually paid is the dirty price (full or invoice price) $=$ clean price $+AI$.

How days are counted is a market day-count convention. For dated G-secs the convention is 30/360 (every month counts as 30 days, a half-year as 180); corporate bonds follow the convention stated in their offer document.

✏️ Example 4 — Paying for a bond between coupon dates
An investor buys ₹5,00,000 face value of a 6.90% G-sec (half-yearly coupons) at a clean price of ₹97.25. Settlement is 82 days (30/360 count) after the last coupon. Find the accrued interest per ₹100, the dirty price and the amount payable.
Coupon $C=6.90/2=$ ₹3.45 per ₹100 each half-year (180 days in the 30/360 count).
Accrued $AI=3.45\times\dfrac{82}{180}\approx$ ₹1.5717 per ₹100.
Dirty price $97.25+1.5717=$ ₹98.8217 per ₹100.
Pay ₹5,00,000 face is 5,000 units of ₹100: $5{,}000\times98.8217\approx$ ₹4,94,108, of which ₹7,858 is accrued interest. At the next coupon date the buyer receives the full ₹17,250 coupon; net of the ₹7,858 already paid, the buyer keeps ₹9,392 — the interest for the 98 days the buyer actually held the bond.
Interactive 1 — The Price–Yield Curve, its Tangent and its Curvature

The blue curve is the price per ₹100 of a bond with half-yearly coupons, for every yield from 0% to 16% (Theorem 2.13). The dot is the bond at the market yield you choose. The red line is the duration estimate (the tangent, Theorem 2.15); switch to "duration + convexity" to see the green parabola of Theorem 2.16 hug the curve. The dashed line is face value ₹100. Press Play to sweep the yield; change the coupon and maturity to see which bonds are most rate-sensitive. (Hypothetical bond.)

7.2%107.4%
price per ₹100—
Macaulay duration—
modified duration—
convexity—
yields +1%: exact—
yields +1%: estimate—

Duration: How Much a Bond Moves — a Derivative

Meera's nephew says a long bond "moves more" than a short one when rates change. To make that precise we need the rate of change of price with respect to yield — the derivative $P'(y)$. Dividing by the price turns it into a percentage sensitivity that can be compared across bonds and funds. Remarkably, this calculus quantity is also an average time.

📘 Definition 2.17 — Macaulay duration, modified duration and PVBP

For a bond priced at yield $y$ ($i=y/m$) with cash flows $CF_k$ at times $t_k=k/m$ years:

  • the Macaulay duration is the present-value-weighted average time to the cash flows, $$D_{\text{mac}}=\sum_{k=1}^{n}t_k\,w_k,\qquad w_k=\frac{CF_k/(1+i)^k}{P},$$ measured in years (the weights $w_k$ are positive and add up to 1);
  • the modified duration is $D_{\text{mod}}=\dfrac{D_{\text{mac}}}{1+y/m}$;
  • a basis point (bp) is $0.01\%=0.0001$, and the price value of a basis point (PVBP, or DV01) of a holding worth $V$ is $D_{\text{mod}}\times V\times0.0001$ — the rupee change for a 1 bp move in yield.
📐 Theorem 2.15 (Formula) — Duration is the (relative) derivative of price

With $P(y)=\sum_k CF_k(1+y/m)^{-k}$,

  1. $$\frac{dP}{dy}=-D_{\text{mod}}\,P,\qquad\text{that is}\qquad D_{\text{mod}}=-\frac{1}{P}\frac{dP}{dy};$$ so for a small change $\Delta y$ in yield, $$\frac{\Delta P}{P}\approx-D_{\text{mod}}\,\Delta y.$$
  2. A zero-coupon bond has $D_{\text{mac}}$ equal to its maturity. A coupon bond has $D_{\text{mac}}$ less than its maturity; at the same maturity and yield, a higher coupon gives a lower duration.
  3. The duration of a portfolio of bonds (all valued at the same yield) is the value-weighted average of their durations: $D=\sum_j \dfrac{V_j}{V}D_j$.

Derivation (the calculus link). Differentiate term by term: $$\frac{dP}{dy}=\sum_{k=1}^{n}CF_k\cdot(-k)\Bigl(1+\frac ym\Bigr)^{-k-1}\cdot\frac1m=-\frac{1}{1+y/m}\sum_{k=1}^{n}\frac km\cdot\frac{CF_k}{(1+i)^k}.$$ The last sum is $\sum_k t_k\,w_kP=D_{\text{mac}}P$, so $P'(y)=-\frac{D_{\text{mac}}}{1+y/m}P=-D_{\text{mod}}P$. The linear approximation $\Delta P\approx P'(y)\Delta y$ of calculus is then (1). For (2): a zero has a single cash flow, with weight 1 at its maturity $T$, so $D_{\text{mac}}=T$; a coupon bond puts some weight on earlier times, so its weighted average is below $T$, and a larger coupon puts more weight early. For (3): the portfolio value is $V=\sum V_j$, so $V'=\sum V_j'=-\sum D_jV_j$, and dividing by $-V$ gives the weighted average. ∎

So duration answers two questions at once. As a time, it is the "centre of gravity" of the cash flows: a 10-year zero has duration 10, but Meera's 10-year 7.20% G-sec at 7.40% has $D_{\text{mac}}\approx7.27$ years, because its coupons return money early. As a sensitivity, $D_{\text{mod}}\approx7.27/1.037\approx7.01$ says that a rise of 1 percentage point in yield cuts its price by about 7%, while the 10-year zero (with $D_{\text{mod}}\approx9.64$ at the same yield) would lose about 9.6%.

