Susan Dudley, Part 4
The case of airbags

2 Nov 2006 in Regulatory Policy, People & Institutions

On Monday we began a multi-part series on Susan Dudley, the Bush administration's nominee to head OMB's Office of Information and Regulatory Affairs. Her confirmation hearing is scheduled for November 13. A campaign is being led by Public Citizen and OMB Watch to scuttle her confirmation. Our series analyzes the merits of their opposition report. Tuesday's installment provided background on Public Citizen and OMB Watch, the activist organizations who authored the report. Wednesday we posted a discussion of the three dominant themes in the introduction to the Opposition Report.

Today we provide an analysis of the airbag case. In 1998 Dudley submitted a public interest comment on the National Highway Traffic Safety Administration's proposed advanced airbag rule. Although it was just one of 33 such public interest comments, the Opposition Report say this document "epitomizes all the reasons that Susan Dudley is unfit to be given power over regulatory policy" (p. 7).

During her tenure at Mercatus, Susan Dudley has authored or co-authored 33 public interest comments. Eighteen (55%) dealt with environmental regulations or guidance; eight (24%) concerned draft OMB guidance or draft reports to Congress by OMB; two each (6%) concerned OSHA's ergonomics rule and a pair of economic regulations. One (3%) concerned an airbag rule. The Opposition Report says that Dudley's comment on airbags "epitomizes all the reasons that Susan Dudley is unfit to be given power over regulatory policy" (p. 7).

It's possible that Dudley's comment on the airbag rule is representative, but this seems unlikely. Dudley's primary interest clearly has been environmental policy. More likely, the Report focuses on her comment on the airbag rule because highway safety is Public Citizen's primary interest. (In the list below, we have identified all 33 of Dudley's public interest comments and highlighted in boldface the 12 [36%] that are cited in the Opposition Report.)

We've already pointed out that airbags are a special regulatory issue to Public Citizen because of president Joan Claybrook's long involvement, including her personal responsibility for signing into law he regulation that mandated these devices in her capacity as Administrator of the National Highway Traffic Safety Administration. We also have suggested that Public Citizen's opposition to John Graham, Dudley's immediate predecessor, may have been motivated in large part because Graham published highly critical reviews of airbag regulations, including several in peer reviewed journals. Indeed, before taking the position at OIRA Graham was something of an expert himself on motor vehicle safety. His 1983 doctoral dissertation at Carnegie Mellon University was titled "Automobile Safety: An Investigation of Occupant-Protection Policies." Graham published several peer reviewed articles based on it, including one in Policy Sciences in 1984.

Graham began by noting the degree to which the practical problem of improving highway safety was affected by ideology:

[A] fundamental scientific and ideological dispute has emerged between those who emphasize the promise of safety technology and those who emphasize the role of driver behavior. A similar tension is apparent in numerous policy areas such as consumer product safety, occupational health, and self-hazardous behavior

Technology and human behavior can influence the effectiveness of safety policies. In the field of traffic safety, rational-choice theorists postulate that automobile safety devices induce increased driver risk taking. Such behavioral responses could partly or totally nullify the lifesaving potential of governmental safety rules for new cars, such as the crashworthiness standards adopted by the United States in the late 1960s and early 1970s. This study explores the behavioral-response hypothesis in the context of a car-vintage model of U.S. car occupant death rates. Results from the model imply that U.S. standards have reduced the occupant death rate by roughly 30 percent, a finding consistent with minimal driver response to safety devices (pp. 141-142).

Graham described the contest as between "technologists," who seek "to bypass the vagaries of driver behavior by adopting measures such as air bags and breakaway sign posts," and "behaviorists," who say "driver behavior is the fundamental cause of traffic accidents" and "recommend educational and incentive-oriented policies designed to encourage less drunken driving and safer driving habits." The behaviorists, led by Sam Peltzman, used crash data to show that drivers adapted to safety devices by taking more risks. The contest between these perspectives is important. If the technologists are right, motor vehicles can be designed to achieve very high levels of safety. If the behaviorists are right, gains from technical improvements are substantially lost because drivers and passengers take them for granted. Where Graham went wrong is he exaggerated the possibility that these opposing views could be reconciled. "In the abstract," he wrote, "one does not need to take sides in this debate" (p. 142).

