Dynamic Pricing (DP) is the practice of varying prices for the same goods over time
or across customer classes/segments in an attempt to increase the total revenues for the
seller. The concept of DP is not new and as the economist, Krugman (2000) mentioned that
DP is merely a new version of the age-old practice of price discrimination. The concept of
DP is not only applicable in different environments, but also commercially feasible. It is
one of the approaches used in Revenue Management (RM) to enhance profit. The RM
practice is most commonly applied when there is a fixed stock of a perishable product, a
product with finite shelf life or a price-sensitive product (Bitran and Caldentey, 2003).
Examples of such products are transportation tickets, seasonal style goods, hotel
bookings, pharmaceutical products with limited shelf life, perishable foods, electronics
products, green vegetables, fruits, poultry products etc. These products can be classified as
time-independent perishable products and time-dependent perishable products.
The time-dependent perishable products have short fixed useful life. However, the
time-independent perishable products are useful to customers or users for a
significant duration, but have very less economic value after short duration. Chatwin
(2000) considered a continuous-time inventory problem in which a retailer sets the price on
a fixed number of perishable assets which must be sold before they perish. The retailer
can dynamically adjust the price between any of a finite number of allowable prices and
the demand for the product is negatively correlated with the price. They extended the
results to (i) the case in which the prices and corresponding demand
intensities depend on the time-to-go; and (ii) the case in which the retailer can restock to meet the demand at
a unit cost after the initial inventory has been sold. Petruzzi and Dada (2002) developed
a dynamic model linking price and found that the nature of demand uncertainty
(i.e., additive or multiplicative) for perishable products plays a significant role
in determining the structure of optimal policy. Raju et al. (2003) investigated the use of reinforcement learning techniques in determining dynamic prices in an electronic
retail market as representative models. They considered a single seller and two seller
market. They formulated the DP problem in a setting that easily generalizes markets with
more than two sellers. Bitran and Caldentey (2003) examined the research results of DP
policies for a perishable and nonrenewable set of resources in a stochastic price-sensitive
demand environment over a finite period of time and studied their relation to RM. Kong
(2004) examined the sellers' strategies for DP in a market for which a seller has a finite
time horizon to sell its inventory. DP strategy was developed by him using neural network
based on online learning called Sales-Directed Neural Network (SDNN) strategy. They
showed that the SDNN strategy exhibits superior performance compared to the other
candidate's DP strategies with similar computational simplicity and lack of assumptions about
the market place. Dasgupta and Hashimoto (2004) addressed the problem of DP in
a competitive online economy where the seller uses a collaborative filtering algorithm
to determine temporal consumer's purchase preferences followed by a DP algorithm
to determine a competitive price for the product.
|