✏️ Example 5 — A duration table for a four-year corporate bond
A four-year bond of the hypothetical "Gamma Infra Ltd" pays an 8% annual coupon and is priced at a yield of 8% (annual compounding). Build the duration table, find $D_{\text{mac}}$ and $D_{\text{mod}}$, estimate the price change if its yield rises by 50 bp, compare with the exact change, and find the PVBP of ₹1 crore face value.
Table Coupon equals yield, so $P=100$ (par).
$t$ (years)$CF$PV $=CF/1.08^t$weight $w$$t\times w$
187.40740.07410.0741
286.85870.06860.1372
386.35070.06350.1905
410879.38320.79383.1753
total100.00001.00003.5771
Durations $D_{\text{mac}}\approx3.577$ years (less than the 4-year maturity); $D_{\text{mod}}=3.577/1.08\approx3.312$.
+50 bp Estimate: $\Delta P/P\approx-3.312\times0.005=-1.656\%$, i.e. the price falls to about ₹98.34. Exact: at 8.5% the price is ₹98.362, a fall of $1.638\%$. The tangent slightly overstates the fall — the curve bends up (convexity).
PVBP ₹1 crore face is worth ₹1 crore at par: $3.312\times1{,}00{,}00{,}000\times0.0001\approx$ ₹3,312 per basis point. A dealer holding this bond gains or loses about ₹3,300 for every 0.01% move in its yield.

Convexity, Debt Funds and Credit Spreads

The tangent line is only the first term of a Taylor expansion. For the large yield moves that matter to investors — 1 or 2 percentage points over a year or two — the second term, the curvature, makes a visible difference. It also explains the asymmetry seen in Example 2.

📘 Definition 2.18 — Convexity

The convexity of a bond at yield $y$ is $$\mathcal C=\frac{1}{P}\frac{d^2P}{dy^2}=\frac{1}{P\,(1+y/m)^2}\sum_{k=1}^{n}\frac{k(k+1)}{m^2}\cdot\frac{CF_k}{(1+i)^k},$$ measured in years². By Theorem 2.14 it is always positive for a bond with positive cash flows. Longer bonds and lower-coupon bonds have larger convexity (roughly, convexity grows like the square of duration).

📐 Theorem 2.16 (Formula) — Duration–convexity approximation

For a change $\Delta y$ in yield, $$\frac{\Delta P}{P}\approx-D_{\text{mod}}\,\Delta y+\frac12\,\mathcal C\,(\Delta y)^2.$$ The convexity term is positive whether yields rise or fall: it makes gains from falling yields larger, and losses from rising yields smaller, than the duration estimate. The error of this approximation is of order $(\Delta y)^3$.

Why. This is Taylor's theorem to second order: $P(y+\Delta y)=P(y)+P'(y)\Delta y$ $+\,\frac12P''(y)(\Delta y)^2+O\bigl((\Delta y)^3\bigr)$. Divide by $P(y)$ and use $P'/P=-D_{\text{mod}}$ (Theorem 2.15) and $P''/P=\mathcal C$ (Definition 2.18). The formula in Definition 2.18 is the second derivative computed in the proof of Theorem 2.14, divided by $P$. ∎

✏️ Example 6 — Rate shocks on the 10-year G-sec
For Meera's 10-year 7.20% G-sec at its YTM of 7.40% (price ₹98.60), the durations are $D_{\text{mac}}\approx7.274$ years, $D_{\text{mod}}\approx7.015$, and the convexity is $\mathcal C\approx63.07$. Compare the exact percentage price change with the duration estimate and the duration–convexity estimate for yield changes of $-2,-1,+1,+2$ percentage points.
Method Exact: reprice with Theorem 2.13 at the new yield. Duration: $-7.015\,\Delta y$. With convexity: add $\frac12\times63.07\times(\Delta y)^2$, which is $+0.32\%$ for a 1-point move and $+1.26\%$ for a 2-point move.
Results
$\Delta y$new price (₹)exactdurationdur. + conv.
−2%113.76+15.38%+14.03%+15.29%
−1%105.84+7.34%+7.01%+7.33%
+1%91.98−6.71%−7.01%−6.70%
+2%85.93−12.85%−14.03%−12.77%
Read Duration alone is good to about 0.3 points for a 1-point move, but misses by more than 1 point for a 2-point move. Adding convexity brings the error below 0.1 point in every case.
Asymmetry A 2-point fall in yields gains 15.4%; a 2-point rise loses 12.8%. For a bond holder, convexity is a pleasant asymmetry — and it is the reason long bonds swing so much more in percentage terms than short ones.

From one bond to a debt fund. A debt mutual fund holds many bonds and values them at market prices every day, so its NAV (§2.2) moves with yields. Fund factsheets usually report the portfolio's YTM, Macaulay duration and modified duration, and SEBI's debt-fund categories (overnight, liquid, short duration, medium duration, long duration, gilt and others) are defined largely by the portfolio's maturity or Macaulay duration — check the current SEBI categorisation rules for the exact bands. Over a holding period, a debt fund earns two things: the carry (the portfolio's yield, minus the fund's expenses) and the price effect of yield changes.

