Do RMs need to become more risk management focussed?

IN-DEPTH: Interview with Dr. Henrik Imhof, Head of Yield Management and Pricing, Sixt Rent a Car

Published: 07 Oct 2010

IN-DEPTH: Interview with Dr. Henrik Imhof, Head of Yield Management and Pricing, Sixt Rent a Car

It is being said that Revenue Managers today need to be able to assign “value” to their sales channels based upon profitability, not just top line numbers.

There is also the need for RMs to become more risk management focussed – commercial value of contracts now need to be assigned risk grades.

Forecasting activity needs to account for these risk grades as does long term business strategy.

Agreeing with the same, Dr. Henrik Imhof, Head of Yield Management and Pricing, Sixt Rent a Car, says most companies do take risk into account when setting up their long-term capacity planning.

“Similar to a return vs. risk assessment in finance, one can look for efficient strategies and evaluate them against one's risk aversity and long term strategy. The same holds true for the evaluation of contracts that bind large chunks of capacity. In general, risk considerations must be applied to decisions that have a major influence on the company as a whole,” says Dr. Imhof, who is scheduled to speak at the forthcoming Revenue Management and Pricing in Travel 2010 conference, scheduled to take place in Amsterdam (23-24 November).

Dr. Imhof added, “Whether risk management should also be applied on the micro-level, like daily price decisions, is currently an interesting topic in academic research but, to my knowledge, not in the development of RMS. Most RM practitioners use some kind of risk management in their daily work, though: In markets with long booking horizons, pricing decisions are often made not just by applying the expected future demand but by factoring in the risk of that demand being lower than usual. As long as this is a deliberate action minimising profit risk, there is nothing wrong with that approach. It is getting dangerous, though, when individual analysts have a strong incentive on high load factors. Then, the risk consideration of the individual differs too much from that of the company and leads to unwanted yield decline.”

Dr. Imhof spoke to EyeforTravel’s Ritesh Gupta about several issues in detail. Excerpts:

Good forecasts don’t always mean increases in revenue. It’s the various decisions that can be derived from the forecasts that count. Firstly, how do you assess this viewpoint?

Absolutely. Especially in industries with a high degree of commoditisation, demand is much more the result of the action of a number of players than the realisation of a well-predictable stochastic process. In car rental, due to transparency of Internet prices and low customer loyalty in general, we must therefore be careful to interpret the meaning of a forecast in the context of different markets.

How do you think the whole approach towards forecasting has evolved?

There have been major technological advances not only in hardware but also in applied methods.

First of all, classical time series forecasts are more and more linked to external variables such as competitor prices. Also, the price elastic nature of demand is modelled in different ways. So-called hybrid forecasts try to split customers into varying degrees of price-dependent behaviour. The main challenge, here, is to control for the size of the underlying demand. I think we are only seeing the beginning of a development of systems with that capability.

The modern RM is no longer the record keeper of the past and instead is far more reactive to market conditions, in tune with sales plans and the RM professionals are being described as decisive forward thinkers who are innovative and creative. How do you think this all is reflecting the approach of RM professionals today as far as forecasting is concerned?

In fact, there is a long tradition of using market information to improve forecast quality. For example, GDS market shares of airlines can reliably be related to schedule quality. Many companies are beginning to realise that the market expertise of the RM department should play a major role in the development of the firm's strategy. After all, RM is, in effect, profit management, and thus will naturally help to find a good balance between the goals of growing market share on the one hand, and growing profit, on the other. Not always do these goals conflict with each other: your RM team might, for example, be involved in the development of new, highly profitable, products.

Automation gives an option to explore new avenues in RM and channel management. Working by exception allows to entrust the RMS with the majority of pricing decisions, leaving the team to concentrate on improving forecast accuracy and dealing with anomalies outside of the RMS remit. How do you assess the situation?

RM systems usually offer the possibility to manipulate forecasts so as to account for market anomalies, and analysts make wide use of it. This used to work pretty well in markets with strong tariff fences, low competition, and predictable seasonal influences. In commoditised, competitive markets, like car rental, though, the situation is different. The underlying market demand, which is the central topic of forecasting, becomes more and more a hidden variable as prices of all players in all channels determine the measurable demand. More effort must be put into understanding past data from the market conditions, and draw the appropriate conclusions channel by channel. This can, in my opinion, best be achieved by:

a) Decentralising market assessment and decision making, and

b) Testing markets systematically, as opposed to relying on historical data alone.

To support this, the RMS must model separately the shared aspects, such as inventory and bid prices, on the one hand, and the individual channel aspects, such as price elasticity, competition, and demand anomalies, on the other.

What according to you is key to forecasting passenger demand in a competitive environment? Which factors need to be taken into consideration?

As there usually is no reliable forecast of competitor prices, the options of applying market information directly to forecasting are very limited. This is not saying that RM should not bother about competition and prices. Only, it is not as simple as reacting to a predicted environment. Much rather, one must think in scenarios and strategies.

For example, when you dominate a certain market, you might decide to also be the price leader in that market, assessing the market as a whole and not caring so much about competition. In other circumstances you might see the need to defend your leadership against a new entrant, and so on. The role of RM, and forecast systems in particular, will then be to predict the result, and support the execution of the strategic options. The relevance of a competitor for a certain channel and market will certainly be an important factor to this. What is important, in my opinion, is that competition-driven price control is not trusted to a system that operates in auto-pilot mode.

A couple of years back, a specialist told me: Current RM systems still use single forecast models to forecast all demand types for all products for all arrival dates for all periods throughout the booking curve. Rather different types of forecast models need to be used and the forecasts in RM systems should determine for each forecast, based on the historical data and past forecasts, what type of forecast model to use. Forecasting techniques need a radical improvement. How do you assess the situation today?

This is true in principal but there is always a trade-off between granularity and statistical validity. The design of existing systems owes largely to the requirements of computational tractability and maintainability.

For example, most airline forecasts are still based on GDS booking classes, thereby ignoring the exact price of a retail product in a particular class, or the share and value of agreed company rates bookable in the same class. The way out is either by applying very sophisticated statistics, or by explicitly modelling the different types of demand.

Revenue Management and Pricing in Travel 2010 conference

Dr. Henrik Imhof is scheduled to speak at the forthcoming Revenue Management and Pricing in Travel 2010 conference, scheduled to take place in Amsterdam (23-24 November).

For more information, click here:

Or contact

Rosie Akenhead
0044 (0) 207 375 7229


Related Reads

comments powered by Disqus