Module 24 — Quantitative Finance for CFOs
Value at Risk, Monte Carlo simulation, fixed income mathematics, and the quantitative techniques modern CFOs need to understand, commission, and challenge.
Learning Objectives
- Understand and interpret Value at Risk (VaR) metrics
- Apply Monte Carlo simulation to financial decision-making
- Master fixed income mathematics: pricing, yield, duration, convexity
- Understand options pricing (Black-Scholes intuition) without being a quant
- Use regression and statistical analysis to support CFO decisions
1. Value at Risk (VaR)
What VaR Measures
VaR answers: "What is the maximum loss over a given period at a given confidence level?"
Example: VaR (95%, 1-day) = PKR 50M
Interpretation: There is a 95% probability that we will NOT lose more than PKR 50M in a single day.
Equivalently: In 1 out of 20 trading days, losses could exceed PKR 50M.
Three VaR Methods
| Method | How It Works | Strengths | Weaknesses |
|---|---|---|---|
| Historical | Use actual past return distribution | No distributional assumption | Past may not represent future |
| Parametric (Variance-Covariance) | Assume normal distribution; use σ and ρ | Fast, simple | Underestimates tail risk; normality assumption |
| Monte Carlo | Simulate thousands of scenarios | Flexible; handles non-linear instruments | Computationally intensive |
VaR Limitations — What Every CFO Must Know
- VaR does not predict the size of losses beyond the VaR threshold — it just says they can happen
- Fat tails: Real return distributions have more extreme events than normal distribution predicts
- Correlation breakdown: Asset correlations spike in crises — diversification disappears when you need it most
- VaR is not a risk limit — it is a measurement tool; Expected Shortfall (CVaR) is more conservative
Expected Shortfall (CVaR)
CVaR = Average loss in the worst (1 - confidence level)% of scenarios
CVaR(95%) = Average of all losses exceeding the 95% VaR threshold
CVaR always exceeds VaR and is the preferred metric for risk limits under Basel III.
2. Monte Carlo Simulation in Finance
What Monte Carlo Does
Simulates thousands of random future scenarios using probability distributions for key inputs, producing a full distribution of outcomes rather than a single-point estimate.
Steps in a CFO Monte Carlo Model
- Identify key uncertain variables (FX rate, EBITDA margin, commodity price)
- Assign probability distributions (normal, lognormal, triangular, uniform)
- Specify correlations between variables
- Run N simulations (10,000–100,000)
- Analyze output distribution: mean, percentiles, probability of target
Python Monte Carlo — FX and Revenue Combined
import numpy as np
n = 100_000
rng = np.random.default_rng(42)
# Correlated inputs: revenue growth and USD/PKR rate
cov = [[0.0064, 0.0048], # revenue growth var, covariance
[0.0048, 0.0144]] # covariance, FX rate var
rev_growth, fx_rate_change = rng.multivariate_normal(
[0.10, 0.05], cov, n).T # 10% mean growth, 5% mean FX move
base_revenue_pkr = 5_000 # PKR millions
usd_payables_ratio = 0.40 # 40% of costs are USD
revenue = base_revenue_pkr * (1 + rev_growth)
usd_cost_increase = base_revenue_pkr * 0.60 * usd_payables_ratio * fx_rate_change
ebitda_margin = 0.25 - usd_cost_increase / revenue
ebitda = revenue * ebitda_margin
print(f"Mean EBITDA: PKR {ebitda.mean():.0f}M")
print(f"5th pct: PKR {np.percentile(ebitda, 5):.0f}M")
print(f"95th pct: PKR {np.percentile(ebitda, 95):.0f}M")
print(f"Prob EBITDA < 1000M: {(ebitda < 1000).mean():.1%}")
3. Fixed Income Mathematics
Bond Pricing — Discounted Cash Flow
Bond Price = Σ [Coupon / (1 + y)^t] + [Face Value / (1 + y)^n]
Where: y = yield to maturity, t = period, n = maturity
Relationship between price and yield:
- When yield rises → bond price falls (inverse relationship)
- A bond priced at par: coupon rate = yield to maturity
- Premium bond: coupon rate > yield
- Discount bond: coupon rate < yield
Duration — Measuring Interest Rate Sensitivity
Duration is the weighted average time to receive the bond's cash flows. It measures the sensitivity of bond price to yield changes.
