How Users Drive Value in Two-sided Markets - Platform Designs That Matter (Appendices)

Appendices A Survey Design We conducted an Amazon MTurk online survey and asked the respondents to code all the categories on Groupon (a total of 78 categories). The respondents were first asked to review the definition and examples of “search” and “experience” goods as follows: In this study, imagine you are browsing a deal website, for example, Groupon. For some goods, you must see, try or experience them at least once to judge their quality. For example, you might need to taste the cuisine at a specific restaurant or try a particular haircut to know if you like the result. These goods are referred to as �experience goods.� For other goods, you can easily judge their value without having to try them. For example, one brand of table salt is like any other brand of table salt. And a gallon of regular gas at one gas station is indistinguishable from a gallon of regular gas at a di↵erent gas station. These goods are referred to as “search goods.” You can often evaluate the quality of search goods simply by reading its product description. Next, you will be provided with 15 goods and asked to evaluate each of them according to the extent they are “experience” versus “search” goods. Please assume that there are no product reviews available. The respondents were then asked to assess for each of the 15 categories, “to what extent it is easy (hard) to evaluate the quality without seeing or trying it.” The 15 questions included “haircut” and “gas,” both of which are NOT the original Groupon categories. These two products are used as the filler questions to check whether respondents paid attention in replying. A total of 818 filler questions and their respondents passed the two attention responses were aggregated to calculate the score for each category. The score closer to “easy” (“hard”) corresponds to search (experience) goods. Based on the scores, we used the median split and code the categories into two groups: experience goods and search goods. We also used the mean split and yielded qualitative similar results. The detailed survey responses are presented in Tables 14 and 15. Table 14: MTurk Online Survey: Experience Goods Groupon Category Haircut (Attention Checking Question) Hair Salons Facials Makeup Services Spa Treatment Health & Beauty Services Restaurants Dance Classes Massage Wine Tasting Comedy Clubs Personal Trainer Concerts Shows Martial Arts Classes Skin Care Nail Care Chiropractor Classes Photography Service Waxing Teeth Whitening Life Skills Classes Boot Camp Dental Clinic and Dental Care Outdoor Adventures City Tours Fitness Classes Bridal Services Kids Activities: Summer camp, play party Museums Bars & Lounges Tanning Dessert Food: seafood orders, healthy snacks Skiing Yoga Home Services Pilates Skydiving Wine N Mean SD Min Max 818 175 159 195 164 153 133 122 145 135 127 109 136 134 128 159 140 132 105 157 168 158 105 129 138 148 119 123 120 144 149 166 115 123 144 125 138 119 127 151 6.46 6.39 6.21 6.13 6.12 6.11 6.11 6.01 5.99 5.88 5.87 5.85 5.77 5.74 5.73 5.67 5.66 5.60 5.54 5.52 5.51 5.50 5.50 5.48 5.47 5.47 5.45 5.44 5.43 5.40 5.33 5.29 5.28 5.18 5.13 5.10 5.09 5.07 5.02 5.01 7 7 7 6 7 6 6 7 7 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 5 6 6 6 0.71 0.93 1.08 1.20 1.18 1.16 1.18 1.46 1.46 1.47 1.27 1.41 1.70 1.45 1.48 1.25 1.47 1.60 1.62 1.60 1.50 1.53 1.66 1.53 1.88 1.63 1.71 1.75 1.84 1.61 1.71 1.93 1.84 1.79 1.76 1.75 1.65 1.93 2.16 1.91 5 2 1 1 2 2 1 1 1 1 2 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 Table 15: MTurk Online Survey: Search Goods Groupon Category Vision and Eye Care Women’s Clothing Travel Packages Gym Membership Men’s Fashion Automotive Services Sweets Home Improvement Women’s Fashion Photo Services Entertainment Goods: movies, magazine subscription Co↵ee & Tea Golf Men’s Clothing Women’s (Clearance Sale) College Test Prep Courses Decor Organic Produce and Food Pet Products and Services Home