Then he threw gasoline on the fire:

Proponents of [the technologists'] view, such as Ralph Nader, have had a powerful, though declining, influence on the evolution of traffic safety policy in the United States (p. 142, emphasis added).

Nader was Claybrook's partner and mentor. Casually dismissing them as past their prime might have inflamed passions across the ideological divide. (Ironically, Graham's empirical research supported the inference that safety regulations prior to NHTSA's first airbag rule had reduced occupant death rates 30%.)

The 7-page section in the Opposition Report on airbags (plus additional material elsewhere) thus appears to be a skirmish in a different war, one in which Dudley is at least a latecomer to the battlefield if not an innocent bystander. The text is an extended defense for technological solutions to highway safety, of which airbags are now the central feature, and a detailed recounting of Claybrook's long battle with automobile manufacturers and the behaviorists (including Graham) who have different views. The authors of the Report appear to consider these different views illegitimate, immoral and corrupt, attitudes that also seem more likely to inflame passions than advance motor vehicle safety.

Leaving aside the decades-long battle between the technologists and behaviorists, we discern three arguments that are relevant to Dudley. We paraphrase them as follows:

1. "The public supports stringent motor vehicle safety standards of the kind Dudley opposes."
2. "Dudley is opposed to any regulation requiring airbags."
3. "Dudley is opposed to NHTSA regulating to provide the same level of protection for everyone."

Let's examine the evidence in the Opposition Report supporting each of these claims.

"The public supports stringent motor vehicle safety standards of the kind Dudley opposes."

The case presented in the Opposition Report supporting stringent motor vehicle regulations is based on public opinion polls sponsored by Advocates for Highway and Auto Safety and performed by pollsters Louis Harris and the Peter Harris Research Group. Advocates for Highway and Auto Safety

is an alliance of consumer, health and safety groups and insurance companies and agents working together to make America's roads safer. Advocates encourages the adoption of federal and state laws, policies and programs that save lives and reduce injuries. By joining its resources with others, Advocates helps build coalitions to increase participation of a wide array of groups in public policy initiatives which advance highway and auto safety.

Its board of directors includes representatives of automobile insurance companies, physicians' associations, and advocacy organizations -- including Public Citizen.

Under conflict of interest principles less stringent than those applied by Public Citizen and OMB Watch to Dudley's work at Mercatus, these polls would be dismissed out of hand because they were commissioned by a special interest group with well-defined policy objectives -- in this case, support for technology-based highway safety regulation. We believe that sponsorship per se should not be used to dismiss research and that it should be evaluated on its merits.

Harris and Harris summarize the major finding as follows:

1. "On a trend basis, federal regulation of highway and auto safety, food and other product safety areas, and safety on the job continues to receive remarkably high levels of public support" (Sec. 1).
2. "Overall, a 6 to 1 majority of the public (84% to 13%) express the view that 'the government should create safety rules that require manufacturers to make all passenger vehicles including SUVs more stable and less likely to roll over'.The split between men and women is decisive: 91% of the women favoring such regulation, compared with a lesser 77% among men. However, the 77% number is still significantly high. SUV owners themselves favor such government action by 82% to 14%".(Sec. 2).
3. "A 55% to 39% majority nationwide favors the government requiring [Electronic Stability Control] devices to be installed in new cars in the future. Several sharp differences emerge from key groups in the population. By 50% to 45%, men oppose requiring ESC devices in new cars, but women carry the day with their 64% to 29% support of ESC devices. By 62% to 33%, SUV owners favor installing ESC devices, while owners of pick-up trucks support it by a smaller 52% to 44% majority" (Sec. 2)
4. "An 82% to 17% majority nationwide would like to see the government require improvements in seat belts to better protect passengers during rollovers" (Sec. 2).
5. "An 83% majority 'wants the government to require a major upgrading of roof safety standards to withstand the weight of the car when it rolls over'." (Sec. 2)
6. "A 72% to 24% majority believes 'the government should set standards for auto manufacturers to prevent people from being ejected from motor vehicles in crashes' " (Sec. 3)

These results appear to show very strong support for more safety, in many cases provided by the government through regulation, but they also show significant diversity in preferences, especially between men and women. According to the survey researchers, 72% of respondent supported making various safety features standard equipment on all vehicles (Table 3). Support was highest among low-income respondents (77% for income between $25,000 and $50,000) and declined with income -- 64% for respondents earning $50,000 to $100,000, and 54% for those earning more than $100,000. According to Harris, "These sentiments seem to indicate that the day might arrive soon when such safety features of cars will take their place along with Social Security and Medicare as part of the built-in health benefits accorded to all parts of the population."

