do common standards promote competition? a market experiment

Do common standards promote competition?A market experiment.

Paolo Crosetto and AlexiaGaudeul

INRA, Université de Grenoble, France andFriedrich-Schiller-Universität Jena, Germany

June 26, 2014Industrial Organization: Theory, Empirics and

ExperimentsAlberobello, Italy

1 of 42

Introduction (1)

• Market experiment where firms can choose not only theirprice but also whether to adopt a common standard.

◦ Extension of the standard model of competition withdifferentiated products (Perloff and Salop, 1985).

◦ Part of the large family of models with two types ofconsumers, naive and savvy (Salop and Stiglitz, 1977).

• Common standard offers are favored by a portion of theconsumers (asymmetric dominance, attraction effect,Huber et al., 1982; Huber and Puto, 1983).◦ We vary how many and how much this portion of consumers

prefer common standard offers (based on Crosetto andGaudeul, 2012).

2 of 42

Introduction (2)

• We find divergent effects in treatments with fullinformation about competitors and in those with noinformation.

◦ In treatments with full information:

• Early phases with strong competition and frequent adoption of acommon standard are followed by later phases with frequentcollusion.

• Welfare decreases as the portion of savvy consumers increases.

◦ In treatments with no information:

• Firms are led to adopt the common standard more often as theportion of savvy consumers increases.

• This leads to an improvement in welfare for all consumers.

3 of 42

Main idea and references (1)

• Main idea:

◦ We test the belief in the self-regulating nature of competitivemarkets.

◦ Will firms choose to compete head-on throughstandardization or employ obfuscatory tactics by avoiding theuse of common standards?

• Other experiments: Kalayci (2011); Kalayci and Potters(2011); Shchepetova (2012); Sluijs et al. (2011).

• Empirical work: Célérier and Vallée (2013); Ellison andEllison (2009); Hossain and Morgan (2007); Wenzel(2013).

4 of 42

Main idea and references (2)• Consumer side: spurious complexity, obfuscation,

transparency, shrouding, confusion, consumer protection,(soft) paternalism.◦ References: Carlin (2009); Chioveanu and Zhou (2013);

Ellison and Wolitzky (2012); Gabaix and Laibson (2006);Gaudeul and Sugden (2012); Piccione and Spiegler (2012);Sitzia and Zizzo (2009); Wenzel (2014).

• Firms side: collusion, industrial organization, competition,oligopoly, standardization and compatibility.◦ References: Aoyagi and Fréchette (2009); Boone et al.

(2012); Bruttel (2009); Davis (2011); Dufwenberg andGneezy (2000); Dugar and Mitra (2009); Huck et al. (2000,2004); Keser (1993, 2000); Wenzel (2014); Wright (2013).

5 of 42

A model of competition with shrouding(1)

• Three firms, A, B and C.

◦ Each firm i choose between its own standard and standard A.◦ Firms set their own prices pi , independently and without

knowing the choice of others.

• Three types of consumers corresponding to each of thethree firms in the market.

◦ Value for the goods is v + εij with i the type of the consumer(A, B or C) and j the label of the firm (A, B or C).

◦ εij takes the value e > 0 if i = j, 0 else.

6 of 42

A model of competition with shrouding(2)Table: Distribution of perceived value by consumers for the goodsof firms on the market.

Value for the good of︸ ︷︷ ︸Firm A Firm B Firm C

Consumers of {Type A v + e v vType B v v + e vType C v v v + e

7 of 42

A model of competition with shrouding(3)• e is to be interpreted as a bias for a specific firm◦ might arise out of genuine preference for its good,◦ or result of a mistake in the consumer’s assessment of the

value of the product of a firm.

8 of 42

A model of competition with shrouding(4)

• (1− µ)% of consumers are “naive” and choose based onv + εij − pi .

• µ% of consumers are “savvy” and buy based on the valueof

v + εij(1− CSi)− pi(1 + (1− CSi)X) (0.1)

• CSi = 1 if firm i adopted a common standard (“CS”), 0else. X > 0 measures the penalty on non-CS offers.

