A Letter in the Window: What Posting Hygiene Grades Actually Changed, and the Part of It That Is Now in Dispute

1. Introduction: the instrument that is not an inspection

The previous article on food safety in this journal examined private audits and found the audited party choosing the auditor, setting the scope and paying the bill. Public inspection has none of those problems and a different one: the result goes into a file.

One county changed that by putting the result in the window, and the consequences have been studied more carefully than almost anything else in this field.

1.1 Why a pest control journal covers it

Because pest evidence is a scored inspection item, and a posted grade turns a sighting into a revenue number. Section 20 is the argument and it is the strongest commercial case this journal has found for preventive work in food premises.

2. What actually changed in 1998

The intervention, which is narrower than people assume.

The county introduced hygiene quality grade cards to be displayed in restaurant windows.¹ The grading system existed before the disclosure requirement.⁴

2.1 Inspection was not the variable

Premises were already being inspected and already being graded. What changed was where the grade went.

So nothing in what follows should be read as an argument for more inspection. It is an argument about what happens to an inspection result already being produced.

That makes this an unusually clean natural experiment: the enforcement regime, the standards and the inspectors stayed put, and only the information moved. Everything attributed to the policy is attributable to publication rather than to scrutiny.

2.2 Why that is rare

Most policy changes alter several things at once, which is why attributing an outcome to any one of them is difficult. Here the standards, the inspectors, the schedule and the penalties were held constant and a piece of card was added to a door.

2.3 The study framing

The work examines the effect of an increase in product quality information to consumers on firms' choices of product quality.¹

Which is a general question in economics that happens to have been asked about restaurant hygiene.

3. The three findings

What the original study reported.

Scores up, demand responsive, hospitalisations down.¹ Taken together the authors concluded that the grade cards cause restaurants to make hygiene quality improvements.¹

3.1 The order of the chain

Information becomes visible, demand responds, revenue depends on the grade, and the operator invests to protect the revenue.

Nothing in that chain requires an enforcement action. It is a regulatory outcome produced by customers rather than by inspectors, which is what makes disclosure attractive as an instrument and why §19's warning is worth reading.

It is also the cheapest form of regulation available, since the enforcement is done for free by people who were going to walk past the door anyway.

4. The sorting question

The methodological problem the authors took seriously, and the reason the study became influential.

Fewer people being ill could mean premises got cleaner, or it could mean the same dirty premises lost customers to clean ones. The second would improve health outcomes without any operator changing anything.

The authors provide evidence that this improvement in health outcomes is not fully explained by consumers substituting from poor hygiene restaurants to good hygiene restaurants.¹

4.1 Why sorting would be less interesting

Not uninteresting. A market in which customers can avoid bad premises is better than one in which they cannot, and the avoidance itself has value.

But a policy justified on the basis that it makes food safer is making a claim about the food, not about who eats it.

4.2 How they separated them

By prediction. If the actual health improvement exceeds the predicted sorting effects, it is likely due to actual hygiene improvement by restaurants, and using econometric techniques both effects do in fact contribute to the decrease in food-related-illness hospitalizations.⁵

4.3 Both, not either

Customers moved and premises improved.⁵ That is the answer one would expect and it is worth noting that the study did not claim the whole effect for the more flattering mechanism.

5. Why that distinction matters

For anybody selling improvement rather than reputation.

If disclosure works only by sorting, then the policy redistributes custom and the total quantity of bad hygiene is unchanged. If it works by improvement, the stock of premises actually got cleaner.

5.1 The pest control stake in that

An improvement effect means operators bought something, and preventive pest control is among the things a premises buys to protect a score. A sorting effect means nobody bought anything and the grade simply moved money around.

So the question of mechanism is also the question of whether disclosure generates demand for this trade's services. That framing is ours.

6. What it did to scores

The least contested finding.

