You Cannot Tell Whether It Worked: Pest Control as a Credence Good, and the Only Remedy the Experiments Support

1. Introduction: the category this trade belongs to

This journal has argued in several articles that customers cannot evaluate what they are buying here. That is not a complaint about the trade. It is a formal category in economics, with a literature, a set of predictions and experiments testing them.

1.1 Why that matters here

Because the two things customers are told to rely on when choosing a pest control company are reviews and competing quotes. Those are reputation and competition, and they are the two remedies the experiments found do least.

2. The classification

Where the term comes from.

The pioneering paper is by Darby and Karni (1973), who introduced the term credence goods and added this type of good to Nelson's (1970) classification of ordinary, search and experience goods.⁷

The key feature of such markets is the informational asymmetry between expert sellers and consumers, since doctors, mechanics, and legal or financial advisors are typically much better informed than the people buying from them. These markets are ubiquitous in daily life and combined, have a huge size in the overall economy.⁶

2.1 The size of the literature

This is not a fringe corner of economics. The bibliography of a single recent working paper on the subject names work on a theory of fraud and overtreatment in expert markets, on credence goods and fraudulent experts in a leading industrial organisation journal, on when experts cheat and whom they target, on the role of verifiability and commitment, on trust building, and on what drives taxi drivers, published across the Review of Economic Studies, the RAND Journal of Economics, the Journal of Economic Literature and the American Economic Journal.³

Half a century of theoretical and experimental work exists on the problem this trade experiences as an ordinary Tuesday.

2.2 The distinction that defines the category

A search good can be assessed before buying. An experience good reveals itself once used. A credence good does neither, because the buyer never learns what they actually needed.

The question is not whether the service was performed. It is whether the service performed was the right one, and that is the thing the buyer has no way to establish.

3. Why pest control qualifies

Working through the criteria rather than asserting the conclusion.

The diagnosis. A customer generally cannot identify the species, cannot estimate the population, and cannot locate the harbourage. This journal has documented misidentification of spiders, of bite marks, of small flies and of wood damage, each time finding that the correct answer required expertise the household did not have.

The prescription. Given a diagnosis, what treatment is appropriate, at what intensity and over how many visits, is a judgement the customer cannot audit.

The execution. A customer generally cannot observe what product was applied, at what rate, or to which surfaces. Most of the work happens in places they do not go.

3.1 The condition that makes it credence rather than experience

All three of those would still allow evaluation after the fact if the outcome were attributable. Section 4 is why it is not.

4. The outcome problem

The reason this trade sits firmly in the credence category rather than at its edge.

Suppose the pests are gone. That is consistent with an excellent treatment, with an adequate one, with an unnecessary one applied to a problem that was resolving anyway, and with a seasonal decline that would have happened regardless.

Suppose they are not gone. That is consistent with a poor treatment, with a reasonable treatment against a hard case, with reinfestation from outside, and with a customer who did not complete the preparation.

4.1 What that does to learning

Neither outcome identifies the quality of the work. A customer who has used the same company for ten years has ten observations and no information about whether any of it was necessary.

This journal reported that handlers of detection dogs uniformly believed their teams performed at 95 per cent while measured performance averaged 44. That is the same mechanism acting on the seller side: absence of feedback produces confident belief rather than calibrated belief, and it applies to both parties in this market.

5. The three fraud types

The literature's categories, which are more precise than the ordinary word.

Credence goods markets typically suffer from the following types of cheating on consumers: undertreatment, that is providing a quality that is insufficient to satisfy the consumer's needs; overtreatment, that is choosing a higher quality than the surplus maximizing one; and overcharging, that is charging for a higher quality than has been provided.²

5.1 Why the precision helps

These are three different failures with three different remedies, and lumping them together as dishonesty obscures that. Sections 11 and 12 show that the institution which fixes one does not fix the others.

5.2 The asymmetry between them

Undertreatment harms the customer by leaving the problem in place. Overtreatment harms them financially while solving the problem. Overcharging harms them financially with no effect on the problem at all.