Modified Duration = Macaulay Duration / (1 + y)
Price change % ≈ −Modified Duration × Δy
Example: 5-year bond, modified duration = 4.2
- Yield rises by 1% (100bps) → price falls ~4.2%
- For PKR 1bn portfolio: price change = −PKR 42M
Convexity — Second-Order Effect
For large yield changes, duration alone underestimates the actual price change. Convexity captures the curvature of the price-yield relationship:
Price change = −Duration × Δy + ½ × Convexity × (Δy)²
Higher convexity = better risk profile (bond falls less on yield rises, gains more on yield falls).
Yield Curve and Its Information Content
| Yield Curve Shape | What It Signals |
|---|---|
| Normal (upward sloping) | Healthy growth expectations, higher long-term rates |
| Inverted (downward sloping) | Recession expected; long-term rates fall below short-term |
| Flat | Uncertainty; transition between environments |
| Humped | Peak rates expected at medium horizon |
Pakistan yield curve: historically inverted during monetary tightening cycles (2022–2023 at peak policy rate).
4. Options Pricing — Intuition Without the Math
Black-Scholes Model — Key Variables
| Variable | Symbol | Effect on Call Price |
|---|---|---|
| Underlying asset price | S | Higher S → higher call value |
| Strike price | K | Higher K → lower call value |
| Time to expiry | T | Longer T → higher value (more time for price to move) |
| Volatility | σ | Higher σ → higher value (more chance of large moves) |
| Risk-free rate | r | Higher r → higher call (cost of carry) |
| Dividends | q | Higher dividends → lower call (stock drops ex-dividend) |
The Greeks — Risk Sensitivities of Options
| Greek | Measures | CFO Implication |
|---|---|---|
| Delta | Sensitivity to underlying price | How much the hedge moves with the FX rate |
| Gamma | Rate of change of delta | How quickly hedge ratio needs adjustment |
| Theta | Time decay | Options lose value as expiry approaches — cost of waiting |
| Vega | Sensitivity to volatility | Higher PKR volatility = more expensive options |
Real Options in Corporate Finance
Real options apply option pricing logic to strategic decisions:
- Option to expand: Value of the ability to scale up if a new market succeeds
- Option to abandon: Value of the ability to exit if conditions deteriorate
- Option to delay: Value of waiting before committing irreversible capital
Real option value is missed by standard DCF. CFOs in high-uncertainty environments (new markets, R&D investments) should include real option value in capital allocation decisions.
5. Statistical Analysis for CFOs
Regression Analysis — Use Cases
Linear regression explains the relationship between a dependent variable (e.g., revenue) and one or more independent variables (e.g., GDP growth, inflation, commodity price):
Revenue = α + β₁(GDP growth) + β₂(FX rate) + ε
- β₁ = 1.5 means: 1% GDP growth → 1.5% revenue growth
- R² = 0.85 means: 85% of revenue variance explained by the model
Time Series Analysis for Forecasting
- Moving averages: Smooth out seasonal fluctuations in sales data
- Exponential smoothing: Weighted moving average that places more weight on recent observations
- ARIMA models: Autoregressive integrated moving average — captures trend, seasonality, and autocorrelation
- Practical use: Revenue forecasting, commodity price forecasting, working capital planning
Correlation and the Limits of Diversification
Portfolio Variance = w₁²σ₁² + w₂²σ₂² + 2w₁w₂ρ₁₂σ₁σ₂
Where ρ₁₂ = correlation between assets. If ρ = +1: no diversification. If ρ = −1: perfect diversification.
CFO application: Correlations between business units' cash flows determine optimal capital allocation and cross-guarantee structures.
Self-Assessment
-
Your treasury portfolio of PKR 2bn has a 1-day VaR (95%) of PKR 15M. Explain to the board what this means, what it does NOT mean, and why the board should also look at Expected Shortfall.
-
A 10-year PKR bond with 15% annual coupon is priced at par. Market yields then rise to 18%. Using duration (estimate: 6.5 years), calculate the approximate new price of the bond. How does convexity affect the actual price change?
-
You are evaluating a new market entry with an NPV of −PKR 20M in the base case. However, if the market succeeds (40% probability), there is an option to double capacity with an NPV of +PKR 150M. How does real option analysis change the investment decision?