Research Papers
Research Papers
Structural Estimation of B2B Demand in Local Food Distribution: Demand Insights and Incentive Design to Improve Farmer Sourcing (with S. Lim and M. Mahmoudi ). 2025.
Abstract: Local food distributors function as essential intermediaries between small and medium farms and institutional and retail buyers such as schools, restaurants, and grocery chains. Yet these distributors face high failure risk due to inconsistent farm supply and rising operating costs, threatening the viability of local food systems. We study this operational challenge in collaboration with Tamarack Holdings (TH), a Michigan-based agri-food distributor. Using detailed transaction data and a Multiple Discrete-Continuous Extreme Value (MDCEV) demand model specification, we analyze business purchasing behavior along two dimensions of customer orders: the number of SKUs selected from the distributor’s assortment and the quantity purchased per SKU.
The demand estimates yield two key insights. First, instead of observing the traditional frequency–quantity trade-off often documented in business-to-consumer settings, business-to-business (B2B) buyers exhibit low satiation for staple products, purchasing frequent items in large volumes. Second, purchasing behavior is highly heterogeneous across buyer segments, limiting a one-size-fits-all sourcing policy.
We embed the estimated demand structure into a principal–agent contract design framework to address inconsistent supply by incentivizing farmers’ capacity expansion. We demonstrate the utility of this structural integration: contracts designed under standard Multinomial Logit (MNL) and Fractional MNL benchmarks generate average profit losses of 86.42% and 75.44%, respectively, relative to the MDCEV-based design. These losses primarily arise from the benchmarks’ inability to account for the volume-driven nature of B2B orders. Lastly, relative to the status quo of no incentives, the optimal contract increases average distributor profit by 25.65%, improves fill rates by 8.12%, and reduces stockout probability by 11.88%.
Non-Profit Support in Education: Resource Allocation and Students’ Lifetime Outcomes with Harish Guda, Milind Dawande, and Ganesh Janakiraman, Major Revision at Manufacturing and Service Operations Management. 2024. [SSRN]
Problem Definition: One of the seventeen United Nations Sustainable Development Goals aims for inclusive and equitable quality education, with lifelong benefits, for all. Our work in this paper focuses on the operations of non-profit organizations (NPOs) that broaden access to high-quality education for underprivileged students. Specifically, we analyze the resource-allocation strategy of an NPO that adopts a two-stage structure in allocating resources to its beneficiaries, e.g., free pre-secondary education (first stage) for all underprivileged students in a target population, followed by sponsorships for post-secondary education (second stage) at leading institutions to those students who demonstrate commendable performance in the first stage. The lifetime outcomes of the beneficiaries depend on their own effort and the quality of the resources that the NPO provides.
Methodology/Results: We adopt a principal multi-agent framework with moral hazard and heterogenous agents in the absence of monetary transfers. We establish the strategic role of an NPO's resource-allocation strategy on the effort beneficiaries invest and their lifetime outcomes. In particular, despite the supportive nature of the resources to both the beneficiaries and the NPO, we show why scarcity of these resources benefits the NPO.
Managerial Implications: Our findings have important implications for the design of such support policies of NPOs. We demonstrate the conditions under which the NPO must withhold resources from the beneficiaries to improve their lifetime outcomes. We also demonstrate the role of exogenous admission criteria on the NPO's resource allocation strategy. Finally, when faced with multiple beneficiary subgroups, we identify how the NPO should earmark resources for each beneficiary subgroup. We compare the optimal allocation strategy vis-a-vis a fair allocation strategy to quantify the price of fairness of the optimal solution.
Optimal Cardinal Contests, with Milind Dawande and Ganesh Janakiraman, Production and Operations Management. Forthcoming. [PDF] .
Abstract: We study the design of crowdsourcing contests in settings where the outputs of the contestants are quantifiable - e.g., a data-science challenge. This setting is in contrast to those where the output is only qualitative and cannot be objectively quantified - e.g. when the contest's goal is to design a logo. The literature on crowdsourcing contests focuses largely on ordinal contests, where the designer ranks the contestants' outputs, and awards are based on relative ranks. Such contests are ideally suited for the latter setting, where output is qualitative. For our setting (quantitative output), it is possible to design cardinal contests, where awards could be based on the actual outputs and not on their ranking alone - thus, the family of cardinal contests includes the family of ordinal contests. We study the problem of designing an optimal cardinal contest. We use mechanism design theory to derive an optimal cardinal mechanism and provide a convenient implementation - a decreasing reward-meter mechanism - of the optimal contest. We establish the practicality of our mechanism by showing that it is "Obviously Strategy-Proof", a recently introduced formal notion of simplicity in the literature. We also numerically examine the benefit of the optimal cardinal contest over the most popular ordinal contest - namely, the Winner-Takes-All (WTA) contest. Our numerical analysis suggests that, for the contest designer, the average improvement provided by the optimal cardinal mechanism over the WTA contest is about 23%. For a given number of contestants, the benefit of the optimal cardinal mechanism is especially appreciable for projects where the ratio of the designer's utility to agents' cost-of-effort falls within a wide practical range. For projects where this ratio is very high, the expected profit of the best WTA mechanism is reasonably close to that of the optimal cardinal mechanism.
Abstract: Most research examining the response of supply chains to environmental regulations assumes that all parties are subject to the same regulations. However, it is uncommon for buyers and vendors to be located in the same place and subject to the same environmental regulations. This paper investigates the inventory and production decisions of a buyer and a vendor in a supply chain with multiple buyers and a single vendor that produces multiple products under various environmental regulations. Our analysis demonstrates that the multiple buyers and multiple products scenario can be simplified to the single buyer and single product case under certain conditions. We find that a cap-and-trade regulation applied to the buyer may increase or decrease the economic order quantity in traditional supply chains, but it consistently reduces the economic order quantity in vendor-managed inventories (VMI). Our results indicate that while the cap-and-trade regulation reduces the carbon emissions of the buyer in both VMI and traditional supply chains, it may increase the carbon emissions of the vendor. Additionally, we discover that carbon cap regulation, although less commonly used than cap-and-trade regulation, may further decrease the carbon emissions of both the buyer and vendor beyond the effects of a green buyer or green vendor. Moreover, we provide recommendations for actions that the buyer or vendor can take to offset the impact of changes in parameters on economic order or production quantities.