Only the first produces the experience of a service that did not work, which is the only signal a customer in this market reliably receives. So the fraud type that is most visible to customers is one of three, and the market's self-correction mechanism can act on that one alone.

5.3 Which we would expect to predominate

On that reasoning, overtreatment should be the commonest form in any credence market with reputational pressure, because it is the type that a reputational mechanism actively rewards: the customer gets their result, pays more than necessary, and leaves a favourable review.

The field measurements in §16 are consistent with that, finding substantial overtreatment alongside overcharging. We flag the expectation as our inference rather than a stated finding.³

6. Undertreatment here

The first category, applied.

A perimeter spray sold against an interior cockroach population. A single visit where the biology requires two. A treatment of one unit in a building where the infestation spans several.

6.1 Why it is attractive to a seller

It costs less to deliver and is invisible at the point of sale. The customer experiences a visit, receives an invoice, and cannot tell that what was needed was different.

6.2 Why it is detected late if at all

The failure presents weeks later as a continuing infestation, which §4 established is consistent with several explanations. A customer will often attribute it to the difficulty of the pest rather than to the adequacy of the service, and frequently that attribution will be correct.

7. Overtreatment here

The second category, which this journal has repeatedly documented without naming it as such.

Whole-structure treatment where one room was involved. A recurring contract maintained after the problem has ended. Chemical application where exclusion was the indicated measure. Environmental spraying for head lice, which this journal argued should not be sold at all.

7.1 The awkward feature of this category

Overtreatment is not obviously harmful to the customer's problem. The pests do go away. It is defined by cost rather than by outcome, which is why the customer has no experience that would reveal it.

It also has costs beyond the invoice: unnecessary pesticide in an occupied building, and the non-target exposures this journal described for pollinators and for predatory wildlife.

7.2 The category that most resembles ordinary practice

We would say honestly that the line between overtreatment and cautious practice is not sharp. A technician who treats a wider area because the infestation may have spread is exercising judgement, not defrauding anyone, and the same action looks identical from outside.

That ambiguity is a feature of credence goods rather than a defect in the definition, and it is why the literature reaches for institutional remedies rather than for detection of intent.

8. Overcharging here

The third category, and the most straightforward.

Billing for a product not applied, for a visit not made, for a duration not worked, or for a higher grade of service than was delivered.

8.1 Why this one is different

Unlike the first two, overcharging concerns a fact rather than a judgement. Either the second visit happened or it did not.

That is what makes it, in principle, the type that verifiability should solve, and §12 reports that the experiments did not find that it does.

9. The two proposed remedies

The framework's central result before any experiment was run.

A survey organizes the different assumptions underlying these papers and provides a unifying framework, whose model shows that any of the two following conditions are sufficient for the efficient provision of credence goods. The first is liability, meaning that the expert is required to provide a good that satisfies the customer's needs. Liability prevents the problem of underprovision, but not necessarily the problems of overprovision and overcharging, and in theory the latter problems are solved by the choice of an appropriate price structure.⁴

Verifiability means that consumers can observe what was provided.¹

9.1 What each is in practice

Liability, in this trade, is a guarantee expressed as an outcome. The company is obliged to deliver a result that solves the problem, and bears the cost of returning until it does.

Verifiability is documentation: a service report stating what was applied, where, at what rate, and what was observed.

10. The experiment

How the two were tested.

Researchers studied in a large experiment with 936 participants the determinants for efficiency in credence goods markets, examining the role of liability, verifiability, reputation, and competition.¹

10.1 Why an experiment rather than field data

Because in the field nobody knows what the customer actually needed, which is the whole problem. A laboratory market can assign a true need and then measure whether the seller met it.

This journal has praised that design logic repeatedly. Here it is being used to measure honesty rather than efficacy, and it is the only way to get a clean answer.

11. The result on liability

The finding that survives.

Liability has a crucial effect.¹

Liability implies the requirement that sellers provide a quality that is sufficient to solve the consumer's problem, and thus prevents undertreatment, but does not preclude overtreatment and/or overcharging.¹

11.1 Why it works

Because it changes what the seller is selling. Under an obligation to solve the problem, an inadequate treatment is not a saving; it is a commitment to return at the seller's expense.

