How eective are reputation systems? A closer look into ethnic discrimination on a platform Work in progress

Xavier Lambin and Emil Palikot

Toulouse School of Economics

June 27, 2018

Xavier Lambin and Emil Palikot (TSE) Reputation and entry June 27, 2018 1 / 36 Never accept a lift in a car from a stranger!

⇒ a matching platform has to overcome a signicant incomplete information problem. ⇒ introducing reputation system is key to the success for many matching platforms (Blablacar, , , Couchsurng, eBay...).

Matching with incomplete information

Basic danger rule for kids:

Xavier Lambin and Emil Palikot (TSE) Reputation and entry June 27, 2018 2 / 36 ⇒ a matching platform has to overcome a signicant incomplete information problem. ⇒ introducing reputation system is key to the success for many matching platforms (Blablacar, Airbnb, Uber, Couchsurng, eBay...).

Matching with incomplete information

Basic danger rule for kids:

Never accept a lift in a car from a stranger!

Xavier Lambin and Emil Palikot (TSE) Reputation and entry June 27, 2018 2 / 36 Matching with incomplete information

Basic danger rule for kids:

Never accept a lift in a car from a stranger!

⇒ a matching platform has to overcome a signicant incomplete information problem. ⇒ introducing reputation system is key to the success for many matching platforms (Blablacar, Airbnb, Uber, Couchsurng, eBay...).

Xavier Lambin and Emil Palikot (TSE) Reputation and entry June 27, 2018 2 / 36 Reputation systems build trust

Blablacar/Arun Sundararajan: Digital trust is a historical breakthrough" Trust is the cornerstone of BlaBlaCar Trust built in on-line communities is unlocking the world's sharing potential When provided with the right set of digital trust tools, users of on-line platforms are able to recreate a sense of trust almost comparable to the level of trust in friends

Xavier Lambin and Emil Palikot (TSE) Reputation and entry June 27, 2018 3 / 36 BUT! Do we trust everyone?

Edelman & Luca (2014) on AirBnB: non-black hosts are able to charge approximately 12% more compared to black hosts; black hosts receive a larger price penalty for having a poor location Laouénan & Rathelot (2017) on Airbnb: minority groups charge 3.2% less for comparable listings Farajallah et al (2017) on Blablacar: drivers with an Arabic name set prices e0.29 lower. e8.6 shortfall in revenue.

Xavier Lambin and Emil Palikot (TSE) Reputation and entry June 27, 2018 4 / 36 Objectives of this study

Research question: What is the eect of reputation on outcomes of minority drivers? Outline:

1 Institutional background and data collection

2 Study performance of drivers on a ride-sharing platform across ethnical groups

3 Reduced form: impact of reputation on outcomes of minority drivers

4 Estimation of a career concern model (Holmström (1999))

Xavier Lambin and Emil Palikot (TSE) Reputation and entry June 27, 2018 5 / 36 Literature

Discrimination in collaborative economy

I Farajallah et al (2016), Edelman and Luca (2014), Edelman et al. (2016), Goddard et al. (2015), Ge et al. (2016), Laouénan and Rathelot (2017) Career concerns

I Holmström (1999), Altonji and Pierret (2001) Entry with reputation system

I Spagnolo (2012), Butler et al. (2017), Kovbasyuk and Spagnolo (2017) Reputation systems

I Roger and Vasconcelos (2014), Nosko and Tadelis (2015), Bar-Isaac and Tadelis (2008), li et al. (2016), Liu and Skrzypacz (2013), Livingston (2015),Jolivet et al (2016), Zervas et al (2015), Mayzlin et al (2014), Jullien and Park (2014)

Xavier Lambin and Emil Palikot (TSE) Reputation and entry June 27, 2018 6 / 36 Matching on Blablacar.com

1.5 million travelers with Blablacar each month, annual growth rate of 200% since 2013.

1 drivers post rides at given time/date. They post price (Blablacar makes suggestion)

2 riders see listings corresponding to their needs, send request to rider Appendix listings

3 driver accepts/rejects request

4 payment is made through Blablacar online system. Fee is ≈ 20% of price asked by driver. Appendix fees

Xavier Lambin and Emil Palikot (TSE) Reputation and entry June 27, 2018 7 / 36 Reputation system

A reputation system helps decision making. After each ride both drivers and riders are asked to send feedback:

1 Grade is 1 ("à éviter"),2,3,4,5 ("parfait")

2 Written comment

3 User is notied that grade was posted, but not the actual grade until she posts hers or time for feedback has elapsed

Appendix prole example

Xavier Lambin and Emil Palikot (TSE) Reputation and entry June 27, 2018 8 / 36 Data collection

1 cross-section: 300K+ observations 2 panel: 4.6K drivers with average 2.5 obs 3 reviews: 3.5 mln a ride is selected randomly over 300+ routes in

I Origin/Destination, price, date, # seats available, baggage/pet/smoking policy, +written description a driver

I Reputation: average grade received, individual grades (with comment and identity of rider), # of published rides, seniority

I Characteristics: name, age, picture, car brand/type

I Checks: ID, phone veried We also collect :

I minority status and gender: F origin of the name: French, Arabic, Spanish...(French gov., legaro, www.signication-prenom.net) & photo recognition; www.kairos.com

I city variables crime in departure/arrival cities (French gov.), poverty rates, median revenue, population

I value of cars and fuel consumption (eBay, , Ademe) I duration of trip by car / public (google.maps) Xavier Lambin and Emil Palikot (TSE) Reputation and entry June 27, 2018 9 / 36 Descriptive statistics (1)-listings

Statistic N Mean St. Dev. Min Max price 333,572 28.79 14.50 6 72 length (km) 335,086 373.50 182.06 55.14 873.33 hours untill ride 290,359 147.35 167.07 3.06 834.17 posted since 338,910 4.94 7.34 0.00 50.96 notice 324,519 10.95 10.40 0.00 52.67 competition (route) 342,405 30.45 35.12 1 723 public transport (hours) 331,278 3.75 2.27 0.14 15.24 median revenue (city) 320,999 18,975 2,114 13,060 30,904 car's price 281,669 6040 5041 603 24400 consumption (fuel) 293,913 4.98 0.75 3.65 7.50 empty seats 342,405 2.42 0.87 0 6 sold seats 342,405 0.27 0.58 0 4 revenue 338,851 6.11 14.42 0 76 number of clicks 338,913 18.09 23.06 0 140

Xavier Lambin and Emil Palikot (TSE) Reputation and entry June 27, 2018 10 / 36 Descriptive statistics (2)-drivers