Inspection scores increased.¹ The distribution moved sharply: when the programme began 58 per cent of restaurants received an A grade, a number that grew to 83 per cent by 2003.⁴

6.1 What 58 per cent means about the starting position

Before the cards went up, more than four premises in ten were operating below an A grade in a system where the grade was already being awarded.⁴

Nobody was hiding that; it simply was not visible at the door, and that is the whole of the intervention.

6.2 Twenty-five points in five years

That is a large movement in a distribution, achieved without changing the standard being applied or the frequency of inspection.

6.3 The interpretation that should be checked

A rising share of A grades could mean premises improved or that grading became more lenient. The original study's separate finding that health outcomes improved was, in part, what distinguished those, and §11 is why that support is now weaker than it was.

We raise this as our own caution rather than as a criticism in the literature.

7. What it did to revenue

The number that makes the incentive real.

After grade posting, restaurants receiving an A grade experienced revenue increases of 5.7 per cent, other factors held constant; B-grade restaurants had increases of 0.7 per cent; and those with a C grade had declines in revenue of 1 per cent.⁴ A separate account puts it as restaurants obtaining an A relative to a B grade having 5 per cent greater revenue.⁷

7.1 Two figures, one result

A 5.7 per cent increase for an A against a 0.7 per cent increase for a B⁴ is a five point gap, which is the same quantity as the reported 5 per cent greater revenue for an A relative to a B.⁷

The two accounts agree. One reports the effect of each grade against the pre-policy baseline and the other reports the difference between grades.

7.2 The spread is the mechanism

Five points of revenue between an A and a C.⁴ On a premises operating at ordinary hospitality margins, a five per cent revenue swing is a very large share of profit, and it recurs every year rather than once.

Against that, the cost of whatever it takes to hold an A is small and known in advance, which is the calculation §20.3 says becomes available.

7.3 Demand became sensitive

The study's second finding was that consumer demand became sensitive to changes in restaurants' hygiene quality.¹

Before the cards it could not be, because the information was not available. Customers were not indifferent to hygiene; they were uninformed about it, which is the distinction the credence goods article in this journal turned on.

8. The incentive that creates

Stated by the original authors.

Grade cards magnify economic incentives for good-quality hygiene.⁵

8.1 Who the incentive now runs through

Before disclosure, the operator answered to an inspector who visited occasionally. After it, the operator answers to every customer who reads the door, continuously.

The inspector still sets the standard and does the assessment. What changed is who applies the consequence, and there are a great many more of them.

8.2 Magnify rather than create

The incentive existed before: an outbreak is catastrophic, and this journal's article on audits described what happens when one occurs. But a catastrophic low-probability event is a weak driver of routine expenditure.

Disclosure replaces it with a continuous, visible, moderate one. That is a better-shaped incentive even if it is smaller, because it acts every day rather than once a decade, and that reading is ours.

9. The chain restaurant result

A detail with implications for how operators behave.

The introduction of grades also improved hygiene at franchised units in chain restaurants.⁴

9.1 The franchisee incentive problem

A franchisee bears the full cost of a hygiene improvement and shares the reputational benefit with every other unit under the same brand. That is a classic under-investment structure, and it predicts that franchised units will run slightly below what the brand would choose.

A posted grade attaches the consequence to the individual premises rather than the brand, which would correct exactly that. The mechanism is our reconstruction; the source reports only that franchised units improved.⁴

9.2 Why that is interesting

A franchised unit already operates under brand standards, audits and a parent company with a reputation to protect. It is the case where one would expect an additional public signal to add least and it still added something.

Which suggests the internal instruments were not delivering everything they were supposed to, and connects directly to the previous article's finding about what a private audit verifies.

10. The health claim

The finding that made the study famous.

The cards caused the number of foodborne illness hospitalizations to decrease,¹ reported as a 20 per cent decrease.⁷ Summaries of the wider literature give decreases ranging from 13 per cent to 20 per cent.⁴

10.1 Why hospitalisations rather than illness

Because they are recorded. Most foodborne illness never reaches a doctor, let alone a hospital, which this journal's article on audits noted when reconciling two counts of the same outbreak that differed by a factor of thirty.

