How frequent are price promotions in food retail?

PP incentivises customers to purchase through reductions below the recommended retail price. In the UK food retail sector, there are predominantly either total price reductions, or volume-based PP that encourage greater quantities to be purchased for the same cost (e.g. buy-one-get-one-free). Data on consumer expenditures (Table 1) show that, in 2017, a third of take-home purchases were made on PP, and products typically considered HFSS (e.g. regular soft drinks) were twice as likely to be bought on promotion in comparison to fruits and vegetables or starchy foods.

Table 1 Share of take-home food and beverage sales volume purchased on price promotion across broad food groups in 2017*

* Data source: Kantar FMCG Panel volume-weighted take-home purchases of foods and non-alcoholic beverages recorded from a nationally representative sample of approximately 30 000 British households annually.

The public health rationale for the DHSC policy proposal follows from this frequent, on-promotion purchasing of unhealthy HFSS products. Even if the policy could be seen as anti-competitive in limiting this frequently used method of competition, regulation might be the only way to proceed as retailers are unlikely to reduce PP unilaterally on voluntary basis.

What evidence exists on price promotions and food buying behaviour?

PP has been studied by researchers in public health, focussing on the nutritional impact of ‘point of sale’ health policies, and marketing, focussing on the sales and revenue impact of PP. The challenge is that existing research is conducted in a retail industry filled with promotions, where high variability in prices boosts purchasing through different consumer behavioural responses.

Do increased sales as a result of price promotions lead to increased consumption?

Considerable effort has gone into identifying how PP increases sales, or the ‘promotion bump’ as often referred to in the marketing literature. Generally, this is attributed to three forms of consumer reactions^{(Reference Hawkes19)}:

1. 1. Consumer switching: purchasing the same quantity but of a different brand. This has little effect on total nutritional consumption.

2. 2. Increased purchasing: promotions causing purchases that otherwise would not have occurred, creating a potential increase in consumption quantity.

3. 3. Stockpiling: increasing purchase quantity to take advantage of a promotion and avoid higher spending on off-promotion purchases in the future. This does not necessarily increase overall consumption, but there is a possibility that it does, notably if it induces a change in consumption habits. When stockpiling is effective, purchases that would otherwise have occurred at a later date are brought forward. This is referred to as ‘purchase acceleration’.

From a health perspective, understanding the relative effects of the last two categories is crucial, particularly whether the ‘additional’ purchases are stockpiled for later use or consumed.

For households, the frequency with which goods are purchased is important: infrequent ‘impulse’ purchases are likely for immediate consumption, but for frequently purchased goods, stockpiling can make the effects less straightforward. Stockpiling creates the opportunity to save the customer money, but it may also lead to unintended consumption. For example, a repeat customer of cola may buy one bottle per week, but with a two-for-one promotion, they might buy two, intending to save money by avoiding future purchases. Once the extra bottle is in the house, it is drunk at a faster rate. If next week the potentially avoided purchase is still made, overall consumption has increased. The increased purchase can, therefore, be decomposed into ‘purchase acceleration’ – a successful use of stockpiling in which future purchases are avoided – and ‘increased consumption’.

Table 2 presents the decomposition of the ‘promotion bump’ into primary demand increases (i.e. increased consumption and purchase acceleration) and secondary demand (i.e. switching brands). It is clear that the ‘promotion bump’ varies a great deal depending on the product: 33–87 % of these increases using the unit sales decomposition approach are increases of category sales, of which 10–56 % are consumption increases (i.e. buying more altogether), and 9–69 % purchase acceleration due to stockpiling. The key evidence, however, comes from two counterfactual analyses^{(Reference Ailawadi and Neslin24,Reference Baohong25)} that are most appropriate from a methodological point of view. These studies conclude that consumption increases of 12–43 % occur as a result of promotions.

Table 2 Product-level sales increases associated with price promotions: decomposition into primary (purchase acceleration and increased consumption) and secondary effects*

Combined (primary) values are the sum of increased consumption and purchase acceleration where they are separately reported in bold.

* With the exception of Nijs et al.^{(Reference Nijs, Dekimpe and Steenkamp26)} and Teunter (2002)^{(Reference Teunter40)}, which were conducted in the Netherlands, all studies used US consumer scanner data.