📐 Theorem 2.17 (Rule) — Approximate return of a debt fund

A debt fund with portfolio YTM $y$, modified duration $D_{\text{mod}}$, convexity $\mathcal C$ and expense ratio $e$ earns over a horizon of $h$ years, if its yields change by $\Delta y$ (all its bonds' yields moving together), approximately $$R\approx(y-e)\,h-D_{\text{mod}}\,\Delta y+\tfrac12\,\mathcal C\,(\Delta y)^2.$$ Ignoring convexity, one horizon's carry is wiped out by a yield rise of about $$\Delta y_{\text{break-even}}\approx\frac{(y-e)\,h}{D_{\text{mod}}}.$$

Where it comes from. If yields did not move, the portfolio would earn its yield $y$ (that is what a YTM is), less the expense ratio deducted from the NAV. A move $\Delta y$ changes the portfolio's market value by the duration–convexity approximation (Theorem 2.16, applied to the portfolio as in Theorem 2.15(3)). Adding the two effects is a first-order approximation: it assumes a parallel move in yields, ignores that the carry is then earned at the new yield, that duration falls as time passes, and any credit events. It is a tool for sizing the rate risk, not a forecast. ∎

✏️ Example 7 — A debt fund in a year of rising yields
A hypothetical medium-to-long duration fund has portfolio YTM 7.6%, modified duration 6.5, convexity 60 and an expense ratio of 0.5%. Estimate its one-year return if its yields (a) stay put, (b) rise by 0.75 percentage points early in the year. What yield rise would wipe out a year's carry? What does this mean for ₹5 lakh invested?
(a) Carry only: $7.6\%-0.5\%=7.1\%$ — about ₹35,500 on ₹5 lakh.
(b) $R\approx7.1\%-6.5\times0.75\%+\tfrac12\times60\times(0.0075)^2=7.1\%-4.875\%+0.169\%\approx2.39\%$ — about ₹11,969 on ₹5 lakh.
Break-even $7.1\%/6.5\approx1.09$ percentage points, i.e. about 109 bp. A rise larger than that in the year gives a negative one-year return, even though every bond in the fund is still paying in full.
Lesson A debt fund's "YTM" is not its return. The return over a year is roughly the YTM minus costs plus or minus duration times the yield change; the longer the duration, the larger the swing.
Interactive 2 — Three Debt Funds in a Simulated Rate Cycle

Three hypothetical funds start at NAV index 100: a liquid fund (YTM 6.6%, modified duration 0.1), a short-duration fund (7.3%, 2.4) and a gilt fund (7.1%, 7.8). Every month each NAV grows by (yield − expenses)/12 and moves by −duration × Δy + ½ convexity × Δy² (Theorem 2.17). The amber line in the lower band is the simulated change in yields (all funds' yields move together). Pick a scenario, or a random path with the volatility you choose; press Play to run 36 months. (Simulated data, not market history.)

15 bp
month—
yield change so far—
liquid NAV—
short-duration NAV—
gilt NAV—
gilt: worst fall from a peak—
📈 Market case — Meera's ₹10 lakh for one year: liquid, short-duration or gilt?

Meera needs the money in about a year for her daughter's wedding. She compares three hypothetical funds (Direct plans; all figures illustrative) and three scenarios for the change in yields over the year: −1, 0 and +1 percentage point. She uses Theorem 2.17 with $h=1$.

fundYTMexpenses$D_{\text{mod}}$$\mathcal C$yields −1%unchangedyields +1%
L — liquid6.6%0.2%0.1≈06.50%6.40%6.30%
S — short duration7.3%0.4%2.489.34%6.90%4.54%
G — gilt7.1%0.5%7.88514.83%6.60%−0.78%
  1. Check one entry. Fund G, yields +1%: $6.6\%-7.8\times1\%+\frac12\times85\times(0.01)^2=6.6\%-7.8\%+0.425\%\approx-0.78\%$.
  2. In rupees on ₹10 lakh. Fund G ends the year anywhere from about ₹9,92,250 (yields +1%) to ₹11,48,250 (yields −1%); Fund L between ₹10,63,000 and ₹10,65,000; Fund S between ₹10,45,400 and ₹10,93,400.
  3. Break-even yield rise $(y-e)/D_{\text{mod}}$: Fund S $6.9\%/2.4\approx2.9$ percentage points; Fund G $6.6\%/7.8\approx0.85$ points — less than one point of yield rise erases the gilt fund's whole year of carry. Fund L's break-even is so large that rates are practically irrelevant to it over a year.
  4. Credit. Funds L and G here hold government or top-rated paper; Fund S's higher YTM includes some corporate bonds, so part of its extra yield is payment for credit risk (Theorem 2.18), not free money.
  5. Matching horizon and duration. A rough rule used by planners: choose a fund whose Macaulay duration is no longer than your holding period. With a one-year goal, a duration of about 8 years means betting the wedding money on the direction of interest rates — which nobody forecasts reliably.

What she learns. The "highest YTM" or the "best last year" fund is not the right choice; the right one fits the horizon. The arithmetic does not choose for her, but it shows that Fund G is a rate bet, Fund L is close to a deposit, and Fund S sits in between. Tax rules for debt funds and for bonds can change — she should check the current rules before comparing after-tax outcomes (§8.2).