A hospitalisation series is a small, biased, but consistently collected sample of the underlying quantity, and consistency is what a before-and-after comparison needs.

10.2 The comparison used

Los Angeles, which enacted grading in 1998, against California from 1995 to 1999.²

10.3 Why the number mattered so much

Because disclosure is cheap. A policy that reduces hospitalisations by a fifth without hiring an inspector is the strongest possible argument for the instrument, and it became the standard citation for it.⁶

11. The challenge

The 2019 re-evaluation, published in a leading field journal.

Researchers state that, expanding hospitalization data and collecting new data on mandatorily reported illnesses, we show that this finding does not hold up under improvements to the original data and methodology.²

11.1 Where it was published

The challenge and the reply appeared as paired articles in the same issue of the same journal.²³ That is how a discipline is supposed to handle a disputed result and it is worth saying so before setting out the substance.

11.2 Two changes at once

More data, and a different method. Either alone could account for a different answer, and the abstract attributes the reversal to both.²

12. The outbreak

The first substantive objection.

The largest salmonella outbreak in state history hit Southern California before Los Angeles implemented grading.²

12.1 The scale of the confound

The largest in the state's history,² in the region containing the treated county, immediately before treatment. For a study whose outcome variable is foodborne illness hospitalisations, that is the worst possible thing to find in the pre-period.

12.2 Why the timing is a problem

A comparison of before and after attributes any difference to what happened in between. If the before period contains an exceptional spike in illness, the after period will look better than normal regardless of policy.

The fall from an abnormal peak is not evidence that anything was fixed. Regression to the mean would produce the same graph with no policy at all.

13. The placebo test

The second and more decisive objection.

Placebo tests detect the same treatment effects for Southern California counties, none of which changed restaurant grading.²

The original specification detects a significant effect not only in Los Angeles but also in the rest of Southern California, where no policy change occurred, which the critics describe as indicating that the model was picking up a regional trend rather than a local policy impact. Once the regional trend is accounted for, the specific policy effect for LA becomes statistically indistinguishable from zero, and under robust specifications, the 20 per cent reduction in hospitalizations disappears.⁶

13.1 Why this is the stronger objection

The outbreak in §12 is a reason to doubt the pre-period. The placebo result is a demonstration that the method produces the finding where the policy did not exist.²

One says the data may be contaminated; the other says the instrument reads positive on a blank. That is a different order of problem.

13.2 What a placebo test does

It runs the analysis where the policy did not happen. If the method finds an effect there, the method is detecting something other than the policy.

This journal's article on detection probability made the same argument about a negative result requiring a positive control. A placebo is the mirror image: a negative control.

14. The reply

The original authors' response, published alongside.

Robustness checks support our original conclusion, foodborne hospitalizations, as defined in the original study, declined in Los Angeles County relative to the rest of California after LA adopted restaurant hygiene grade cards in 1998. More precisely, the decline in LA is pronounced against central and Northern California, but insignificant when compared with the rest of Southern California.³

14.1 That is a concession and a defence at once

They accept the specific pattern the critics identified: the effect is not significant against neighbouring Southern California counties.³

Their defence is that the effect is pronounced against the rest of the state, and that the neighbouring comparison is the one that should be doubted.

15. Spillover or confound

The core of the disagreement, in one sentence from each side.

The original authors offer that one possible explanation is that the LA regulation has generated spillovers in Southern California.³ The critics read the same pattern as a regional trend the model mistook for a policy effect.⁶

15.1 The structure of the disagreement

It is about the control group. Every difference-in-differences design rests on the assumption that the comparison area would have followed the same path absent the policy, and the dispute is entirely about which comparison area satisfies that.

15.2 Both readings fit the data

If the policy genuinely spilled over, neighbouring counties are contaminated controls and using them understates the effect. If it did not, neighbouring counties are valid controls and the effect is not there.

