The Market Failure in Men's Business Suits:
A note on the threshold for regulation
13 Nov 2006 in Regulatory Economics, Regulatory Policy, People & Institutions
In our recent multi-part series on Susan Dudley, whose nomination to be the next Administrator of OMB's Office of Information and Regulatory Affairs comes up today, we offered a few comments on the meaning of market failure.
Executive order 12866, the presidential directive Dudley would be responsible for enforcing, says that market failure must be "significant." Dudley and her critics differ concerning how much market imperfection qualifies as "significant." Of the 19 public interest comments she authored or co-authored on specific proposed regulations, she said the agency had failed to meet this test of significance in 10 cases. Her critics appear to believe (but do not say directly) that the agencies satisfied the significance test in these 11 cases. Whereas Dudley accepts the market failure justification for regulation in some cases, her critics appear to accept it in all cases.
Dudley and her critics also disagree concerning whether regulation is always the best approach for remedying a market failure deemed to be "significant." In her Primer on Regulation Dudley wrote: "Government should only impose regulations in the case of a clear market failure that cannot be adequately addressed by other means" (pp. 38-39)." The other means she has in mind include clearer assignment of property rights. Her critics say regulation should be the first public policy choice.
The weekend edition of the Wall Street Journal included a long article on men's business suits. Their research showed the existence of a market failure. Did they find a "significant" market failure"? If so, will the market correct itself or must government intervene to correct it?
Men's business suits are often touted by the quality of wool yarn use to make them. Reporter Ray A. Smith summarizes (subscription may be required) gives a short background into what distinguishes the various numbers for superfine wool fiber, abbreviated "Super" and followed by a number (usually three digits) representing different gradations:
Just as sheets trumpet thread counts and gas has its premium octane, suit manufacturers are using these numbers to tout their wool. Higher numbers translate to narrower fibers, which makers say are softer to the touch. It's one of several tactics the suit industry is using to combat slowing sales. On the high end, makers are pushing suits in the Super 220s range for thousands of dollars. Discounters are also adopting the system, hoping to convince shoppers that a superior suit can be had for $300 or less.
The Journal sponsored laboratory tests of the fabric used in 10 men's business suits, one from each of 10 makers. These suits varied in price and the quality claims made for them (most notably the S-number). They tested for strength, wrinkle recovery, and fiber diameter. It is the last of these margins on which manufacturers recently have competed. The so-called "S" number describes fiber diameter: the higher the number, the finer the thread, the higher the quality -- and presumably, the better the justification for higher prices.
There are three underlying economic issues relevant to regulation, and it's important to keep them separate.
(1) DO HIGHER S-NUMBERS MEAN HIGHER QUALITY?
It turns out the answer is a firm maybe. The higher the S-number, the more it costs to make the fabric and the more luxurious it feels. But higher production cost is a limited proxy for quality; it also could be a sign of inefficiency or the absence of economies of scale. More importantly, the quality of a suit has more attributes than just fiber diameter. For most men, durability and wrinkle resistance are vital quality attributes. High S-numbers mean less durability and greater susceptibility to wrinkling, so there is an optimal S-number beyond which quality, as measured by consumer willingness to pay, goes down.
Not all suit buyers understand this before purchase, and Green's story provides an anecdote of a consumer who thought he'd be happiest with a very high S-number and was disappointed by wrinkling and lack of durability. Green also point out that manufacturers are aggressively marketing their S-numbers to stimulate sales.
Sellers can be expected to emphasize those quality attributes on which they are most competitive and deemphasize those attributes on which they aren't. No one reasonably expects sellers to advertise the shortcomings of their products. But that does not excuse misrepresentation.
(2) ARE MANUFACTURERS' CLAIMS ACCURATE?
The second issue is whether manufacturers' claims about the S-numbers of the fabric in their suits are accurate. This is the issue addressed by the Journal-sponsored laboratory study. In four of the 10 suit tested, fibers were thicker than advertised and in one case they were thinner. Green's implication is that consumers were misled. When asked to explain the discrepancies, one manufacturer said the Journal's test results conflicted with its own laboratory data, another said the test was inappropriate, and a third said the text description of the suit on its web site was wrong (and corrected it). The maker whose suit was made of finer fabric than advertised declined to comment but arguably didn't have an obvious public relations need to do so.