9 of 42

A model of competition with shrouding(5)Example

• Firms A and B chose standard A, Firm C chose standard C.• A savvy consumer of type C chooses to buy from

argmax [v − pA, v − pB, v + e − pC(1 + X)].• Maintains a preference for firm C (measured by e).• Counterbalanced by preference for common standard

offers: a penalty of X is applied to price pC.

10 of 42

A model of competition with shrouding(6)

• Modeling inspired by patterns of consumer choice insimilar tasks.

• See Crosetto P. and Gaudeul A. (2011): Do ConsumersPrefer Offers that are Easy to Compare? An ExperimentalInvestigation, Jena Economic Research Papers 2011-044.

11 of 42

Example with no common standard (1)

• Compute the quantity (surface).• Divide price by quantity.• Buy the cheapest product, keep what remains of budget.12 of 42

Example with a common standard (1)

13 of 42

Treatments (1)

Table: Treatments in the experiment, with labels

µ0% 10% 20%

Limited information X 10%L0

L11 L1220% L21 L22

Full information X 10%F0

F11 F1220% F21 F22

14 of 42

Best response dynamics (1)

15 of 42

Simulations (1)

Table: Simulated results, effective price (mean, sd) and % ofperiods with a CS.

µ0% 10% 20%

X 10% 1.63 (0.34)NA

1.62 (0.34)41%

1.56 (0.35)42%

20%1.59 (0.34)

40%1.53 (0.36)

45%

16 of 42

The experiment (1)

• Run in November and December 2013 at the laboratory ofthe Max Planck Institute of Economics in Jena.

• 300 subjects over 10 sessions, each with 30 subjects.• Each session lasted about 1 hour and 30 minutes overall

and participants earned 12 euros on average.• Each subject matched three times with different market

players (perfect stranger matching).• Each matching lasted several periods, random termination

time.

17 of 42

The experiment (2)

18 of 42

The experiment (3)

19 of 42

The experiment (4)

20 of 42

Some patterns (1)

21 of 42

Some patterns (2)

22 of 42

Some patterns (3)

23 of 42

Some patterns (4)

24 of 42

Experimental findings (1)

1. Average revenues for firms by treatment.2. Prevalence of collusion.3. The link between collusion and non-adoption of a

common standard.4. The link between competition and adoption of a common

standard.5. Welfare analysis.6. Individual differences in the strategies of firms.7. Robustness tests.

25 of 42

Experimental findings (2)

• Only when collusion was difficult to sustain (limitedinformation) was there any improvement in welfare forconsumers overall as µ and X increased.

• No improvement in case with full information, due to twocounter-acting factors:

◦ Periods with a CS became more competitive and/or morefrequent as µ and X increased, but

◦ Periods with collusion and thus higher prices also becamemore frequent.

26 of 42

Revenues (1)

Table: Experimental results, effective price (mean, sd) and % ofperiods with a CS.

µ0% 10% 20%

Limitedinformation X 10% 1.76 (0.62)

40%

1.71 (0.57)48%

1.62 (0.66)54%

20% 1.67 (0.55)57%

1.66 (0.60)68%

Fullinformation X 10% 1.86 (0.85)

33%

1.95 (0.83)42%

2.05 (0.90)44%

20% 1.77 (0.83)40%

2.14 (1.06)37%

27 of 42

The prevalence of collusion (1)

DefinitionWe say a firm is colluding if it could have increased its profitand lowered the profit of at least another firm by changingits decisions given the decisions of its competitors in a givenperiod.

• This is a reasonable definition IF collusion is established(individual deviation).

• As long as one firm did not conform to collusion, collusionis not established.

28 of 42

The prevalence of collusion (2)

Table: Experimental results, % of periods with all firms colludingand price during those periods.

µ0% 10% 20%

Limitedinformation

X 10% 2.1942%

2.0256%

2.0153%

20%1.9852%

1.8469%

Fullinformation

X 10% 2.4643%

2.2960%

2.3370%

20%2.2847%

2.5165%

29 of 42

The prevalence of collusion (3)

• More periods with collusion when µ > 0 than when µ = 0.• Because more opportunities to undercut if µ > 0: not

taking advantage of this is deemed collusion.• Under limited information, firms appear as likely to

collude but that collusion is less effective

30 of 42

Collusion and not adopting a CS (1)

Table: Correlation between not adopting a CS and collusion.

