Personalised Pricing. A comprehensive and critical examination of first-degree price discrimination

Table of Contents

1 Introduction

2 Personalised Pricing
2.1 Is it possible?
2.2 What’s the customer’s willingness to pay?
2.3 To increase or decrease the price?
2.4 Is it worth to adopt it?

3 Conclusion

Appendix

List of References

List of Tables

Table 1: Values of a personalised pricing equation.

Table 2: Conceivable variables for price discrimination.

List of Figures

Figure 1: Exploiting customer’s willingness to pay (1)

Figure 2: Exploiting customer’s willingness to pay (2)

Figure 3: Price elasticity and revenue.

1 Introduction

This essay is about pricing, a core area of marketing. More specifically, it is about personalised pricing, which must not be confused with dynamic pricing. Personalised pricing describes adjusting the price for every single customer individually, while dynamic pricing describes adjusting the price for all customers subject to external factors like the current demand as of this moment, for example. If an airline company for instance lifts prices on weekends because demand is stronger on weekends in general, this is dynamic pricing. If the airline company however increases the price only for one particular customer, because they find out, for instance, that the customer uses a certain computer type which makes him likely to be wealthier than other customers, this is personalised pricing. The underlying motivation of this essay is to critically assess how personalised pricing is carried out and whether it should be adopted.

Therefore, this essay takes the following approach and structure. Firstly, it is examined whether personalised pricing is legally permitted. Only if it is legally permitted to personalise prices it is worthwhile to further investigate this topic. Secondly, the customer’s willingness to pay is analysed. In order to personalise prices, it is necessary to know a customer’s exact willingness to pay. Thirdly, the topic of price elasticity is elaborated. It is necessary to assess whether profit is increased via increasing prices or decreasing prices and therefore higher demand. Fourthly, resulting retaliation as a consequence is explored. It is critically examined whether personalised pricing should be adopted, and empirical evidence is gathered to determine a retribution effect of personalised pricing which might end up making this practice unprofitable.

2 Personalised Pricing

2.1 Is it possible?

Before evaluating personalised pricing in-depth it is necessary to answer the questions of whether personalising prices is legally permitted, at all. This is essential, because otherwise the discussion of personalised pricing would be merely theoretical and no real marketing implications for companies could be deduced. This constitutes the necessity of this chapter.

The idea of personalised pricing is equivalent to first-degree price discrimination after Pigou, (2013) and dates back to his book’s first edition of 1920. Second-degree price discrimination can be described as non-linear pricing. Third-degree price discrimination differentiates groups charged with different prices. Evidently, first-degree price discrimination plays the most pivotal role for this essay, however third-degree price discrimination will play a role in chapter 2.3 (To increase or decrease the price?).

Botta and Wiedemann (2020) find, for the European Union, that personalised pricing is so recent as a topic, that no sound regulation is yet in place. Furthermore, personalised pricing is difficult to prove and its abuse has rarely been investigated by regulators, yet. Hence, the practice of personalised pricing would need to be evaluated on a case-by-case basis. What is more, the most recent General Data Protection Regulation (GDPR) of 2016 does not contain specific regulations to consider about personalised pricing (Poort & Zuiderveen Borgesius, 2020). The only noticeable regulation in that context is that the GDPR could indirectly trigger notifications if personalised data of a customer is processed therefor. Current focus of regulatory investigations is rather on predatory pricing (Botta & Wiedemann, 2020).

This essay, however, also investigates both directions of price adjustments, thus potentially the opposite of predatory pricing, namely increasing prices. As for marketing the above findings can have two different implications. On one hand, if a company is conservative, adopting personalised pricing appears not as a good idea because of the ongoing uncertainty about it. On the other hand, if a company is less risk-averse, adopting it early could bear a first-mover advantage.

Another requirement that must be fulfilled to allow and sustain personalised pricing is limiting arbitrage (Ezrachi & Stucke, 2016). This means it must not be possible for customers to buy cheaply and sell it at a higher price to other customers who else would have paid more. To briefly look at it, for example the entire flight industry would fulfil this criterion since tickets are always issued on a specific name. However, further discussion extends the scope of this essay and can thus only be an impulse for further research on this topic.

2.2 What’s the customer’s willingness to pay?

We have found that it is, at least for now, before lawmakers impose restrictions, legally permitted to personalise prices and we expect to increase a company’s profit from it. Now that we know personalised pricing is possible to practice, the next logical question we have to answer is: How should we adjust the price, so that we increase a company’s profit with it?

For instance, Ho, Liang, Weinberg and Yan (2018) take on the uniform pricing puzzle, which argues that movies are of different quality and hence must be price individually. So far, the only observable price discriminations in cinemas are discounts for students, a premium for 3D movies and the like. The idea of this essay, however, is to not only to price differently for different qualities, but to price differently for the very same product or service.