The observed discrepancy between actual and advertised fiver diameter could be explained many ways, ranging from manufacturer misrepresentation, to manufacturer ignorance, to normal variability in fiber diameter, to statistical sampling error. The principle defect in the Journal's report of its laboratory study is that it admits to the possibility of only the first two of these explanations, and it doesn't allow any of these competing inferences to be subjected to hypothesis testing. For example, if we knew how much variation in fiber diameter was normal -- both within a given suit and across suits in the same lot -- we could test the hypothesis that the laboratory's observations were within normal variation. The degree to which the lab results depart from the norm tells us the likelihood that variability could be ruled out. Without this information, however, we are left to speculate darkly -- which, unfortunately, is what Green's article invites.
For this reason, Green is too quick to imply that manufacturers are misrepresenting their products. This is an occupational hazard of reporters who attempt to do scientific research. They often lack the requisite training and skills to do research correctly, and because of this they are prone to ignore variability and uncertainty and overinterpret the data they have. Also, had the laboratory found no discrepancies it seems likely that Green would not have written about it. For that reason, both Green and the Journal have an incentive to believe that the observed discrepancies are significant, meaningful, and probably indicative of wrong conduct.
There is also ambiguity in the nature of manufacturers' S-number claims. The way suits are marketed, it's not clear what percentile of the fiber diameter distribution they intend any designation to mean. For example, is a suit labeled "Super 120s" intended to meet the fiver diameter standard on average, or is a higher percentage required to qualify? If it's a higher percentage, how much higher? 75%? 90%? 95% 100%? The definition is crucial for correctly interpreting the Journal's test results.
Green's story suggests that the Journal's lab tests were a follow-up to a trade association study:
In part because of the profusion of S-numbers across all price levels, the Cashmere and Camel Hair Manufacturers Institute decided to test the veracity of these claims. The Boston-based trade group that includes makers of superfine wool fabrics started testing finished products two years ago, at the urging of some of its members who noticed lower-priced suits were increasingly being labeled with high S-numbers.
In its test of 20 suits, the institute found that roughly two-thirds of them were mislabeled, including suits purchased from Men's Wearhouse, Filene's (now part of Macy's) and Sierra Trading Post. The group then notified the stores and manufacturers, which either stopped selling the suits or changed the labels.
Green does not say, and we cannot find a copy of this study on CCMI's web site, but we don't know if the study targeted manufacturers that bought its fabrics from firms that don't belong to CCMI. Even if the study sample was unbiased, it isn't clear from Green's story that the CCMI study ruled out variability or uncertainty.
So what did the Journal-sponsored laboratory study find?
Six of the suits we tested passed with flying colors, with results that matched or exceeded their promised S-numbers, including two of our least expensive choices -- an Arnold Brant and a Jos. A. Bank, both Super 110s that cost less than $300. A pricier Hickey Freeman Super 120s was found to have Super 130s fibers.
The Wool Products Labeling Act of 1939 prohibits garment "misbranding," which it defines as containing less than 95% wool exclusive of declared ornamentation. Interestingly, the Wool Act specifically recognizes variability by providing an exception to this rule in the case where a "person charged with misbranding proves such deviation resulted from unavoidable variations in manufacture and despite the exercise of due care to make accurate the statements on such stamp, tag, label, or other means of identification" (Sec 68b(a)(2)A)).
Green says the House of Representatives approved a bill in September that would codify guidelines issued by the International Wool Textile Organization. We can't find the guidelines online, but we have found H.R. 4583, the "Wool Suit Fabric Labeling Fairness and International Standards Conforming Act." The bill would resolve the part of the definitional question by specifying that each Super- designation corresponds to an average fiber diameter. For example, Super 130s would be defined to include garments whose average diameter is 17.25 microns or finer, with each 10-unit increase in Super-designation requiring a 0.5 micron reduction in fiber diameter.