Statistic N Mean St. Dev. Min Max picture 342,405 0.88 0.33 0 1 bio (# words) 338,791 14.66 16.65 0 79 Arabic 342,405 0.11 0.31 0 1 African 319,090 0.05 0.22 0 1 male 338,276 0.73 0.44 0 1 minority 342,405 0.15 0.36 0 1 automatic acceptance 342,405 0.48 0.50 0 1 driver's age 338,317 37.21 12.60 18 68 reviews (#) 338,960 38.15 61.48 0 410 talkative 342,403 2.21 0.48 1 3 published listings (#) 338,957 54.95 90.54 2 646 reputation 315,402 4.60 0.27 3.60 5.00 seniority (# months) 338,916 42.62 26.89 1 114 posts per month 342,405 1.77 3.13 0.01 29.86 strike 342,405 0.06 0.25 0 1

Xavier Lambin and Emil Palikot (TSE) Reputation and entry June 27, 2018 11 / 36 Performance gap (1)

Set lower prices: 28.97EUR vs. 27.79EUR Minorities' oers receive fewer views: 18.29 vs. 17 Minorities sell fewer seats: 0.267 vs. 0.265 Minority drivers make lower revenue: 6.2 EUR vs 5.59 EUR

Xavier Lambin and Emil Palikot (TSE) Reputation and entry June 27, 2018 12 / 36 Performance gap (2)

IV

Dependent variable: number of clicks sold seats revenue driver's age −0.060∗∗∗ (0.004) −0.001∗∗∗ (0.0001) −0.013∗∗∗ (0.003) reviews (#) 0.014∗∗∗ (0.001) 0.001∗∗∗ (0.00003) 0.014∗∗∗ (0.001) talkative 0.447∗∗∗ (0.092) 0.006∗∗ (0.003) 0.134∗∗ (0.066) minority −0.606∗∗∗ (0.128) −0.015∗∗∗ (0.004) −0.523∗∗∗ (0.092) male −1.349∗∗∗ (0.100) −0.001 (0.003) −0.137∗ (0.072) seniority (# months) −0.015∗∗∗ (0.002) −0.0003∗∗∗ (0.0001) −0.004∗∗∗ (0.001) hours until ride −0.006∗∗∗ (0.002) −0.00001 (0.0001) −0.001 (0.001) posted since 2.345∗∗∗ (0.046) 0.028∗∗∗ (0.001) 0.642∗∗∗ (0.033) posts per month −0.553∗∗∗ (0.021) −0.007∗∗∗ (0.001) −0.121∗∗∗ (0.015) picture 1.000∗∗∗ (0.147) 0.013∗∗∗ (0.004) 0.447∗∗∗ (0.106) bio (# words) 0.016∗∗∗ (0.003) 0.0004∗∗∗ (0.0001) 0.012∗∗∗ (0.002) car's price −0.002 (0.009) −0.0002 (0.0002) −0.007 (0.006) competition 0.022∗∗∗ (0.002) 0.001∗∗∗ (0.00004) 0.015∗∗∗ (0.001) median revenue 0.001∗∗∗ (0.0002) 0.00002∗∗∗ (0.00001) 0.001∗∗∗ (0.0001) public transport (# hours) 1.154∗ (0.665) −0.026 (0.019) −2.756∗∗∗ (0.479) length (# km) 0.003 (0.003) −0.0002∗ (0.0001) 0.013∗∗∗ (0.002) length (# hour) 0.065∗∗∗ (0.010) 0.002∗∗∗ (0.0003) 0.019∗∗∗ (0.007) day 0.625∗∗∗ (0.101) 0.016∗∗∗ (0.003) 0.435∗∗∗ (0.073) notice −0.684∗∗∗ (0.045) −0.014∗∗∗ (0.001) −0.318∗∗∗ (0.032) automatic acceptance −1.947∗∗∗ (0.090) 0.097∗∗∗ (0.003) 2.149∗∗∗ (0.065) sncf strike 7.331∗∗∗ (0.185) 0.123∗∗∗ (0.005) 2.737∗∗∗ (0.134) Constant −3.817 (2.723) −0.123 (0.077) −3.525∗ (1.967) Observations 197,301 198,985 196,976 Xavier LambinR and2 Emil Palikot (TSE) 0.266Reputation and entry 0.076 0.080June 27, 2018 13 / 36 Impact of the reputation system

We show that: performance gap is concentrated at the beginning of driver's career disappears entirely when reputation is collected not driven by exit of underperforming minority drivers

1 cross-section

2 panel data

3 matching

Xavier Lambin and Emil Palikot (TSE) Reputation and entry June 27, 2018 14 / 36 Cross section: Minority gap as experience increases

Dependent variable: number of clicks reviews(#) 1:3 4:49 50:300 driver's age −0.081∗∗∗ (0.007) −0.067∗∗∗ (0.005) −0.009 (0.007) talkative −0.096 (0.198) 0.315∗∗ (0.138) 0.679∗∗∗ (0.172) minority −0.986∗∗∗ (0.267) −0.775∗∗∗ (0.193) 0.091 (0.244) seniority (# months) −0.005 (0.004) −0.011∗∗∗ (0.003) −0.030∗∗∗ (0.004) hours until ride −0.006 (0.004) −0.006∗∗ (0.003) −0.002 (0.004) posted since 2.281∗∗∗ (0.097) 2.515∗∗∗ (0.068) 2.343∗∗∗ (0.090) posts per month −0.325∗∗∗ (0.064) −0.454∗∗∗ (0.036) −0.556∗∗∗ (0.030) picture 0.976∗∗∗ (0.318) 0.885∗∗∗ (0.231) 0.341 (0.279) bio (#words) 0.001 (0.006) 0.016∗∗∗ (0.004) 0.013∗∗∗ (0.005) male −1.482∗∗∗ (0.204) −1.337∗∗∗ (0.144) −0.790∗∗∗ (0.211) car's price −0.033∗ (0.018) 0.003 (0.013) 0.007 (0.016) competition 0.021∗∗∗ (0.003) 0.020∗∗∗ (0.002) 0.024∗∗∗ (0.003) median revenue 0.0003 (0.0004) 0.001∗∗∗ (0.0003) 0.001∗∗∗ (0.0004) public transport 1.960 (1.334) 2.007∗∗ (0.986) 0.746 (1.539) length (# km) 0.001 (0.007) 0.001 (0.005) 0.004 (0.006) length(# hour) 0.041∗ (0.021) 0.036∗∗ (0.016) 0.101∗∗∗ (0.019) day −0.135 (0.215) 0.517∗∗∗ (0.151) 1.373∗∗∗ (0.190) notice −0.582∗∗∗ (0.096) −0.714∗∗∗ (0.067) −0.817∗∗∗ (0.089) automatic acceptance −1.354∗∗∗ (0.194) −1.841∗∗∗ (0.133) −3.270∗∗∗ (0.175) sncf strike 7.424∗∗∗ (0.335) 7.215∗∗∗ (0.272) 7.827∗∗∗ (0.445) Constant 7.628 (5.228) −6.018 (4.052) −10.145 (6.522) Observations 43,894 89,105 51,090 R2 0.273 0.275 0.271 Trip xed eects YES YES YES Xavier Lambin and Emil Palikot (TSE) Reputation and entry June 27, 2018 15 / 36 Cross section: Minority gap as experience increases