While Harris and Harris report the decline in support with rising income, they can't explain it. We can.

When economists encounter goods for which demand decreases as income increases, they call them "inferior goods." Standard economics textbooks give as examples of inferior goods "hamburger, cheap wine (André), and public transportation," and "rotgut whiskey, potatoes, and second-hand clothing." (For an on-line explanation of inferior goods, see the Wikipedia entry.)

Leaving aside the dismissive tone of the textbooks, what these commodities have in common is that consumers buy less of them as their income increases. But no serious economist believes that safety is an inferior good. Yet that's what we'd have to believe if the data in the Harris Survey are meaningful: as income rises, people want less safety.

A more plausible explanation is that survey respondents' answers about safety features were colored by different expectations concerning who would pay for them. Lower-income respondents may have been more likely to think that additional safety features could be had at no increase in price. If confidence in the ability to have one's preferences subsidized declines with income, we'd expect declining support with rising income. 

This fatal defect in the Harris and Harris survey becomes even more glaring when the survey's responses on willingness to pay are considered. Harris and Harris did not obtain data on actual willingness to pay based on revealed preference. This is the preferred technique of economists, and not incidentally, it's also the preferred technique in Circular A-4, OMB's guidance to agencies on performing regulatory impact analysis. A carefully conducted economic study would have been required to do that, not a public opinion poll.

The next-best alternative to a revealed preference study is a stated preference survey . Stated preference surveys are hard to do well, and often they're controversial because they can yield results that are incompatible with economic theory (such as, for example, declining willingness to pay with income). That's why Circular A-4 has extensive guidance (p. 23) concerning stated-preference methods. We've highlighted in boldface those elements in OMB's guidance that Harris and Harris clearly did not follow.

• willingness-to-pay questions should be designed to focus the respondent on the reality of budgetary limitations and alerted to the availability of substitute goods and alternative expenditure options;
• the survey instrument should be designed to probe beyond general attitudes (e.g., a "warm glow" effect for a particular use or non-use value) and focus on the magnitude of the respondent's economic valuation;
• the analytic results should be consistent with economic theory using both "internal" (within respondent) and "external" (between respondent) scope tests such as the willingness to pay is larger (smaller) when more (less) of a good is provided;
• the subjects being interviewed should be selected/sampled in a statistically appropriate manner. The sample frame should adequately cover the target population. The sample should be drawn using probability methods in order to generalize the results to the target population;
• response rates should be as high as reasonably possible. Best survey practices should be followed to achieve high response rates. Low response rates increase the potential for bias and raise concerns about the generalizability of the results. If response rates are not adequate, you should conduct an analysis of non-response bias or further study. Caution should be used in assessing the representativeness of the sample based solely on demographic profiles. Statistical adjustments to reduce non-response bias should be undertaken whenever feasible and appropriate;
• the mode of administration of surveys (in-person, phone, mail, computer, Internet or multiple modes ) should be appropriate in light of the nature of the questions being posed to respondents and the length and complexity of the instrument;
• documentation should be provided about the target population, the sampling frame used and its coverage of the target population, the design of the sample including any stratification or clustering, the cumulative response rate (including response rate at each stage of selection if applicable); the item non-response rate for critical questions; the exact wording and sequence of questions and other information provided to respondents; and the training of interviewers and techniques they employed (as appropriate);
• the statistical and econometric methods used to analyze the collected data should be transparent, well suited for the analysis, and applied with rigor and care.

It's likely that respondents were reacting to a hypothetical offer, and so they responded hypothetically. When asked "Should these improvements be required by the federal government to be offered as standard equipment on all vehicle makes and models or not?", the implication is that they could get something for nothing. Who (besides an economist) would not agree to this deal? What's amazing in the Harris and Harris poll is that when presented with free safety features, up to 26% said they didn't want them.

