Skip to main content

Dynamic pricing: Indian Railways cope with the uncertainties of business

The Indian Railway has introduced dynamic pricing model in some of its more well-known trains Photograph: (Others)

WION Delhi, India Sep 16, 2016, 09.50 AM (IST) Harish Rao

Indian Railways recently announced the introduction of dynamic pricing (DP) on trains, such as the Rajdhani Express, Duronto and Shatabdi with the fares increasing by 10 per cent after every 10 per cent berths being sold, but a cap of maximum 50 per cent hike has been placed on the original fare. Across segments of the population, DP has been painted as evil and designed to hurt travelers. Dynamic pricing, most famously introduced by the American Airlines (AA) in 1988, is being scrutinised by the whole nation. AA estimated the benefit to be around $1.4 billion over three years (at 1990s prices) and expected a steady state contribution of over $500 million through the implementation. This has given rise to a completely new area of research and application in Revenue Management (RM). The field finds wide usage and adoption in many industries, including airlines, railways and hotels amongst others.  

Fundamentals of dynamic pricing
Humans love certainty. All of our efforts go into trying to convert uncertain situations into certain ones. And we Indians love prices that are not only certain but also low! But in the real world, certainty is a myth. Businesses operate under huge uncertainties. 

The cost of inputs, the nature of demand and the cost of servicing the demand are all exposed to uncertainties. Any change in any of these parameters severely affects the cost of satisfying the demand. Especially in hospitality and transportation (or logistical) businesses. This is primarily due to the “perishable” nature of the products that they sell. 

Perishability: In this context, we use the term perishable to refer to its shelf life. An airline cannot sell a seat once the flight takes off. A hotel that has an empty room for a day cannot go back in time and sell the room. The revenue foregone due to this perishable nature is detrimental to the ranking of the firm. It would do well to remember that the additional cost incurred by the firm on selling the seat or selling the room at a low price is way less than the additional revenue it would have earned from it. 

So, the logical question comes up. If that’s the case, why increase prices? Just sell it at any price more than the cost of providing the service.

Yes, that is what DP is actually meant to achieve. In times of low demand, the system tries to increase occupancy by selling seats or rooms at lower prices (obviously above the floor price) to reduce the losses. The losses are made up by charging a premium in situations of high demand.

The popular assumption is DP is time-based, that is, if a ticket is bought closer to the departure time, the price of the ticket would necessarily be higher. But true DP models are not. They are purely demand based.

How does dynamic pricing work? 
The primary objective of such a model is to maximise load factors or occupancy factors for the trains while maximising revenue.  Occupancy or load factors are a widely used metric in the hospitality and logistics industry. DP algorithms work to ensure higher load factors via pricing differentiation. This is done by rewarding customers with certain travel plans and extracting a rent from customers with uncertain travel plans. However, this rent extraction is only possible if there is high demand.

DP categorises the resources available into categories or buckets. For ease of understanding, let us consider just three buckets; low fare (deep discount), regular fare and premium fare. All these three buckets are for service at the same class level, providing the same facilities. 

The first bucket sells resources at low fares (40-50 per cent of the actual fare) to achieve high occupancy. But to keep the service viable, a minimum occupancy factor is needed; think of it like the minimum customers that a hotel must have to sustain itself. However, these discounts come with a caveat. These fares are non-refundable because the service provider is selling you at a cheaper fare to ensure occupancy levels. In case you back out, the occupancy level of the service goes down and the firm needs to reduce its losses. 

If you don’t want to take up this offer from the service provider, you can choose to pay the regular fare with the regular refund restrictions. Once the first bucket is full, no more low fares are on offer. Once the second bucket is full, even the regular fares are exhausted. Only premium fares are available. 

Generally, premium fare bucket differs from the other buckets in the sense that there is no fixed fare. For example, given high demand, the 10th person who buys a ticket would be paying more than the 9th person who bought the ticket. The sale of this ticket indicates the existence of heavy demand. I am sure you remember how touts earlier sold train tickets priced originally at Rs. 500 for Rs. 3000? It was informal DP at work. 