Sold seats Revenue Xavier Lambin and Emil Palikot (TSE) Reputation and entry June 27, 2018 16 / 36 Panel data: denition

4.6K drivers observed at least 2 times and at least once when number of reviews smaller than 10 entrant when number of reviews less than 4

Dependent variable: number of clicks Pooled Between Random Within Mundlak method minority 0.502 (0.727) 0.315 (0.961) 0.218 (0.835) 0.203 (0.20) minority*entrant −4.516∗∗∗ (1.348) −5.773∗∗∗ (1.92) −3.943∗∗∗ (1.405) −2.604 (2.014) −4.617∗ Panel IV: minority 0.513 (0.727) − 2.104 (2.538) −0.132 (1.534) minority*entrant −4.527∗∗∗ (1.348) −8.689∗∗ (3.503) −3.155∗∗ (1.487) −7.978 (11.530) Note: instrument- fuel consumption

Full results: Number of clicks Sold seats Revenue

Xavier Lambin and Emil Palikot (TSE) Reputation and entry June 27, 2018 17 / 36 Matching

1 Calculation of the propensity score-> drop 5% extreme values car price driver's age posts per month Pr(minorityi ) = α1 i + α2 i + α3 i picture bio gender consumption driver blabla +α4 i + α5 i + α6 i + α7 i + α8 i + i

2 Exact matching-8636 minority drivers matched with 19039 non-minority drivers 3 Coarsened matching-14146 minority drivers with 45959 nonminority

Xavier Lambin and Emil Palikot (TSE) Reputation and entry June 27, 2018 18 / 36 Exact matching-Entrants

Dependent variable Number of clicks Revenue Taken seats minority -1.4211∗∗ (0.002 ) -0.8294∗∗ (0.007) -3.00E-02∗∗ (0.005) hours until ride -0.0234 ∗∗ (0.009) -0.0025 (0.686) -9.20E-05 (0.672) posted since 2.2056 ∗∗∗ (<2e-16) 0.6555 (9.00E-06)∗∗∗ 2.40E-02 ∗∗∗ (5.00E-06) competition 0.0461 ∗∗∗( 8e-10) 0.0029 (0.569) 1.10E-04 (0.559) day 0.7465 ( 0.124 ) 0.9003 (0.006)∗∗ 2.80E-02∗ (0.018) night 1.9899 ∗∗ ( 0.006 ) -1.1733 (0.018)∗ -4.90E-02∗∗ (0.006) notice -0.3299 (0.122 ) -0.3203 (0.027)∗ -1.20E-02∗ (0.023) Matched Observations 9,555

Note: Trip xed eects not reported ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01

Xavier Lambin and Emil Palikot (TSE) Reputation and entry June 27, 2018 19 / 36 Exact matching-Incumbents

Dependent variable Number of clicks Revenue Taken seats minority -0.0200 (0.971 ) 0.9449 (0.035)∗ 0.02679 ( 0.142 ) hours untill ride 0.0091 (0.408 ) 0.0180 (0.044)∗ 0.00110∗∗ (0.002) posted since 3.2570∗∗∗ (<2e-16) 1.5431 (2e-12 )∗∗∗ 0.06835 ∗∗∗ ( 3e-14) competition 0.0285 ∗∗( 0.002) 0.0237 ( 0.001)∗∗ 0.00061∗(0.041) day 1.2000∗ ( 0.041 ) 0.7038 ( 0.140 ) 0.04531 ∗ (0.020) night 0.8739 ( 0.357 ) -1.8608 (0.016)∗ -0.10491∗∗∗ (8e-04 ) notice -1.1991 ∗∗∗ ( 5e-06 ) -0.9369 (1e-05)∗∗∗ -0.04648 ∗∗∗ ( 9e-08 ) Matched Observations 5,316

Note: Trip xed eects not reported ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01

Xavier Lambin and Emil Palikot (TSE) Reputation and entry June 27, 2018 20 / 36 Coarsened matching

Dependent variable Number of clicks Revenue Taken seats minority (entrant) -0.9869 ∗∗ (0.005) -0.52608∗ (0.029) -0.02070∗ ( 0.016 ) minority (incumbent) 0.2507 ( 0.459 ) 0.2327 (0.390) 0.01684 ( 0.134 ) Matched Observations (both models) 60,105

∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01

Xavier Lambin and Emil Palikot (TSE) Reputation and entry June 27, 2018 21 / 36 Not driven by selection

No.listings

Xavier Lambin and Emil Palikot (TSE) Reputation and entry June 27, 2018 22 / 36

No.listings A model of career concerns -preliminary

Reputation system works; but how? Passengers observe proles of drivers and form expectations Reviews are used to update beliefs Drivers respond to market beliefs about their quality by taking eort Further questions: Which part of underperformance of minority drivers can we attribute to statistical discrimination and which one to taste-based? Can we nd beliefs about minority drivers' type associated with their outcomes? What is the net eect of reputation system? (accounting for additional eorts)

Xavier Lambin and Emil Palikot (TSE) Reputation and entry June 27, 2018 23 / 36 Holmström (1999)- driver as a manager: Output (measure in grades)

yi,t = ηi,j + ei,t + i,t

,where ηi,j is intrinsic quality (type) of driver i coming from population j, ηj ∼ N(mj , 1/hj ), incompletely known both to the driver and the market.

Assume yi,t = wi,t . Driver's problem:

( ∞ ) X t−1 max β [wi,t − g(ei,t )] et t=1 Optimal eort: ∞ X s−t 0 ∗ β αi,s = g (ei,t ) s=t

h , where αti = hi +th

Xavier Lambin and Emil Palikot (TSE) Reputation and entry June 27, 2018 24 / 36 Holmström's career concerns model: Types, eorts and errors: Output

Eort

Type

t∗ 0 Time

Xavier Lambin and Emil Palikot (TSE) Reputation and entry June 27, 2018 25 / 36 Decreasing grades

Figure: Distributions of mean grade across experience

Xavier Lambin and Emil Palikot (TSE) Reputation and entry June 27, 2018 26 / 36 Looking for the cuto- t∗ When does the downward trend stop?

gradei = α0 + α1 ∗ cutoff + ....

Dependent variable: grade g_15 0.011∗∗∗ (0.001) minority −0.080∗∗∗ (0.001) gender −0.019∗∗∗ (0.001) car price −0.000 (0.000) driver's age −0.001∗∗∗ (0.00003) posts per month 0.004∗∗∗ (0.0001) picture 0.080∗∗∗ (0.001) length bio 0.0004∗∗∗ (0.00002) number of reviews −0.0003∗∗∗ (0.00002) Constant 4.572∗∗∗ (0.002) Observations 3,435,064

Xavier Lambin and Emil Palikot (TSE) Reputation and entry June 27, 2018 27 / 36 Parameters of the model

How do we recover the parameters of the model?