Experienced stated-preference researchers would say that results from a poll overstate respondents actual willingness to pay, quite possibly by a lot. So whatever results Harris obtained provide a plausible upper bound on reality. According to Harris and Harris, 91% of respondents say they'd be be willing to pay $200 to $300 for more safety features. For that small sum, Harris offered them "safety improvements that would prevent rollovers, provide better roof crush protection, and upgrade side impact protection." That's a lot of safety improvement for very little money -- especially the complete prevention of rollover. What's striking is that 7% of respondents (and 17% in 1996) were willing to say $200 to $300 was too costly for this rich basket of safety features.

We think all credible practitioners of regulatory analysis would dismiss these poll results as useless.  They help us only reach the banal conclusion that most (but possibly not all) consumers value safety. They don't provide any evidence of market failure. Assuming that a market failure exists, they don't illuminate which options should be examined. And they don't illuminate which option would be best irrespective of what policy goal we decide to achieve. NHTSA could not legitimately use these poll results to justify regulation, and OIRA staff would be expected to oppose -- on elementary technical grounds alone -- any draft regulation based on them.

"Dudley is opposed to any regulation requiring airbags."

According to the Opposition Report, "it was precisely a lack of regulation that led to the air bag tragedies Congress required NHTSA to address in the advanced air bag rulemaking." The Report alleges that "manufacturers chose to install cut-rate air bags in the absence of a stringent regulation requiring advanced designs" (p.7). These "cut-rate designs" complied with the performance standard Claybrook made mandatory during her tenure as NHTSA Administrator.

To make this inference requires that the facts of the case be ignored. Airbags have saved many lives -- Dudley herself credits NHTSA's estimate that they prevent 3,200 deaths per year. But for small-stature adult drivers, and both small-stature adults and children riding as front-seat passengers, airbags also can kill. By requiring airbags, Claybrook's regulation is responsible for both. The Report suggests that Public Citizen is happy to claim the lifesaving benefits from making airbags mandatory but loathe to accept responsibility for the deaths attributable to this regulatory mandate. Dudley's public interest comment is not critical of airbags, but rather a regulatory policy that forces all consumers to live or die with them.

Part of the problem was the testing procedure used to craft the performance standard. Claybrook's rule required airbags powerful enough to protect an unbelted 164-pound adult. Airbags this powerful are not necessary to protect belted occupants, and they are way overpowered for protecting small-stature adults and children. By imposing by regulation a testing procedure biased against belted occupants in general and small-stature adults and children in particular, Claybrook's rule made an implicit judgment that the lives of belted small-stature adults and children were less valuable than the lives of unbelted adult men.

As the airbag death toll mounted, consumers rebelled against this value judgment. Many consumers wanted to turn the airbags off, but doing do was illegal. Dudley summarized the history of NHTSA's efforts to manage regulatory failure with more regulation and provides the context for NHTSA's advanced airbag rule.

NHTSA has taken several actions to reduce the risks air bags pose to certain occupants. In May 1995, NHTSA temporarily allowed (until August 1996) vehicle manufacturers to install on-off switches for passenger air bags in new cars and light trucks without rear seats or with rear seats that were too small to properly accommodate a child seat. In December 1996, this policy was extended until Aug. 31, 2000.

In March 1997, NHTSA temporarily amended the unbelted dummy test to include a sled test, in which a vehicle is stopped quickly, but not crashed into a barrier (as an alternative to the fixed-barrier test), which would allow manufacturers to reduce the force of the air bags during deployment. NHTSA recently released information on 115 crashes which reveals that the depowering that was permitted by this amendment has virtually eliminated air bag-induced deaths of short women in minor, low-speed accidents.

In November 1997, NHTSA issued a notice allowing dealers and repair shops to install an on-off switch to temporarily deactivate air bags for consumers fitting certain risk profiles who obtain prior authorization from the agency.6 NHTSA considered, but rejected in November 1997, a rulemaking to permit all air bag vehicles to be equipped with manual cutoff switches (p. 3, footnotes omitted).

In a footnote within the preceding text, Dudley explains how difficult NHTSA made it for consumers to exercise their own judgment concerning how to balance the risks and benefits of airbags:

To obtain [an on-off] switch, consumers must: (1) obtain a NHTSA information brochure and a request form, (2) fill out the request form, certifying that they have read the information brochure and that they fit one of four profiles of people at risk, and send it to NHTSA, (3) receive NHTSA authorization via a letter, and (4) take the letter to a dealership or other service outlet to have an on-off switch installed. The four eligibility profiles are: (1) those who cannot avoid placing rear-facing infant seats in the front passenger seat. (2) Those who have a medical condition that places them at specific risk. (3) Those who cannot adjust their driver's position to keep back approximately 10 inches from the steering wheel. (4) Those who cannot avoid situations -- such as a car pool -- that require a child 12 or under to ride in the front seat.