That is the mechanism this journal identified in the bed bug pricing article, where an outcome contract aligns incentives while a per-visit contract does not. The credence goods literature is the formal version of the same argument.

11.2 What it does not fix

Overtreatment and overcharging. A company obliged to solve the problem still has an incentive to solve it with more product and more visits than necessary, and to bill for them.

12. The result on verifiability

The finding that surprised the theorists.

While theory predicts that liability or verifiability yield efficiency, verifiability has at best a minor effect.¹ Other work reports that liability clauses, preventing undertreatment, are key for the efficient provision of credence goods, whereas verifiability, preventing overcharging, fails to improve efficiency, although in theory it should.²

12.1 The implication for documentation

A detailed service report is verifiability. The experiments suggest it does less to make the market work than the theory predicts it should.

We are not reading that as a reason to stop documenting. Documentation has other uses, including the resistance and monitoring arguments this journal has made elsewhere. But it is a reason not to treat a thorough report as evidence that the right work was done, because the report records what was done rather than what was needed.

12.2 Why it might underperform

Verifiability addresses overcharging, which is the type concerning facts. It does nothing about whether the treatment was appropriate, which is the judgement the customer cannot make.

A perfectly documented unnecessary treatment is fully verifiable and still wrong. That is our reading of why the effect is small rather than a mechanism the papers state.

13. The result on reputation

The finding most at odds with how this market actually operates.

Allowing sellers to build up reputation has little influence, as predicted.¹

A survey concludes that the effects of an opportunity to build up a reputation as a reliable seller are generally weak and only present in an environment where there are no institutional remedies against fraudulent behavior of sellers.⁴

13.1 What that does to the review economy

Online reviews are a reputation mechanism, and reputation mechanisms are the ones found to have weak effects.

The reason follows from §4. A reviewer is reporting an experience, and in a credence good the experience does not reveal quality. A customer whose problem returned may leave a poor review for good work, and a customer who received an unnecessary treatment will leave a good one.

13.2 The qualification

Reputation has some effect where no other remedy operates.⁴ In a market with no guarantees and no oversight, it is better than nothing.

The honest statement is that it is a weak instrument being asked to do the whole job.

14. The result on competition

The finding with the clearest consumer implication.

Seller competition drives down prices and yields maximal trade, but does not lead to higher efficiency as long as liability is violated.¹

The same paper notes that competition operates without substantially increasing efficiency where neither liability nor verifiability applies, having only redistribution effects, shifting the gains from trade from sellers to consumers.⁷

14.1 What three quotes achieves

A lower price. That is a real benefit and we are not dismissing it.

What it does not achieve is a better match between what is done and what was needed. Comparing three quotes for a treatment nobody has established is necessary selects the cheapest version of a possibly unnecessary job.

14.2 The direction competition can push

There is a worse possibility the result hints at. If buyers compare on price and cannot assess quality, then the pressure falls on the dimension they cannot see, and undertreatment is the cheapest way to meet a lower price.

That is our inference rather than a finding of the experiment, which reports competition as efficiency-neutral rather than efficiency-reducing.¹ We raise it because it is the mechanism by which a price-competitive market in a credence good could get worse rather than merely failing to improve.

15. Does this hold for real experts

The obvious objection to a laboratory result about honesty, which has been tested.

Researchers compared the behavior of university students and car mechanics as sellers in an experimental credence goods market, finding that both subject pools react qualitatively in the same way to changes in the informational and institutional framework, and specifically that liability, but not verifiability, has a strong effect on the likelihood of interaction and overall market efficiency.²

15.1 Why this matters

The result is not an artefact of inexperienced subjects in an abstract task. Practising professionals in a real expert trade responded to the same institutions in the same direction.

15.2 The reading we would resist

That mechanics are dishonest. The finding is that professionals respond to institutional structure the way the model predicts anyone would, which is a statement about incentives rather than about character.

