Diminishing Returns to Promotion Depth in Grocery: Evidence from Dominick’s Cereal (1989–1997)

Read the full article See related articles

Discuss this preprint

Start a discussion What are Sciety discussions?

Listed in

This article is not in any list yet, why not save it to one of your lists.
Log in to save this article

Abstract

This study examines how the depth of temporary price cuts is related to weekly unit sales in the ready-to-eat cereal category of the Dominick’s Finer Foods scanner data for 1989–1997. Promotion depth is modeled as a dose–response treatment in a store × UPC × week panel. The baseline specification regresses ln(1+MOVE) on discount-depth bins while absorbing store × UPC, store × week, and UPC × week fixed effects, with two-way clustered standard errors by store and UPC. The cleaned and trimmed panel contains 4.64 million observations, and the preferred estimating sample uses the top 100 UPCs ranked by cumulative sales. Deeper discounts are consistently associated with larger same-week lift. Relative to the 0–5% bin, the preferred top-100 estimates imply approximately 4.7% higher weekly sales for 5–10% discounts, 10.4% for 10–20% discounts, and 28.0% for discounts above 20%. Additional episode evidence indicates that most promotions are short, although a long right tail motivates treating very long runs as price regimes. A corrected event study around promotion starts shows a large contemporaneous spike but a significantly positive near lead, so those dynamics are best read as descriptive. By contrast, a promotion-end event study documents a persistent post-promotion dip of roughly 2–5% for at least eight weeks after promotions end, consistent with inventory drawdown. The evidence therefore supports strong contemporaneous lift and dynamic payback, although the coarse-bin static design does not, by itself, establish concavity in the depth–sales relationship.

Article activity feed