Firms colluding0 1 2 3

Limitedinformation

Firmswith CS

0 1% 1% 20% 25%2 0% 2% 16% 23%3 0% 1% 4% 6%

Fullinformation

Firmswith CS

0 1% 2% 23% 35%2 0% 1% 12% 17%3 0% 0% 3% 5%

31 of 42

Collusion and not adopting a CS (2)

• Close link between collusion and not adopting a CS sinceadopting a CS makes sense only if one wishes to undercutanother firm.

32 of 42

Competition and adoption of a commonstandard (1)

Table: Mean effective prices, by number of CS offers in the marketand by treatment.

µ0% 10% 20%

Limitedinformation X 10% 1.83; 1.66; 1.66

1.87; 1.58; 1.38 1.81; 1.53; 1.0820% 1.80; 1.58; 1.56 1.87; 1.59; 1.48

Fullinformation X 10% 1.88; 1.82; 1.79

2.18; 1.66; 1.42 2.37; 1.71; 1.4020% 1.88; 1.64; 1.53 2.47; 1.67; 1.37

33 of 42

Competition and adoption of a commonstandard (2)

• Adoption of a CS was associated with punishment periods• Periods without a CS were periods of tacit collusion.• Not adopting a CS may have been seen/used as a signal

that one wished to collude.• The availability of a CS thus provided a coordination

mechanism that supported higher levels of collusion in fullinformation treatments.

34 of 42

Welfare analysis (1)

Table: Price paid by savvy and by naive consumers, by treatment.

µ0% 10% 20%

Limitedinformation

X 10% 1.711.76

1.641.72

1.541.64

20%1.571.68

1.541.69

Fullinformation

X 10% 1.821.86

1.881.95

1.982.06

20%1.701.78

2.082.15

35 of 42

Welfare analysis (2)

• Under limited information, savvy consumers are the oneswho derive most benefits from standardization

• Under full information, savvy consumers suffer less thannaive consumers, but their own existence makes their ownsituation worse.

• The more savvy consumers there are and the stronger aretheir preferences, the worse they fare under fullinformation.

36 of 42

Analysis of strategies from the feedback ofsubjects (1)

• We collected a wide range of feedback from subjectsregarding the strategy they followed in the game as wellas information regarding their level of risk aversion,perception of fairness and trust in others.

37 of 42

Robustness checks (1)

• Consider only phase 2 and 3.• Consider only the first 9 periods in each phase to control

for time effect.• Consider only intermediary periods to control for initial

adjustments but also avoid end game effects.• Control for group and individual effects with panel

regressions.

38 of 42

Conclusion (1)

• Firms understand the benefit of not adopting the commonstandard and are able to collude in shrouding their offers.

• This effect plays out when firms can see prices andstandards of other firms.

• Paradoxical result whereby being able to choose to makeprices transparent to consumers could help collusion.◦ Not choosing a common standard served as a signal that one

wished to make peace with others.◦ Having more savvy consumers made the punishment phases

harder.

39 of 42

Bibliography (1)

Aoyagi, M. and G. Fréchette (2009). Collusion as public monitoring becomes noisy: Experimental evidence. Journal ofEconomic Theory 144(3), 1135–1165.

Boone, J., M. J. L. Aylwin, W. Müller, and A. R. Chaudhuri (2012). Bertrand competition with asymmetric costs:Experimental evidence. Economics Letters 117, 134–137.

Bruttel, L. V. (2009). Group dynamics in experimental studies—The Bertrand Paradox revisited. Journal of EconomicBehavior & Organization 69(1), 51–63.

Carlin, B. I. (2009). Strategic price complexity in retail financial markets. Journal of Financial Economics 91(3), 278–287.

Célérier, C. and B. Vallée (2013). What drives financial complexity? A look into the retail market for structured products.HEC Paris Research Paper No. FIN-2013-1013.

Chioveanu, I. and J. Zhou (2013). Price competition with consumer confusion. Management Science 59(11), 2450–2469.