But the bill addresses only average fiber diameter. It doesn't address the issues of uncertainty or variability. These it punts to the Federal Trade Commission:
[T]he average fiber diameter of such wool product may be subject to such standards or deviations as adopted by regulation by the Commission.
No doubt there will be extensive jockeying over what deviations from the mean are permitted. For purely competitive reasons, makers of "high" quality fabrics will want the FTC to set and strictly enforce very low tolerances. Makers of "low" quality fabrics, in turn will want the FTC to permit wide variations and enforce the standards lightly.
Two general rules will apply. First, each firm will want the standard set slightly below what it can profitably achieve. That drives out the most competitors and keeps prices as high as possible. Second, each firm will defend its preferred standard by stating that it's needed to protect consumers (makers of "high" quality fabrics) or to satisfy diverse consumer tastes (makers of "low" quality fabrics).
Meanwhile, it's not at all clear what "consumers" want, or whether the resulting regulations will reduce ignorance or informational asymmetries. The independent regulatory analyst prefers a different approach: simple disclosure of both the mean fiber diameter and the standard deviation.
(3) IS ASYMMETRIC INFORMATION PER SE A JUSTIFICATION FOR REGULATION?
Markets provide a lot of the information consumers are looking for, if they care enough to look. That turns out to be a crucial test because some apparent informational asymmetries result from consumers' lack of effort to obtain information. It's entirely rational for inframarginal consumers, for whom the value of better information is not worth the effort to obtain it, to rely on the market signals sent by those who passionately care about such things as fiber diameter. Rational ignorance is not evidence of asymmetric information.
For those who really want to know, there's a lot that can be learned without too much effort. For example, Holland and Sherry, a Saville Row fabric maker, informs consumers what various Super grades mean in terms of the range of fiber diameters included:
14 to 15 micron fibre thickness = Super 170's
15 to 15.5 micron fibre thickness = Super 150's
15.5 to 16 micron fibre thickness = Super 140's
16 to 16.5 micron fibre thickness = Super 130's and so on
Holland and Sherry also provides useful performance information. Indeed, the firm is so solicitous of consumer demand that it will weave the buyer's name within the pinstripes. If you have to ask the price, you probably can't afford it.
No one would suggest that Holland and Sherry intends to mislead its customers. Yet the diameter ranges they cite differ from those in H.R. 4583. If the bill is enacted, Holland and Sherry will have to change its labeling or it could be found guilty of misbranding if a US retailer sells a suit made from its fabric in the US.
An important source of information to overcome informational asymmetry is independent testing, such as the small study sponsored by the Wall Street Journal. Consumer Reports used to have something approaching a monopoly on product testing, but today they face stiff competition. Myriad review services have arisen and thrived to overcome notorious informational asymmetries in lodging and restaurants. In addition to the classic, professional guides by Michelin, Mobil, Zagat and Gayot, local newspapers publish their own dining guides and hire reporters to cover the restaurant beat. The Internet has created a wealth of opportunities for regular travelers to post comments on hotels, cruises, and even frequent flyer programs.
In 1970, the study of informational asymmetries in economics started in earnest with the publication of George Akerlof's paper "The Market for 'Lemons'," a study that secured for Akerlof a Nobel prize in 2001. Akerlof developed principles for determining when informational asymmetries were so severe that markets could not form. He studied the market for used cars, which has long been plagued by informational asymmetries that impair buyers' ability to discern quality, and suggested that a market for truly defective used cars ("lemons") could not exist. But markets are creative organisms, and the list of genuine "lemon" markets is thin at best.
Even the used car market works amazingly well. Sellers of high quality used cars voluntarily reveal information to help buyers distinguish quality. Automakers operate certification programs to assure minimum quality standards are met, and both direct and third-party warranties can be obtained. Other services help consumers learn of serious potential problems such as prior collisions, flood or fire damage, or prior service as a rental.
