Browsing by Author "Mela, Carl F"

Television (TV) is the predominant advertising medium, and recent technological advances such as digital video recorders (DVRs) and set-top boxes (STBs) have the potential to transform this industry by enabling household-specific advertising. Since exposure to TV represents a substantial share of consumer time and attention, this potential to micro-target communications represents an enormous opportunity for the TV advertising market.

This paper outlines an approach to facilitate the micro-targeting of TV advertising. We employ a unique dataset, integrating TV program and advertisement viewing at the household level with purchase data, to address the question of how advertisers can achieve better advertising targeting in the digital context. Based on this dataset, we first develop a model of household TV viewing behavior. The viewing model comprises three integrated components: TV show sampling and watching, TV show recording, and advertising viewing. All three components are motivated by the theoretical concept of flow utility, that is, the moment-by-moment enjoyment a household derives from different activities: watching a TV show, watching a TV advertisement, and other non-TV activities. This model has decent out-of-sample prediction power on show choices and time spent on each selected show. We then link household advertising exposure with purchase. Finally, the viewing model and identified advertising-sales relationship are utilized to conduct counterfactual policy experiments on advertising targeting. We consider several household-level targeting scenarios by manipulating: 1) whether the advertising purchase is made in advance; and 2) whether the objective function is to minimize costs for a given set of exposures or to maximize revenues from advertising. Results indicate micro-targeting can lower advertising costs and raise incremental revenue.

The key contributions of this paper are as follows. Theoretically, we develop an integrated model on TV show viewing, TV advertising viewing, purchasing and advertising targeting. Methodologically, we propose a new modeling framework on media consumption by explicitly accounting for the role of uncertainty, and propose targeting strategies leveraging household-level data. Substantively, we offer policy recommendations to advertisers on micro-targeting which can be of great potential.

This dissertation presents two essays at the intersection of quantitative marketing and empirical industrial organization. The first essay, based on a joint research project with Andres Musalem, studies how consumers respond to subsidy designs with different spending restrictions in the context of SNAP (commonly known as the Food Stamps Program), with implications to the optimal design of consumer subsidies. In the second essay, I empirically examine how supply-side heterogeneity on a crowd-based platform moderates policy outcomes of platform initiatives.

In the past five years, the number of Americans using the Internet as their main source of news has doubled to more than 40%, while those choosing newspapers has dropped to just over 30%. Although this trend signals a shift in the consumption of information in favor of hyper-media, wherein excerpting and linking is commonplace, research regarding how consumers acquire news in online environments has not proceeded apace. This dissertation presents a model of forward-looking consumers who gain knowledge and utility through the search and consumption of hyper-media, and explore its implications for welfare and site policy.

The model is estimated using comScore browsing data for five celebrity news and gossip sites, supplemented with link data scraped from site archives. The model is estimated by coupling Imai, Jain, and Ching's (2009; Econometrica 77(6):1865-1899) method with the Riemann manifold sampling algorithm of Girolami and Calderhead (2011; JRSS-B 73(2):123-214). The pairing of these recent advances enables researchers to estimate dynamic models with many more variables and states than previously considered.

Results indicate consumers are heterogeneous in their preferences for certain types of celebrity sites (e.g., females prefer sites that emphasize gossip over pictorials of female models and entertainers). Results also indicate that links to other sites are informative about the target site's potential quality; after observing one link to another site, consumers' uncertainty about the linked site's quality is about 20% lower.

Counterfactual analyses show that network throttling, in which access to all sites is slowed down uniformly in an effort to reduce congestion, has an unequal effect on site traffic. For example, sites frequented by consumers with higher than average browsing costs lose a greater share of their traffic. They also show how a change in fair use law, in which excerpts and links become less informative of the linked sites' quality, causes consumers with more extreme tastes to curtail their searches, consequently decreasing traffic at sites with mainstream content and many inbound links. A third counterfactual experiment uses the model to explore the implications of pay walls for site traffic.

