Rho And Interest Rate Sensitivity

Every other Greek asks "what happens if the stock or its volatility moves?" Rho asks the odd one out: what happens if the interest rate moves, stock unchanged? It sounds like a footnote, and for years it was treated as one — desks quoted delta, gamma, vega and theta by heart and looked rho up if needed. Then rates moved from near zero to several percent within about two years, and portfolios that never bothered hedging rho found it quietly the biggest line on their P&L.

Rent versus a deposit

Buying a call is a bit like renting the right to buy a house later at today's price, instead of borrowing to buy it outright now. If rates rise, borrowing to buy now gets more expensive, so the rented option — pay only later, and invest your cash at the higher rate meanwhile — becomes relatively more attractive, and its price rises. A put works the other way: rent the right to sell later, and if rates rise, the cash received at expiry is worth less once discounted back, so the put's value falls.

The formula

Rho is the sensitivity of an option's price $V$ to a small change in the risk-free rate $r$:

$$\rho \;=\; \frac{\partial V}{\partial r}.$$

In plain English: if the interest rate moves by a small amount, the option's price moves by rho times that amount. For a European call and put under Black-Scholes,

$$\rho_{\text{call}} = K T e^{-rT} N(d_2), \qquad \rho_{\text{put}} = -K T e^{-rT} N(-d_2),$$

where $K$ is the strike, $T$ is time to expiry in years, $e^{-rT}$ discounts the strike to today, and $N(d_2)$ is the probability the option finishes in the money. In plain English: rho is largest for options far from expiry and deep in the money, and it's quoted per 1.00 (100 percentage points) of rate change, so desks usually scale to "per basis point" by dividing by 10,000.

Worked example 1: computing rho for a one-year call

Take a one-year at-the-money call: stock and strike both at $100, $r = 4\%$, $\sigma = 20\%$. First, $d_1 = \frac{\ln(S/K) + (r+\sigma^2/2)T}{\sigma\sqrt{T}} = \frac{0 + (0.04+0.02)(1)}{0.20} = 0.30$, and $d_2 = d_1 - \sigma\sqrt{T} = 0.30 - 0.20 = 0.10$. Looking up the normal CDF, $N(0.10) = 0.5398$.

$$\rho_{\text{call}} = K T e^{-rT} N(d_2) = 100 \times 1 \times e^{-0.04} \times 0.5398 = 100 \times 0.9608 \times 0.5398 = 51.87.$$

That is rho per 1.00 of rate move (i.e., per 100 percentage points), which is never realistic; scale it to a 1 percentage point move by dividing by 100: $\rho = 0.5187$. So if rates rise from 4% to 5%, the call gains roughly $0.52, holding everything else fixed.

Worked example 2: rho for the matching put, and put-call parity check

$N(-d_2) = 1 - 0.5398 = 0.4602$.

$$\rho_{\text{put}} = -K T e^{-rT} N(-d_2) = -100 \times 1 \times 0.9608 \times 0.4602 = -44.22,$$

scaled per 1 percentage point: $-0.4422$. Check with put-call parity, which says $C - P = S - Ke^{-rT}$, so $\rho_{\text{call}} - \rho_{\text{put}}$ should equal the rate-sensitivity of $-Ke^{-rT}$, which is $KTe^{-rT} = 100 \times 1 \times 0.9608 = 96.08$ (per 1.00 of rate), or $0.9608$ per 1 percentage point. Indeed $0.5187 - (-0.4422) = 0.9609$, matching to rounding — the put and call rhos are tied together by the same no-arbitrage relationship that ties their prices.

What this means in practice

Rho matters most for long-dated options — LEAPS, long-maturity structured notes, anything on a rate-sensitive underlying — and for books large enough that even a small per-option sensitivity adds up. When aggregating Greeks across a portfolio, desks bucket rho by currency and tenor, the same way vega gets bucketed by expiry, since a shift in short rates and a shift in long rates are genuinely different risks.