We would apply the same reading to our own trade, and §21 does.

16. What the field measurements found

How much fraud actually occurs where it can be measured.

In the market for taxi services, researchers quantified the extent of fraudulent behaviour by taxi drivers in the form of overtreatment, taking unnecessary detours in order to increase the total fare, and overcharging, artificially increasing the fare by charging fake additional charges or manipulating the taximeter, reporting that about 45 per cent of the taxi drivers in their sample took at least a short detour, while 11 per cent engaged in overcharging, requesting a higher price than the one corresponding to the distance actually driven.⁵

Several natural field experiments document the prevalence of fraud in computer repair markets, with robust evidence for fraud, mostly in the form of overcharging in the working time dimension, but also some evidence for overtreatment, replacing parts that did not need replacement, in two separate national markets.⁵

16.1 Why the taxi market is measurable and this one is not

A taxi route has a correct answer that the researcher knows in advance. A pest control job does not, which is precisely why no equivalent field measurement exists for this trade.

We would expect the fraud rates here to be unmeasurable rather than low, and we flag that as an inference. The absence of evidence about this trade is a consequence of the asymmetry rather than evidence of its absence.

16.2 The insurance finding

One study investigated how informing a repair shop that an insurance agency will cover the repair costs affects the provision and charging behaviour of repair shops.⁵

That is the structure of any pest control work paid by a landlord, a property manager or an insurer rather than by the occupant, and the question it raises is one this trade should be comfortable having asked.

17. The informed customer

The one thing that changes the category itself.

In theory, repair services are credence goods for consumers who are unable to self-diagnose the problem, but ordinary goods for other consumers.⁶

17.1 What that implies

Knowledge does not merely help the buyer negotiate. It moves the transaction out of the credence category and into one where ordinary market discipline works.

That is the strongest justification we have for this journal existing, and we recognise the self-serving shape of that sentence.

17.2 The limit

A field experiment in which customers offered a vague but correct guess about their fault found that given that the conjecture is vague, it still leaves room for a dishonest expert seller to overtreat or overcharge, so that whether prices fall remains an empirical question.⁶

Partial knowledge is not the same as the ability to evaluate. A customer who knows roughly what is wrong can still be sold the wrong remedy for it.

18. Separating diagnosis from treatment

An institution the literature names which this trade almost never uses.

A review of the field covers, among the features of market structure and institutional environment, separation of diagnosis and treatment, liability, verifiability, reputational concerns, competition between experts and second opinions.⁸

18.1 Why separation addresses the root

An expert who diagnoses and also sells the treatment has an interest in the diagnosis. An expert who only diagnoses does not.

This journal reached that conclusion independently in the detection dog article, arguing that the party verifying an alert should not be the party selling the treatment. The literature had the institution named and studied, and we had not read it.

18.2 What it would look like here

Paid inspection by a party that does not perform treatment, with the report taken to whoever does the work.

The obstacle is cost. A household facing a treatment bill is reluctant to pay a second party for an opinion, which is the same undersupply of detection this journal identified in the bed bug pricing article. The institution that would help is the one nobody wants to buy.

18.3 The places it already exists

Pre-purchase building inspections are a partial version in some jurisdictions, where an inspector reports on a property without tendering for the repairs. Home inspection generally works that way.

Whether an equivalent could be sustained for pest work depends on whether enough buyers exist to support inspectors who do not treat, and the free inspection that most companies offer as a route to a quote is the competing product. A free diagnosis from the party selling the treatment will always undercut a paid one from a party who is not.

18.4 The free inspection as an economic object

Worth naming plainly, because this trade offers it universally and we do too. A free inspection is a diagnosis given away in order to sell a treatment, which is the exact opposite of the separation the literature recommends.

That does not make it dishonest. It makes the incentive structure explicit, and a customer should understand that they are receiving an opinion from the party who benefits from one answer to it.

19. What this says about guarantees

Pulling the threads together, because this is the actionable part.

