INBAL YAHAV WOLFGANG JANK R.H. SMITH SCHOOL OF BUSINESS, UNIVERSITY OF MARYLAND E-Loyalty Networks in Online Auctions.

INBAL YAHAV WOLFGANG JANK

R.H. SMITH SCHOOL OF BUSINESS, UNIVERSITY OF MARYLAND

E-Loyalty Networks in Online Auctions

Motivation

Sellers Bidders

Objective High profit High conversion rate

Get the product? Low price?

Get the product Get the product (quality)

Means Trust Feedback score Lit

Auction design (e.g., open price, duration, etc.)

IS THAT ENOUGH??

Actors

Research Questions

1. How to define and measure e-loyalty?

3. What factors drive loyalty in online auctions?

2. How does loyalty impact auction outcome (price, conversion)?

Data

~350 Sellers~700 Repeating Buyers

Loyalty in the Literature

Definition: repeating purchases

Brand-switch literature: Probability of switching to another brand Distribution of purchases across different brands

(commonly 2 brands)

Research Questions

1. How to define and measure e-loyalty?

3. What factors drive loyalty in online auctions?

2. How does loyalty impact auction outcome (price, conversion)?

Define and Measure eLoyalty

Three steps measurements

Construct eLoyalty network Transform network into loyalty distribution Transform the distribution into quantifiers using PC

analysis

Define and Measure eLoyalty

eLoyalty networks

Bipartite graph with: First nodes set: sellers (red) Second node set: buyers (white)

Arcs: purchases, with the width corresponding to the number of interactions

Define and Measure eLoyalty

eLoyalty disribution

SellersBuyers

100%

100%

100%

70%

80%

2. Measure the perceived loyalty per seller (~distribution of the weighted in-degree)

1. Measure proportion of interactions per buyer (~normalized distribution of out-degree)

30%

Define and Measure eLoyalty

Transform the distribution into two quantifiers (PC1, PC2) that measure the difference between the sellers’ perceived loyalty.

m sellers

(discrete grid)

First & Second PCA Scores

(~80% of the variation)

InputPCA

Sellers’ Perceived eLoyalty: PCAs

Most weight on medium-scores PC2 contrasts the moderate loyalty distributionfrom the extremes – distinguishes sellers that have neither extremely loyal norextremely disloyal bidders

Very little weight on low scores , very large weight on high scores (between 0.8 and 1 PC1 contrasts distributions of sellers with extremely loyal bidders with those that are little loyal

Research Questions

1. How to define and measure e-loyalty?

3. What factors drive loyalty in online auctions?

2. How does loyalty impact auction outcome (price, conversion)?

Modeling eLoyalty : Effect of eLoyalty on Price

OLS/ WLS regression High volume sellers have multiple, inter-dependent

auctions Low-volume sellers have only few auctions Violates regression assumption

Modeling eLoyalty : Effect of eLoyalty on Price

Random-effects regression model Account for seller-specific variation

Heteroscedasticity

Modeling eLoyalty : Effect of eLoyalty on Price

Segment sellers into three groups

Modeling eLoyalty : Effect of eLoyalty on Price

Segment sellers into three groups: model fit

Low volume Medium volume High volume

R2=0.81 R2=0.77 R2=0.83

Effect of eLoyalty on Price

The effect of loyalty depends strongly on size of the seller: High volume sellers can extract huge price-premiums from loyal

bidders The impact of loyalty is much smaller for sellers of smaller scale

Coefficient Medium volume Low volume High volume(Intercept) -0.21 0 4StartPrice 0.08 0.04 0.05AuctionDuration 0 0 0log(ItemQuantity) 0.12 0.14 0.08Bidcount 0.11 0.13 0.07log(Pieces) 0.19 0.08 0.36Size 0.03 0.03 0.07log(SellerFeedback) 0.04 0.04 0.12log(Volume) -0.04 0.01 -0.81PC1 0.21 -0.24 2.74PC2 0.02 -1.72 -15.41

Summary

Define and measure eLoyalty eLoyalty network Buyers loyalty ~ normalized distribution of out-degree Seller perceived loyalty ~ distribution of the weighted

in-degree Transform the distribution into quantifiers using PC

analysis

Modeling eLoyalty: data segmentationConclusions

Loyalty has higher impact on high volume sellers Saturated market

Discussion

The analysis can be replicated to other products; the results might change

Temporal networks Examine the evaluation of eLoyalty We did not observe temporal effect in our data

More Information?

Inbal [email protected]

http://www.rhsmith.umd.edu/faculty/phd/inbal/