Credit spreads. A bond of Beta Finance Ltd and a G-sec of the same maturity promise similar cash flows, but only one of them can default. Investors therefore demand a higher yield on the corporate bond.

📘 Definition 2.19 — Credit spread

The credit spread of a bond is its yield to maturity minus the yield of a government security of the same maturity, usually quoted in basis points. It compensates the investor for the expected loss from default, for the uncertainty about that loss, and for the bond's lower liquidity (it may be harder to sell quickly at a fair price). Credit ratings (AAA, AA, A, BBB …, assigned by SEBI-registered credit rating agencies) are opinions about default risk; lower ratings usually come with wider spreads, and a downgrade widens a bond's spread and lowers its price.

📐 Theorem 2.18 (Rule) — Break-even default probability

Consider a one-year zero-coupon corporate bond with yield $y_c$ and a one-year government bill with yield $y_g$. Suppose that if the company defaults, the holder recovers a fraction $R$ of what was due. A risk-neutral investor is indifferent between the two when the default probability is $$p^{*}=\frac{1}{1-R}\left(1-\frac{1+y_g}{1+y_c}\right)\approx\frac{y_c-y_g}{1-R}.$$ So a spread $s=y_c-y_g$ "pays for" a default probability of about $s/(1-R)$ a year.

Why. ₹1 invested in the bond becomes $1+y_c$ if there is no default (probability $1-p$) and $R(1+y_c)$ if there is (probability $p$). The expected payoff $(1+y_c)\bigl(1-p(1-R)\bigr)$ equals the government's $1+y_g$ exactly when $p(1-R)=1-\frac{1+y_g}{1+y_c}$. For small rates the right-hand side is close to $y_c-y_g$. ∎ In reality investors are not risk-neutral and corporate bonds are less liquid, so actual default rates are usually well below this break-even — which is exactly what the extra spread rewards.

✏️ Example 8 — What a 180 bp spread is paying for
A one-year zero-coupon bond of a hypothetical AA-rated company yields 8.6%; a one-year T-bill yields 6.8%. Assume 40% recovery on default. Find the credit spread and the break-even default probability. Then: the bond's rating is cut and its spread widens by 60 bp; if the bond (now a different, five-year bond) has modified duration 3.9, estimate the price change.
Spread $8.6\%-6.8\%=1.8\%=180$ bp.
Break-even $p^{*}=\dfrac{1}{0.6}\left(1-\dfrac{1.068}{1.086}\right)\approx2.76\%$ a year; the shortcut $0.018/0.6=3.0\%$ is close.
Read The bond is the better deal only if its true default probability is below about 2.8% a year (ignoring risk aversion and liquidity). For a highly rated issuer that may well be so — but if it defaults, the loss is 60% of the money, not a few percent.
Downgrade A wider spread is a higher yield, so the price falls: $\Delta P/P\approx-3.9\times0.006=-2.34\%$. Credit-risk funds can lose value this way without any default — and lose much more if a default happens.
⚠️ Common mistakes and myths
  1. "The coupon rate is my return." Only if you buy at par. Buy a 7.20% bond at ₹98.60 and hold it to maturity, and the yield is about 7.40% (Example 3); buy at a premium and it is below the coupon.
  2. "Government bonds are risk-free, so their prices can't fall." The rupees are safe; the price before maturity is not. A 10-year G-sec loses about 7% if yields rise 1 point (Example 6).
  3. "Prices and yields move together." They move in opposite directions (Theorem 2.14). A rate cut is good news for existing bondholders, bad news for the yield on new money.
  4. "Current yield = YTM." Current yield ignores the pull to par. For a discount bond it understates the YTM, for a premium bond it overstates it.
  5. "Duration is just maturity." Only for a zero-coupon bond. A coupon bond's duration is shorter, and a fund's duration is an average over its bonds (Theorem 2.15).
  6. "The duration estimate is exact." It is the tangent line. For moves of 1–2 points it overstates losses and understates gains; add the convexity term.
  7. "A debt fund with a higher YTM is better." A higher YTM may reflect longer duration (more rate risk) or lower credit quality (more default risk). Compare like with like, and after expenses.
  8. "A debt fund is like an FD." An FD's value does not move with market yields; a debt fund's NAV is marked to market every day and can fall over months (Interactive 2), and credit events can cause sudden falls.
  9. "The T-bill discount rate is its yield." The discount rate divides the gain by the face value; the yield divides by the price paid and is higher (Theorem 2.12).
  10. Forgetting accrued interest. The quoted clean price is not what you pay between coupon dates; you also pay the accrued interest (Example 4) — and you get it back with the next coupon.
  11. "A AAA rating is a guarantee." A rating is an opinion that can change; downgrades widen spreads and lower prices even before any default.

Exercises

All prices, yields and fund figures are illustrative. Face value ₹100 and half-yearly coupons unless stated; T-bill yields on the price basis with a 365-day year. Give prices to two decimals and yields to two decimals of a percent.

