Price elasticity’s credibility has been stretched!
Friday 06 October 2017Article
But in reality, as soon as we leave the ideal textbook bubble where the theoretical concepts of economics seem to aggregate millions of decisions in one straight line, the concept of price elasticity of demand (PED) becomes somewhat useless. The assumption that all of us react in the same way and with the same magnitude to a certain price change is, most of the time far away from reality – think of the last time you had to find a restaurant that fits everybody’s budget.
Historically, PED was first coined in the 1890s here in London by Alfred Marshall and served as a useful concept to gauge the direction and magnitude of shifts in demand. But even though we are fond of all things Victorian, similar to steam engines, cholera, and dark satanic mills, we have grown out of price elasticity of demand.
What we actually see today in terms of consumer behaviour to price, is a number of realities far removed from a classroom notion of price elasticity:
- Price elasticities vary hugely at different points of the demand curve (ie: at different prices) so asking “what’s our price elasticity?” is almost a meaningless question. We've seen an example where the price elasticity of a product was -1.5 at £1.20 and -4.0 at £1.60
- We’ve seen that consumers don’t behave with a steady linear anyway. What actually happens is that different customer segments have a zone of indifference to pricing and that small changes in price have no effect on purchasing. Each customer group has a given point of inflexion where a reaction to a price change (increase or decrease) is triggered. This creates an interesting opportunity for differential pricing by customer segment
- Race to be the lowest price, e.g. in tobacco and certain drink categories, PED will not explain behaviours as it is simply the lowest priced SKUs in the category to whom volume flows
- Relative price, i.e. the price of one SKU relative to another, means a traditional PED calculation is only looking at one side of the story
- SKU mix prices and compressed price ladders. Just recently, at a conference, a zesty drink brand’s marketing leader quipped that his greatest competitor was in the next cubicle, launching another product. In other words, his biggest concern was cannibalization of his SKUs from others in the same business!
Often there is no easy definition of price, e.g. B2B telco bundles and bolt-ons? Thereis no such thing as a pure “price.”The price of a product reflects many factors, depending on the context. In e-commerce, for example, the price incorporates such things as the minimum order value required for free delivery, the mix of products offered by the manufacturer, and the use of promotions and loyalty incentives (reward points). Sometimes a company undermines its own pricing effectiveness.
HOW does a price change actually effect sales? You need to break down the impact of any given sale into its component parts. When prices go down, for example, how many new customers are acquired? How many trial buyers switch to becoming repeat purchasers? What happens to the frequency of purchases? Are customers likely to spend the money they save elsewhere, perhaps on categories they haven’t bought before? Does a price decrease make them less likely to buy from competitors?
Consumer behaviour has significantly changed in the last five years in multiple categories. New business models enabled by information technology that sell goods and services through new platforms and channels, have been the primary drivers of this change. Take for example Uber, and the criticism that it receives whenever ‘surge pricing’ happens in the case of emergencies, terrorist attacks and major festivals or occasions. According to traditional PED thinking, an increase in price (surge pricing) should lower demand for Uber cars. Apart from extremes, this never happens. People accept the surge pricing and order cars. This leads to higher demand and even higher surge pricing. So, there are two things about Uber’s value equation one needs to consider. First, it means that elasticity is low, or even non-existent, when it comes to high demand periods. Second, Uber is capable of dynamically price the available capacity at hand and therefore unlock large portions of the consumer surplus (the difference between the maximum price someone is willing to pay and the actual price) almost like a reverse auction.
Because of that PED curves have stopped being smooth curves, and have started resembling the typical waterfall charts. Consumers have a higher tolerance level for small price increases or when it happens within a range. We can call this as the ‘zone of indifference’. Demand remains relatively stable within a zone, and then drops when the upper limit of that zone is exceeded. Then it remains stable again within the boundaries of the new zone, and so on and so forth.
We have altogether stopped reaching the bottom of PED curves (whether they are smooth or waterfall). Zero demand scenarios never happen, unless and until a brand has been killed or there are significant distribution issues. Going back to the previous point, the tolerance levels for price increases is significantly high in specific categories. Technology, fast fashion, sportswear, eating out and branded hot and cold beverages are a few of them.