Liability is the remedy with the crucial effect.¹ A guarantee expressed as an outcome, rather than as a number of visits, is liability.

19.1 The form of words that matters

A guarantee of further visits within a period is not liability if the obligation is discharged by attending. A guarantee that the problem will be resolved, with the company bearing the cost of whatever that takes, is.

The difference is whether the seller's obligation ends with an action or with a result.

19.2 Why the market supplies little of it

This journal's bed bug economics article found that outcome pricing aligns incentives and is suppressed by competition, because a company offering it prices above one that does not and loses the comparison.

Section 14 explains why that dynamic is not self-correcting. Competition drives prices down without improving efficiency, so a market competing on price will select against the one institution that works.

19.3 What follows

That the remedy has to come from somewhere other than customer choice. Regulation requiring outcome obligations in certain classes of work, or an insurer or landlord large enough to impose the term, are the candidates, and we are not aware of either operating here.

20. What a customer can do

Ask what the guarantee obliges. Attendance or resolution. Liability is the remedy with the effect.¹

Treat reviews as weak evidence. Reputation had little influence.¹

Use three quotes for price, not for quality. Competition redistributes rather than improving.¹⁷

Ask what was needed, not what was done. Verifiability covers the second question.²

Get a second opinion on the diagnosis. Named in the literature.⁸

Learn enough to change the category. Self-diagnosis makes it an ordinary good.⁶

21. What this says about us

This section is the point of publishing the article, and it would be dishonest to write the preceding twenty and omit it.

APC Exterminators is an expert seller in a credence goods market. Every incentive described in §5 operates on us. We diagnose the problem, we recommend the treatment, we perform it, and we bill for it, with no party in the transaction able to check any step.

21.1 The use we are not making of this

An article about fraud in expert markets could easily be an attack on competitors. It is not one, and we have named no company and alleged no conduct.

The literature's finding is about structure rather than character.² Nothing here suggests that pest control operators are worse than other people, and the car mechanic experiment found professionals behaving like everyone else.

21.2 The uncomfortable implication we accept

Section 19 says the remedy is an outcome obligation, and a company writing that should expect to be asked whether it offers one. That is a fair question to put to us and to anybody else, and it is the question this article is recommending that customers ask.

21.3 What we think the honest position is

That a seller in this market cannot demonstrate honesty by asserting it, because assertion is exactly what an asymmetry permits. What a seller can do is prefer the institutions that constrain them: outcome terms rather than visit counts, declining work that is not indicated, and publishing the reasoning so that customers can evaluate the argument even when they cannot evaluate the treatment.

That last one is the weakest of the three and we should not overclaim it. A published argument is still an assertion by an interested party, and a customer who cannot assess a treatment is not obviously better placed to assess an essay about treatments. What it does offer is something checkable: the sources are named, so the reasoning can be audited by someone other than us even when the work cannot be.

22. Limitations and open questions

No study in this literature examines pest control. The markets measured are taxis, computer repair, car repair, medicine and financial advice. The application to this trade is ours.

The core results are experimental. A laboratory market assigns a known need, which is what makes measurement possible and also what makes it abstract.¹

The car mechanic replication is one study. It establishes qualitative agreement between student and professional subjects rather than equivalence.²

Several findings reach us through surveys and overviews. The taxi and computer repair figures, and the 1973 and 2006 framework papers, come from review articles and topic summaries rather than from the primary papers read in full.⁴⁵⁷

The verifiability result is about market efficiency. It does not establish that documentation is useless, and §12.1 says why we would keep it.

We have not addressed regulation. Manitoba licensing, label law and consumer protection all constrain conduct here, and none of it appears in this analysis. A licensed trade is not an unregulated market and the model does not account for that.

Sections 3, 4, 6, 7, 8, 12.2, 16.1, 19.2 and 21 are our reasoning. The qualification of pest control as a credence good, the outcome-attribution argument, the three worked examples, the explanation for the verifiability result, the unmeasurability inference and the application to our own position are ours rather than sourced findings.