I
T-bills, G-secs and bond basics.
  • 1 A 182-day T-bill is bought at ₹96.55. Find its yield and its discount rate.
  • 2 At what price does a 91-day T-bill yield 6.5%?
  • 3 A 364-day T-bill yields 7.00%. Find its price and its effective annual yield.
  • 4 How much does an investor receive at each coupon date on ₹5,00,000 face value of a 7.26% G-sec?
  • 5 Find the current yield of an 8.50% bond priced at ₹104.00. Is the YTM higher or lower than the current yield? Why?
  • 6 Price a zero-coupon G-sec STRIPS maturing in exactly 3 years at a yield of 7.00% (half-yearly compounding).
  • 7 Meera buys the 7.20% G-sec at ₹98.60 and yields then rise. If she holds it to maturity, which of these change: the coupons she receives, the amount repaid, the bond's market price, the rate at which she can reinvest coupons? Explain.
1 Gain ₹3.45. Yield $\frac{3.45}{96.55}\times\frac{365}{182}\approx7.17\%$; discount rate $\frac{3.45}{100}\times\frac{365}{182}\approx6.92\%$.
2 $P=\dfrac{100}{1+0.065\times91/365}\approx$ ₹98.41.
3 $P=100/(1+0.07\times364/365)\approx$ ₹93.47; effective $(100/93.47)^{365/364}-1\approx7.00\%$ — for a bill of almost exactly a year the two agree.
4 $5{,}00{,}000\times0.0726/2=$ ₹18,150 every half-year (₹36,300 a year).
5 $8.50/104\approx8.17\%$. The bond trades at a premium, so at maturity the holder gets back only ₹100 for ₹104 paid: the YTM is below the current yield (and below the coupon rate, by Theorem 2.13).
6 $100/1.035^6\approx$ ₹81.35.
7 The coupons (₹3.60 per ₹100 every half-year) and the ₹100 repayment are fixed — they do not change. The market price falls (Theorem 2.14), which matters only if she sells early. The reinvestment rate rises, which actually helps a holder who reinvests coupons. Held to maturity, her realised return stays close to 7.40% (Practice corner E4 makes this precise).
II
Pricing and yield to maturity (on coupon dates).
  • 1 Price a 7-year 7.10% G-sec at yields of 7.10%, 6.60% and 7.60%.
  • 2 Use $P-F=(C-iF)a_n$ to price a 10-year 8.00% bond at a yield of 7.00%.
  • 3 A 3-year corporate bond with a 9.00% annual coupon is priced at ₹102.50. Find its YTM (annual compounding).
  • 4 Find the YTM of a 5-year 6.50% G-sec priced at ₹95.80.
  • 5 (Pull to par.) A 6-year 8.00% bond is priced at a constant 7.00% yield. Find its price now, with 4 years left and with 2 years left (just after a coupon each time), and say what it is worth at maturity.
  • 6 Show from Theorem 2.13 that if a bond is priced at par then its YTM equals its coupon rate, and conversely.
  • 7 Bond A: 4 years, 9.00% coupon, price ₹101.00. Bond B: 4 years, 6.00% coupon, price ₹90.00. Which has the higher YTM? Compute both.
1 $n=14$, $C=3.55$: at 7.10% the price is ₹100.00; at 6.60% ($i=0.033$) ₹102.77; at 7.60% ($i=0.038$) ₹97.32.
2 $i=0.035$, $n=20$, $a_{20}=\frac{1-1.035^{-20}}{0.035}\approx14.2124$; $P=100+(4.00-3.50)\times14.2124\approx$ ₹107.11.
3 Solve $\frac{9}{1+y}+\frac{9}{(1+y)^2}+\frac{109}{(1+y)^3}=102.50$: $y\approx8.03\%$ (below the coupon, since the price is above par).
4 $n=10$, $C=3.25$; solving numerically gives $y\approx7.52\%$ a year (half-yearly compounding).
5 $C=4$, $i=0.035$: 12 periods left ₹104.83; 8 left ₹103.44; 4 left ₹101.84; at maturity it is redeemed at ₹100. The premium melts away as maturity approaches.
6 $P-F=(C-iF)a_n$ with $a_n\gt0$. $P=F\iff C=iF\iff cF/m=yF/m\iff c=y$. Since the YTM is unique (Theorem 2.14), the yield at which $P=F$ is the YTM.
7 A: $y\approx8.70\%$; B: $y\approx9.03\%$. B has the higher YTM: its deep discount (₹10 below par) adds about 2.4 points a year of pull to par on top of a 6.67% current yield. A higher coupon does not mean a higher yield.
III
Duration and convexity.
  • 1 Find the Macaulay and modified durations of a 3-year 10% annual-coupon bond at a 10% yield (annual compounding).
  • 2 Find the Macaulay and modified durations of a 5-year zero-coupon bond at a yield of 7.00% (half-yearly compounding).
  • 3 A bond with modified duration 6.2 is priced at ₹101.40. Estimate its new price if its yield rises by 25 bp, and find the PVBP of ₹1 crore face value.
