Santiago R. Balseiro

An Ant Colony algorithm hybridized with insertion heuristics for the Time Dependent Vehicle Routing Problem with Time Windows

This paper presents an Ant Colony System algorithm hybridized with insertion heuristics for the Time-Dependent Vehicle Routing Problem with Time Windows (TDVRPTW). In the TDVRPTW a fleet of vehicles must deliver goods to a set of customers, time window constraints of the customers must be respected and the fact that the travel time between two points depends on the time of departure has to be taken into account. The latter assumption is particularly important in an urban context where the traffic plays a significant role. A shortcoming of Ant Colony algorithms for capacitated routing problems is that, at the final stages of the algorithm, ants tend to create infeasible solutions with unrouted clients. Hence, we propose enhancing the algorithm with an aggressive insertion heuristic relying on the minimum delay metric. Computational results confirm the benefits of more involved insertion heuristics. Moreover, the resulting algorithm turns out to be competitive, matching or improving the best known results in several benchmark problems.

Yield Optimization of Display Advertising with Ad Exchange

It is clear from the growing role of ad exchanges in the real-time sale of advertising slots that Web publishers are considering a new alternative to their more traditional reservation-based ad contracts. To make this choice, the publisher must trade off, in real-time, the short-term revenue from ad exchange with the long-term benefits of delivering good spots to the reservation ads. In this paper we formalize this combined optimization problem as a multiobjective stochastic control problem and derive an efficient policy for online ad allocation in settings with general joint distribution over placement quality and exchange prices. We prove the asymptotic optimality of this policy in terms of any arbitrary trade-off between the quality of delivered reservation ads and revenue from the exchange, and we show that our policy approximates any Pareto-optimal point on the quality-versus-revenue curve. Experimental results on data derived from real publisher inventory confirm that there are significant benefits ...

The cost of moral hazard and limited liability in the principal-agent problem

In the classical principal-agent problem, a principal hires an agent to perform a task. The principal cares about the tasks output but has no control over it. The agent can perform the task at different effort intensities, and that choice affects the tasks output. To provide an incentive to the agent to work hard and since his effort intensity cannot be observed, the principal ties the agents compensation to the tasks output. If both the principal and the agent are risk-neutral and no further constraints are imposed, it is well-known that the outcome of the game maximizes social welfare. In this paper we quantify the potential social-welfare loss due to the existence of limited liability, which takes the form of a minimum wage constraint. To do so we rely on the worst-case welfare loss--commonly referred to as the Price of Anarchy--which quantifies the (in)efficiency of a system when its players act selfishly (i.e., they play a Nash equilibrium) versus choosing a socially-optimal solution. Our main result establishes that under the monotone likelihood-ratio property and limited liability constraints, the worst-case welfare loss in the principal-agent model is exactly equal to the number of efforts available.

Dynamic Revenue Sharing

Many online platforms act as intermediaries between a seller and a set of buyers. Examples of such settings include online retailers (such as Ebay) selling items on behalf of sellers to buyers, or advertising exchanges (such as AdX) selling pageviews on behalf of publishers to advertisers. In such settings, revenue sharing is a central part of running such a marketplace for the intermediary, and fixed-percentage revenue sharing schemes are often used to split the revenue among the platform and the sellers. In particular, such revenue sharing schemes require the platform to (i) take at most a constant fraction \alpha of the revenue from auctions and (ii) pay the seller at least the seller declared opportunity cost c for each item sold. A straightforward way to satisfy the constraints is to set a reserve price at c / (1 - \alpha) for each item, but it is not the optimal solution on maximizing the profit of the intermediary. While previous studies (by Mirrokni and Gomes, and by Niazadeh et al) focused on revenue-sharing schemes in static double auctions, in this paper, we take advantage of the repeated nature of the auctions. In particular, we introduce dynamic revenue sharing schemes where we balance the two constraints over different auctions to achieve higher profit and seller revenue. This is directly motivated by the practice of advertising exchanges where the fixed-percentage revenue-share should be met across all auctions and not in each auction. In this paper, we characterize the optimal revenue sharing scheme that satisfies both constraints in expectation. Finally, we empirically evaluate our revenue sharing scheme on real data.