The observation is identical under both hypotheses. Distinguishing them requires an argument about mechanism rather than more of the same data.

15.3 What a spillover would require

Chains operating across county lines standardising practice, suppliers serving the whole region, and operators anticipating that grading would spread. All are plausible and none is demonstrated in what we read.

That assessment is ours.

16. How we would hold it

A position, stated as such.

The health effect is unproven rather than disproven. A challenge that survives peer review at a leading journal, and a reply that concedes the neighbouring comparison, together describe a finding that is no longer safe to cite as established.

16.1 What would settle it

Another jurisdiction adopting posted grading, with a comparison area far enough away to be uncontaminated and no regional outbreak in the pre-period. Section 18.3's later evidence on point-of-service grading and outbreak rates is a step in that direction and uses a different outcome measure.⁷

Which is the general answer to a contested finding: not more argument about the original data, but a second natural experiment.

16.2 Why we are not picking a winner

Because we are not equipped to. Evaluating a triple-difference specification against its alternatives requires access to the data and the expertise to judge the modelling, and this journal has neither.

Reporting that two competent groups disagree, and why, is what we can honestly offer. It is also, we would argue, more useful than a confident summary of whichever side happened to be read last.

16.3 Why the dispute is worth an article anyway

Because the contested part is one of three findings, and the other two are not in dispute. Section 17 is what remains.

17. What survives regardless

The findings nobody has challenged.

Inspection scores rose. With the A share moving from 58 per cent to 83.¹⁴

Demand became sensitive to hygiene. Which it could not have been before.¹

Revenue tracked the grade. Roughly five to six points between the best and the worst.⁴

Chain units improved too. Despite existing brand standards.⁴

And operators improved, not just customers. Both sorting and genuine improvement contributed, on the original analysis.⁵

17.1 Why the challenge did not touch them

The re-evaluation concerns the health outcome specifically.²⁶ It does not dispute that scores rose, that demand responded or that revenue tracked the grade.

A reader encountering the headline that the famous finding collapsed could reasonably conclude the whole study was overturned, and that would be wrong.

17.2 Those are behavioural findings

They concern what operators and customers did, measured directly, over a defined period in a defined place. The disputed finding is the one that required linking a policy to a health statistic through several steps and a comparison group.

The further a claim travels from the thing measured, the more assumptions it carries, which is why the revenue result is solid and the hospitalisation result is arguable.

18. The wider disclosure literature

The context that makes this case important beyond restaurants.

Researchers have studied whether disclosures can persuade people to avoid costly loans, reduce credit card fees, and avoid visiting restaurants with poor hygiene. In general, these studies tend to find null results or small benefits from disclosures.⁷

The restaurant case was a notable exception, reported to yield large reductions in foodborne illness hospitalizations.⁷

18.1 Why disclosure usually underperforms

Because it assumes the information reaches a decision, is understood, and outweighs the other things driving the choice. A calorie count competes with hunger; a loan disclosure competes with needing money today.

A letter grade in a window has advantages those lack: it is at the point of decision, it requires no numeracy, and the thing it warns about is something people already care about. That may be why this case looked exceptional. Our reasoning.

18.2 Which explains the stakes

Disclosure as a policy instrument mostly does very little, and the one celebrated success is the one now under challenge. That is why the exchange appeared in a leading journal rather than as a technical footnote.

18.3 The later evidence

More recent work indicates that publicly accessible disclosure systems can improve inspection scores and that point-of-service grading and posting practices are associated with lower foodborne outbreak rates at the jurisdictional level.⁷

Associated with is weaker than caused, and it is a different outcome measure from the contested one. We record it as partial support rather than as resolution.

19. The warning in that literature

A caution we had not anticipated.

Without careful design, disclosure policies supplementing inspection programs may inadvertently undermine the regulatory efforts they were meant to support.⁷

19.1 The general shape of the concern

Disclosure is described as supplementing inspection programmes rather than replacing them,⁷ and the worry is that adding the public layer changes how the underlying layer behaves.