  • 4 A portfolio has ₹40 lakh in a bond fund of modified duration 2 and ₹60 lakh in one of modified duration 9 (same yields). Find the portfolio's modified duration and the approximate loss if yields rise by 0.5 point.
  • 5 A bond has $D_{\text{mod}}=7.0$ and $\mathcal C=63$. Estimate its percentage price change for a fall of 150 bp in its yield, with and without convexity.
  • 6 A perpetual bond pays a coupon $C$ once a year for ever, priced at yield $y=7\%$: $P=C/y$. Find $D_{\text{mod}}$ by differentiating, and $D_{\text{mac}}$.
  • 7 Explain, using the weights $w_k$, why at the same maturity and yield a 12% coupon bond has a shorter duration than a 4% coupon bond. (For a 10-year bond at 8% the Macaulay durations are about 6.54 and 7.99 years.)
1 PVs $9.0909$, $8.2645$, $82.6446$ (total ₹100). $D_{\text{mac}}=\frac{1\times9.0909+2\times8.2645+3\times82.6446}{100}\approx2.736$ years; $D_{\text{mod}}=2.736/1.10\approx2.487$.
2 $D_{\text{mac}}=5$ years (one cash flow); $D_{\text{mod}}=5/1.035\approx4.831$.
3 $\Delta P\approx-6.2\times0.0025\times101.40\approx-1.57$, so the price falls to about ₹99.83. ₹1 crore face is worth ₹1.014 crore; PVBP $=6.2\times1{,}01{,}40{,}000\times0.0001\approx$ ₹6,287.
4 $D=0.4\times2+0.6\times9=6.2$. Loss $\approx6.2\times0.005\times1{,}00{,}00{,}000=$ ₹3,10,000 (about 3.1%).
5 Duration only: $+7.0\times1.5\%=+10.50\%$. With convexity: $+10.50\%+\frac12\times63\times(0.015)^2\approx+10.50\%+0.71\%=+11.21\%$.
6 $P'(y)=-C/y^2$, so $D_{\text{mod}}=-P'/P=\dfrac{C/y^2}{C/y}=\dfrac1y\approx14.29$, and $D_{\text{mac}}=(1+y)D_{\text{mod}}=\dfrac{1+y}{y}\approx15.29$ years — finite, although the bond never matures.
7 $D_{\text{mac}}$ is the average of the times $t_k$ with weights $w_k$ proportional to the present values. The final payment $C+F$ is the same order of size for both bonds, but the 12% bond's coupons are three times larger, so more of its weight sits on the early dates and its average time is shorter: about 6.54 years against 7.99.
IV
Debt funds, credit spreads and accrued interest.
  • 1 A fund has YTM 7.4%, modified duration 4.2 and expense ratio 0.6%. Estimate its one-year return if yields are unchanged, and if they rise by 0.5 point early in the year (ignore convexity).
  • 2 For the fund in part 1, what yield rise over a year would make its one-year return zero (ignore convexity)?
  • 3 A gilt fund's NAV is ₹52.40; its modified duration is 8.5 and convexity 90. Overnight, yields jump by 15 bp. Estimate the new NAV (ignore one day's accrual).
  • 4 A five-year AA bond yields 8.45% and the five-year G-sec yields 7.05%. Find the credit spread in bp. After a downgrade the spread widens by 60 bp; with modified duration 3.9, estimate the bond's price change.
  • 5 A 7.30% G-sec is bought at a clean price of ₹101.15, 117 days (30/360) after its last coupon. Find the accrued interest, the dirty price and the amount payable for ₹2,00,000 face value.
  • 6 A one-year zero-coupon bond of a hypothetical housing finance company yields 9.2%, and the one-year T-bill yields 6.9%. With 50% recovery on default, find the break-even default probability (exact and approximate).
1 Unchanged: $7.4\%-0.6\%=6.8\%$. Yields +0.5: $6.8\%-4.2\times0.5\%=4.7\%$.
2 $6.8\%/4.2\approx1.62$ points (about 162 bp).
3 $\Delta N/N\approx-8.5\times0.0015+\frac12\times90\times0.0015^2\approx-1.275\%+0.010\%=-1.265\%$; new NAV $\approx52.40\times(1-0.01265)\approx$ ₹51.74. At a carry of about 7% a year, that is more than two months of carry lost in one day.
4 Spread $8.45-7.05=1.40\%=140$ bp. Price change $\approx-3.9\times0.006=-2.34\%$.
5 $C=3.65$; $AI=3.65\times117/180\approx$ ₹2.3725; dirty price ₹103.5225; for 2,000 units: $2{,}000\times103.5225=$ ₹2,07,045.
6 $p^{*}=\dfrac{1}{0.5}\left(1-\dfrac{1.069}{1.092}\right)\approx4.21\%$; approximate $\dfrac{0.023}{0.5}=4.6\%$. The shortcut overstates a little when spreads are wide.
🎯 Practice corner NISM-styleCase studyInterview