1 type-> mean and variance of grades after they stabilise, per population- 16 groups: minority (Yes/No), male (Yes/No), Age (above/below median), car price (above/below median).

2 error term->variance of grades after they stabilise

3 eort-> dierence between type and grade, for the rst 15 grades

Xavier Lambin and Emil Palikot (TSE) Reputation and entry June 27, 2018 28 / 36 Populations

Dependent variable: Mean Variance minority −0.075∗∗∗ (0.0003) 0.081∗∗∗ (0.0002) car's price 0.010∗∗∗ (0.0002) −0.002∗∗∗ (0.0001) driver's age −0.026∗∗∗ (0.0002) 0.022∗∗∗ (0.0001) male 0.004∗∗∗ (0.0002) 0.020∗∗∗ (0.0001) Constant 4.562∗∗∗ (0.0002) 0.338∗∗∗ (0.0001) Observations 51,902 51,902 R2 0.672 0.856 F Statistic (df = 4; 51897) 26,602.260∗∗∗ 76,835.680∗∗∗

Note: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01

Xavier Lambin and Emil Palikot (TSE) Reputation and entry June 27, 2018 29 / 36 Cost function estimation Holmström relates eort over time to measurement errors and discount rate: ∞ X h g 0(e ) = βs−t α , with α = it is is h + th s=t i  γ 2 we assume g(eit ) = 2 (eit ) .   1 0 h 1 h eortj,t,i = α + β + β + .... + j,t,i γ hi + th hi + (t + 1)h

Dependent variable: eort const −0.089∗∗∗ (0.005) 1/γ 0.108∗∗∗ (0.013) β 0.8418∗∗∗ (0.049) Observations 1,467,645 Residual Std. Error 0.664 (df = 1467645)

Xavier Lambin and Emil Palikot (TSE) Reputation and entry June 27, 2018 30 / 36 Expected quality

We calculate expected quality from Holmström:

h1m1 E[wt ] = ht t−1 X E ∗ + hht (m1 + es − [es ]) s=1 E ∗ + [et ]

Xavier Lambin and Emil Palikot (TSE) Reputation and entry June 27, 2018 31 / 36 Predicting demand

estimate model: IV q = a1 + a2 ∗ p + a3 ∗ quality + a4 ∗ route−specific +  using non-minorities

Dependent variable: taken seats OLS OLS IV

∗ ∗∗∗ wt 0.113 (0.060) 0.041 (0.061) 0.840 (0.310) price −0.010∗∗∗ (0.0005) 0.103∗∗ (0.043) hours until ride 0.00003 (0.0001) 0.0001 (0.0001) −0.001∗∗ (0.0003) posted since 0.022∗∗∗ (0.002) 0.025∗∗∗ (0.002) 0.008 (0.007) competition 0.0004∗∗∗ (0.0001) 0.0004∗∗∗ (0.0001) 0.0001 (0.0001) median revenue 0.00001∗∗∗ (0.00000) 0.00001∗∗∗ (0.00000) 0.00001∗∗∗ (0.00000) public transport −0.001 (0.001) −0.001 (0.001) −0.008∗∗∗ (0.003) km −0.00003∗∗ (0.00001) 0.001∗∗∗ (0.00004) −0.008∗∗ (0.003) hour 0.001∗∗∗ (0.0004) 0.001∗∗ (0.0004) 0.001∗∗ (0.001) notice −0.012∗∗∗ (0.002) −0.013∗∗∗ (0.002) 0.002 (0.007) number of reviews 0.007∗∗∗ (0.001) 0.005∗∗∗ (0.001) 0.021∗∗∗ (0.006) Constant −0.452 (0.280) −0.093 (0.283) −3.844∗∗∗ (1.451) Observations 66,537 65,400 63,393 R2 0.044 0.052 −0.854 Adjusted R2 0.044 0.051 −0.855 Residual Std. Error 0.489 (df = 66526) 0.490 (df = 65388) 0.684 (df = 63381) F Statistic 306.236∗∗∗ (df = 10; 66526) 323.079∗∗∗ (df = 11; 65388)

Note: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01

Xavier Lambin and Emil Palikot (TSE) Reputation and entry June 27, 2018 32 / 36 Beliefs predict expected q for minorities; calculate the gap between predicted and actual seats sold-> 3%

Xavier Lambin and Emil Palikot (TSE) Reputation and entry June 27, 2018 33 / 36 Holmström model next step

Next steps: Rene quality estimate by population Improve demand model Calculate loss in economic outcomes absent a reputation system

Possible extensions: Study social Homophily Impact of strikes on minority outcomes

Xavier Lambin and Emil Palikot (TSE) Reputation and entry June 27, 2018 34 / 36 Conclusion:

Minority drivers experience performance gap upon entry to Blablacar Building reputation allows minority users to narrow the gap Preliminary study of a career concerns model shows that statistical updating of belifs about expected quality explains a part of the gap

Xavier Lambin and Emil Palikot (TSE) Reputation and entry June 27, 2018 35 / 36 Trip

back

Xavier Lambin and Emil Palikot (TSE) Reputation and entry June 27, 2018 35 / 36 Listings

back

Xavier Lambin and Emil Palikot (TSE) Reputation and entry June 27, 2018 35 / 36 Driver proles

back

Xavier Lambin and Emil Palikot (TSE) Reputation and entry June 27, 2018 35 / 36 Service fees of Blablacar

back

Xavier Lambin and Emil Palikot (TSE) Reputation and entry June 27, 2018 35 / 36 Not driven by selection

back

Xavier Lambin and Emil Palikot (TSE) Reputation and entry June 27, 2018 35 / 36 Cost function estimation

Holmström relates eort over time to measurement errors and discount rate:

∞ 0 X s−t h g (et ) = β αs , with αs = (1) h1 + h s=t 

we assume γm 2, 1 , g(et ) = 2 (et ) h = var(individual ratings|experienced) 1 h1 = var(average rating|experienced) ⇒ aim at estimating β, γm, γnm

Xavier Lambin and Emil Palikot (TSE) Reputation and entry June 27, 2018 36 / 36 Documenting discrimination- IV