The really good DP algorithms have a few additional features:

 Dynamic bucket sizes: Based on historical data and expected demand, buckets sizes are not static. This means that the same fare bucket would have different number of seats available at different times of the year.

Migration between buckets: Based on factors, such as current booking trends compared to historical data, closeness to departure time, seats can shift between buckets. For example, if the time is close to departure and a large number of seats in the premium bucket are unsold, these seats would be shifted towards regular fare bucket or even the low fare bucket. Hence, it is possible to get low fare seats even minutes before departure due to low demand.

Number of buckets: Based on the levels of price discrimination that the service wants to offer, it can have more buckets. This helps to gain better revenues for the service provider.

 Price discrimination within bucket: Seats getting booked quickly within a bucket implies high demand. The same price discrimination system applied in the premium category can be applied within other buckets as well. So the first ticket sold in the low fare bucket would be quite cheap as compared to the last seat in the same bucket.

How is it different in the Indian Railways?
Firstly, there is no additional advantage for a traveler with a certain plan. Regular fares are same as low fares, the traveler gets no discount. The argument given for this is, regular fares are already below par. But these fares are refundable. So from the perspective of the IR, is this helpful in guaranteeing occupancy?  No. 

What’s the flip side? Rationalisation of demand is very much needed to meet supply-demand mismatch. Because refunds are nearly full and overbooking is rampant. Trains with waitlisted tickets sometimes have occupancy as low as 60 per cent. Revoking the Ticket Deposit Receipt, submitted in case of inability to undertake the journey, to obtain a refund has been a logical step in this direction. 

Secondly, the DP model of IR occurs in flat and established discrete steps: 10 per cent increment after 10 per cent seats gets full. This happens irrespective of the rate of demand and is totally contrary to the underlying principles of DP.

Thirdly, there is no true dynamism in the prices. Prices can move only up and not down. Here, they make an assumption of demand being greater than supply at all times. While this may be true for regular fares during normal periods, this may not hold true for premium fares. There is a high chance of seats going unoccupied in such cases, defeating the primary purpose of dynamic pricing. 

There is no migration between regular and premium fares according to demand. A fixed quota of seats is earmarked. There are several possible reasons behind this policy, such as:

Bureaucratic approach: A mindset entrenched in certainty that believes in fixed quotas mean fixed prices. This is what happens when a business is viewed as a government department. This is highly harmful to businesses operating in an uncertain environment. Fixing important parameters leave seats unsold. Bureaucrats don’t answer for unsold seats and unrealised revenue. Additionally, the unsold seats are used to appease the unauthorised “stakeholders”.

 Citizens' approach: We have taken Indian Railways for granted. Anything that departs from the traditional way of operations is taken as taking away some fundamental right. 

What the Indian Railways should do?
The first step is to introduce non-refundable deep discount fares. Announce this move, but do not disclose the number of seats in this bucket. Airlines do not disclose their pricing structure. Even Air India doesn’t.

People from the lower-middle class who travel long distances for vacations (infrequent travel) will be hugely benefited. This is because their travel plans are mostly certain and they can take advantage of the deep discount fares. Adding the no refund clause also reduces bogus bookings by touts. 

It is necessary to introduce partially refundable discount fares. Refund rules have to be stringent than regular fare class, but lenient when compared to deep discount fares. 

IR has to introduce dynamic bucket allocation. They have to adopt the dynamic pricing model in its entirety and not piecemeal. It needs to be introduced across all express trains, in all classes, not for a few months, but for a whole year. 

All quota need to be removed. If needed, coupons or vouchers have to be introduced. Every quota is a leakage that needs to be plugged. 

The first two steps help in covering up occupancy issues and appeasing the masses. The next two steps help in garnering the revenue needed for future investments. 

Yes, of course, we must ask Indian Railways to cut costs and manage itself more efficiently. But that’s a different topic, needing a different article. 

 

 

 

Harish Rao

Harish Rao is assistant professor of International Management Institute, Delhi. He takes interests in operational research applications in transport, project management and supply chain management.

Show Comments
  • delete