The Opposition Report quotes selectively from Dudley's public interest comment on the advanced airbag rule. When the text quoted in the Report (below in boldface) appears in context, its meaning changes dramatically:

NHTSA estimates that air bags have reduced fatalities in frontal crashes by about 30 percent. Moreover, judging from vehicle manufacturers’ pre-regulation actions and ongoing advertising, which lists dual air bags as a positive attribute in new vehicles, consumers appear to prefer vehicles equipped with air bags. These facts, however, are not sufficient to justify federal regulation requiring air bags. If air bags protect lives, and consumers demand them, it is reasonable to assume that automobile manufacturers would have installed air bags in the absence of federal requirements to do so. In fact, the preamble confirms that a “strong market preference for air bags led manufacturers to move toward installing them in all of their passenger cars and light trucks” before federal law required them (pp. 8-9).

Dudley's point was that consumer demand alone was providing airbags to the market. She said changes in personal behavior (dramatic increases in seat belt use) since the 1970s combined with market forces (high and rising consumer demand for safety and variety in safety features) may have rendered NHTSA's airbag rule superfluous as well as counterproductive. "Public safety may well be maximized," she wrote, "if NHTSA simply required manual lapshoulder belts, and left further decisions to the market place" (p. 19). In short, NHTSA could achieve greater motor vehicle safety if it simply got out of the way.

In his 1984 Policy Sciences article cited earlier, Graham identified two warring camps -- the technologists and the behaviorists. Dudley's comment sided with the behaviorists but left room for the technologists. Behavioral changes among consumers, combined with their strong and rising demand for motor vehicle safety, could achieve more lifesaving and injury prevention than would continued fine-tuning of federal regulations. NHTSA's determination to control the market for safety through regulation stifled technical innovation. Freed of NHTSA's rules, automotive engineers would be able to respond to consumer demands for a rich variety of safety features and provide what consumers want..

"Dudley is opposed to NHTSA regulating to provide the same level of protection for everyone."

On this point the Opposition Report is correct, but only in part. In her public interest comment, Dudley objected to NHTSA's use of fixed engineering performance standards for all vehicles, what she called a "one-size-fits-all" approach to safety (pp. ES-1, 6, 14, 18, 19, and 21). She argued that this approach was both ineffective (meaning that it could not achieve equal protection for all motor vehicle occupants) and inefficient (meaning that it would not maximize net social benefits). Effectiveness requires that expenditures on safety actually achieve it. Efficiency is achieved when consumers get exactly the safety improvements they want for the amount they want to pay. Dudley argued that NHTSA's proposed advanced airbag rule failed on both counts.

The authors of the Opposition Report seem to prefer equal protection for all motor vehicle occupants on equity grounds. This is an entirely reasonable aspiration. Dudley says that because consumers are different in size, age, and driving habits, equal protection is technically infeasible. It's impossible for regulators to achieve equal protection no matter how hard they try. The Opposition Report does not offer any insight concerning how regulators could overcome this inherent limitation in safety regulation. The authors seem satisfied that as long as safety inputs are constant, safety outputs and safety outcomes will be constant, too.

Besides saying equal protection is infeasible, Dudley also says it's wrong because consumers don't want it. She says motor vehicle safety ought to be a matter of consumer choice, and she trusts buyers to make decisions consistent with their own preferences. The Opposition Report criticizes (indeed, ridicules) the view that consumers know best. The Report addresses the question of consumer choice only obliquely, and on equity grounds. Leaving safety to the market, they write,

places safety features financially out of reach for many customers, depriving lower- and middle-income consumers and their families the safety afforded to the wealthy. And even when offered independently of other features, safety technologies may still be out of the financial reach of many consumers (p. 12).

For Public Citizen and OMB Watch, motor vehicle safety is a right, and rights should not be traded in markets. Allowing safety to be traded in markets permits the rich to buy more of it than the poor, and that means the rich get more rights. So it's unethical to allow markets to decide how much safety is produced, even if markets could produce more of it.

