Drivers of Price Elasticity

ConversionBijmolt, T. H. A.; Van Heerde, H. J.; Pieters, R. G. M. · 2005Journal of Marketing Research

Topicsprice elasticity·meta-analysis·pricing·promotional pricing·endogeneity·cpg·durables

The annual plan uses one price elasticity assumption to forecast revenue across brands. Before approving it, ask when the underlying data were collected, which price change was measured, and whether the model separated pricing decisions from customer demand. The pooled evidence shows that mismatched choices can materially alter the estimate. Make benchmark quality part of the pricing review, not an invisible analytics decision.

Treat every price elasticity benchmark as conditional, not universal.

Price elasticity, how much sales fall when price rises 1%, varies with the era of the data, the product's life-cycle stage, the time horizon, which price was measured, and how the model was built. A benchmark from one context misleads pricing decisions in another, so validate elasticities before making pricing decisions.

Data chart

Measured price elasticity changes when pricing decisions are accounted for

Price elasticity estimates are markedly larger when models separate customer demand from managers' own pricing reactions.

Action guide

Set a validation standard for every price elasticity benchmark.Require enterprise pricing reviews to check data era, product stage, time horizon, price definition, and model design.
Use recent, category-matched evidence in planning.Older benchmarks understate how strongly sales respond to price, while early-stage products respond more strongly than mature products.
Separate demand from managers' pricing decisions.Require pricing models to account for managers' reactions to customer demand before their estimates inform enterprise plans.
Require price elasticity to be estimated alongside advertising and promotion.Estimates that leave either activity out overstate price elasticity by crediting price with effects that belong to those activities, so a number produced that way should not carry a pricing decision.
Separate promotional and actual-price forecasts.Promotional pricing shows the stronger short-term response, while the actual price paid shows the stronger long-term response; validate the latter because long-term evidence is limited.

Evidence

Across 1,851 measurements, a 1% price increase was associated with a 2.62% decrease in sales, share, or choice; profit was not measured.
The 1988 review averaged a 1.76% decrease in sales, share, or choice per 1% price increase; with every 25 years of newer data, sales fall about 1 percentage point more per 1% price increase, while share and choice show no reliable trend.
Models separating customer demand from managers' pricing reactions estimated a 3.74% decrease in sales, share, or choice per 1% price increase, versus 2.47% otherwise.
Models including advertising estimated a 1.98% sales, share, or choice decrease per 1% price increase, versus 2.83% without it; models including sales promotion estimated 2.22%, versus 3.01% without it.
Price elasticity is larger for products in introduction or growth than in maturity or decline, across groceries and durables.
A 1% promotional price cut was associated with a 3.63% short-run rise in sales, share, or choice, versus 2.36% for a cut in the actual price paid; the long-run rise was 3.17% for promotional price and 3.78% for actual price, based on 98 long-run measurements.

Key takeaway

Price elasticity benchmarks travel poorly; match the evidence to the decision before using it.

Source

Bijmolt, T. H. A., Van Heerde, H. J., & Pieters, R. G. M. (2005). New Empirical Generalizations on the Determinants of Price Elasticity. Journal of Marketing Research.

Evidence strength: Strong (pools 1,851 published business-to-consumer price elasticity measurements from 81 studies using data collected 1956-1999, mostly groceries). Covers brand- and individual-product sales, share, and choice response; does not establish profit, ROI, or payback and is less certain for durables and economies with extreme conditions. Durables account for only 33 of the 1,851 measurements.