Coming back to the price and demand dynamics at play for the recently launched iPhone X and iPhone 8. Priced at close to $1,000, the iPhone X is the more premium of the two, with the iPhone 8 priced at around $699. The delay in the shipping of the more premium X to November 2017 is already hurting demand for the iPhone 8, which was a natural progression from the previous iPhone 7. From a portfolio perspective, the traditional PED curve does not fit in here. Demand (or in this case pre-orders) are not picking up for the lower priced iPhone 8, as aspirational equity (and consequently pre-orders) is skewed towards the iPhone X. Not only for the iPhone, but also for many other technology products that go through a cycle of updates or launches of newer versions, traditional PED models do not hold.
If we look at consumers, we witness similar types of behaviour. An early consumer, who is still in the experimenting and exploration mode, is going to be more price sensitive compared to a long time consumer who keeps on coming back. From an online retailer’s perspective, this means that to keep a relatively newer consumer engaged, there is a need for more incentives, which could be bonus reward points with every purchase or higher level of discounts. For a more loyal consumer, incentives can be less or almost nil as they are now coming back due to the brand’s equity. Growing your consumer base over a long term using brand equity results in a PED curve that is relatively flat (very low sensitivity to price increases), compared to that of a new consumer.
Fast fashion is another interesting category that demands a fresh thinking of PED. Fast fashion essentially democratised luxury fashion from the point of view of giving everyone the chance to buy and wear the latest styles (minus the designer label). Because fast fashion recreates latest runway trends and brings them to the shop floor quickly, there is a very fast turnover of stock in outlets. Higher frequency of stock refresh turnover numbs price sensitivity to a large extent. Demand will be high and stable until the price range is significantly lower than a designer created clothing item of the same style. In the fast fashion business model, high demand is maintained not through pricing strategies, but more by keeping with runway trends across the world. Price is a factor, but for a buyer hankering after the latest from New York runways it is an afterthought.
Our sensitivity to price changes has increasingly become very non-linear and unstructured (akin to decision journeys). Structured and careful decision-making is a behaviour that still falls in line with traditional PED models. But increasingly our decision-making has become fast, erratic and unstructured. The examples of Uber, iPhone 8, iPhone X and fast fashion exemplify that need for price elasticity thinking to incorporate demand stages and not demand as a linear variable.
So how are we able to tie this all together? Part of the answer lies within the technology that enabled customers to make these better informed, faster and more erratic decisions. Leaving the Uber example and moving on to a more common business model, the most interesting question is, ‘What is my customer’s zone of indifference and how can I maximise my share of the consumer surplus?’ In order to achieve this, one need to know two things. First, which customers are alike with respect to their price sensitivity? Second, how big is the ‘zone of indifference’ and what is ultimate drop to the next step?
There is no one truth to achieve this, but given the following three premises, there is one preferred way. First, considering the Pareto principle, stating that 20% of the work creates 80% of the results, we don’t want to overengineer or overcomplicate things. Second, the solution needs to be audible in order to be credible. Third, because of audibility we prefer a rule based approach, rather than a distance based clustering algorithm in combination with a complex discrete choice model. Considering these three premises we are left with Decision Trees and OLS or Logit regression.
Decision trees are probably among the most intuitive algorithms that exist. The overall idea here is, similar to our credo as consultants to be MECE in each leaf. This means that we want to split up the data in a way that we have 100% purity in each leaf and therefore be mutually exclusive and commonly exhaustive. The decision tree will allow to categorize every customer by his/her attributes and assign him/her to the right cluster, with a similar ‘zone of indifference’.
Based on the created clusters, we can use OLS or Logit regression to first, understand the size of the ‘zone of indifference’ and second to predict the likelihood of purchase at a given price within a certain cluster.
We pointed out numerous reasons why there is a certain indifference towards price and Apple customers are considered to be the most indifferent kind. Interestingly, the foundation for this phenomenon was created by a person who most likely never sat in an economics class and also most likely didn’t care about the concept of price elasticity of demand. The iPhone is an icon of the modern age, just like the steam engine was for the Victorians, or indeed new dangled ideas like price elasticity of demand. We talk a lot about Apple pricing, but price elasticity won't help us.
OC&C contributors
- James Walker, Partner at OC&C Strategy Consultants, and Global Head of Analytics
- Johannes Barth, Associate Consultant
