Increasing trademark protection lowers product quality

The central argument in favor of trademark protection is that trademarks incentivize firms to produce high quality products and prevent a race to the bottom in product quality. We investigate this central pillar of the FTDA’s effects on firm strategy using a broad and objective measure: recalls of unsafe products.16

The Consumer Product Safety Commission (CPSC) conducts safety testing and enforces recalls of most consumer products in the United States; the National Highway Traffic Safety Administration (NHTSA) conducts safety testing and enforces recalls of automotive products in the United States. Unsafe products are detected through several channels. First, firms are required by law to conduct appropriate internal safety tests and report defects as soon as they are discovered. Second, the agencies conduct proactive testing in their own labs. Third, the agencies operate hotlines and Web sites that allow consumers to report unsafe products. We collected data on all product recalls announced by the CPSC and NHTSA between 1982 and 2005.

Table 3 presents difference-in-differences estimates of the FTDA’s effects on product recalls. Column 1, using a logit model, shows that post-FTDA treated firms were almost twice as likely (+81%) to have a product recalled in any given year. Column 2 shows that the results are similar using a linear probability model: The estimated treatment effect is +1.5 pp compared to a base rate of 0.9 pp, and Columns 3 and 4 show similar results using the number of product recalls, either in a generalized negative binomial model or in logs. Columns 5 and 6 show that the number of units recalled rose 17% and the dollar value of products recalled rose 31% for treated firms relative to control firms in the same industry-year.17

These results suggest that the FTDA’s added trademark protection resulted in lower product quality among treated firms. An alternative explanation is that the rise in recall frequency reflects greater caution by treated firms; for example, if the increased value of their brands led treated firms to conduct more careful testing and proactively recall unsafe products. In Table 3, Column 7, the dependent variable Voluntary is the fraction of recalls that were voluntarily initiated by the firm itself. The estimated treatment effect is not statistically significant, and the point estimate is negative (-5.1 pp); that is, treated firms were slightly less likely to initiate a recall voluntarily.

Figure 3 shows graphs of the year-by-year coefficients for the number of recalled products surrounding the TLRA, FTDA, and Moseley decision. Panel A shows that the trends for the treatment and control groups in the pretreatment period prior to the FTDA are similar. Unlike the observed effects on firms' profits in Figure 2, the effect of the FTDA on product quality is more gradually realized. The apparent lag in the treatment effect is plausibly due to the time needed for an unsafe product to be produced, sold, detected, and recalled.18 Consistent with the hypothesis that the increase in product recalls was an effect of the FTDA, panel B shows no treatment effect around the placebo event of the TLRA.

In sum, across a range of specifications, post-FTDA treated firms declared more product recalls covering more units with a higher dollar value and were less likely to do so voluntarily. These results suggest that treated firms lowered their product quality in the post-FTDA period and are broadly consistent with the quiet life (Bertrand and Mullainathan 2003) and trademarks-foster- monopoly hypotheses (Chamberlin 1933). In the absence of an incentive for higher quality, stronger trademark protection will lead firms to raise prices and cut costs as they face fewer competitive threats. If product quality is costly to maintain, via more stringent internal testing and lower production yields, then firms are likely to accept a cut on these margins as well.

Our results are also potentially consistent with the quality-incentive hypothesis (e.g., Shapiro 1983), because we measure the effects of dilution protection on treated firms, which were incumbents that owned famous brands. The quality-incentive hypothesis suggests that firms will produce high-quality products initially to build a reputation. Once reputation is established, an incumbent will trade off the costs of higher quality (hiring more inspectors, lower production yields) against the business lost by producing a lower quality product. Strengthening trademark protection for incumbents might increase their temptation to produce lower quality at a higher markup. Our results are also notably consistent with those of Qian (2008) in a very different setting. Qian finds that weaker protection against counterfeits in China led brand-name shoe firms to produce higher quality goods; we find that stronger trademark protection in the United States led treated firms to produce lower quality goods.