Somya Singhvi

In spite of their pivotal role in the global rural economy and their significant contribution as a primary source of employment for women in the developing world, artisanal supply chains continue to face issues of low productivity and high poverty rates in the upstream. Identifying effective and feasible solutions to enhance the productivity of artisans remains a formidable challenge, primarily due to the distributed structure of the supply chain. In this chapter, we highlight four research questions that have emerged from close collaboration with a prominent exporter of handmade rugs in India. These research questions shed light on the pivotal role that analytics and operational considerations can play in elevating the productivity of the workforce operating within resource-constrained environments.

Despite their vital role in the global rural economy, and as a major source of employment for women in the developing world, artisanal supply chains continue to be plagued by low productivity and high poverty levels. Identifying effective and implementable solutions to improve artisan productivity is a challenging task due to high fragmentation in upstream parts of the supply chain, where artisans often work from their individual households. This study presents research conducted in close collaboration with one of the leading exporters of handmade rugs in India. Leveraging insights from the field visits and analysis of detailed supply chain data, we provide robust empirical evidence that frequent supervisor visits can play a crucial role in improving artisans' productivity. Our results from Instrumental Variables analysis indicate that a one-day decrease in the average number of days between supervisor visits to remote weavers can decrease weaving times by 13.1%-14.1%, which can lead to a 15%-17% increase in monthly income for weavers. Our analysis also suggests that this impact is heterogeneous, with visits to difficult-to-weave rugs, and visits that are more consistently scheduled, leading to maximum productivity gains for the weavers. To capitalize on these insights, we propose a novel predict-then-optimize framework for optimizing supervisor visits in the supply chain. Finally, we demonstrate that the proposed framework can significantly increase weaver productivity even after accounting for various operational and scheduling constraints. This research offers valuable insights into other distributed supply chains and highlight how supply chain considerations can play a critical role in improving the productivity of the workforce in resource-constrained settings.

To improve the welfare of smallholder farmers, multiple countries (e.g., Ethiopia and India) have launched online agri-platforms to transform traditional markets. However, there is still mixed evidence regarding the impact of these platforms and more generally how they can be leveraged to enable more efficient agricultural supply chains and markets. This paper describes work conducted in close collaboration with the state government of Karnataka, India, to design, implement, and assess the impact of a new two-stage auction on the state's online agri-platform, the United Market Platform (UMP). To ensure implementability and protect farmers' revenue, the auction design is guided by practical operational considerations as well as semi-structured interviews with a majority of the traders in the field. A new behavioral auction model informed by the field insights is developed to determine when the proposed two-stage auction can generate a higher revenue for farmers than the traditional single-stage, first-price, sealed-bid auction. The new auction mechanism was implemented on the UMP for a major market of lentils in February 2019. By the end of May 2019, commodities worth more than $6 million (USD) had been traded under the new auction. A difference-in-differences analysis demonstrates that the implementation has yielded a significant 3.6% price increase (corresponding to a 55%--94% profit gain), benefiting over 10,000 farmers who traded in the treatment market. The results from this paper offer tangible lessons on how innovative price discovery mechanisms could be enabled by online agri-platforms in resource-constrained environments. Importantly, the success of these designs critically depends on careful considerations of systemic operational and behavioral factors that affect trades in the physical markets.

To improve smallholder farmers’ welfare, several governments have led reforms in improving market access for these farmers through online agri-platforms. This paper empirically evaluates the impact of such a reform—the Unified Market Platform (UMP) in Karnataka, India—on market prices and farmers’ profitability. The analysis shows an average 5.1%, 3.6%, and 3.5% increase in the modal prices of paddy, groundnut, and maize due to UMP. However, the analysis also indicates a lack of statistically significant impact on the modal prices of cotton, green gram, and tur. We provide evidence that commodities with a significant price increase differ from those without systemic features related to logistical challenges, bidding efficiency, in-market concentration, and the price discovery process used.

We consider the problem of personalized recommendations on online platforms, where user preferences are unknown, and users interact with the platform through a series of sequential decisions (such as clicking to watch on video platforms or clicking to donate on donation platforms). The platform aims to maximize the final outcome (e.g., donations on donation platforms or viewing duration on video platforms). However, the platform only observes the final outcome for users who complete the first stage (clicking on the recommendation). The final outcome for users who do not complete the first stage (not clicking on the recommendation) remains unobserved (also referred to as \textit{funneling}). This censoring of outcomes creates a selection bias issue, as the observed outcomes at different stages are often correlated. We demonstrate that failing to account for this selection bias results in biased estimates and sub-optimal recommendations. In fact, well-performing personalized learning algorithms perform poorly and incur linear regret in this setting. Therefore, we propose the Sample Selection Bandit (SSB) algorithm, which combines Heckman's two-step estimator with the ``optimism under uncertainty" principle to address the sample selection bias issue. We show that the SSB algorithm achieves a rate-optimal regret rate (up to logarithmic terms). Furthermore, we conduct extensive numerical experiments on both synthetic data and real donation data collected from GoFundMe (a crowdfunding platform), demonstrating significant improvements over benchmark state-of-the-art learning algorithms in this setting.