But let's follow up on the implications of their equity argument. It's not clear how Public Citizen and OMB Watch expect automobile manufacturers to recover the cost of safety features that are more expensive than lower- and middle-income consumers are willing to pay. Manufacturers would have to cross-subsidize vehicles preferred by lower- and middle-income consumers with profits earned from higher-income consumers. Market forces alone will never yield that result. NHTSA lacks the authority to compel automobile manufacturers to make buyers of expensive cars pay for the safety features in cheap cars. Public Citizen might support giving NHTSA that authority.

Until then, however, Public Citizen's preference for technology-based regulations ensures that the poor pay more than they want for motor vehicles, because manufacturers can't legally sell these consumers what they want to buy, or they have to go without. This result, which makes the poor worse off, is justified by Public Citizen as a matter of equity.

Mercatus Center Regulatory Studies Program Public Interest Comments Authored or Co-authored by Susan Dudley
Boldface = Cited in the Opposition Report

1. OMB's 2005 Benefit-Cost Report to Congress (6/25/05)
2. EPA's Proposed Willingness to Pay Survey: Phase III Cooling Water Intake Structures (4/25/2005)
3. EPA's Phase III Cooling Water Intake Structures (1/31/05)
4. OMB's 2004 Benefit-Cost Report to Congress (5/11/04)
5. Application of the SDWA to Submetered Properties (10/16/03)
6. Cooling Water Intake (6/2/03)
7. RIA Guidelines (5/5/03)
8. Office of Management and Budget's Draft 2003 Report to Congress on the Costs and Benefits of Federal Regulation (4/29/03)
9. Withdrawal of TMDL Water Quality Plans (1/17/03)
10. Office of Management and Budget's 2002 Draft Report to Congress on the Costs and Benefits of Federal Regulation (5/28/02)
11. The Environmental Protection Agency's Request for Comments on National Drinking Water Regulations for Arsenic (10/31/01)
12. Office of Management and Budget's 2001 Draft Report to Congress on the Costs and Benefits of Federal Regulations (8/14/01)
13. EPA's National Primary Drinking Water Regulations: Arsenic and Clarifications to Compliance and New Source Contaminants Monitoring (5/7/01)
14. DOE Clothes Washer Addendum - Poll Results (12/4/00)
15. EPA and DOJ's Worst Case Scenario Proposal (6/8/00)
16. The U.S. Postal Service's Proposed Delivery of Mail to a Commercial Mail Receiving Agency (4/13/00)
17. The Occupational Safety And Health Administration's Proposed Ergonomics Program Standard (2/25/00)
18. Office of Management and Budget's 2000 Draft Report to Congress on the Costs and Benefits of Federal Regulations (1/19/00)
19. TRI Reporting of Lead and Lead Compounds (12/15/99)
20. EPA's Economic Incentive Program Guidance (12/10/99)
21. Tier 2 Supplemental Notice (12/1/99)
22. Economic Benefit Component of Civil Penalties (9/30/99)
23. EPA's Tier 2 Standards for Vehicle Emissions and Gasoline Sulfur Content (7/23/99)
24. Working Draft of a Proposed Ergonomics Program Standard (6/16/99)
25. DOL's Proposal Governing Helpers on Davis-Bacon Act Projects (6/8/99)
26. Environmental Enforcement and Compliance Assurance Activities (4/16/99)
27. Toxic Release Inventory (4/1/99)
28. Minimum Security Devices and Procedures and Bank Secrecy Act (3/8/99)
29. Advanced Air Bags (12/17/98)
30. Proposal to Issue and Modify Nationwide Permits (Wetlands) (11/30/98)
31. Office of Management and Budget's 1998 Draft Report to Congress on the Costs and Benefits of Federal Regulations (10/16/98)
32. NOx Emissions Trading (6/25/98)
33. National Ambient Air Quality Standard for Ozone (3/12/97)

ADDITIONAL REFERENCES

Browning, Edgar K., and Mark A. Zupan. 1996. Microecononomic Theory and Applications, 5th Ed. Harper Collins [8th Ed.].

Nicholson, Walter. 1978 Microeconomic Theory: Basic Principles and Extensions, 2d Ed. Dryden Press [10th Ed.]

Peltzman, Sam. 1975. "The Effects of Automobile Safety Regulation," Journal of Political Economy, 83: pp. 677-725.

[add a comment]

Add a Comment