Publications: 84 | Followers: 0

Empirical Research on Sketchy Pricing_ Discussion

Publish on Category: Birds 0

Empirical Research onSketchy Pricing:Discussion
JonathanZinmanDartmouth College, IPA, NBERFTC Drip Pricing ConferenceMay 21, 2012
Terms of Engagement Today
“Sketchy” Pricing ===Multiple margins*questionable (non-)disclosure practicesPedagogical Approach:To panel’s papers: take-aways, not quibblesUse take-awaysto:Highlight state of evidence (vispotential applications)Identify key evidentiary gaps: “cliffhangers”Motivate a new research design that would fill some gapsUse this to highlight how/why policy should support R&D
Ellison and Ellison:Take-awaysand Cliffhangers
Take-aways:Innovation that promotes transparency may also promote obfuscation.Important to analyzemarket outcomes(e.g.,equilibria)Cliffhangers re: longer-run dynamics.Do we see an arms race between transparency engines and obfuscation strategies? What does this look like?If no arms race, why not?
Morwitzand Santana:Take-awaysand Cliffhangers
Take-aways:Drip pricing matters a lot… sometimes upon sometimes. E.g.,Depends on experience. Sometimes.Depends on what included in base. Sometimes.Drip pricing doesn’t(does) affect choices ifmandatory surcharges included in car (airline) base priceCliffhangers:Are there regularities in how consumers respond to information/framing?Not just content, but timing, source, other aspects of context?General challenge for models and applications of “nudging”,debiasing
Busseet al:Take-awaysand Cliffhangers
Average car buyer grasps that both new vehicle price and trade-in value affect net purchase price, and negotiates close to full offset.Is thisoffsetan empirical regularity for(consumer responses to) sketchypricing in auto purchase market?No.Nearopposite pattern holds on other key margin. Car buyers who pay higher margins on car also pay much higher margins on financing (and vice versa).Cliffhangers:Can a single model of consumer choice explain multiple (seemingly disparate) phenomena?What explains equilibrium and whether/how it evolves?Growth (dearth?) of negotiation-free options (Saturn RIP)?Growth (dearth?) of unbundled financing?
(Slightly) Bigger Picture:Bodies of Evidence on Sketchy Pricing
Is sketchy pricing prevalent? Prevalent enough.Does it affect (market) outcomes? Presumably.Does it create worse outcomes? Maybe. (Probably?)Why persist (why doesn’t competition solve)? Don’t really know.Why does it “work”? (cognitive/behavioral channelsvisconsumer decision making). Don’t really know.How “work” (search, upfront choice quality, downstream usage quality)? Don’t really know.How improve outcomes? Don’t really know.Many policy levers (including some less-obvious ones)Does intervention that improves outcomes in partial equilibrium work in general equilibrium? Don’t really know.Enforcement costsCountervailing investments in deceptionOverall evidentiary state:humbling
So where do we go from here?
A research approachSome policy approaches
Research Approach:A Sketch
One way to tackle problems with many moving parts is build theory model and test itA good theory yields distinct, testable predictionsIf those predictions supported can use model for equilibrium/policy analysisExample:Gabaix-Laibson(2006 QJE)Application: creditcards. Interesting economically (if not jurisdictionally to FTC?)Price discriminationMulti-homingIntensive as well as extensive margin
Theory:Gabaix-Laibson(As Applied to Credit Cards)
Base price: printer (contract rate)*(Could also/instead be float, teaser rate)Add-on price: cartridge (penalty fees)*(Could also/instead be contract rate)Some consumers (myopes) don’t infer thatshroudedadd-on prices are high pricesAnd/or they underestimate future use of add-onWhy don’t issuers compete byunshrouding/debiasing? Because it turnsmyopesinto unprofitablesophisticates[Shroudedequilibrium (“curse ofdebiasing”) more stable if:Debiasingcosts higher?Switch costs higher? (“Thanks but no thanks effect”)See alsoHeidhueset al (2012)Importantto develop testable hypotheses re: innovations that would destabilize a shroudedequilibrium]
General Setup for Proposed Test in Credit Card Market
Key pieces of research design:Issuer willing to experiment withdebiasingin its direct marketingOr could be 3rd-party (advice provider, agency)On sample of consumers for whom researcher observes full set of credit card accountsVia issuer’s ability to pull credit reportsFrom consent obtained to do soft pullsFrom participation in a market research panel (a laLightspeed, Mintel) where consumers provide access to account/transaction/solicitation dataTest hypotheses thatunshroudingwill:Change consumer behavior: lower use of add-onBe (weakly) unprofitable for issuerBe unprofitable for issuer’s competitors: when try to steal customers bydebiasing, they simply change behaviorin their existing accounts(Does not) affect competitor shrouding behavior?Effects on shopping/advice engines?
Proposed Research Design:Finer Points
“Treatment” effectiveness on consumer choices largely unknown. Need to test different versions.Focus on different add-ons (contract rate; penalty fees)Information typesCompetitor pricesOwn pricesCosts based on typical usageCosts based on projected usage (“our model predicts you will…”)Cost horizonDirect mail/marketing (dominant channel in card market) is conducive todebiasingresearch. Tight control over content:Cheap to do randomized-control testingWith less worry than usual than information treatments are undone or diluted by high-touch marketing
Models Highlight Rationales for Government-Supported R&D
Underinvestment indebiasinginnovations:subsidize*Takes costly experimentationThat may be unprofitable in expectation, even when socially beneficialNon-excludable even when profitable: public good problem?Also suggest another research design: have 3rd-party disseminatedebiasingstrategies to some suppliers (thru e.g., randomizing rollout timing), track all supplier responsesSharp tests may require outcome data from multiple providers: coordination problemSome innovations may rely on machine-readable data (“smart disclosure”): standardsproblem*Caveat: doesdebiasingR&D help deceptive R&D?
Wrapping Up
Panel papers make important contributionsBut we still have a long way to go (visempirical evidence-based policy)I suggested some meta-strategies for navigation:Empirical research that focuses on theory-testingPolicy levers that focus on supporting R&DAnd also sketched a research design for implementing R&D in the credit card market





Make amazing presentation for free
Empirical Research on Sketchy Pricing_ Discussion