Case of the Day: Counterintuitive Price Behavior of Seasonal Goods
The basic theory of supply and demand in competitive markets tells us that short-run increases in demand should lead to increases in price (if the supply curve slopes upward) or no change in price (if supply is horizontal). An increase in demand should only lower price in the competitive model under the (rather unlikely) condition that the short-run supply curve slopes downward. (Note: We will discuss some reasons why supply curves might slope downward later on, but these seem plausible only in the long run, not the short run.)
In most cases, we do see prices rising as demand increases. Yet retail businesses sometimes put seasonal items on sale, effectively lowering their prices, at precisely the time when demand is highest. We see "back-to-school" sales on clothing and supplies, bargains on turkeys before the Thanksgiving feast at which they are traditionally served, and discount offers on all kinds of attractive gift items during the shopping season prior to the Christmas holiday.
In a paper published in the March 2003 American Economic Review, Judith Chevalier, Anil Kashyap, and Peter Rossi (CKR) examined this behavior using data from a large Chicago supermarket chain over several years. They found that there were indeed negative correlations between price and seasonal demand fluctuations for some items, and that the fluctuations that occurred were due to changes in the store's markup rather than in the wholesale cost of the goods.
CKR chose seven categories of products and identified likely periods of peak demand for each. Beer was expected to have high demand when the weather was hot and around holidays. Oatmeal should sell well when it is cold, and tuna during Lent (when devout Catholics eat no meat on Fridays). Snack crackers and cheese were expected to be popular at Thanksgiving and Christmas holidays. They examined two different kinds of soups: eating soups and cooking soups, the latter being mostly broths used in recipes. The eating soups were expected to be popular during cold weather, but the cooking soups would peak around holiday feasts. As a control, they also examined a set of goods that they did not think would have a seasonal pattern: analgesics, cookies, crackers, and dish soap. For each category of seasonal or non-seasonal goods, the CKR examined scanner data over seven years to select the 6-9 highest-selling UPC codes, thus focusing on the largest selling brands, sizes, and varieties.
CKR found substantial evidence that retail prices for many of these seasonal goods fell during peak periods, in contrast to the prediction of simple market theory. When they looked at the corresponding wholesale prices, there are peak-season price reductions for only a few goods. Thus, the retail store must be reducing its markup on goods during times of peak demand.
The authors proposed and tested three models that would explain this. The first model is one in which demand is more sensitive to price during high-demand periods. We will see in a few weeks that firms with some monopoly power charge lower prices when consumer demand is more price-sensitive, so this could explain why retailers behave more competitively during high-demand periods.
The second model is based on collusion among sellers. In this model, under "normal" demand conditions, sellers make tacit agreements not to compete agressively on price. However, when demand is temporarily high, each firm sees greater potential gains from cheating on its rivals and cutting its price, leading to lower prices during the peak-demand period.
The final model is called the "loss-leader advertising" model. It argues that firms may advertise very low sale prices on high-demand products in order to draw customers into the store, knowing that while they are in the store they are likely to buy other, non-sale items as well.
While it is impossible to fully rule out the first two models, CKR find no evidence that demand is more price sensitive during the peak-demand periods. They do, however, find that their supermarket chain significantly increased its advertising on products during the high-demand periods, lending support to the loss-leader hypothesis.
- One of the assumptions of perfectly competitive markets is that the good being traded is perfectly homogeneous. This means that buyers are completely indifferent between the goods offered by various sellers. Why might all buyers not be completely indifferent between can of Swanson's vegetable broth at Fred Meyer and an identical can at Safeway? How do factors of geography and transportation costs affect the degree to which various sellers' "identical" products are homogeneous or differentiated in an economic sense?
- If buyers are not completely indifferent between broth at Fred Meyer and broth at Safeway, how can the stores use this product differentiation to exercise a limited degree of monopoly power over their customers? In other words, how does this imply that grocery stores are not perfect price takers?
- Would a perfectly competitive firm ever spend money on advertising? What would it gain from advertising or why wouldn't it gain anything?
- CKR argue that the loss-leader strategy lies behind the offering of sale prices by grocery stores during periods of peak demand. If the true cause of this behavior is loss-leader behavior, would you expect the same behavior from gas stations, airlines, hotels, or specialty stores? Why? Based on your limited and casual observations, does this seem to be the case?