Competition and Yield Optimization in Ad Exchanges

Competition and Yield Optimization in Ad Exchanges Santiago R. Balseiro Ad Exchanges are emerging Internet markets where advertisers may purchase display ad placements, in real-time and based on specific viewer information, directly from publishers via a simple auction mechanism. The presence of such channels presents a host of new strategic and tactical questions for publishers. How should the supply of impressions be divided between bilateral contracts and exchanges? How should auctions be designed to maximize profits? What is the role of user information and to what extent should it be disclosed? In this thesis, we develop a novel framework to address some of these questions. We first study how publishers should allocate their inventory in the presence of these new markets when traditional reservation-based ad contracts are available. We then study the competitive landscape that arises in Ad Exchanges and the implications for publishers’ decisions. Traditionally, an advertiser would buy display ad placements by negotiating deals directly with a publisher, and signing an agreement, called a guaranteed contract. These deals usually take the form of a specific number of ad impressions reserved over a particular time horizon. In light of the growing market of Ad Exchanges, publishers face new challenges in choosing between the allocation of contract-based reservation ads and spot market ads. In this setting, the publisher should take into account the tradeoff between short-term revenue from an Ad Exchange and the long-term impact of assigning high quality impressions to the reservations (typically measured by the click-through rate). In the first part of this thesis, we formalize this combined optimization problem as a stochastic control problem and derive an efficient policy for online ad allocation in settings with general joint distribution over placement quality and exchange bids, where the exchange bids are assumed to be exogenous and independent of the decisions of the publishers. We prove asymptotic optimality of this policy in terms of any arbitrary trade-off between quality of delivered reservation ads and revenue from the exchange, and provide a bound for its convergence rate to the optimal policy. We also give experimental results on data derived from real publisher inventory, showing that our policy can achieve any Pareto-optimal point on the quality vs. revenue curve. In the second part of this thesis, we relax the assumption of exogenous bids in the Ad Exchange and study in more detail the competitive landscape that arises in Ad Exchanges and the implications for publishers’ decisions. Typically, advertisers join these markets with a prespecified budget and participate in multiple second-price auctions over the length of a campaign. We introduce the novel notion of a Fluid Mean Field Equilibrium (FMFE) to study the dynamic bidding strategies of budget-constrained advertisers in these repeated auctions. This concept is based on a mean field approximation to relax the advertisers’ informational requirements, together with a fluid approximation to handle the complex dynamics of the advertisers’ control problems. Notably, we are able to derive a closed-form characterization of FMFE, which we use to study the auction design problem from the publisher’s perspective focusing on three design decisions: (1) the reserve price; (2) the supply of impressions to the Exchange versus an alternative channel such as bilateral contracts; and (3) the disclosure of viewers’ information. Our results provide novel insights with regard to key auction design decisions that publishers face in these markets. In the third part of this thesis, we justify the use of the FMFE as an equilibrium concept in this setting by proving that the FMFE provides a good approximation to the rational behavior of agents in large markets. To do so, we consider a sequence of scaled systems with increasing market “size”. In this regime we show that, when all advertisers implement the FMFE strategy, the relative profit obtained from any unilateral deviation that keeps track of all available information in the market becomes negligible as the scale of the market increases. Hence, a FMFE strategy indeed becomes a best response in large markets.

Dynamic Double Auctions: Towards First Best

We study the problem of designing dynamic double auctions for two-sided markets in which a platform intermediates the trade between one seller offering independent items to multiple buyers, repeatedly over a finite horizon, when agents have private values. Motivated by online platforms for advertising, ride-sharing, and freelancing markets, we seek to design mechanisms satisfying the following properties: no positive transfers, i.e., the platform never asks the seller to make payments nor are buyers ever paid and periodic individual rationality, i.e., every agent derives a non-negative utility from every trade opportunity. We provide mechanisms satisfying these requirements that are asymptotically efficient and budget-balanced with high probability as the number of trading opportunities grows. Moreover, we show that the average expected profit obtained by the platform under these mechanisms asymptotically approaches first best (the maximum possible welfare generated by the market). We also to extend our approach to general environments with complex, combinatorial preferences.

Dynamic Mechanisms with Martingale Utilities

We study the dynamic mechanism design problem of a seller who repeatedly sells independent items to a buyer with private values. In this setting, the seller could potentially extract the entire buyer surplus by running efficient auctions and charging an upfront participation fee at the beginning of the horizon. In some markets, such as Internet advertising, participation fees are not practical since buyers expect to inspect items before purchasing them. This motivates us to study the design of dynamic mechanisms under successively more stringent requirements that capture the implicit business constraints of these markets. We first consider a periodic individual rationality constraint, which limits the mechanism to charge at most the buyer’s value in each period. While this prevents large upfront participation fees, the seller can still design mechanisms that spread a participation fee across multiple initial auctions. These mechanisms have the unappealing feature that they provide close-to-zero buyer util...

Bounds on the Welfare Loss from Moral Hazard with Limited Liability

We study a principal–agent problem with discrete outcome and effort level spaces. The principal and the agent are risk neutral and the latter is subject to limited liability. Quantifying welfare loss as the ratio between the first-best social welfare and that arising from the principals optimal pay-for-performance contract, we provide simple parametric bounds for problem instances with moral hazard. Relying on that, we compute the worst-case welfare loss ratio among all problem instances with a fixed number of effort and outcome levels as a function of the number of possible effort levels and the likelihood ratio evaluated at the highest outcome. As extensions, we look at linear contracts and at cases with multiple identical tasks. Our work constitutes an initial attempt to quantify the losses arising from moral hazard when the agent is subject to limited liability, and shows that these losses are non-negligible in the worst case.