IV

Dependent variable: number_of_views taken_seats ride_price 11.870∗∗∗ (3.023) 0.311∗∗∗ (0.080) driver_age −0.237∗∗∗ (0.046) −0.005∗∗∗ (0.001) number_avis 0.063∗∗∗ (0.013) 0.002∗∗∗ (0.0003) driver_blabla 0.121 (0.250) −0.003 (0.007) f_minority −7.101∗∗∗ (1.682) −0.186∗∗∗ (0.045) f_gender 0.997 (0.653) 0.061∗∗∗ (0.017) seniority_months 0.035∗∗∗ (0.013) 0.001∗∗∗ (0.0003) hours_till_ride −0.048∗∗∗ (0.012) −0.001∗∗∗ (0.0003) posted_since 1.536∗∗∗ (0.241) 0.008 (0.006) post_per_month −0.757∗∗∗ (0.085) −0.014∗∗∗ (0.002) picture −2.126∗∗ (0.873) −0.069∗∗∗ (0.023) length_bio 0.079∗∗∗ (0.017) 0.002∗∗∗ (0.0005) car_price −0.361∗∗∗ (0.094) −0.010∗∗∗ (0.002) competition −0.020∗ (0.012) −0.0005 (0.0003) revenu_median 0.003∗∗∗ (0.001) 0.0001∗∗∗ (0.00002) duration_public_transport −18.924∗∗∗ (5.293) −0.516∗∗∗ (0.136) km −0.873∗∗∗ (0.223) −0.023∗∗∗ (0.006) hour 0.115∗∗∗ (0.030) 0.003∗∗∗ (0.001) night_dayday 0.896∗∗∗ (0.266) 0.024∗∗∗ (0.007) night_daynight −2.152∗∗∗ (0.763) −0.128∗∗∗ (0.021) notice −0.049 (0.200) 0.001 (0.005) acceptation_automatique 18.283∗∗∗ (5.142) 0.624∗∗∗ (0.135) sncf_strike −10.841∗∗ (4.638) −0.354∗∗∗ (0.123) Constant 15.228∗ (8.306) 0.238 (0.214) Observations 189,455 191,060 R2 −3.635 −4.237 Adjusted R2 −3.641 −4.244 Residual Std. Error 48.367 (df = 189208) 1.310 (df = 190813) Xavier Lambin and Emil Palikot (TSE) Reputation and entry June 27, 2018 36 / 36 Note: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01 Cross section: Minority gap as experience increases: taken seats

Sold seats

Dependent variable: taken seats reviews(#) 1:3 4:49 50:300 driver_age −0.001∗∗∗ (0.0002) −0.001∗∗∗ (0.0002) −0.0001 (0.0002) driver_blabla −0.005 (0.005) 0.002 (0.004) 0.018∗∗∗ (0.006) f_minority −0.019∗∗∗ (0.007) −0.019∗∗∗ (0.005) −0.005 (0.008) seniority_months −0.0001 (0.0001) −0.0004∗∗∗ (0.0001) −0.001∗∗∗ (0.0001) hours_till_ride −0.0002∗∗ (0.0001) 0.0001 (0.0001) 0.00004 (0.0001) posted_since 0.016∗∗∗ (0.002) 0.031∗∗∗ (0.002) 0.036∗∗∗ (0.003) post_per_month 0.0003 (0.002) −0.004∗∗∗ (0.001) −0.008∗∗∗ (0.001) picture −0.001 (0.008) 0.003 (0.007) 0.009 (0.009) length_bio 0.0002 (0.0002) 0.0004∗∗∗ (0.0001) 0.0001 (0.0002) f_gender −0.013∗∗∗ (0.005) 0.007∗ (0.004) 0.012∗ (0.007) car_price −0.0003 (0.0005) −0.0001 (0.0004) −0.0002 (0.001) competition 0.001∗∗∗ (0.0001) 0.001∗∗∗ (0.0001) 0.001∗∗∗ (0.0001) revenu_median 0.00002∗∗ (0.00001) 0.00002∗∗∗ (0.00001) 0.00004∗∗∗ (0.00001) duration_public_transport −0.013 (0.032) −0.014 (0.028) 0.013 (0.050) km −0.0004∗∗ (0.0002) −0.0001 (0.0001) −0.0003 (0.0002) hour 0.001 (0.001) 0.001∗∗∗ (0.0004) 0.002∗∗ (0.001) night_dayday 0.010∗∗ (0.005) 0.013∗∗∗ (0.004) 0.027∗∗∗ (0.006) night_daynight −0.041∗∗∗ (0.008) −0.057∗∗∗ (0.007) −0.076∗∗∗ (0.010) notice −0.004∗ (0.002) −0.017∗∗∗ (0.002) −0.020∗∗∗ (0.003) acceptation_automatique 0.097∗∗∗ (0.005) 0.089∗∗∗ (0.004) 0.101∗∗∗ (0.006) sncf_strike 0.104∗∗∗ (0.008) 0.125∗∗∗ (0.008) 0.166∗∗∗ (0.015) Constant 0.077 (0.128) −0.096 (0.114) −0.361∗ (0.213) Observations 44,303 89,789 51,479 R2 0.067 0.068 0.085 Xavier LambinAdjusted and R2 Emil Palikot (TSE) 0.062Reputation and entry 0.066June 0.080 27, 2018 36 / 36 Residual Std. Error 0.467 (df = 44060) 0.545 (df = 89546) 0.619 (df = 51235) F Statistic 13.013∗∗∗ (df = 242; 44060) 27.107∗∗∗ (df = 242; 89546) 19.544∗∗∗ (df = 243; 51235) Trip xed eects YES YES YES Cross section: Minority gap as experience increases-revenue

Revenue

Dependent variable: revenue reviews(#) 1:3 4:49 50:300 driver_age −0.020∗∗∗ (0.005) −0.014∗∗∗ (0.004) −0.0002 (0.006) driver_blabla −0.215 (0.132) 0.112 (0.101) 0.316∗∗ (0.133) f_minority −0.455∗∗ (0.179) −0.686∗∗∗ (0.142) −0.241 (0.189) seniority_months −0.006∗∗ (0.003) −0.010∗∗∗ (0.002) −0.013∗∗∗ (0.003) hours_till_ride −0.008∗∗∗ (0.003) 0.001 (0.002) −0.0004 (0.003) posted_since 0.356∗∗∗ (0.065) 0.747∗∗∗ (0.049) 0.789∗∗∗ (0.070) post_per_month 0.008 (0.043) −0.063∗∗ (0.027) −0.135∗∗∗ (0.023) picture 0.035 (0.213) 0.128 (0.169) 0.482∗∗ (0.217) length_bio 0.011∗∗ (0.004) 0.010∗∗∗ (0.003) 0.002 (0.004) f_gender −0.424∗∗∗ (0.137) 0.059 (0.106) 0.047 (0.163) car_price −0.012 (0.012) −0.005 (0.010) −0.002 (0.013) competition 0.015∗∗∗ (0.002) 0.013∗∗∗ (0.002) 0.015∗∗∗ (0.002) revenu_median 0.0004∗ (0.0002) 0.001∗∗∗ (0.0002) 0.001∗∗ (0.0003) duration_public_transport −2.050∗∗ (0.890) −1.839∗∗ (0.722) −5.715∗∗∗ (1.221) km 0.005 (0.005) 0.013∗∗∗ (0.004) 0.020∗∗∗ (0.005) hour 0.012 (0.014) 0.017 (0.011) 0.006 (0.015) night_dayday 0.203 (0.144) 0.445∗∗∗ (0.110) 0.579∗∗∗ (0.147) night_daynight −0.869∗∗∗ (0.229) −1.168∗∗∗ (0.183) −1.233∗∗∗ (0.232) notice −0.072 (0.064) −0.392∗∗∗ (0.048) −0.416∗∗∗ (0.068) acceptation_automatique 2.352∗∗∗ (0.130) 2.007∗∗∗ (0.097) 1.977∗∗∗ (0.136) sncf_strike 2.765∗∗∗ (0.224) 2.552∗∗∗ (0.200) 3.710∗∗∗ (0.346) Constant 1.371 (3.504) −4.611 (2.973) 2.588 (5.191) Observations 43,940 88,889 50,879 R2 0.067 0.078 0.103 Residual Std. Error 12.758 (df = 43697) 14.063 (df = 88646) 14.495 (df = 50635) F Statistic 12.971∗∗∗ (df = 242; 43697) 30.938∗∗∗ (df = 242; 88646) 23.861∗∗∗ (df = 243; 50635) Xavier Lambin and Emil Palikot (TSE) Reputation and entry June 27, 2018 36 / 36 Trip xed eects YES YES YES Panel: Minority gap as experience increases-number of clicks