19.2 How that could happen

An inspector aware that a marginal score becomes a public letter grade with revenue consequences faces a pressure that did not previously exist, which could soften scoring at the margin. Section 6.2 raised the same possibility from the distribution.

A grade that is easier to obtain is a grade that carries less information, and the instrument erodes its own signal. That mechanism is our reconstruction of the warning rather than the source's explanation of it.

20. What this means for a pest programme

The application, and the reason this article is in a pest control journal.

Evidence of pests is an inspection item. Where grades are posted, a scored deduction becomes a public letter, and the letter moves revenue by several per cent.⁴

20.1 The scored-item assumption

We are assuming that evidence of pests is an inspection item and that a deduction for it behaves like a deduction for anything else. That is how inspection schedules generally work and we have not found it demonstrated, which §22 records.

20.2 Why that changes the argument for prevention

Most of what this trade sells is insurance against an event the client hopes will not happen, which is the hardest thing to sell and the easiest to defer. A posted grade converts it into protection of a continuously visible asset.

The client is no longer buying the absence of an outbreak. They are buying the letter in the window, which they and every customer can see.

20.3 The return is calculable

An operator can put a figure on the revenue difference between grades and compare it with the cost of a programme, which is an arithmetic this journal has generally found unavailable.

The credence goods article argued that this trade's central problem is a customer who cannot assess what they bought. Disclosure does not solve that, but it supplies a proxy that is public, regular and independently produced.

20.4 What a client should actually be told

That a programme protects a score, that the score becomes public in jurisdictions with posted grading, and that the published revenue effect gives them a number to weigh the cost against. Not that pest control prevents outbreaks, which is true and unmeasurable and therefore unpersuasive.

20.5 The honest qualification

A good grade is not the same as an absence of pests, for exactly the reasons the audit article set out about documentation against outcome. An inspection is a point in time and this journal has spent several articles on what a single observation is worth.

So the grade is a better proxy than the ones previously available, and it remains a proxy. A premises can hold an A and have a developing problem in a void, which is the gap between a scored inspection and an actual population that this journal has now described from four different directions.

21. The Manitoba position

What applies here, which is less than the above implies.

Public health inspection of food premises operates in this province under provincial legislation, and inspection results are published in some Canadian jurisdictions, but we have not established what disclosure regime applies locally, whether results are posted at the premises, or whether any letter grading equivalent exists.

21.1 The variable that matters

Whether the result is at the door or on a website. A grade in the window is encountered by somebody already deciding whether to go in; a database entry is encountered by somebody who went looking.

Section 18.1 argues that point-of-decision placement is probably why this case looked stronger than other disclosure interventions, and if that is right then online publication is a much weaker instrument than a card.

21.2 Why we are not asserting more

Because it would be guesswork. The specific instrument matters enormously to whether any of §20 transfers, and the difference between a result published online and a card in the window is precisely the difference this article is about.

An operator wanting to know should ask their regional health authority what is published, where, and how a member of the public would find it. The answer determines whether §20 is a business case or an interesting fact about California.

22. Limitations and open questions

The central health finding is disputed and we cannot adjudicate it. Stated in §16.1. Both papers appeared in the same journal issue and we have read abstracts and a summary rather than the modelling.²³

The revenue figures come from a secondary account. The 5.7, 0.7 and minus 1 per cent figures are as reported by a policy project summarising the research rather than from the paper itself.⁴

The grade distribution figures are similarly secondary. Fifty-eight to 83 per cent is from the same summary.⁴

One jurisdiction, one period. A county in 1998, with all the specifics that implies about its restaurant market, its inspection regime and its population.

Nothing here is about pests specifically. We have assumed that pest evidence is a scored item and that a deduction for it behaves like any other, and we have not found a study isolating pest-related deductions from other hygiene failures.

The Canadian position is unresearched. Section 21 says so and the gap is a real one for any local reader.