Based on this thought, a theoretical proof is deduced which shows how profit is maximised by adopting personalised pricing. A company should match every single customer’s individual willingness to pay, to maximise its profit. This answers the above outlined question of how to adjust prices and is illustrated by the two following figures.

Abbildung in dieser Leseprobe nicht enthalten

Figure 1: Exploiting customer’s willingness to pay (1)¹

In this fictive example there are twenty customers and willingness to pay of each is known. Calculations can be found in the appendix. The company sells when profit is greater than zero, and price steps (and willingness to pay) are discrete (integer variables). The willingness to pay of customer 4 and 16 are too low, and hence cannot buy. Two boundaries arise. The lowest possible price where the company makes a profit is 5 (e.g. customer 12). The highest possible price of 11 would bear customer 6. Evidently, the optimum lies in between. The optimisation problem shall not be topic of discussion for this essay. The optimal price is 7, leading to a profit of 39.

Now we turn to personalised pricing. In this scenario the company applies personalised pricing. Again customer 4 and 16 do not bear a profit (i.e. profit <0 and =0 respectively). All other customers are priced their exact willingness to pay . This leads to a profit of 61, and thus more than without personalised pricing (61>39).

Abbildung in dieser Leseprobe nicht enthalten

Figure 2: Exploiting customer’s willingness to pay (2)²

If a company can set only one price for all customers, it will be between and , and it will be determined by price elasticity (see chapter 2.3). If the company is capable of pricing customers individually, it should set price , subject to and equal to each customers’ willingness to pay.

Remotely inspired by Ezrachi and Stucke (2016) a model to determine customer’s willingness to pay is deduced. Customers’ willingness to pay can be broken down into single variables that contribute towards the final price. A model is conceivable where there is an ordinary price (e.g. the price a company charges before it adopts personalised pricing) extended by multiple variables that either increase or decrease the price for e.g. customer 1 which would be . The ordinary price must be divided by the amount of all customers due to the mathematical distributive property of the equation (i.e. in order to not automatically increase the discriminatory price with adding new variables). Variations of this approach are imaginable, where every variable is allocated an individual weight . For instance, it could be empirically measured that variable 73 has a much higher overall impact than variable 28. In this simple example every variable receives a uniform weight .

Abbildung in dieser Leseprobe nicht enthalten

Table 1: Values of a personalised pricing equation³

The variables must oscillate around the value of one, with a value smaller than one (but greater than zero) to decrease the price, and a value of greater than one to increase the price. This leads to the domain of . There can be a vast variety of different variables. For instance, variable could take into account the proximity of the closest competitor for customer 1, which could incentivise the company to decrease the price in order to not lose customer 1’s purchase to the competitor in proximity (Ezrachi & Stucke, 2016). The underlying calculation could be: Is there a competitor in proximity of 20 kilometres or less for customer 1? If no, (i.e. no change to price). If yes, the proximity of the company’s closest competitor for customer 1 is taken into account with where z denotes distance in kilometres. The effect must be a decrease to the price as outlined above, hence the value must be between zero and one. If the closest competitor was 18 kilometres away from customer 1, this should lower the price less, as compared to being only 3 kilometres away vs.

In this simple example, and if a competitor within 18 kilometres proximity of customer 1 would lower the price by 10% (=) times its weight 1%, hence by 0.1%. In contrast, a competitor within 3 kilometres proximity of customer 1 would lower the price by 85% times its weight 1%, hence 0.85%. In addition to this extensive example, 99 further factors would have to be computed. Victor, Fekete-Farkas and Lakner (2019) outline further conceivable variables, as depicted by table 2.

Abbildung in dieser Leseprobe nicht enthalten

Table 2: Conceivable variables for price discrimination⁴

However, demographics appear to be not the most pivotal area for personalised pricing. Shiller (2020) conducted a study which revealed that Netflix could generate only 0.8% more profit by personalising prices based on demographics, but 12.2% based on 5,000 web-browsing variables. Acquisti and Varian (2005) list different means to track relevant customer’s data which include cookies, credit card numbers, direct customer login and static IP addresses. In contrast, Ezrachi and Stucke (2016) argue that none of these variables necessarily needs to be observable directly but all can be deduced indirectly. A conceivable means could be A/B testing. While a variable might not be quantifiable a priori, it might well be possible to converge to the variable’s true value by A/B testing random groups of customers. This leads to the next chapter, which elaborates on how much the price should be adjusted.

2.3 To increase or decrease the price?

To answer that question, we have to look at price elasticity, because price elasticity is the measurement which gives insights about how demand is affected by changing the price and vice versa. This essay will not derive price elasticity from its root, which is presupposed for the reader.

[...]

¹ Data fictive, figure self-created.

² Data fictive, figure self-created.

³ Data fictive, table self-created.

⁴ (Victor, Fekete-Farkas & Lakner, 2019) p.142.