2. Profiting from Uncertainty
Financial markets recognize uncertainty and capitalize on expected variances
“Risk” justifies higher expected returns in equities and fixed income securities
Investment counselors encourage diversification to reduce overall portfolio risk
Exotic derivatives (puts, calls, collars, floors) are designed specifically to manage, or exploit, risk
Risk is managed more crudely in “real” markets and daily business decisions
Investment decisions, where the financial markets are best represented, typically have 1 hurdle rate based on the whole firm’s
risk profile (D/E ratio, cost of capital); risk differences among projects are often assessed more subjectively
Marketing, pricing, product, and distribution decisions, typically, do not include a sophisticated risk metric
At airlines, however, risk is incorporated explicitly in many commercial and operational decisions
Optimal aircraft sizes and fleet structures incorporate the expected variance in demand across days-of-week and seasons
Pricing – or more particularly revenue management – is based on both expected demand of each fare level on each flight and
on the uncertainty of the forecast for that specific fare level
There is an opportunity for more industries to take a more disciplined approach to Risk
3. Risk-based Decision Making
Airlines have incorporated risk into key commercial/operational decisions
Sophisticated models using Big Data Analytics and linear programming optimization
Exploiting Risk for increased profits is well developed in:
• Supply / Capacity Planning:
• Aircraft and scheduling decisions factor both average daily demand and projected variance in demand
• Pricing:
• Pricing is dynamic, based on forecast demand by flight by fare level
• Forecast uncertainty for each fare level within each flight is explicitly built into optimization models
• Lost Sales:
• Overbooking recognizes high variance in no-shows across flights and days
• Purchasing:
• Sophisticated fuel hedging is designed around specific goals and operating models
4. Capacity Planning: Fleet Decisions
The optimal size of an aircraft on a route cannot solely be based on “average” demand.
For example, average demand of 100 can come from multiple alternative profiles, ranging from a
constant, predictable 100 passengers to random fluctuations between zero and 200+
The aircraft sizing algorithm assesses the Demand Distribution (passengers & fares) and
compares to the operating cost of different aircraft types
Expected marginal revenue versus marginal cost for each incremental seat
The profit maximizing aircraft rarely accommodates average demand
In fact, due to high volatility in demand, 100 passengers on “average” may justify a 110 seat aircraft but this aircraft may only
capture 80 “observed” passengers (a 72% “observed” load factor) with 20 passengers on average unaccommodated
Every flight has its own unique distribution of demand.
High variance, with higher fares, will justify larger aircraft to meet more peak demand.
Lower variance will allow more “perfect” aircraft sizing and higher observed load factors
Optimization models predict “observed” loads and “spilled” (unmet) demand for each flight
5. Pricing Decisions
A typical flight may have 100+ different fares
Airline objective is for each passenger to pay his Maximum Willingness to Pay
Discrete market segments are defined based on different elasticities and behavioral proxies
Lower fares are only offered when demand for higher fares is forecast to be low
Inventory allocations for each fare incorporate demand variance
High variance combined with high upsell opportunity justifies setting aside more inventory for the higher fare demand
Variance is measured, and applied in linear program optimization techniques, for each fare level on each flight
Forecasts and optimized allocation recommendations are updated each night – over a million
distinct forecasts for a 50 aircraft fleet
Such revenue management adds 5-7% revenue to the airline
More discrete pricing based on distributions of demand at different price points – or
between market segments - can improve revenue results in a multitude of industries
6. Lost Sales: Overbooking Decisions
In addition to the demand forecast, additional uncertainty for airlines occurs with no-shows
Passengers who change their travel plans or discard non-refundable tickets
No-shows can be 10-20% of bookings on certain flights on certain days
Without overbooking, no shows result in empty seats and foregone revenue
However, how many or which passengers will no show on a given flight isn’t known
Airlines measure the average and variance of no-show behavior by flight
The expected revenue from selling an extra seat is compared against the probability and cost of oversales
Efficient management of oversales can drive very high overbooking rates when the variance is high
Statistically-based overbooking can add 2% to total airline revenue
7. Other Risk-related Decisions
Fuel hedging became common beginning in 2008 when Southwest earned more from its
hedges than from operating its fleet
Fuel hedging, like financial options, is highly efficient and offers exotic alternatives including collars,
floors, etc. …for a price
However, fuel hedging needs to be tied into operations to avoid being merely speculative
The consolidated industry is now better positioned to pass on higher fuel prices to customers
American Airlines believes any fuel hedging is speculative
Different fleet strategies drive differences in exposure to fluctuating fuel prices
Allegiant airlines operates older aircraft which it grounds when fuel prices are higher or demand lower
Delta manages a fleet of both old and new aircraft, allowing it, too, to manage capacity in response to market changes
Overall risk (operational and financial) needs to be measured and included to meet overall corporate risk/return objectives
Fuel hedges are designed to reduce volatility; not to “make money”
8. Risk-oriented Culture
A firm that relies on heavy statistical modeling and that includes calculated risk in
commercial and operating decisions must adopt different organizational processes
Big Data-based statistical models require special skills and oversight
The “Wall Street” trader mentality within a commercial organization
Specific features of a successful organization built around “Risk” include:
Recruiting and training of skilled analysts
Model transparency and ease-of-use
Checks and balances on analyst decisions and model interventions
Standard metrics for both model and analyst performance; accountability
Within a commercial organization, the Risk group cannot act as a silo
Cross-functional collaboration insures model inputs & outputs (decisions) are “real world”
Learning between the “quants” and the operators is continuous
9. Risk-oriented Decision Process
Point
Forecast
Plan / Act
Forecast
Average
Plan / Act
Type of
Distribution
Economic Assessment of all Outcomes
Overforecast vs. Underforecast
Optimization
Forecast
Volatility
Identification
of Outliers
Goals
Scenario Planning
Traditional
Forecast Process
Risk-oriented Forecast and Decision Process
Operational
constraints/
flexibility
10. Does your Firm Exploit Risk?
Does your firm incorporate forecast uncertainty into its operational decision making?
Analysis of, and metrics for, demand volatility
Economic assessment of under- vs. over-forecasting
Mathematical optimization across distribution of potential outcomes
Capitalizing on high pay-off outcomes, even when they aren’t the most probable
Do you prepare forecasts, along with associated uncertainty, at a sufficiently granular level?
Discrete segmentation of customer market segments based on differing elasticities
Millions of SKU’s including, for services, date- and time-of-day-specific demand
Updated algorithms, coefficients, and the forecasts themselves with real-time orders
Are Operations and Analytics aligned around risk-taking?
11. Can you further exploit Risk?
Tom Bacon has applied Big Data Analytics and Risk Management to support commercial
decision-making at numerous airlines
Led Capacity Planning, Pricing, and other commercial functions as an executive at 5 carriers
Restructured carriers in changing markets or facing new competition
All sectors: global legacy carriers, LCC’s, regionals, and niche carriers
Global: North America, Asia, Middle East, Europe
Assessed analytical systems & modules; achieved track record of success in exploiting risk
Oversaw delivery and deployment of over 150 regional jets, transforming a turboprop carrier into a >$1 B airline
Developed new processes for Pricing Analytics for a bankrupt airline during world recession
Integrated analytical systems for merger of two major airlines
Launched travel start-up designed to manage customers’ risk of fare increases
Persistent advocate for cross-functional collaboration between Analytics and Operations
Thought leader in Big Data Analytics; regular contributor to travel publication and speaker at
industry events
To implement commercial and operational risk management in your organization please contact
Tom at tom.bacon@yahoo.com