Our commercial position. Stated at length in §21 rather than in a footnote here. A reader should note in particular that §17.1 gives this journal a self-serving justification, and that we have said so in the text.

23. Conclusion

Economists call a service a credence good when the buyer cannot assess its quality even after consuming it, a category introduced in 1973 and formalised in a 2006 survey, with medicine, repair, legal services and financial advice as the standard examples.⁶⁷ Pest control belongs there, because the customer cannot establish the diagnosis, cannot audit the prescription, cannot observe the execution, and cannot attribute the outcome.

The literature identifies three fraud types and tests four remedies. A large experiment found that liability has a crucial effect, verifiability at best a minor one, reputation little influence, and competition lowers prices without improving efficiency so long as liability is violated.¹ A replication with practising car mechanics found real experts responding the same way.² Where the answer can be known in advance, as on a taxi route, about 45 per cent of drivers took a detour.⁵

Which leaves customers being advised to do the two things that work least. Reviews are reputation and three quotes are competition, and the experiments found both weak. The one that works is an obligation to deliver a result rather than a visit, which is harder to sell, costs more to offer, and loses the comparison to a company that promises attendance instead. We are one of the companies in that comparison, which is why §21 exists and why the fair response to this article is to ask us what our guarantee actually obliges.

References

1. Dulleck, U., Kerschbamer, R. and Sutter, M. (2011). The Economics of Credence Goods: An Experiment on the Role of Liability, Verifiability, Reputation, and Competition. American Economic Review, 101(2), 526 to 555. Principal source. Used for the statements that credence goods markets are characterised by asymmetric information between sellers and consumers that may give rise to inefficiencies such as under- and overtreatment or market breakdown; for the study of the determinants of efficiency in credence goods markets in a large experiment with 936 participants; for the findings that while theory predicts that liability or verifiability yield efficiency, liability has a crucial but verifiability at best a minor effect, that allowing sellers to build up reputation has little influence as predicted, and that seller competition drives down prices and yields maximal trade but does not lead to higher efficiency as long as liability is violated; and for the definitions that liability implies the requirement that sellers provide a quality sufficient to solve the consumer's problem, thus preventing undertreatment but not precluding overtreatment or overcharging, and that verifiability means consumers can observe what was provided. https://www.aeaweb.org/articles?id=10.1257%2Faer.101.2.526
2. Beck, A., Kerschbamer, R., Qiu, J. and Sutter, M. (2014). Car mechanics in the lab: investigating the behavior of real experts on experimental markets for credence goods. Journal of Economic Behavior and Organization, 108, 166 to 173. Used for the statement that the informational asymmetries prevalent in markets for credence goods invite fraudulent behaviour by sellers, making the search for efficiency-increasing institutions a relevant topic; for the definitions of the three types of cheating on consumers, being undertreatment, that is providing a quality insufficient to satisfy the consumer's needs, overtreatment, that is choosing a higher quality than the surplus maximising one, and overcharging, that is charging for a higher quality than has been provided; for the finding from prior studies that liability clauses preventing undertreatment are key for the efficient provision of credence goods whereas verifiability preventing overcharging fails to improve efficiency although in theory it should; and for the authors' own result comparing university students and car mechanics as sellers, finding that both subject pools react qualitatively in the same way to changes in the informational and institutional framework, with liability but not verifiability having a strong effect on the likelihood of interaction and overall market efficiency. https://www.sciencedirect.com/science/article/abs/pii/S0167268114002467
3. Doing Less for More: Consumer Search and Undertreatment in Credence Service Markets. Working paper. Cited for its bibliography, which we used to identify the principal literature relied on here, including Darby and Karni (1973); Dulleck and Kerschbamer (2006) on doctors, mechanics and computer specialists in the Journal of Economic Literature 44(1); Emons (1997) on credence goods and fraudulent experts in the RAND Journal of Economics; Alger and Salanie (2006) on a theory of fraud and overtreatment in expert markets; Balafoutas, Beck, Kerschbamer and Sutter (2013) on what drives taxi drivers, in the Review of Economic Studies 80(3); Fong, Liu and Wright (2014) on the role of verifiability and commitment in credence goods markets; and Fong, Liu and Meng (2022) on trust building in credence goods markets. https://arxiv.org/pdf/2503.21175
