 The
Contribution of Operational Research |
From figure 1,
it can be seen that the contribution of Operational Research
in RM is important in managing the vast number of decision
variables to achieve the right yield. Operational Research
provides a mechanism to model scenarios for a great number
of factors. Operational Research has made a significant contribution
to the development of RM for capacity constrained organisations
in the service sector. As Cross (1997a) comments, Operational
Research is the Rocket Science behind RM.
The airline industry has been at the leading edge in the
development and application of Operational Research applications
for Revenue Management of which British Airways (Anon 1999)
is one of the most successful and largest OR groups applying
and developing RM. A variety of methods have been tried from
the simplest rule-based heuristics to highly sophisticated
mathematical programming techniques involving large multiple
decision variables. In general airline companies which have
adopted sophisticated algorithms to aid RM decision making.
Cross (1997a) states that revenue management has contributed
$500 million annually to American Airlines profits, $100
million annually to Marriott hotels, and in the case of National
Car Rental in the USA, Revenue Management is credited with
turning the company around from near bankruptcy to a profitable
high growth organisation.
Raeside (1997) categorised OR quantitative approaches to
Revenue Management across a spectrum of service industries.
- Mathematical programming - this approach is best for
static problems, it is not best suited to dynamic problems
where continual additional information becomes available
and the system requires updating. Linear programming is
probably limited to the application of allocating resources
to different classes. This could be aircraft seat classes,
hire car categories, or hotel room grades for example.
However, Raeside (1997) notes that in studies into optimisation
of airline booking systems, dynamic programming and network
models have been used (Rothstein 1974, Alstrup 1986, Glover
et al 1982). Williamson and Belobaba (1988) have made use
of greedy algorithms to allow virtual nesting of bookings
in which the value of each fare depends on the overall
ticket revenue.
- Marginal Revenue Approach - Helps the manager to decide
how many units of inventory to sell at discount rates and
how many to reserve for full price customers. The method
may help to solve the problem of decision making in respect
of overbooking. Pfeifer (1989), Brumelle et al (1990),
Bodily and Weatherford (1995) advocate the use a simple
decision rules to maximise profit in marginal revenue approaches.
- Threshold curves - the construction of threshold curves
is based largely upon historical data, past experience
and observation. The curve should show how units of inventory
offered at discount prices are shut out as the date of
use gets closer. This is a relatively simple method, available
to even the smallest company. However, the method is reliant
upon accurate data, intelligent forecasting and continual
updating of the curve. Applying this method (Gu & Caneen
1998) a hotel will use its segmented demand forecast to
determine the optimal mix of guests for a given day. The
hotel (Sheel 1994) would allocate rooms in a preference
order, ranking from the highest rate segment to the lowest
rate segment. The hotel would then take this allocation
to determine threshold curves.
|