This dissertation considers monetization strategies in the context

of online product and display ad marketplaces.

The first chapter considers online marketplace platforms that trade-off

their fees from advertising with commissions from product sales. While

featuring advertised products can make search less efficient (lowering

transaction commissions), it incentivizes sellers to compete for better

placements via advertising (increasing advertising fees). We consider

this trade-off by modeling both sides of the platform. On the demand

side, we develop a joint model of browsing (impressions), clicking,

and purchase. On the supply side, we consider sellers' valuation and

advertising competition under various fee structures (CPM, CPC, CPA)

and ranking algorithms.

Using buyer, seller, and platform data from an online marketplace

where advertising dollars affect the order of seller items listed,

we explore various product ranking and ad pricing mechanisms. We find

that sorting items below the fifth position by expected sales revenue

while conducting a CPC auction in the top 5 positions yields the greatest

improvement in profits (181%) because this approach balances the

highest valuations from advertising in the top positions with the

transaction revenues in the lower positions.

The second chapter considers how a publisher should set reserve prices

for real-time bidding (RTB) auctions when selling display advertising

impressions through ad exchanges. Through a series of field experiments,

we show that a reserve price set based on an imputed demand curve

(in the absence of constraints) can increase publisher's revenues

by 32%, thereby affirming the importance of reserve price in maximizing

publisher's revenues. Further, we find that advertisers increase their

bids in response to an experimental increase in reserve price and

show this behavior is consistent with the use of a minimum impression

constraint to ensure advertising reach.

Based on this insight, we construct an advertiser bidding model and

use it to infer the overall demand curve for advertising as a function

of reserve prices. Using this constraint-based demand model, we solve

the publisher pricing problem. Incorporating the minimum impression

constraint into the reserve price setting process yields a 50% increase

over a solution that does not incorporate the constraint and an additional

increase in profits of 9 percentage points.

Central to the explosive growth of the Internet has been the desire

of dispersed buyers and sellers to interact readily and in a manner

hitherto impossible. Underpinning these interactions, auction

pricing mechanisms have enabled Internet transactions in novel ways.

Despite this massive growth and new medium, empirical work in

marketing and economics on auction use in Internet contexts remains

relatively nascent. Accordingly, this dissertation investigates the

role of online auctions; it is composed of three essays.

The first essay, ``Online Auction Demand,'' investigates seller and

buyer interactions via online auction websites, such as eBay. Such

auction sites are among the earliest prominent transaction sites on

the Internet (eBay started in 1995, the same year Internet Explorer

was released) and helped pave the way for e-commerce. Hence, online

auction demand is the first topic considered in my dissertation. The

second essay, ``A Dynamic Model of Sponsored Search Advertising,''

investigates sponsored search advertising auctions, a novel approach

that allocates premium advertising space to advertisers at popular

websites, such as search engines. Because sponsored search

advertising targets buyers in active purchase states, such

advertising venues have grown very rapidly in recent years and have

become a highly topical research domain. These two essays form the

foundation of the empirical research in this dissertation. The third

essay, ``Sponsored Search Auctions: Research Opportunities in

Marketing,'' outlines areas of future inquiry that I intend to

pursue in my research.

Of note, the problems underpinning the two empirical essays exhibits

a common form, that of a two-sided network wherein two parties

interact on a common platform (Rochet and Tirole, 2006). Although

theoretical research on two-sided markets is abundant, this

dissertation focuses on their use in e-commerce and adopts an

empirical orientation. I assume an empirical orientation because I

seek to guide firm behavior with concrete policy recommendations and

offer new insights into the actual behavior of the agents who

interact in these contexts. Although the two empirical essays share

this common feature, they also exhibit notable differences,

including the nature of the auction mechanism itself, the

interactions between the agents, and the dynamic frame of the

problem, thus making the problems distinct. The following abstracts

for these two essays as well as the chapter that describes my future

research serve to summarize these contributions, commonalities and

differences.

