Differentiation inFood Safety

Ginger Z. JinUniversity of Maryland

(based on joint work withPhillip Leslie at Stanford)

Food Safety = the impact of food intake on health risk

Short run – throw-up, food poisoning hygiene

Long run – obesity, heart attack, diabetes nutrition contents production methods

What Do I Mean by Food Safety?

Two Meanings of Differentiation

Actual differenceinfood safety

A Case Study of Los Angeles Restaurants

Nov. 16-18, 1997 CBS 2 News “Behind the Kitchen Door” January 16, 1998, LA county inspectors start issuing

hygiene grade cards A grade if score of 90 to 100 B grade if score of 80 to 89 C grade if score of 70 to 79 score below 70 actual score shown

Grade cards are prominently displayed in restaurant windows

Score not shown on grade cards

Actual Differentiation

First Cut Major impacts after grade cards (GC)

dramatic increase in hygiene quality decrease in the dispersion of hygiene

quality revenue more responsive to hygiene

grade food-borne illnesses drop 20%

More information less differentiation

Why Differentiate After GC? Information is equal

Different cost to maintain good hygiene Burger, Chinese cuisine, Sushi Bar

Different benefit from good hygiene consumer willingness to pay for good hygiene local competition

Why Differentiate Before GC? Consumers know nothing

no restaurant bothers to maintain good hygiene

pure noise Consumers know everything

restaurants choose to “be dirty” or “be clean” no response to GC

Consumers have lousy information equally lousy everywhere dispersion in the amount of information noise

Depends on the extent of consumer learning chain affiliation

=> possible free-riding for franchisees degree of repeat customers in local region

=> regional clustering in hygiene quality

How Could Information Differ Before GC?

Basic evidence - chain affiliation Before

GC After GC

All restaurants 76.77 89.62 In Zagat 77.43 88.97 Chains 82.5 92.76 Company-owned chains

82.94 92.70

Franchised chains 81.84 92.87

Variation Across Chains Before

GC After GC

com-owned 86.98 94.04

franchised 82.09 94.14

com-owned

81.49 90.83

franchised 78.12 92.04

com-owned

81.09 91.50

franchised 81.78 92.69

Statistically ... chains have better hygiene than independent

restaurants

company-owned chain units have better hygiene than franchised units

better hygiene if a chain has a greater number of units in LA county

better hygiene if a chain has a greater % of units in LA county

Repeat Customers-- Santa Monica before GCUpper 1/3 Lower 1/3

Statistically ... better hygiene in heavy retail districts better hygiene in hotel districts worse hygiene in recreational districts no difference in white-collar employment

districts no difference as to whether competes

with at least one chain in the same census tract

Region clustering before GC

Regional clustering after GC

Statistically ... Significant regional clustering in

information structure

Different information structures lead to different reputation incentives, thus different hygiene quality

Summary - Information Matters! Large impact of GC suggests low degree of consumer

learning for most restaurants before GC No voluntary revelation before GC, although the

inspection records are public Zagat restaurants only slightly better in hygiene Chain affiliation is an effective source of information A small degree of franchisee free-riding Regional differences in the degree of consumer

learning impact hygiene quality for independent restaurants

Bottom line: only 25% “A” restaurants before GC, now is over 80%

1. Why is National Restaurant Association against GC?

2. Why don’t other counties adopt the same GC policy?

Two Remaining Questions

Lessons From Other Markets Voluntary disclosure of HMO quality is

incomplete and provides extra tools for HMOs to differentiate (Jin RAND)

Grade card regulation may lead to patient selection (Dranove et al. JPE) or inspector bias (Jin and Leslie in progress)

Private certifiers have strong incentives to differentiate in grading precision and grading criteria (Jin, Kato and List 2004)