Panel1

Dependent variable: number of clicks Pooled Between Random Within f_minority 0.502 (0.727) 0.315 (0.961) 0.218 (0.835) f_gender −1.959∗∗∗ (0.561) −2.923∗∗∗ (0.727) −1.928∗∗∗ (0.638) entrant 2.267∗∗ (0.971) 0.685 (1.440) 2.231∗∗ (1.007) driver_age −0.054∗∗∗ (0.016) −0.061∗∗∗ (0.020) −0.056∗∗∗ (0.019) −1.515 (1.724) driver_blabla 0.580 (0.439) 1.061∗ (0.544) 0.697 (0.512) 0.511 (3.012) seniority_months 0.008 (0.009) 0.014 (0.011) 0.010 (0.010) 0.052 (1.030) picture 0.006 (0.987) −0.247 (1.151) 0.402 (1.120) length_bio 0.041∗∗∗ (0.012) 0.043∗∗∗ (0.016) 0.038∗∗∗ (0.014) 0.030 (0.068) car_price −14.160 (32.387) 0.351 (39.792) −15.371 (37.429) 217.906 (223.283) comp 0.059∗∗∗ (0.008) 0.061∗∗∗ (0.012) 0.031∗∗∗ (0.008) 0.047∗∗∗ (0.011) revenu_median 0.0001 (0.0001) 0.0001 (0.0002) 0.00003 (0.0001) 0.00005 (0.0002) duration_public_transport −5.474 (6.869) 1.248 (10.192) −8.143 (7.238) −13.485 (10.667) night_dayday 0.596 (0.487) 0.850 (0.765) 0.375 (0.502) −0.018 (0.659) night_daynight 0.898 (0.765) 1.675 (1.218) 0.533 (0.784) −0.567 (1.008) acceptation_automatique −1.984∗∗∗ (0.426) −1.867∗∗∗ (0.571) −1.836∗∗∗ (0.472) −1.780∗∗ (0.902) notice −0.642∗∗∗ (0.036) −0.430∗∗∗ (0.059) −0.536∗∗∗ (0.025) −0.986∗∗∗ (0.067) notice2 0.001∗∗∗ (0.0003) −0.001∗∗ (0.001) 0.001∗∗∗ (0.0001) 0.006∗∗∗ (0.001) km 0.009∗∗∗ (0.001) 0.010∗∗∗ (0.002) 0.009∗∗∗ (0.002) −0.002 (0.003) posted_since 2.201∗∗∗ (0.051) 1.955∗∗∗ (0.081) 2.175∗∗∗ (0.045) 2.844∗∗∗ (0.083) posted_since2 −0.008∗∗∗ (0.0005) −0.004∗∗∗ (0.001) −0.008∗∗∗ (0.0003) −0.019∗∗∗ (0.001) as.numeric(number_avis) 0.023 (0.042) 0.122 (0.092) 0.021 (0.050) time −0.043 (0.170) −0.211 (0.271) −0.184∗∗∗ (0.025) 0.138 (0.256) sncf_strike 6.035∗∗∗ (0.975) 5.600∗∗∗ (1.674) 5.340∗∗∗ (1.188) f_minority:entrant −4.516∗∗∗ (1.348) −5.773∗∗∗ (1.920) −3.943∗∗∗ (1.405) −2.604 (2.014) f_gender:entrant −1.231 (1.098) 1.671 (1.567) −1.392 (1.137) −4.306∗∗∗ (1.606) Constant 10.566 (7.956) 14.799 (12.536) 16.815∗∗∗ (2.959) Observations 8,849 3,898 8,849 8,849 R2 0.284 0.278 0.270 0.323 Adjusted R2 0.281 0.269 0.268 −0.235 XavierF Statistic Lambin and Emil Palikot 79.447∗∗∗ (TSE)(df = 44; 8804) 33.639Reputation∗∗∗ (df = and 44; 3853) entry 135.664∗∗∗ (df = 24; 8824)June 60.902 27,∗∗∗ 2018(df = 38; 4849) 36 / 36 Note: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01 Panel: Minority gap as experience increases-taken seats