Online Auction Demand

With $40B in annual gross merchandise volume, electronic auctions

comprise a substantial and growing sector of the retail economy. For

example, eBay alone generated a gross merchandise volume of $14.4B

during the fourth quarter of 2006. Concurrent with this growth has

been an attendant increase in empirical research on Internet

auctions. However, this literature focuses primarily on the bidder;

I extend this research to consider both seller and bidder behavior

in an integrated system within a two-sided network of the two

parties. This extension of the existing literature enables an

exploration of the implications of the auction house's marketing on

its revenues as well as the nature of bidder and seller interactions

on this platform. In the first essay, I use a unique data set of

Celtic coins online auctions. These data were obtained from an

anonymous firm and include complete bidding and listing histories.

In contrast, most existing research relies only on the observed

website bids. The complete bidding and listing histories provided by

the data afford additional information that illuminates the insights

into bidder and seller behavior such as bidder valuations and seller

costs.

Using these data from the ancient coins category, I estimate a

structural model that integrates both bidder and seller behavior.

Bidders choose coins and sellers list them to maximize their

respective profits. I then develop a Markov Chain Monte Carlo (MCMC)

estimation approach that enables me, via data augmentation, to infer

unobserved bidder and seller characteristics and to account for

heterogeneity in these characteristics. My findings indicate that:

i) bidder valuations are affected by item characteristics (e.g., the

attributes of the coin), seller (e.g. reputation), and auction

characteristics (e.g., the characteristics of the listing); ii)

bidder costs are affected by bidding behavior, such as the recency

of the last purchase and the number of concurrent auctions; and iii)

seller costs are affected by item characteristics and the number of

concurrent listings from the seller (because acquisition costs

evidence increasing marginal values).

Of special interest, the model enables me to compute fee

elasticities, even though no variation in historical fees exists in

these data. I compute fee elasticities by inferring the role of

seller costs in their historical listing decision and then imputing

how an increase in these costs (which arises from more fees) would

affect the seller's subsequent listing behavior. I find that these

implied commission elasticities exceed per-item fee elasticities

because commissions target high value sellers, and hence, commission

reductions enhance their listing likelihood. By targeting commission

reductions to high value sellers, auction house revenues can be

increased by 3.9%. Computing customer value, I find that attrition

of the largest seller would decrease fees paid to the auction house

by $97. Given that the seller paid $127 in fees, competition

offsets only 24% of the fees paid by the seller. In contrast,

competition largely in the form of other bidders offsets 81% of the

$26 loss from buyer attrition. In both events, the auction house

would overvalue its customers by neglecting the effects of

competition.

A Dynamic Model of Sponsored Search Advertising

Sponsored search advertising is ascendant. Jupiter Research reports

that expenditures rose 28% in 2007 to $8.9B and will continue to

rise at a 26% Compound Annual Growth Rate (CAGR), approaching half

the level of television advertising and making sponsored search

advertising one of the major advertising trends affecting the

marketing landscape. Although empirical studies of sponsored search

advertising are ascending, little research exists that explores how

the interactions of various agents (searchers,

advertisers, and the search engine) in keyword

markets affect searcher and advertiser behavior, welfare and search

engine profits. As in the first essay, sponsored search constitutes

a two-sided network. In this case, bidders (advertisers) and

searchers interact on a common platform, the search engine. The

bidder seeks to maximize profits, and the searcher seeks to maximize

utility.

The structural model I propose serves as a foundation to explore

these outcomes and, to my knowledge, is the first structural model

for keyword search. Not only does the model integrate the behavior

of advertisers and searchers, it also accounts for advertisers

competition in a dynamic setting. Prior theoretical research has

assumed a static orientation to the problem whereas prior empirical

research, although dynamic, has focused solely on estimating the

dynamic sales response to a single firm's keyword advertising

expenditures.

To estimate the proposed model, I have developed a two-step Bayesian

estimator for dynamic games. This approach does not rely on

asymptotics and also facilitates a more flexible model

specification.

