37.5 Consumer Search
37.5.1 (Jiang and Zou 2020) Consumer Search and Filtering on Online Retail Platforms
Findings
Lower search cost reduces seller’s profit margin (due to increased competition), but also attract more consumers to the platform
If referral fee is exogenously determined, a decrease in search cost can either increase and decrease the platform’s profit
Hence, platform should compensate by raising its referral fee, but it actually depends on the effect of search cost on demand elasticity:
If the search cost increases the demand elasticity, platform should lower its referral fee
If the search cost decreases the demand elasticity, the platform should increase its referral fee.
If referral fee is set optimally, the platform’s profit will increase (kinda intuitive)
With the filtering feature, consumers will be more likely to buy after searching due to reduction in uncertainty about the product match value for consumers
Motivation:
The search costs tend to decrease overtime
However, there is not study that study the platform’s optimal referral-fee in respect to search costs, filtering.
Research questions:
How search cost affects optimal referral fee and profit?
How filtering affects the platform, sellers, and consumers?
How filtering impact differs from that of a reduction of the search cost?
Results are robust to
heterogeneous search costs
heterogeneous product quality
platform charges a fixed per-unit referral fee (instead of percentage)
Related Literature
Consumer search: mixed literature on whether to increase or decrease search cost
Information structure effect on consumer search and firm profits
This paper:
Add filtering (different from target search)
search cost affects platform’s and sellers’ optimal pricing decision
search cost affects platform’s optimal referral fee
Setup
Horizontally differentiated products
After search (cost), consumers learn the price and match value (with the product)
Match value = f(filterable match value + unfilterable match value)
Consumer utility
\[ u_{ij} = M_{ij} + p_i \]
where
\(M_{ij} = \mu_{ij} + m_{ij}\) = match value of product \(i\) to consumer \(j\)
\(\mu_{ij} = \delta \times \hat{\mu}\) = filterable match value
\(\delta >0\) is a cosntnat scale factor
\(\hat{\mu}\) = mean-zero discrete random variable with \(K\) realizations
\(m_{ij}\) = unfilterable match value
It’s ok if \(\mu_{ij}\) correlates with \(m_{ij}\)
\(p_i\) = product price
Consumer search
- Consumer know her outside option \(u_{0j}\) and \(\mu_{ij}\) (filterable match value)
- Consumer do not know \(m_{ij}\) (match value) and price before hand, only after search cost \(\tau\)
Consumers decisions
- Whether to search on the platform
- Search sellers sequentially and learn \(m_{ij}\) and \(p_i\)
- Consumers can buy, search more (to the next product with highest \(\mu_{ij}\)), and leave
Game Timing
- Retail platform sets a percentage referral fee \(r \in (0,1)\)
- Sellers set price where profit is \((1-r)p_i-c\)
- Consumers make search and purchase decisions.
Notes:
\(p^*(r)\) = equilibrium retail price
\(r^*\) = optimal percentage referral fee
In equilibrium, consumers will continue search until the next product expected utility beyond \(u_{ij}\) is less than \(\tau\) (search cost) (Wolinsky 1986)
Result 1: consumer purchase if \(m_{ij} \ge \bar{m} (\tau)\) where \(\bar{m}(\tau)\) satisfies \(\int_{\bar{m}(\tau)}^{m_{max}} (m - \bar{m}(\tau)) dF(m) = \tau\)
Result 2: \(\bar{m}(\tau)\) decreases with \(\tau\)
In equilibrium, \(\bar{m}(\tau)\) is the equilibrium acceptance threshold for the unfilterable match value
For exogenous referral fee
Lemma 1: retail price, sellers’ profit, total demand
Result 3: As the cost of searching reduces, a consumer’s expected number of searches before purchasing increases, as does the expected aggregate match value \(M_{ij}\) of the product she will buy.
Lemma 2: Market demand and sellers’ profits change with the search cost (\(\tau\)). Lower search cost intensify sellers’ competition, which reduces their prices and increase profits due to a demand-expanding effect (increase the whole pie). The platform’s equilibrium profit can also increase or decrease as the consumer’s search cost decreases
Endogenous referral fee
Proposition 1: A decrease in search cost increases the platform’s expected profit
Proposition 2: As the consumer’s search cost \(\tau\) decreases, the platform will
- Increase its referral fee if the search does not increase the demand elasticity (but communication is important here)
- Decreases its referral fee if the search cost increase the demand elasticity
Effects of Filtering
Consumer’s optimal search strategy
Lemma 3: Filtering increases the consumers’ acceptance threshold for the aggregate match value by less than \(\mu_K\)
Filtering allows consumers to have smaller consideration set, which raises acceptance threshold for the aggregate match value
Filtering reduces the direct benefit of searching. Hence, consumers are more likely to stop searching at a lower acceptance threshold.