Sections 2.1, 3.1, 5, 6.2, 7.1, 8.1, 9.1, 13.1, 15.1, 15.2, 16, 17.1, 19.1 and 20 are our reasoning. The natural experiment framing, the chain description, the mechanism stake, the leniency caution, the margin argument, the incentive shape, the franchise reading, the negative control comparison, the identical-observation point, the spillover requirements, the position taken, the distance-from-measurement argument, the reconstruction of the warning and the application to preventive programmes are ours rather than sourced positions.

Our position. Section 20 is an argument that disclosure increases the value of what this trade sells, which is a conclusion in our interest. We have tried to offset that by putting the strongest version of the challenge to the underlying evidence in §11 to §16 rather than summarising it away.

23. Conclusion

The grading already existed. What changed in 1998 was that the result went into the window, and after it did, the share of premises holding an A went from 58 per cent to 83, consumer demand became sensitive to hygiene quality for the first time, and an A rather than a C was worth something like five points of revenue.¹⁴ Even franchised units of chains, already operating under brand standards and private audits, improved.⁴

The famous part, a twenty per cent fall in foodborne illness hospitalisations, is now genuinely in dispute. A re-evaluation found that the largest salmonella outbreak in the state's history preceded the policy and that placebo tests detect the same effect in counties that changed nothing.² The original authors replied that the decline holds against central and northern parts of the state and proposed that the policy spilled over into the neighbouring counties used as controls.³ Both readings fit the same observation, and we are not able to choose between them.

What is not in dispute is the part that matters for anybody keeping a food premises clean. Making the result public changed what operators did, and it did so by attaching money to a letter that every customer walks past. For a trade whose central difficulty is a customer who cannot see what they bought, that is the most useful thing in this article: not a proof that hygiene disclosure saves lives, which is contested, but a demonstration that when quality becomes visible, people start paying for it.

