Asymmetric information in programmatic advertising: Three studies on adverse selection, mechanism choices, and fee structures

The display advertising market has evolved dramatically in recent times. Firms can leverage sophisticated real-time supply chains that go under the umbrella term of emph{programmatic advertising} to deliver their promotional content to the right customer, at the right moment, and in the right place.

Despite these premises and the ever increasing economic weight of the advertising industry, programmatic still suffers from a diffused lack of transparency that originates from the vast and often confusing competitive and technological landscape that advertisers (buyers), publishers (sellers) and Ad Exchanges (auction platforms) have to navigate. In this dissertation we present three studies that investigate the role of asymmetric information in shaping the relationship between advertisers, publishers and Ad Exchanges.

In the first study (Chapter 3), we use a game-theoretical model and an empirical analysis to investigate the implications of asymmetric information in the relationship between advertisers and publishers. We study the welfare consequences of publishers' heterogeneous adoption of the quality signaling practice of allocating higher-quality impressions to premium contracts (PMP). Our analytical results show that premium contracts lead to adverse selection in the open market (RTB) while not increasing advertisers' utility. Our empirical findings support our analytical framework by providing strong evidence of adverse selection in RTB. Our findings challenge the current understanding of PMPs as an effective signaling device and highlight the threat they pose to the integrity of programmatic.

In the second study (Chapter 4), we examine ad exchanges' (ADX) mechanism choices between the second-price to the first-price auction. We address the effect that mistrust towards the ADXs' commitment to the second-price rules, and the ADXs' fee-setting policies have on their mechanism choices. First, we present descriptive empirical evidence from a large proprietary dataset of bid requests. Then, we construct a game-theoretical model of the interactions between publishers, advertisers, and ADXs.
We find that, while publishers generally prefer trading via the first-price auction, ADXs that offer second-price auctions may still conquer a significant market share depending on their competitors' equilibrium fees. Furthermore, we find that the prevalence of second-price auctions depends on the publishers' size and the likelihood that ADXs engage in shill bidding in second-price auctions. Our findings reconcile an ongoing debate on the mechanism switch by clarifying the seemingly problematic timeline of the ADXs' move to the first-price auction.

In the third study (Chapter 5), we investigate the incentives for ADXs in deciding their business model by setting buy-side and sell-side fees. In this study, we use an analytical model to research the impact of recent changes in publishers' strategy for clearing unsold inventory on such incentives of ADXs. In recent years, efficiency concerns have pushed publishers towards abandoning the legacy sequential Waterfall mechanism in favor of the newer parallel Header Bidding mechanism to manage the competition between affiliated ADXs. Despite being commonly viewed as a revenue-increasing move, the publisher's mechanism change eliminates all incentives for ad exchanges to keep sell-side fees low, thereby hurting publishers' revenues. Our findings explain the recent surge in sell-side fees, commonly referred to as the Ad Tech-Tax.

The three studies of this dissertation bring substantial theoretical and practical contributions to the digital advertising literature. From the theoretical perspective, our work brings new insight into how asymmetric information and conflicting incentives shape the relationships between advertisers, publishers, and ADX. From the managerial perspective, our work brings three surprising results that challenge the current understanding of the mechanisms and contractual tools in programmatic.