Crosetto, P. and A. Gaudeul (2012). Do consumers prefer offers that are easy to compare? An experimental investigation.Working Paper, GAEL-INRA Grenoble and Friedrich-Schiller-Universität, Jena.

Davis, D. (2011). Behavioral convergence properties of cournot and bertrand markets: An experimental analysis. Journalof Economic Behavior & Organization 80(3), 443–458.

Dufwenberg, M. and U. Gneezy (2000). Price competition and market concentration: An experimental study. InternationalJournal of Industrial Organization 18, 7–22.

Dugar, S. and A. Mitra (2009). The size of the cost asymmetry and Bertrand competition: Experimental evidence.http://dx.doi.org/10.2139/ssrn.1328662.

Ellison, G. and S. Ellison (2009). Search, obfuscation, and price elasticities on the internet. Econometrica 77(2), 427–452.

Ellison, G. and A. Wolitzky (2012). A search cost model of obfuscation. The RAND Journal of Economics 43(3), 417–441.

40 of 42

Bibliography (2)

Gabaix, X. and D. Laibson (2006). Shrouded attributes, consumer myopia, and information suppression in competitivemarkets. The Quarterly Journal of Economics 121(2), 505–540.

Gaudeul, A. and R. Sugden (2012). Spurious complexity and common standards in markets for consumer goods.Economica 79(314), 209–225.

Hossain, T. and J. Morgan (2007). ...plus shipping and handling: Revenue (non) equivalence in field experiments oneBay. The B.E. Journal of Economic Analysis & Policy 6(2).

Huber, J., J. W. Payne, and C. Puto (1982). Adding asymmetrically dominated alternatives: Violations of regularity andthe similarity hypothesis. Journal of Consumer Research 9(1), 90–98.

Huber, J. and C. Puto (1983). Market boundaries and product choice: Illustrating attraction and substitution effects.Journal of Consumer Research 10(1), 31–41.

Huck, S., H.-T. Normann, and J. Oechssler (2000). Does information about competitors’ actions increase or decreasecompetition in experimental oligopoly markets? International Journal of Industrial Organization 18(1), 39 – 57.

Huck, S., H.-T. Normann, and J. Oechssler (2004). Two are few and four are many: number effects in experimentaloligopolies. Journal of Economic Behavior & Organization 53(4), 435 – 446.

Kalayci, K. (2011). Price complexity and buyer confusion in markets. Netspar Discussion Paper 10/2011-082.

Kalayci, K. and J. Potters (2011). Buyer confusion and market prices. International Journal of IndustrialOrganization 29(1), 14–22.

Keser, C. (1993). Some results of experimental duopoly markets with demand inertia. The Journal of IndustrialEconomics 41(2), 133–151.

Keser, C. (2000). Cooperation in symmetric duopolies with demand inertia. International Journal of IndustrialOrganization 18(1), 23–38.

41 of 42

Bibliography (3)

Perloff, J. M. and S. C. Salop (1985). Equilibrium with product differentiation. The Review of Economic Studies 52(1),107–120.

Piccione, M. and R. Spiegler (2012). Price competition under limited comparability. The Quarterly Journal ofEconomics 127(1), 97–135.

Salop, S. and J. Stiglitz (1977). Bargains and ripoffs: A model of monopolistically competitive price dispersion. The Reviewof Economic Studies 44(3), 493–510.

Shchepetova, A. (2012). Strategic limitation of price comparison by competing firms: An experimental study. WorkingPaper, Toulouse School of Economics.

Sitzia, S. and D. J. Zizzo (2009). Does product complexity matter for competition in experimental retail markets? Theoryand Decision 70(1), 65–82.

Sluijs, J. P., F. Schuett, and B. Henze (2011). Transparency regulation in broadband markets: Lessons from experimentalresearch. Telecommunications Policy 35(7), 592–602.

Wenzel, T. (2013). Naive consumers, banking competition, and ATM pricing. Review of Network Economics 12(3),271–286.

Wenzel, T. (2014). Consumer myopia, competition and the incentives to unshroud add-on information. Journal ofEconomic Behavior & Organization 98, 89–96.

Wright, J. (2013). Punishment strategies in repeated games: Evidence from experimental markets. Games and EconomicBehavior 82, 91–102.

42 of 42