E1. NISM-style An investor holds a fixed-coupon bond. If market interest rates rise, the price of the bond will generally (A) rise (B) fall (C) remain unchanged, since the coupon is fixed (D) rise only if the coupon rate is above the yield.

Idea The price is the present value of fixed cash flows; a higher discount rate lowers every term (Theorem 2.14). (C) confuses fixed cash flows with a fixed price; (D) confuses the price level (premium/discount) with its direction of change.
Answer (B).

E2. NISM-style A 91-day Treasury bill is bought at ₹98.25 per ₹100 face value. Its yield (price basis, 365-day year) is closest to (A) 7.00% (B) 7.02% (C) 7.14% (D) 1.78%.

Compute $\dfrac{1.75}{98.25}\times\dfrac{365}{91}\approx0.01781\times4.011\approx7.14\%$.
Distractors (B) is the discount rate $1.75\%\times365/91\approx7.02\%$, computed on face value; (D) is the unannualised 91-day return; (A) is a round guess.
Answer (C).

E3. NISM-style A debt fund's portfolio has a modified duration of 5 years. If yields fall by 0.40 percentage points, its NAV will approximately (A) rise by 2% (B) fall by 2% (C) rise by 0.4% (D) rise by 5%.

Compute $\Delta N/N\approx-D_{\text{mod}}\,\Delta y=-5\times(-0.004)=+2\%$ (convexity adds only a few hundredths of a percent).
Answer (A). (B) has the wrong sign; (C) ignores duration; (D) treats duration itself as a percentage change.

E4. Case study The treasury of a hypothetical cooperative society holds ₹2 crore face value of the 10-year 7.20% G-sec of Example 3, bought on a coupon date at ₹98.60 (YTM 7.40%; $D_{\text{mod}}\approx7.015$, $\mathcal C\approx63.07$). (a) Find the market value and the PVBP of the holding. (b) Estimate the loss if yields rise by 50 bp, with duration only and with convexity, and compare with the exact loss. (c) If the society holds to maturity and reinvests every coupon at 7.40%, 6.40% or 8.40% (half-yearly), what does it have at maturity and what is its realised annual yield in each case? (d) What do (b) and (c) together say about "interest-rate risk"?

(a) Value $2\times0.9860=$ ₹1.972 crore (₹1,97,20,000). PVBP $=7.015\times1{,}97{,}20{,}000\times0.0001\approx$ ₹13,833 per basis point.
(b) Duration: $-7.015\times0.005\times1{,}97{,}20{,}000\approx-$₹6,91,632. With convexity: $\left(-0.035073+\tfrac12\times63.07\times0.000025\right)\times1{,}97{,}20{,}000\approx-$₹6,76,085. Exact (repricing at 7.90%): about −₹6,76,340. The convexity term brings the estimate within ₹300.
(c) Per ₹100 face, the 20 coupons of ₹3.60 grow at the reinvestment rate to maturity, plus ₹100: at 7.40% the total is ₹203.93, at 6.40% ₹198.73, at 8.40% ₹209.45. Realised yield $y_R$ solves $98.60(1+y_R/2)^{20}=$ total: 7.40%, 7.13% and 7.68%. On ₹2 crore face: about ₹4.08 crore, ₹3.97 crore and ₹4.19 crore.
(d) Rate changes hit a bondholder in two opposite ways: price risk (higher yields lower today's value) and reinvestment risk (lower yields lower what coupons earn). A seller in the short run feels mostly price risk; a holder to maturity feels mostly reinvestment risk. The YTM is a realised return only if coupons are reinvested at the YTM itself. (The two risks balance for a horizon equal to the Macaulay duration — the idea behind immunisation in E5.)

E5. Case study A hypothetical trust must pay ₹1 crore in exactly 7 years. The yield curve is flat at 7% (annual compounding). The trust buys a 2-year zero-coupon bond and a 12-year zero-coupon bond, half of the money in each, so that the asset duration equals 7 years. (a) How much must it invest in total, and what face value of each zero does it buy? (b) Immediately afterwards all yields move to 6%, 8% or 9%. Find the value of the assets and of the liability in each case. (c) Interpret.

Duration Zeros have durations 2 and 12; with weights $\frac12,\frac12$ the portfolio has $D_{\text{mac}}=7$ = the liability's duration (Theorem 2.15(3)).
(a) PV of the liability: $1{,}00{,}00{,}000/1.07^7\approx$ ₹62,27,497. Invest ₹31,13,749 in each: face of the 2-year zero $31{,}13{,}749\times1.07^2\approx$ ₹35,64,931; of the 12-year zero $31{,}13{,}749\times1.07^{12}\approx$ ₹70,12,759.
(b)
new yieldassets (₹)liability (₹)surplus (₹)
6%66,57,90266,50,571+7,331
8%58,41,21758,34,904+6,313
9%54,93,81054,70,342+23,468
(c) Matching durations makes assets and liability move almost together: a change in yields of up to 2 points moves each by lakhs, but the surplus by only thousands. The surplus is positive both ways because the two-bond "barbell" has more convexity than the single 7-year liability. This is immunisation. Its limits: it protects only against parallel moves of a flat curve, and it must be rebalanced as time passes and yields change.

E6. Interview "A perpetual bond never matures. What is its duration at a yield of 8% (annual coupons)? Is it infinite?"

Idea The price is the perpetuity value $P=C/y$ (§1.2). Duration is a sensitivity, not a maturity.
Compute $D_{\text{mod}}=-\frac{P'}{P}=\frac1y=12.5$; $D_{\text{mac}}=(1+y)D_{\text{mod}}=\frac{1+y}{y}=13.5$ years.
Why finite The present-value weights of far-off coupons shrink geometrically, so the weighted average time converges (it is $\sum_k k\,v^k/\sum_k v^k=\frac{1}{1-v}$ with $v=\frac{1}{1.08}$, which is again 13.5). A good answer adds: a 1-point rise in yield cuts the price by about 12.5% (more exactly, from $C/0.08$ to $C/0.09$, by 11.1%) — the convexity again.
🧑‍🏫 For the teacher

Suggested plan — 2 lectures × 60 min.