References

1. Jin, G. Z. and Leslie, P. (2003). The Effect of Information on Product Quality: Evidence from Restaurant Hygiene Grade Cards. Quarterly Journal of Economics, working paper record. Principal source. Used for the framing that the study examines the effect of an increase in product quality information to consumers on firms' choices of product quality; for the account that in 1998 Los Angeles County introduced hygiene quality grade cards to be displayed in restaurant windows; for the three findings that the grade cards cause restaurant health inspection scores to increase, consumer demand to become sensitive to changes in restaurants' hygiene quality, and the number of foodborne illness hospitalizations to decrease; for the further evidence that the improvement in health outcomes is not fully explained by consumers substituting from poor hygiene restaurants to good hygiene restaurants; and for the authors' conclusion that these results imply the grade cards cause restaurants to make hygiene quality improvements. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=322883
2. Ho, D. E., Ashwood, Z. C. and Handan-Nader, C. (2019). New Evidence on Information Disclosure through Restaurant Hygiene Grading. American Economic Journal: Economic Policy, 11(4), 404 to 428. doi:10.1257/pol.20180230. Used for the statements that the case of restaurant hygiene grading occupies a central role in information disclosure scholarship; that comparing Los Angeles, which enacted grading in 1998, with California from 1995 to 1999, the original study found that grading reduced foodborne illness hospitalizations by 20 per cent; that expanding hospitalization data and collecting new data on mandatorily reported illnesses shows this finding does not hold up under improvements to the original data and methodology; that the largest salmonella outbreak in state history hit Southern California before Los Angeles implemented grading; and that placebo tests detect the same treatment effects for Southern California counties, none of which changed restaurant grading. https://dho.stanford.edu/wp-content/uploads/AEJ_Published.pdf
3. Jin, G. Z. and Leslie, P. (2019). New Evidence on Information Disclosure through Restaurant Hygiene Grading: Reply. American Economic Journal: Economic Policy, 11(4), 429 to 443. doi:10.1257/pol.20180543. Used for the authors' response that robustness checks support their original conclusion, that foodborne hospitalizations as defined in their study declined in Los Angeles County relative to the rest of California after the county adopted restaurant hygiene grade cards in 1998; for the more precise statement that the decline in Los Angeles is pronounced against central and Northern California but insignificant when compared with the rest of Southern California; and for their suggestion that one possible explanation is that the Los Angeles regulation generated spillovers in Southern California. https://www.aeaweb.org/articles?id=10.1257/pol.20180543
4. Restaurant Hygiene. Transparency policy research project. Secondary account summarising the research, cited as attributed material. Used for the statements that the grading system existed before the disclosure requirement; that after grade posting, restaurants receiving an A grade experienced revenue increases of 5.7 per cent with other factors held constant, B-grade restaurants had increases of 0.7 per cent, and those with a C grade had declines in revenue of 1 per cent; that the introduction of grades also improved hygiene at franchised units in chain restaurants; that studies found significant decreases in foodborne-illness hospitalizations ranging from 13 per cent to 20 per cent; that when the programme began 58 per cent of restaurants received an A grade, a number that grew to 83 per cent by 2003; and for the background figure that foodborne pathogens cause an estimated 48 million cases of illness, 128,000 hospitalizations and 3,000 deaths each year in the United States. https://www.transparencypolicy.net/restaurant-hygiene.php
5. Jin, G. Z. and Leslie, P. (2005). The Case in Support of Restaurant Hygiene Grade Cards. Choices, agricultural economics association magazine. Used for the account of how sorting and improvement effects were separated, namely that if the actual health improvement exceeds the predicted sorting effects it is likely due to actual hygiene improvement by restaurants, and that using econometric techniques both effects were shown to contribute to the decrease in food-related-illness hospitalizations; for the authors' argument that restaurant owners made efforts to improve hygiene after the introduction of grade cards; and for their statement that grade cards magnify economic incentives for good-quality hygiene. https://www.choicesmagazine.org/2005-2/safety/2005-2-02.pdf
6. Summary and analysis of the re-evaluation of restaurant hygiene grading evidence. Research summary service, cited as attributed material. Used for the account that information disclosure is frequently championed as a low-cost mechanism to nudge consumer behaviour and enhance public health, with the foundational evidence for nearly two decades being the original study reporting a 20 per cent reduction in foodborne illness hospitalizations; for the description of the re-evaluation's critical discovery regarding the timing of a large salmonella outbreak before implementation; for the explanation that the original specification detects a significant effect not only in Los Angeles but also in the rest of Southern California where no policy change occurred, described as a placebo reveal indicating the model was picking up a regional trend rather than a local policy impact; and for the statement that once this trend is accounted for the specific policy effect becomes statistically indistinguishable from zero, so that under robust specifications the 20 per cent reduction in hospitalizations disappears. https://lacuna.tiptreesystems.com/work/new-evidence-on-information-disclosure-through-restaurant-hygiene-grading/wrk_a3b358432bcf5d2b4af736c92006a94d
7. Repository record of the original study together with citing literature on disclosure policy. Used for the reported finding that restaurants obtaining an A relative to a B grade have 5 per cent greater revenue and that grade cards cause a 20 per cent decrease in foodborne illness hospitalizations not fully explained by consumer substitution; for the summary of the wider disclosure literature, in which researchers have studied whether disclosures can persuade people to avoid costly loans, reduce credit card fees and avoid visiting restaurants with poor hygiene, with these studies in general tending to find null results or small benefits, the restaurant hygiene case being a notable exception reported to yield large reductions in hospitalizations; for the warning that without careful design, disclosure policies supplementing inspection programs may inadvertently undermine the regulatory efforts they were meant to support; and for the note that more recent evidence indicates publicly accessible disclosure systems can improve inspection scores and that point-of-service grading and posting practices are associated with lower foodborne outbreak rates at the jurisdictional level. https://www.researchgate.net/publication/24091818_The_Effect_Of_Information_On_Product_Quality_Evidence_From_Restaurant_Hygiene_Grade_Cards