Fudging the nudge: information disclosure and restaurant grading.

• Author: Ho, Daniel E.
• Position: IV. Fudging by Noise: New York through VI. Intended and Unintended Effects, p. 616-650

4. FUDGING BY NOISE: NEW YORK

   We begin, again, with some regulatory background about New York's restaurant-inspection system in Section IV.A, focusing particularly on the system as it existed immediately before and after grading was instituted in July 2010. Section IV.B discusses the inspections data, with Appendices A and B detailing data integrity issues and the classification algorithm we developed to address the fact that New York fails to disclose types of inspections. Section IV.C presents results. (Appendix C shows that findings are robust to scoring changes, different types of inspections, the time period or inspection cycles examined, and the role of administrative hearings.)

   1. Regulatory Background

1. Inspections 2005 to 2010

   The Bureau of Food Safety and Community Sanitation (BFSCS) in the City Department of Health and Mental Hygiene (DOHMH) administers the city's food safety program. (202) The Bureau inspects thirty-two different types of facilities (e.g., senior centers, public schools, correctional institutions, apartment window guards (203)), but roughly three-quarters of its inspections are of restaurants. (204) Its 2007 budget was $11.7 million, (205) and its staff consists of roughly 180 full-time positions. (206)

   The frontline employees are the health inspectors ("Public Health Sanitarians"). (207) Qualifications are comparable to San Diego's. Inspectors must pass a civil service exam and possess either baccalaureate degrees with at least thirty semester credits in biological or physical sciences, or associate degrees with twelve semester credits in biological or physical sciences and five years of experience as public health technicians. Starting salaries range roughly from $40,000 to $50,000. (208)

   As in San Diego, substantive violations in New York City largely derive from the state sanitary code, with local officials bearing primary responsibility for enforcement. (209) In contrast to San Diego's longstanding grading system, however, New York's method of scoring inspections is relatively recent. New York first began to use a numerical scoring system (without grades) in March 2003, (210) the procedures for which were formalized in a rulemaking in February 2005. (211) The point scoring system, which has remained largely the same since 2005, aimed in part to "provide a more objective method of evaluating ... public health risks." (212)

   The point scoring system works as follows. Violations are classified as either "critical" or "general" violations. Critical violations are those "more likely ... to contribute to food contamination, illness, or environmental degradation." (213) As of 2005, sixty critical violations fell into one of six substantive categories: administration, food temperature, food source, food protection, facility design, and personal hygiene. Thirty-eight general violations fell into one of seven substantive categories: vermin or garbage, food source, facility maintenance, documentation, and three tobacco-related categories. Each violation could range in severity ("condition") from I to V; the conditions determined point scores, which ranged from two to twenty-eight points per violation.

   In an initial inspection--a full, unannounced sanitary inspection conducted roughly once a year--inspectors chose whether to cite a particular violation and determined the severity (or condition) of the violation. Table 3 provides examples of common violations. For example, a 2B violation for failure to hold a hot food item at or above 140[degrees]F could be assigned seven to twenty-eight points, depending primarily on the number of food items so held. "Two hot food items out of temperature," such as "8 chicken wings and cooked rice," would constitute a 2B Condition II violation with eight points, while "[t]hree hot food items out of temperature," such as "8 chicken wings, cooked rice and roast beef," would constitute a 2B Condition III violation with nine points. (214) Failure to correct any public health hazard during the course of the inspection would result in an automatic twenty-eight points. (215)

   When an inspection resulted in one or more critical violations or fourteen or more points, inspectors issued a notice of violation. (216) Each violation cited therein would carry a penalty of between $200 and $2,000. (217) An inspection resulting in twenty-eight or more points was considered a failed inspection, triggering, "whenever practicable," a "compliance inspection." (218) DOHMH policy was to conduct compliance inspections fifteen to forty-five days after the failed initial inspection. (219) Failure to remedy violations after two such compliance inspections would increase the chance of DOHMH shutting down the restaurant, although in principle a restaurant could be shut down at any time for posing a public health hazard.