Theater Personal Electronics Furniture Gadgets: Headphone, speaker dock, electronic clock Bowling Men’s (Clearance Sale) Women’s Accessories Cameras Men’s Accessories Baby Products: Clothing, Diapers, Monitors Home Goods T-Shirts GPS & Navigation Computers & Hardware Cooking Gadgets Groceries Sports Tickets Holiday Decor Gifts: candles, phone cases, stationary Office Supplies Gas (Attention Checking Question) N Mean SD Min Max 142 136 125 114 150 135 133 123 119 139 153 99 119 138 124 143 142 153 117 138 111 147 128 133 148 144 115 140 133 138 144 120 144 141 150 141 124 142 132 818 4.94 4.85 4.65 4.58 4.53 4.50 4.48 4.46 4.40 4.38 4.35 4.32 4.28 4.22 4.14 4.11 4.08 4.07 4.01 3.97 3.81 3.79 3.78 3.75 3.72 3.65 3.64 3.43 3.38 3.36 3.34 3.21 3.21 3.21 3.07 3.01 2.98 2.60 1.98 1.24 5 5 5 5 5 5 5 5 5 4 4 5 5 4 4 4 4 4 4 4 3 4 4 4 4 4 4 3 3 3 3 3 3 3 3 2 2 2 2 1 1.81 1.62 2.02 1.84 1.83 1.89 1.95 1.82 1.99 1.81 2.05 1.91 2.05 1.74 1.96 1.85 2.14 1.98 1.63 1.89 1.95 1.91 1.96 2.00 1.92 2.01 1.98 1.66 1.91 1.79 1.63 1.78 1.86 1.68 1.69 2.09 1.87 1.68 1.34 0.50 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 B Validation Survey We do a parallel o✏ine survey at a major U.S. university to check the robustness of our results. Twenty-eight business-school undergraduate students participated in the survey. Each student was asked to assess the product/service categories on Groupon. The category is coded as the experience type if more than half of the respondents coded so. Based on the survey, we re-estimate the consumer-side model for search and experience goods, respectively, and present the results in Table 16. Similarly, the estimates for the merchant-side model for search and experience goods are presented in Table 17. Our main conclusions hold using category coding from this validation survey. C Robust Analysis: Non-local Network E↵ects In this analysis, we replace the focal market’s SS 1 and CS0 e↵ects with the non- local counterparts. The parameter estimates of our main model are presented as follows. Table 16: Validation Survey: Parameter Estimates for Consumer-Side Model FE Experience Goods Est Lagged consumer base ( 1 ) Merchant base ( 1 ) Deal price (⇢1 ) Rival consumer base (✓1n ) Rival merchant base (✓1m ) Returning merchants Deal duration Week Week-square Sample size R-square Number of instruments Hansen test of overidentification Di↵erence-in-Hansen test for customer base Di↵erence-in-Hansen test for merchant base Di↵erence-in-Hansen test for price (SE) 0.030** (0.013) 1.238*** (0.029) -0.241*** (0.015) 0.071*** (0.012) -0.012 (0.034) -0.453*** (0.073) 0.109*** (0.010) -0.007*** (0.001) 1.9E-5 (4.2E-5) 5,354 0.737 Est (0.038) (0.061) (0.093) (0.018) (0.050) (0.181) (0.022) (0.001) (5.6E-5) 5,242 110 (101) = 101.71, p = 0.461 2 (50) = 14.02, p > 0.999 2 (9) = 1.11, p = 0.999 2 (51) = 12.17, p > 0.999 FE Est (SE) 0.128*** 1.449*** -0.365*** 0.068** -0.053 -0.570*** 0.118*** -0.007*** 6.5E-5 2 Search Goods Lagged consumer base ( 1 ) Merchant base ( 1 ) Deal price (⇢1 ) Rival consumer base (✓1n ) Rival merchant base (✓1m ) Returning merchants Duration Week Week-square Sample size R-square Number of instruments Hansen test of overidentification Di↵erence-in-Hansen test for consumer base Di↵erence-in-Hansen test for merchant base Di↵erence-in-Hansen test for price DGMM DGMM (SE) -0.019 (0.013) 2.099*** (0.044) -0.706*** (0.025) -0.018 (0.027) -0.041 (0.076) -0.163** (0.08) -0.032** (0.015) -0.012*** (0.002) -2.8E-4*** (9.5E-5) 3,662 0.551 Est -0.018 2.077*** -0.694*** -0.004 -0.084 0.063 -0.030 -0.015** -1.8E-4 (SE) (0.029) (0.136) (0.153) (0.057) (0.156) (0.397) (0.031) (0.007) (2.7E-4) 3,213 110 (101) = 91.85, p = 0.731 2 (50) = 23.3, p > 0.999 2 (9) = 6.15, p = 0.725 2 (51) = 32.25, p = 0.981 2 FE is the fixed-e↵ect estimator and DGMM is the first-di↵erence GMM estimator. *** p<0.01, ** p<0.05 Table 17: Validation Survey: Parameter Estimates for Merchant-Side Model FE Experience Goods Est Lagged merchant base ( 2 ) Consumer base ( 2 ) Deal price (⇢2 ) Rival consumer base (✓2n ) Rival merchant base (✓2m ) Returning merchants Week Week-square Sample size R-square Number of instruments Hansen test of overidentification Di↵erence-in-Hansen test for consumer base Di↵erence-in-Hansen test for merchant base Di↵erence-in-Hansen test for price SE 0.182*** (0.016) 0.278*** (0.006) 0.099*** (0.007) 0.013** (0.006) 0.158*** (0.016) 0.149** (0.036) 0.004*** (0.0004) -8.3E-5*** (2.0E-5) 5,354 0.882 Est (0.056) (0.019) (0.048) (0.010) (0.027) (0.064) (0.001) (2.0E-5) 5,242 111 (103) = 106.05, p = 0.399 2 (10) = 2.34, p = 0.993 2 (50) = 14.77, p > 0.999 2 (51) = 9.65, p > 0.999 FE Est SE 0.398*** 0.420*** 0.360*** -0.028*** 0.149*** 0.236*** 0.001*** -8.3E-5 2 Search Goods Lagged merchant base ( 2 ) Consumer base ( 2 ) Deal price (⇢2 ) Rival consumer base (✓2n ) Rival merchant base (✓2m ) Returning merchants Week Week-square Sample size R-square Number of instruments Hansen test Di↵erence-in-Hansen test for consumer base Di↵erence-in-Hansen test for merchant base Di↵erence-in-Hansen test for price DGMM DGMM SE 0.025 (0.017) 0.275*** (0.006) 0.263*** (0.01) 0.037*** (0.01) 0.065** (0.029) 0.081** (0.036) 0.010*** (0.001) 3.8E-4*** (3.8E-5) 3,662 0.833 Est 0.192*** 0.369*** 0.362*** 0.031* 0.024 0.078 0.006* 7.0E-5 SE (0.061) (0.019) (0.058) (0.015) (0.045) (0.057) (0.003) (1.3E-4) 3,213 111 (103) = 92.88, p = 0.753 2 (10) = 2.34, p = 0.993 2 (50) = 14.77, p > 0.999 2 (51) = 25.33, p = 0.999 2 FE is the fixed-e↵ect estimator and DGMM is the first-di↵erence GMM estimator. *** p<0.01, ** p<0.05, * p<0.1 Table 18: Parameter Estimates: Customer-Side Model for Non-local E↵ects FE Experience Goods Est Lagged customer base ( 1 ) Merchant base ( 1 ) Deal price (⇢1 ) Rival customer base (✓1n ) Rival merchant base (✓1m ) Returning merchants Deal duration Week Week-square Sample size (SE) Est (SE) 0.0006* (0.0004) 0.0140*** (0.0010) -0.265*** (0.017) 0.050*** (0.013) -0.032 (0.039) -0.417*** (0.08) 0.061*** (0.012) -0.014*** (0.001) 1.86E-05 (4.71E-05) 5,324 0.0010** (0.0004) 0.0130*** (0.0010) -0.473*** (0.098) 0.052** (0.022) 0.015 (0.061) -0.164 (0.114) 0.092*** (0.023) -0.016*** (0.001) 3.40E-05 (5.85E-05) 5,212 FE DGMM Search Goods Est Lagged customer base ( 1 ) Merchant base ( 1 ) Deal price (⇢1 ) Rival customer base (✓1n ) Rival merchant base (✓1m ) Returning merchants Deal duration Week Week-square Sample size DGMM (SE) -0.0008** (0.0004) 0.0132*** (0.0009) -0.271*** (0.023) 0.031 (0.029) -0.101 (0.085) -0.582*** (0.092) 0.031* (0.016) -0.007*** (0.002) 5.32E-05 (1.07E-04) 4,175 Est (SE) -0.0008* (0.0004) 0.0128*** (0.0016) -0.099 (0.176) 0.070 (0.049) -0.200 (0.128) -0.442*** (0.151) -0.027 (0.048) 0.001 (0.007) -3.48E-04 (2.71E-04) 3,801 FE is the fixed-e↵ect estimator and DGMM is the first-di↵erence GMM estimator. *** p<0.01, ** p<0.05 Table 19: Parameter Estimates: Merchant-Side Model for Non-local E↵ects FE Experience Goods Est Lagged merchant base ( 2 ) Customer base ( 2 ) Deal price (⇢2 ) Rival customer base (✓2n ) Rival merchant base (✓2m ) Returning merchants Week Week-square Sample size SE Est SE -0.0003 (0.0004) 0.0036*** (0.0001) 0.036*** (0.008) 0.001 (0.006) 0.059*** (0.018) 0.083** (0.038) -0.0002487 (0.001) -6.3E-05*** (2.21E-05) 5,324 -0.0004 (0.0004) 0.0035*** (0.0002) 0.186*** (0.044) -0.012 (0.007) 0.112*** (0.022) 0.207** (0.087) -0.001 (0.001) -2.07E-05 (3.07E-05) 5,212 FE DGMM Search Goods Est Lagged merchant base ( 2 ) Customer base ( 2 ) Deal price (⇢2 ) Rival customer base (✓2n ) Rival merchant base (✓2m ) Returning merchants Week Week-square Sample size DGMM SE 0.0010** (0.0005) 0.0029*** (0.0001) 0.078*** (0.009) 0.021* (0.012) 0.035 (0.035) -0.017 (0.038) 0.008*** (0.001) 9.54E-05** (4.26E-05) 4,175 Est SE -0.0005 (0.0008) 0.0035*** (0.0002) -0.098** (0.043) 0.019 (0.017) 0.035 (0.049) 0.027 (0.062) 0.01*** (0.003) 1.56E-05 (1.17E-04) 3,801 FE is the fixed-e↵ect estimator and DGMM is the first-di↵erence GMM estimator. *** p<0.01, ** p<0.05, * p<0.1