But the first effect > the second effect.
Proposition 3: filtering increases a consumer’s probability of buying a product after searching it, and reduces the consumer’s expected number of search.
Seller’s pricing decisions: filtering reduces retail price
Proposition 4: Filtering reduces consumer’s uncertainty about the product’s match value
Consumers only search products with the highest filterable attributes, reducing the match-value differentiation among product (increase price competition, and lower equilibrium price)
filtering decreases consumers’ equilibrium probability of continuing search after searching a product, reducing competition and higher equilibrium prices.
Proposition 5: Filtering increases the platform’s demand, seller’s profit, and platform’s profit, and consumer surplus under small scale factor for the match value and referral fee is exogenous
Conclusion
Reduction in search cost makes consumers search more, while filtering reduces the benefit of search making consumers search power products
Decrease in search cost intensifies seller’s competition, leading to lower retail prices. While filtering softens seller competition, and lead to higher retail prices
When the referral fee is exogenous, a lower search cost can either be good or bad for the platform, and consumers. Whereas filtering will always benefit the sellers, the platform, and consumers if filtering only reveals a small amount of the variations in the match value.
Heterogeneous search costs:
- Proposition 6: Optimal retail price is the weighted average of the homogeneous-search cost equilibrium prices
Heterogeneous Product quality:
- Premium sellers charge a higher price (than non-premium sellers), but the difference is smaller than their quality difference
Fixed Per-unit referral fee:
- When the platform’s referral fee is exogenous, a reduction in search costs will always increase the platform’s profit under a set per-unit referral price, but it will lower the platform’s profit under a percentage referral fee.
37.5.2 (Yanbin Chen et al. 2021) Consumer Search with blind buying
Objective and Motivation
Consumers don’t have relevant info and must invest effort to purchase a suitable product
Previous search literature, consumers pay a fixed search cost to visit a store and fully informed about the price and match value at the store
These models do not capture
Online search technology that lower the cost of visiting stores and collecting info about prices
Consumers cannot physically inspect and try products.
This paper designs a sequential search model to capture these two aspects
Sequence:
Before searching, consumers observe the prices and prior values of both firms, and they pay a search cost to learn the match value
Consumers have to decide product to search firs, and whethe to continue searching after inspecting the first product. If consumers decide to stop searching, they can either buy the insepcted product or bldinly buy an unispected product
Model construction
Duopoly market, 2 firms sell horizontally differentiated products in price competition.
The random utility (net of nay search cost) fo consumer \(l\), who purchases the product from firm \(i\) is \(\tilde{u}_{il} = \tilde{\eta}_{il} + \tilde{\epsilon}_{il} - p_i\)
\(\tilde{\eta}_{il}\) = observable match value
\(\tilde{\epsilon}_{il}\) = unobservable match value
The key feature is that consumers can blindly buy a product
At any stage of search process, consumers have the option of purchasing a product without inspecting its match value \(\tilde{\epsilon}_{il}\) where no search cost is incurred
The expected utility for consumer blind search is \(E(\tilde{u}_{il}) = \eta_{il} + E(\tilde{\epsilon}_{il}) - p_i\)
Search is assumed to be costly
Game sequence
- Firms determine their prices
- Each consumer costlessly observes both the prices and the prior values of the products and decides whether to blindly buy a product. the game ends if a blind buying takes place
- The game continue and after learning product match value, the costumer decides among three 3 options
buy product without uncertainty
blindly buy product without inspecting its match value
discover both realized value
Results
The effects of blind buying on equilibrium prices
- In price-directed search models without the blind buying option, search friction has 2 effect on prices competition.
When the match value has a symmetric distribution, both consumer and firms are indifferent to the search order, conditioning on searching
An increase int eh first sample search cost has no effect on market prices in existing price-directed search models, reducing equilibirum prices in the model
37.5.3 (X. Li and Xu 2022) Superior knowledge, price discrimination, and customer inspection
Customer inspection : learn about their intrinsic preferences.
Superior knowledge: firms know customers preferences
Setup
A monopolistic firm selling a product to a continum of customers of unit measure
Customer have heterogeneous match values with the firm’s product (either high or low)
the prior probability of a high match value is \(\alpha\)
Information structure:
Customers know their expected match value, they are ex ante uninformed about their idiosyncratic shock and do not know their match value beyond its prior distribution
Customer inspection: the cost that consumer make for an informed purchase. The inspection cost is independent of rational inferences
firm knows both its product characteristics and preference of individual customers.
Timeline
- Each customer has their match
- Firm offers each customer a personalized price according to her preference. The firm chooses the probability distribution of its price
- Observing the personalized price, customer decides whether to inspect to find out her true preference by incurring an inspection cost with probability
- Based on the inspection outcome, if the customer chooses to inspect, the customer decides whether to make a purchase