  1. Prices and yields. Open with the 📍 situation; ask the class whether a bond quoted at ₹98.60 with a 7.20% coupon earns more or less than 7.20%. T-bills (Definition 2.14, Theorem 2.12, Example 1) as the simplest bond. Bond price as annuity + zero (Theorem 2.13) with Example 2 on the board; stress the premium/discount rule. Then YTM as an equation to be solved (Theorem 2.14, the 🧭 strategy, Example 3), connecting to XIRR in §1.3. Close with accrued interest (Example 4). Project Interactive 1 with the convexity option off, sweep the yield and ask why the curve falls and bends.
  2. Rate risk. Duration as derivative and as average time (Theorem 2.15, Example 5); convexity as the second derivative (Theorem 2.16, Example 6) — switch Interactive 1 to "duration + convexity". Debt funds (Theorem 2.17, Example 7) and the 📈 Market case; project Interactive 2 with the "sudden +1%" scenario, then a random path. End with credit spreads (Theorem 2.18, Example 8) and Practice corner E4.

Class activity. Give each group a hypothetical bond (different coupons and maturities). Each group computes its price at 7% and 8% yields, then its duration. Line the groups up by duration and check that the price changes line up in the same order — then ask which bond a one-year saver and a twenty-year pension fund should prefer.

Misconceptions to watch for. "Coupon = return"; "G-secs can't lose value"; prices and yields moving together; current yield as YTM; duration as maturity; the discount rate of a T-bill as its yield; "higher YTM fund is better"; forgetting accrued interest.

Suggested homework. I(1, 3, 5, 7), II(1, 3, 5, 7), III(1, 3, 5, 6), IV(1, 3, 4, 6); Practice corner E2, E3, E5.

🛠️ Do it yourselfSpreadsheet mini-project · about 60 minutes

Project. Build a "bond calculator" sheet that prices any bond from its cash flows, solves for its YTM, computes its durations and convexity from first principles, and checks them against the spreadsheet's built-in bond functions and against a real repricing.

Why it matters. Every debt-fund factsheet and every bond dealer's screen shows YTM, duration and PVBP. Building them yourself from a column of cash flows shows exactly what those numbers mean — and when the duration shortcut fails.

What to do.

  1. Inputs: coupon rate (B1), years to maturity (B2), coupons per year (B3), yield (B4), face value 100 (B5).
  2. In A10 down, the period numbers $k=1,\dots,n$ (=B2*B3 rows). Column B: time =A10/$B$3; C: cash flow =$B$1*$B$5/$B$3+IF(A10=$B$2*$B$3,$B$5,0); D: discount factor =(1+$B$4/$B$3)^(-A10); E: PV =C10*D10.
  3. Price =SUM(E:E). Weights in F: =E10/price; Macaulay duration =SUMPRODUCT(B10:B..,F10:F..); modified duration =MacD/(1+B4/B3); convexity =SUMPRODUCT(A10:A..*(A10:A..+1),E10:E..)/(price*B3^2*(1+B4/B3)^2).
  4. Check against built-ins (dates for a coupon-date settlement): =PRICE(settle,maturity,B1,B4,100,B3), =DURATION(...), =MDURATION(...); and the YTM of a quoted price with =YIELD(settle,maturity,B1,price,100,B3) or =RATE(n,C,-price,100)*B3.
  5. Shock table: for $\Delta y=-2\%,-1.5\%,\dots,+2\%$ reprice exactly, and compare with =-ModD*dy and =-ModD*dy+0.5*Conv*dy^2. Chart the three columns against $\Delta y$.
  6. Download the latest monthly factsheet of any two debt funds of different categories from their AMC websites. Record the YTM, Macaulay and modified duration and expense ratio, and use Theorem 2.17 to tabulate each fund's approximate one-year return for yield changes of −1, 0, +1 point.

Python version: k=np.arange(1,n+1); cf=np.full(n,c*100/m); cf[-1]+=100; v=(1+y/m)**-k; P=(cf*v).sum(); D=(k/m*cf*v).sum()/P; MD=D/(1+y/m); C=(k*(k+1)*cf*v).sum()/(P*m**2*(1+y/m)**2); ytm=scipy.optimize.brentq(lambda r: price(r)-Pm, 1e-6, 1).

Hand in: the shock chart for a 10-year and a 30-year bond; the largest yield move for which the duration-only estimate is within 0.5 percentage point of the exact change for each; and your table for the two funds with one paragraph on which suits a one-year goal and why.

Going further. Price a bond between coupon dates (fractional first period and accrued interest), and verify that =PRICE returns the clean price. Build a two-bond barbell whose duration matches a 7-year liability, as in E5, and plot surplus against the new yield.

Instructor note: for a 10-year 7.2% bond at 7.4% the duration-only error stays below 0.5 point for moves up to about ±1.2 points; for a 30-year bond at the same yield (modified duration ≈ 12.0, convexity ≈ 237) it fails beyond about ±0.6 point, because convexity grows roughly like duration squared. Students often forget to divide convexity by $m^2$ when $y$ is an annual rate — check that their 10-year value is near 63, not 250.

🧪 Self-check quiz

0/5 correct

Q1. A bond is quoted at ₹104 per ₹100 of face value. Which must be true?

Q2. A bond's modified duration is 6. If its yield rises from 7.0% to 8.0%, its price changes by approximately

Q3. The Macaulay duration of an 8-year zero-coupon bond is

Q4. Which hypothetical fund's NAV is likely to react most to a 0.5-point change in interest rates?

Q5. A G-sec is bought between two coupon dates. The buyer pays