In short, when informational is asymmetrical, opportunity arises for innovative entrepreneurs to fill these gaps. Newspapers like the Wall Street Journal may be in the primary business of selling newspapers, but to do so they need to provide information its customers want. To a great extent, and especially in the Internet age, that means helping their customers overcome asymmetric information.
IS THIS A "SIGNIFICANT" MARKET FAILURE WARRANTING REGULATION?To some, the fiber diameter of yarn used to make men's business suits might seem to be a trivial matter. To others, and apparently to a majority of Members of the House of Representatives, who wear business suits every day, it's serious enough to earn approval by voice vote. Members likely had different motivations for supporting the bill ranging from consumer protection ("ensur[ing] that unscrupulous garment manufacturers don't dupe consumers with simple phony labels") to trade protectionism ("We owe that to" ... "the great American textile industry that produces these fine products", which mostly come from Italy). (Both arguments were provided by Rep. Stearns in his defense of the bill on the House floor.)
The published data show that some men's suits contain some fibers that have diameters greater than advertised. The data aren't sufficient to enable us to distinguish between intentional "misbranding"(which certainly could be going on) and normal variability (which we know exists) and sampling error (which is inherent in all statistical studies). Possibly in part because a market failure is assumed and not demonstrated, the proposed regulatory remedy problem is
Evidence of consumer dissatisfaction appears to be limited, and based on Green's article, mostly associated with trade-offs among quality attributes. Green quotes no suit purchasers who felt cheated by discrepancies in fiber diameter or who judged apparent discrepancies to be meaningful. The real problem appears to be that makers of expensive, narrow diameter wools are unhappy with low-price competition that they perceive (but have not demonstrated) to be the result of cheating. They have issued industry guidelines but cannot enforce those guidelines on nonmembers, so they turn to government to enforce industry these guidelines on their behalf.
This is an example of the most conventional use of regulation -- the use by industry of the coercive powers of government, usually to reduce competition. Dudley wrote about it in her Primer on Regulation. It's in her summary of the theory of regulation exposited by Nobel prize winner George Stigler in 1970:
- The basic resource of the government is the power to coerce.
- An interest group that can convince the government to use its coercive power to its benefit can improve its well-being at the expense of others.
- Agents (firms, individuals, government officials, interest groups) are rational and try to maximize their own utility (well-being) (p. 9.)
Probably not. To see why, let's first assume that industry advocates of regulation are exactly right about what's going on: disreputable manufacturers are misbranding their fabrics, calling them something they aren't. Government is about to implement exactly the regulatory solution they have asked for: standards for average fiber diameter for each superfine term ranging from Super 80s to Super 250s. This will work only if the problem us average fiber diameter. Disreputable manufacturers will respond by achieving the required average fiber diameter but increasing variation. If the real problem is variation in fiber diameter, regulation will both fail to achieve its intended purposes and layer a new governmental failure on top.
If industry advocates are wrong, and the underlying problem is not the disreputability of their competitors but variability in fiber diameter, the regulation also will fail -- and for an analogous reason. It's not targeted at the real problem. In that case, we can expect a new round of industry complaints and demands for more regulation to fix the governmental failure as well as the original (alleged) market failure.
Finally, it's possible that the real purpose of the regulation is just to reduce competition, especially from low-cost Asian imports. But it probably will fail to do that, too. Regulatory standards won't trump low production costs in Asia unless they impose design requirements that low-cost producers cannot duplicate.
Earlier we suggested an alternative regulatory approach: full disclosure of both average fiber diameter and variance, but no regulation of the meaning of marketing terminology. If the point is really to overcome consumers' asymmetric information, this would actually work. Consumers can be easily educated understand the significance of fiber diameter -- indeed, manufacturers of high quality suits want to educate them. But consumers also are smart enough to learn that lower variances in fiber diameter mean greater uniformity in quality for every average fiber diameter.
Reader note: The final two sections were
inadvertently omitted from our original post.
REFERENCES
Akerlof GA. 1970. “The Market for ‘Lemons’: Quality Uncertainty and the Market Mechanism.” Quarterly Journal of Economics, 84(3), pp. 488-500.