The product range studied is clearly restricted, which makes generalisation of these estimates difficult, although Nijs et al.^{(Reference Nijs, Dekimpe and Steenkamp26)}, who used a large range (n 560) of products, found (without a decomposition analysis) that promotions lead to an increase in primary demand for more than half (58 %) of these products. Importantly, these results are not restricted to volume-based promotions, but include simple price reductions as well. At present the DHSC’s proposal mentions volume-based promotion only which is a small part of PP as a marketing strategy.

While increases in primary demand due to promotions appear prominent, we must question whether increased purchasing necessarily leads to increased consumption – which is what leads to detrimental effects on public health^{(Reference Hawkes19)}. There is some evidence in behavioural and economics research that actual consumption rates can be affected by stockpiled food (or inventory)^{(Reference Wansink and Deshpande27)}. This is through a number of mechanisms, including uncertainty about future prices^{(Reference Assunção and Meyer28–Reference Soman and John30)}, scarcity – concerns of running out before the next shop would reduce consumption rates^{(Reference Folkes and Wheat31,Reference Wansink32)} , increased storage costs – stockpiling leads to crowded kitchens and pantries, increasing holding costs and the desire to consume^{(Reference Chandon and Wansink33)}, replacement costs – when prices fluctuate, stockpiled goods are replaced only when on promotion^{(Reference Chandon and Wansink33)} and convenience – the presence of food in the kitchen, in the fridge or on counter tops^{(Reference Wansink and Deshpande27,Reference Wansink, Brasel and Amjad34,Reference Raghubir and Krishna35)} .

What is the evidence relating to a restrictive policy on price promotions?

This evidence, together with Public Health England’s estimated ‘effect’ from PP of up to 22 % increase in purchases, presents a rationale for intervening to reduce PP on unhealthy foods^{(Reference Chisholm, Moro and Bertram17)}. However, these methods still do not answer the question at hand: ‘what if PP on unhealthy foods was restricted or banned altogether?’ Without a direct evidence, it is difficult, ex ante, to quantify the potential benefits, as well as identify the potential risks from unknown consumer and retailer response, because:

• Existing evidence largely ignores a crucial aspect of PP: their efficacy relies on their repeated use (i.e. consumers may expect PP and factor this into their purchasing decisions). In the current retail markets, PP is frequent and shoppers are likely to stockpile during sales and delay purchases when they are not on^{(Reference Neslin and van Heerde21)}.

• Consumers respond asymmetrically to price changes^{(Reference Cornelsen, Mazzocchi and Smith36)} meaning that the effect of price increase is not necessarily the opposite of the effect of a discount.

• Few studies have looked at what happens to the demand for a product once a promotion is withdrawn (rather than added)^{(Reference Baohong25)}. But removing all promotions on similar products with no promoted substitutes available altogether has never been addressed.

• The effects of this policy depend on the response from retailers who will act to maintain profitability. Will the new pricing strategy be a switch to pre-regulation non-promotional prices? Or a regular low price? It could be that if retailers reduce their regular, everyday prices enough, the policy will have little effect.

There are techniques that allow researchers to deal with these dynamic difficulties. Structural demand estimation^{(Reference Baohong25,Reference Hendel and Nevo37–Reference Liu and Balachander39)} can identify the effect of price expectations on current purchasing and consumption decisions. These are difficult to implement but workable. Without their use, the analysis will overestimate the consumption effects of PP. In simplest terms, this is because there is no incentive to stockpile if shoppers know the price will be the same in a week’s time; people can better plan their purchasing, allowing them to take control of their diets. The extent to which this occurs, as well as retailer response, could be estimated through dynamic structural modelling.

What can we conclude for current policy?

Existing evidence suggests that PP might lead to significant increases in purchases that, in turn, can lead to greater consumption and likely overconsumption, but the evidence is not sufficient to know the extent to which banning or significantly restricting promotions would reduce consumption. This requires more studies to simulate the effects of promotions removal. On the other hand, this is not new in public health policies, especially major government initiatives that are often based on a combination of evidence related to the problem and its solutions (e.g. public indoor smoking ban). It is rare to have a priori direct evidence on policy impact, especially if the scope for an experimental investigation is limited.

Given the seriousness of adult and childhood obesity, it is clear that the usual playbook of individual-focused interventions and policies has not worked. More radical and structural policy initiatives that rely less on consumer agency might, therefore, be exactly what we need, even if the evidence is less-than-perfect. In this instance, the rationale and logic for the policy of restricting PP is clear. The evidence of intended and unintended consequences is of utmost importance and should be carefully monitored and evaluated when a policy is implemented. However, the lack of direct evidence now should not cause a missed opportunity.