4. Kerschbamer, R. and Sutter, M. The Economics of Credence Goods: a Survey of Recent Lab and Field Experiments. CESifo Economic Studies, 63(1), 1. Used for the account of the Dulleck and Kerschbamer (2006) survey, which organises the assumptions underlying the literature and provides a unifying framework whose model shows that either of two conditions is sufficient for the efficient provision of credence goods, the first being liability, meaning the expert is required to provide a good that satisfies the customer's needs, which prevents underprovision but not necessarily overprovision and overcharging, the latter problems being solved in theory by the choice of an appropriate price structure; and for the conclusion that the effects of an opportunity to build up a reputation as a reliable seller are generally weak and only present in an environment where there are no institutional remedies against fraudulent behaviour of sellers. https://academic.oup.com/cesifo/article/63/1/1/2992734
5. Credence Goods: an overview. Reference topic summary compiling findings from the primary literature. Cited as a compilation rather than a primary report. Used for the account of Balafoutas and colleagues (2013, 2017) quantifying fraudulent behaviour by taxi drivers in the form of overtreatment, taking unnecessary detours to increase the total fare, and overcharging, artificially increasing the fare by charging fake additional charges or manipulating the taximeter, with about 45 per cent of drivers in the sample taking at least a short detour and 11 per cent engaging in overcharging; and for the account of natural field experiments documenting fraud in computer repair markets, including work by Kerschbamer and colleagues (2016) investigating how informing a repair shop that an insurance agency will cover the costs affects provision and charging behaviour, with robust evidence for fraud mostly in the form of overcharging in the working time dimension and some evidence for overtreatment through replacement of parts that did not need replacement, and similar evidence reported for a second national market. https://www.sciencedirect.com/topics/economics-econometrics-and-finance/credence-goods
6. Credence goods markets, online information and repair prices: a natural field experiment. Journal of Public Economics. Used for the statements that markets for credence goods are ubiquitous in daily life, including markets for health care, repair and legal services as well as financial advice and fund management, and that combined these markets have a huge size in the overall economy; that their key feature is the informational asymmetry between expert sellers and consumers, with doctors, mechanics and legal or financial advisors typically much better informed; that in theory repair services are credence goods for consumers unable to self-diagnose the problem but ordinary goods for other consumers; and for the experimental treatment in which a customer offered a vague but correct conjecture about the fault, with the authors noting that because the conjecture is vague it still leaves room for a dishonest expert seller to overtreat or overcharge, so that whether prices fall remains an empirical question. https://www.sciencedirect.com/science/article/pii/S0047272723000737
7. Dulleck, U., Kerschbamer, R. and Sutter, M. (2011), working paper version hosted by Queensland University of Technology. Used for the account of the pioneering paper by Darby and Karni (1973), who introduced the term credence goods and added this type of good to Nelson's (1970) classification of ordinary, search and experience goods, studying how market conditions such as the presence or absence of idle capacities and regulation, and reputation concerns, affect the equilibrium amount of fraud comprising under- and overtreatment and overcharging; and for the statement that competition operates without substantially increasing efficiency where neither liability nor verifiability applies, having only redistribution effects that shift the gains from trade from sellers to consumers. https://eprints.qut.edu.au/75882/1/75882(pub).pdf
8. Credence goods in the literature: what the past fifteen years have taught us about fraud, incentives, and the role of institutions. Journal of Behavioral and Experimental Finance. Used for the scope statement that the review considers various markets for credence goods and evidence on the extent of fraud, and reviews theoretical and empirical contributions on the determinants of seller and consumer behaviour, with topics including informational asymmetries, pro-social motivations and seller characteristics, and features of the market structure and institutional environment comprising separation of diagnosis and treatment, liability, verifiability, reputational concerns, competition between experts and second opinions. https://www.sciencedirect.com/science/article/pii/S2214635020300265