Panel2

Dependent variable: taken seats Pooled Between Random Within f_minority 0.009 (0.021) 0.016 (0.025) 0.010 (0.022) f_gender −0.007 (0.016) −0.010 (0.019) −0.004 (0.017) entrant 0.024 (0.027) 0.020 (0.038) 0.025 (0.028) driver_age −0.001∗∗ (0.0005) −0.001 (0.001) −0.001∗ (0.0005) −0.108∗∗ (0.053) driver_blabla −0.018 (0.012) −0.003 (0.014) −0.017 (0.013) −0.080 (0.092) seniority_months 0.0004 (0.0002) 0.0004 (0.0003) 0.0004 (0.0003) 0.015 (0.032) picture −0.007 (0.028) −0.025 (0.030) 0.003 (0.029) length_bio 0.00003 (0.0003) −0.0002 (0.0004) −0.0001 (0.0004) 0.001 (0.002) car_price −0.776 (0.919) −0.705 (1.039) −0.953 (0.966) −10.395 (6.826) comp 0.001∗∗∗ (0.0002) 0.001∗∗∗ (0.0003) 0.001∗∗∗ (0.0002) 0.002∗∗∗ (0.0003) revenu_median 0.00001∗∗∗ (0.00000) 0.00001∗∗ (0.00000) 0.00001∗∗ (0.00000) 0.00000 (0.00001) duration_public_transport −0.143 (0.195) −0.115 (0.269) −0.124 (0.199) −0.428 (0.325) night_dayday 0.006 (0.014) 0.048∗∗ (0.020) 0.004 (0.014) −0.034∗ (0.020) night_daynight −0.065∗∗∗ (0.022) −0.029 (0.032) −0.070∗∗∗ (0.022) −0.086∗∗∗ (0.031) acceptation_automatique 0.085∗∗∗ (0.012) 0.073∗∗∗ (0.015) 0.091∗∗∗ (0.013) 0.113∗∗∗ (0.028) notice −0.013∗∗∗ (0.001) −0.013∗∗∗ (0.002) −0.010∗∗∗ (0.001) −0.017∗∗∗ (0.002) notice2 0.00004∗∗∗ (0.00001) 0.00002 (0.00002) 0.00002∗∗∗ (0.00000) 0.0001∗∗∗ (0.00002) km 0.0001 (0.00004) 0.00003 (0.0001) 0.0001∗ (0.00004) 0.0001 (0.0001) posted_since 0.027∗∗∗ (0.001) 0.025∗∗∗ (0.002) 0.025∗∗∗ (0.001) 0.034∗∗∗ (0.003) posted_since2 −0.0001∗∗∗ (0.00001) −0.0001∗∗∗ (0.00002) −0.0001∗∗∗ (0.00001) −0.0003∗∗∗ (0.00003) as.numeric(number_avis) 0.004∗∗∗ (0.001) 0.009∗∗∗ (0.002) 0.004∗∗∗ (0.001) time −0.008∗ (0.005) −0.009 (0.007) −0.007∗∗∗ (0.001) −0.006 (0.008) sncf_strike 0.140∗∗∗ (0.028) 0.098∗∗ (0.044) 0.200∗∗∗ (0.025) 0.137∗∗∗ (0.036) f_minority:entrant −0.058 (0.038) −0.067 (0.050) −0.057 (0.039) −0.049 (0.062) f_gender:entrant −0.028 (0.031) −0.002 (0.041) −0.024 (0.032) −0.040 (0.049) Constant 0.451∗∗ (0.225) 0.406 (0.329) 0.333∗∗∗ (0.079) Observations 8,939 3,907 8,939 8,939 R2 0.100 0.105 0.079 0.095 Adjusted R2 0.095 0.095 0.077 −0.640 XavierF Statistic Lambin and Emil Palikot 22.369∗∗∗ (TSE)(df = 44; 8894)Reputation 10.318∗∗∗ (df = and 44; 3862)entry 30.620∗∗∗ (df = 25; 8913)June 13.659 27,∗∗∗ 2018(df = 38; 4930) 36 / 36 Note: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01 Panel: Minority gap as experience increases-revenue

Panel3

Dependent variable: revenue Pooled Between Random Within f_minority −0.235 (0.607) 0.136 (0.772) −0.153 (0.657) f_gender −0.481 (0.470) −0.544 (0.586) −0.489 (0.504) entrant 0.511 (0.810) 0.595 (1.169) 0.425 (0.830) driver_age −0.025∗ (0.013) −0.018 (0.016) −0.025∗ (0.015) −3.382∗∗ (1.545) driver_blabla −0.332 (0.367) 0.041 (0.438) −0.313 (0.400) −2.364 (2.688) seniority_months 0.007 (0.007) 0.006 (0.008) 0.008 (0.008) 0.384 (0.917) picture −0.046 (0.828) −0.571 (0.927) 0.292 (0.885) length_bio 0.006 (0.010) 0.005 (0.013) 0.002 (0.011) 0.022 (0.060) car_price −19.957 (27.151) −21.467 (31.993) −26.927 (29.400) −218.782 (198.394) comp 0.030∗∗∗ (0.007) 0.020∗∗ (0.010) 0.018∗∗∗ (0.007) 0.031∗∗∗ (0.010) revenu_median 0.0002∗∗∗ (0.0001) 0.0003∗∗ (0.0001) 0.0002∗∗ (0.0001) 0.0002 (0.0002) duration_public_transport −3.613 (5.754) −5.975 (8.269) −3.930 (5.936) −6.941 (9.456) night_dayday 0.248 (0.408) 1.489∗∗ (0.616) 0.106 (0.418) −1.046∗ (0.589) night_daynight −1.393∗∗ (0.638) 0.070 (0.978) −1.654∗∗ (0.652) −2.660∗∗∗ (0.901) acceptation_automatique 1.859∗∗∗ (0.357) 1.466∗∗∗ (0.460) 2.025∗∗∗ (0.379) 2.562∗∗∗ (0.807) notice −0.400∗∗∗ (0.030) −0.410∗∗∗ (0.049) −0.306∗∗∗ (0.021) −0.467∗∗∗ (0.059) notice2 0.001∗∗∗ (0.0003) 0.001 (0.0005) 0.001∗∗∗ (0.00004) 0.002∗∗∗ (0.001) km 0.021∗∗∗ (0.001) 0.021∗∗∗ (0.002) 0.022∗∗∗ (0.001) 0.023∗∗∗ (0.002) posted_since 0.781∗∗∗ (0.042) 0.808∗∗∗ (0.065) 0.725∗∗∗ (0.037) 0.933∗∗∗ (0.073) posted_since2 −0.003∗∗∗ (0.0004) −0.002∗∗∗ (0.001) −0.003∗∗∗ (0.0002) −0.007∗∗∗ (0.001) as.numeric(number_avis) 0.068∗ (0.035) 0.200∗∗∗ (0.074) 0.061 (0.040) time −0.168 (0.142) −0.043 (0.219) −0.191∗∗∗ (0.021) −0.182 (0.228) sncf_strike 3.017∗∗∗ (0.816) 1.664 (1.350) 4.583∗∗∗ (0.729) 3.105∗∗∗ (1.061) f_minority:entrant −0.511 (1.130) −1.444 (1.550) −0.498 (1.161) 0.504 (1.800) f_gender:entrant −1.108 (0.917) −0.692 (1.269) −0.819 (0.938) 0.044 (1.422) Constant 3.098 (6.649) −3.869 (10.131) 2.175 (2.388) Observations 8,939 3,907 8,939 8,939 R2 0.130 0.153 0.110 0.100 Adjusted R2 0.125 0.143 0.108 −0.632 XavierF Statistic Lambin and Emil Palikot 30.151∗∗∗ (TSE)(df = 44; 8894)Reputation 15.861∗∗∗ (df = and 44; 3862)entry 44.113∗∗∗ (df = 25; 8913)June 14.428 27,∗∗∗ 2018(df = 38; 4930) 36 / 36 Note: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01 Panel: Minority gap as experience increases-number of clicks IV

