The Politics of Food Labeling and Certification - The Oxford Topic of Food, Politics, and Society - page 624

error in data from self-reports because of a respondent's wish to project a favorable image
(Fisher 1993, p. 303). Some surveys attempt to mitigate the effects of social desirability bias
by using anonymous, mail-in surveys (e.g., De Pelsmacker et al. 2005). However, because
social desirability bias comprises a self-deceptive aspect in addition to an other-deceptive
aspect (Nederhoff 1985), there is reason to suspect that all self-reported data will overesti-
mate consumer willingness to pay a premium on ethically certified food.
A small but growing number of empirical studies avoid problems of social desirability
bias by measuring actual consumer purchasing behavior in response to the presence
or absence of ethical labels on products in retail settings. The best of these studies are
experimental, since their research design avoids the selection bias that often plagues
observational studies. In a study of 150,000 eBay auctions that resemble “natural experi-
ments,” Elfenbein et al. found that auction participants were willing to pay 6 percent
more for an identical product when it was advertised that the auction was for charity
(Elfenbein et al. 2010). In a two-part field experiment of 26 stores of a major U.S. grocery
chain, sales of coffee rose by 10 percent when a Fair Trade label was added, as consum-
ers switched from unlabeled coffee (sales for unlabeled coffee dropped 9 percent when
the labeled option appeared). This suggests that when consumers have a choice between
labeled and unlabeled coffees that are otherwise identical and equally priced, consumers
respond to the label. In the second part of the experiment, the addition of a price pre-
mium to two Fair-Trade-labeled coffees led to a 2 percent increase in sales in the case of
the higher-end coffee and a 30 percent decrease in sales in the lower-end coffee, suggest-
ing that some consumers are willing to pay a premium on Fair Trade labeled coffee, but
that this willingness may be limited to higher-income shoppers (Hainmueller, Hiscox,
and Sequeira 2011).9
Similar experiments on nonfood items support the idea that some types of consum-
ers are willing to spend extra on ethically labeled products. In an experiment in a home
goods store, sales of towels and candles rose by 10-20 percent when labels promising
good labor standards were added (Hiscox and Smyth 2007). In an eBay experiment,
consumers were willing to pay on average a 45 percent premium for Fair-Trade-labeled
shirts over unlabeled shirts (Hainmueller et al. 2011). In an experiment in 111 Banana
Republic factory outlet stores, the addition of labels conveying a fair-labor message
increased sales of a high-priced women's clothing item by 14 percent, whereas labels had
no significant effect on the sales of low-priced women's or men's10 items (Hainmueller
and Hiscox 2012a). In another experiment in 419 Banana Republic stores and 155 Gap
Outlet stores, women proved willing to buy eco-labeled denim jeans sold in nonoutlet
stores (sales increased by 8 percent in the presence of a label with an environmental
message), but the labels had no effect on sales of men's jeans sold in nonoutlet stores or
of women's jeans sold in outlet stores (Hainmueller and Hiscox 2012b).
While experimental studies offer the most rigorous tests of consumer demand for
certified products to date, they, too, have limitations. The most important caveat con-
cerns issues of generalizability. Demand for ethically certified products may be product
specific. Additionally, none of the studies—even those conduct in outlet stores—effec-
tively targeted low-income shoppers.