I fit this model to a proprietary data set provided by an anonymous

search engine. These data include a complete history of consumer

search behavior from the site's web log files and a complete history

of advertiser bidding behavior across all advertisers. In addition,

the data include search engine information, such as keyword pricing

and website design.

With respect to advertisers, I find evidence of dynamic

bidding behavior. Advertiser valuation for clicks on their sponsored

links averages about $0.27. Given the typical $22 retail price of

the software products advertised on the considered search engine,

this figure implies a conversion rate (sales per click) of about

1.2%, well within common estimates of 1-2% (gamedaily.com). With

respect to consumers, I find that frequent clickers place a

greater emphasis on the position of the sponsored advertising link.

I further find that 10% of consumers perform 90% of the clicks.

I then conduct several policy simulations to illustrate the effects

of change in search engine policy. First, I find that the

search engine obtains revenue gains of nearly 1.4% by sharing

individual level information with advertisers and enabling them to

vary their bids by consumer segment. This strategy also improves

advertiser profits by 11% and consumer welfare by 2.9%. Second, I

find that a switch from a first to second price auction results in

truth telling (advertiser bids rise to advertiser valuations), which

is consistent with economic theory. However, the second price

auction has little impact on search engine profits. Third, consumer

search tools lead to a platform revenue increase of 3.7% and an

increase of consumer welfare of 5.6%. However, these tools, by

reducing advertising exposure, lower advertiser profits by 4.1%.

Sponsored Search Auctions: Research Opportunities in Marketing

In the final chapter, I systematically review the literature on

keyword search and propose several promising research directions.

The chapter is organized according to each agent in the search

process, i.e., searchers, advertisers and the search engine, and

reviews the key research issues for each. For each group, I outline

the decision process involved in keyword search. For searchers, this

process involves what to search, where to search, which results to

click, and when to exit the search. For advertisers, this process

involves where to bid, which word or words to bid on, how much to

bid, and how searchers and auction mechanisms moderate these

behaviors. The search engine faces choices on mechanism design,

website design, and how much information to share with its

advertisers and searchers. These choices have implications for

customer lifetime value and the nature of competition among

advertisers. Overall, I provide a number of potential areas of

future research that arise from the decision processes of these

various agents.

Foremost among these potential areas of future research are i) the

role of alternative consumer search strategies for information

acquisition and clicking behavior, ii) the effect of advertiser

placement alternatives on long-term profits, and iii) the measure of

customer lifetime value for search engines. Regarding the first

area, a consumer's search strategy (i.e., sequential search and

non-sequential search) affects which sponsored links are more likely

to be clicked. The search pattern of a consumer is likely to be

affected by the nature of the product (experience product vs. search

product), the design of the website, the dynamic orientation of the

consumer (e.g., myopic or forward-looking), and so on. This search

pattern will, in turn, affect advertisers payments, online traffic,

sales, as well as the search engine's revenue. With respect to the

second area, advertisers must ascertain the economic value of

advertising, conditioned on the slot in which it appears, before

making decisions such as which keywords to bid on and how much to

bid. This area of possible research suggests opportunities to

examine how advertising click-through and the number of impressions

differentially affect the value of appearing in a particular

sponsored slot on a webpage, and how this value is moderated by an

appearance in a non-sponsored slot (i.e., a slot in the organic

search results section). With respect to the third area of future

research, customer value is central to the profitability and

long-term growth of a search engine and affects how the firm should

allocate resources for customer acquisition and retention.

Organization

This dissertation is organized as follows. After this brief

introduction, the essay, ``Online Auction Demand,'' serves as a

basis that introduces some concepts of auctions as two-sided

markets. Next, the second essay, ``A Dynamic Model of Sponsored

Search Advertising,'' extends the first essay by considering a

richer context of bidder competition and consumer choice behavior.

Finally, the concluding chapter, which outlines my future research

interests, considers potential extensions that pertain especially to

sponsored search advertising.