Panel1IV

Dependent variable: number of clicks Pooled Between Random Within ride_price 6.477 (5.158) −0.364 (0.353) −13.751 (40.527) f_minority 0.513 (0.727) −2.104 (2.538) −0.132 (1.534) f_gender −1.957∗∗∗ (0.561) −2.501∗∗ (1.203) −1.704 (1.169) entrant 2.267∗∗ (0.971) −0.231 (2.460) 3.548∗∗∗ (1.054) 6.587 (7.965) driver_age −0.054∗∗∗ (0.016) −0.159∗∗ (0.076) −0.053 (0.038) −5.756 (5.674) driver_blabla 0.579 (0.439) 1.501 (0.984) 0.689 (0.961) −2.603 (13.168) seniority_months 0.009 (0.009) 0.020 (0.017) 0.014 (0.020) −4.627 (14.502) picture 0.006 (0.988) −0.216 (1.868) 0.393 (2.182) length_bio 0.041∗∗∗ (0.012) 0.058∗∗ (0.026) 0.044∗ (0.026) 0.248 (0.607) car_price −14.289 (32.390) 29.730 (70.574) 627.361 (1,451.442) comp 0.059∗∗∗ (0.008) 0.063∗∗∗ (0.019) 0.049∗∗∗ (0.008) 0.039 (0.030) revenu_median 0.0001 (0.0001) 0.0002 (0.0003) 0.0001 (0.0001) 0.0001 (0.0004) duration_public_transport −5.488 (6.869) −3.649 (16.172) −11.702 (7.826) −41.515 (89.841) night_dayday 0.598 (0.487) 0.029 (1.351) 0.0003 (0.498) −1.234 (3.384) night_daynight 0.898 (0.765) −4.686 (5.239) −0.648 (0.769) −3.506 (8.138) acceptation_automatique −1.986∗∗∗ (0.426) 6.121 (6.131) −2.658∗∗∗ (0.877) −29.684 (82.412) notice −0.642∗∗∗ (0.036) −0.409∗∗∗ (0.105) −0.803∗∗∗ (0.041) −0.848 (1.026) notice2 0.001∗∗∗ (0.0003) −0.002 (0.001) 0.003∗∗∗ (0.0004) 0.001∗∗∗ (0.0002) km 0.009∗∗∗ (0.001) −0.476 (0.386) 0.028 (0.026) 1.027 (3.035) posted_since 2.201∗∗∗ (0.051) 1.656∗∗∗ (0.278) 2.514∗∗∗ (0.052) 2.817∗∗ (1.124) posted_since2 −0.008∗∗∗ (0.0005) −0.002 (0.002) −0.012∗∗∗ (0.001) −0.012∗∗∗ (0.002) as.numeric(number_avis) 0.023 (0.042) 0.301 (0.225) −0.050 (0.058) −0.431 (1.183) time −0.043 (0.170) −1.436 (0.964) 0.101 (0.177) 2.371 (7.205) sncf_strike 6.027∗∗∗ (0.975) 1.161 (4.783) 5.886∗∗∗ (0.918) 14.959 (21.977) f_minority:entrant −4.527∗∗∗ (1.348) −8.689∗∗ (3.503) −3.155∗∗ (1.487) −6.766 (12.392) f_gender:entrant −1.233 (1.098) 3.899 (3.072) −3.566∗∗∗ (1.195) −7.978 (11.530) Constant 10.585 (7.956) 54.393∗ (32.565) 10.585 (8.667) Observations 8,848 4,163 8,641 8,641 Xavier Lambin and Emil Palikot (TSE) Reputation and entry June 27, 2018 36 / 36 R2 0.284 0.049 0.324 0.038 Adjusted R2 0.281 0.039 0.320 −0.748 F Statistic 79.431∗∗∗ (df = 44; 8803) −45.187 (df = 44; 4118) 459,099.800∗∗∗ (df = 45; 8595) −147.070 (df = 23; 4757) Note: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01 Panel: Minority gap as experience increases- taken seats IV

Panel1IV

Dependent variable: taken seats Pooled Between Random Within ride_price 0.739 (1.617) 0.191 (0.157) −0.033∗ (0.020) 0.065 (0.781) f_minority −0.439 (0.979) −0.065 (0.078) 0.029 (0.089) f_gender 0.121 (0.289) 0.006 (0.036) −0.004 (0.068) entrant −0.157 (0.423) −0.009 (0.071) 0.050 (0.033) 0.038 (0.103) driver_age −0.013 (0.026) −0.003 (0.002) −0.001 (0.002) −0.225∗ (0.116) driver_blabla 0.039 (0.142) 0.014 (0.029) −0.046 (0.047) −0.032 (0.337) seniority_months 0.002 (0.003) 0.001 (0.0005) 0.0005 (0.001) 0.048 (0.300) picture −0.042 (0.159) −0.025 (0.054) 0.016 (0.130) length_bio 0.001 (0.003) 0.0001 (0.001) 0.001 (0.001) 0.001 (0.014) car_price −19.828 (42.141) −4.105 (3.497) −14.410 (31.275) comp −0.00005 (0.003) 0.001∗∗ (0.001) 0.002∗∗∗ (0.0003) 0.001∗∗ (0.001) revenu_median 0.00002 (0.00003) 0.00001 (0.00001) 0.00001 (0.00000) 0.00001 (0.00001) duration_public_transport −0.929 (1.830) −0.138 (0.473) −0.549∗∗ (0.249) −0.361 (1.721) night_dayday −0.035 (0.112) 0.030 (0.040) −0.039∗∗ (0.016) −0.036 (0.067) night_daynight −0.285 (0.489) −0.206 (0.149) −0.102∗∗∗ (0.024) −0.086 (0.143) acceptation_automatique 1.103 (2.221) 0.300 (0.186) 0.051 (0.044) 0.257 (1.589) notice −0.017∗∗ (0.008) −0.011∗∗∗ (0.003) −0.017∗∗∗ (0.001) −0.006 (0.020) notice2 0.0001 (0.0001) −0.00000 (0.00004) 0.0001∗∗∗ (0.00001) 0.00001∗∗∗ (0.00000) km −0.055 (0.121) −0.014 (0.012) 0.003∗ (0.001) −0.005 (0.059) posted_since 0.006 (0.047) 0.016∗ (0.008) 0.032∗∗∗ (0.002) 0.024 (0.019) posted_since2 −0.0001 (0.0001) −0.00002 (0.0001) −0.0002∗∗∗ (0.00002) −0.0001∗∗∗ (0.00003) as.numeric(number_avis) 0.019 (0.034) 0.015∗∗ (0.007) −0.002 (0.002) 0.001 (0.022) time −0.128 (0.261) −0.042 (0.028) −0.004 (0.006) −0.020 (0.142) sncf_strike −0.113 (0.584) −0.047 (0.146) 0.162∗∗∗ (0.029) 0.176 (0.426) f_minority:entrant 0.159 (0.506) −0.130 (0.104) −0.064 (0.047) −0.035 (0.228) f_gender:entrant 0.096 (0.316) 0.063 (0.090) −0.039 (0.038) −0.019 (0.183) Constant 4.106 (7.921) 1.476 (0.929) 0.487 (0.327) Observations 8,731 4,172 8,731 8,731 R2 0.001 0.00003 0.095 0.040 Adjusted R2 −0.004 −0.011 0.091 −0.733 Xavier LambinF Statistic and Emil Palikot−183.911 (TSE) (df = 45; 8685) Reputation−62.299 (df =and 44; entry4127) −105.191 (df = 45; 8685) June−2.104 27, (df 2018 = 23; 4838) 36 / 36 Note: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01