   The pre-2010 inspection system contained one additional carrot and stick. The carrot was a "Golden Apple" award issued to any restaurant that, among other requirements, "passed two consecutive annual inspections with no critical violations and fewer than four general violations." (220) The stick was an "Accelerated Inspection Program," which increased the frequency of initial inspections for restaurants that had failed two consecutive initial inspections. (221)

   Between 2005 and 2010, DOHMH made only relatively minor scoring changes to the inspection system. (222) The only notable change, implemented in July 2009, was that DOHMH stopped scoring administrative and documentation violations. (223) Violation 1A in Table 3, for example, would still be cited, but no longer scored.

   2. The 2009 Comptroller Audit

      In 2009, the City Comptroller conducted a performance audit of the restaurant-inspection system for the 2008 fiscal year. (224) The Comptroller interviewed officials and reviewed samples of fifty restaurants that were not inspected during the year, sixty-two restaurants that failed one regular inspection, thirty-nine restaurants that failed three or more consecutive inspections, and twenty-one restaurants in the Accelerated Inspections Program. (225) The Comptroller made numerous findings critical of the system, but, most relevant for our purpose, concluded that "DOHMH did not adequately track its inspectors or supervisors to ensure that inspections were being properly conducted and monitored." (226) Examining the internal DOHMH inspections database, the Comptroller found major limitations to the database. For example, "DOHMH officials provided a list of 194 inspectors," but the "database file identified 280 inspector codes," (227) an unexplained excess of eighty-six inspectors. DOHMH's cryptic explanation for the deviation was the presence of "input errors." (228) More shocking, given that inspectors are ostensibly assigned randomly, was the variation across sixty-seven inspectors who conducted more than one hundred inspections in the year. The average inspection score was 25 points, but the audit uncovered some inspectors with average scores of 15 and others with average scores of 50. (229)

   3. Letter Grading

      In July 2010, as part of the Bloomberg Administration's push for government transparency and digital modernization, DOHMH instituted letter grading. (Mayor Bloomberg's slogan: "In God we trust. Everyone else, bring data." (230)) The primary change was to convert each inspection score into a letter grade, required to be posted in a location visible to passersby. (231) In principle, the conversion was simple: fewer than 14 points resulted in an 'A'; 14-27 points resulted in a 'B'; and 28 or more points resulted in a 'C.' (232)

      In practice, the implementation was more complicated. First, DOHMH introduced a system of "reinspection" for grading purposes. If an initial inspection resulted in fewer than 14 points, a restaurant received an 'A.' If, however, the restaurant scored above 14 points on the initial inspection, DOHMH scheduled a reinspection to occur roughly a month (and "no sooner than 7 days" (233)) after the initial inspection. (234) Such reinspections were, in theory, plenary inspections conducted by a new inspector, the score of which would determine the restaurant's grade. (235) For example, a restaurant scoring 20 points in the initial inspection could be issued a final grade of 'A,' 'B,' or 'C,' depending entirely on the score upon reinspection. As before, restaurants scoring 28 or more points on any inspection were, in principle, subject to compliance inspections. (236)

      Second, the 2010 revision changed the timing of the inspection cycle. For any restaurant receiving 28 or more points on either the initial inspection or reinspection, "[a]n initial inspection commencing a new cycle shall be conducted 90 to 150 days after the" last full inspection. (237) For any restaurant receiving 14 to 27 points on either the initial inspection or reinspection, "[a]n initial inspection commencing a new cycle shall be conducted 150 to 210 days after the" last full inspection. (238) All remaining restaurants received initial inspections roughly once a year. (239)

      Third, restaurants that did not receive an 'A' upon reinspection were provided the option of posting either the final grade (of 'B' or 'C') or a "grade pending" card until a hearing in front...