Sunil Tiwari

Healthcare supply chains are among the most consequential and least well-understood in operations management. When a hospital runs short of surgical consumables, or medication reaches a ward late, the cost is not financial — it is clinical. At the same time, healthcare institutions face relentless pressure to reduce operating costs from the same budgets that fund care, making the tension between service level and efficiency structurally difficult to resolve. My work in this stream asks how simulation, mathematical optimisation, machine learning, and analytical modelling can help healthcare systems design supply chains that are simultaneously more efficient and more reliable.

My engagement with these questions began in Singapore, where I led a multi-year research and industrial programme examining whether hospitals could replace manual material handling with Automated Guided Vehicles without disrupting patient care — and if so, how many vehicles, routing where, and on what operational schedules. The answer required building a simulation-based digital twin of each hospital's internal logistics system in Simio. These models replicated the real-time flow of materials — consumables, pharmaceuticals, sterile instruments, linen — through wards, pharmacies, sterilisation units, and operating theatres, capturing the stochastic variability in demand arrival rates, travel times, and service durations that characterise hospital operations. Coupled with simulation-based optimisation, the twin was used to evaluate different AGV fleet configurations, routing strategies, and scheduling rules across a range of patient-volume scenarios. The outcome was concrete: a set of operational recommendations specifying the optimum number of AGVs required for daily operation and the routing and scheduling policies that minimise disruption risk while meeting service-level targets — delivered directly to the hospitals and incorporated into their automation deployment plans.

A related project applied machine learning to a different bottleneck: the unpredictability of consumable demand within a patient stay. Using inpatient health records — patient diagnoses, clinical pathways, procedure logs, and historical ward consumption patterns — I developed prediction algorithms that estimate, at the point of admission, what consumables a patient is likely to need during their stay. The challenge was not purely predictive accuracy but clinical deployability: the model had to be fast enough for point-of-admission use, interpretable enough for procurement teams to trust, and robust enough to hold across different ward types and patient populations. The resulting system was deployed as a web-based application built in Dash-Plotly, making consumables forecasts actionable by supply and procurement teams without specialist data science infrastructure. A parallel AI Singapore project applied data-driven methods to bed management — using real-time clinical data to predict occupancy patterns and optimise bed allocation decisions across hospital wards.

Current theoretical work extends this programme in three directions. The first develops an analytical queueing model of teleconsultation capacity allocation in tiered healthcare systems, examining how platforms should optimise physician time across consultation types — balancing profitability against access equity and wait-time congestion, and characterising the divergence between profit-maximising and socially optimal allocation. The second uses game-theoretic analysis to study physician recruitment on digital health platforms: the choice between full-time and part-time contracts, and how it shapes platform quality, physician incentives, and patient welfare under different market structures. The third applies robust optimisation, with feature-driven uncertainty sets derived from patient demographic and prescription data, to the siting of medication self-collection lockers — producing location plans that remain near-optimal across a wide range of demand scenarios. I am actively extending this programme to NHS supply chain contexts in the UK.

Financial mechanisms and governance structures are not merely background conditions for supply chain behaviour — they are among its most powerful determinants. Who finances a supplier's working capital, how ESG ratings are constructed and interpreted, whether a blockchain record produces genuine accountability or a formal simulacrum of it — these are governance questions with direct operational consequences. My work in this stream examines how financial design, ESG frameworks, and digital transparency tools shape supply chain performance, resilience, and ethical conduct, drawing on large-sample empirical methods, game theory, and mechanism design.

A central current project concerns ESG rating divergence — the fact that major rating agencies frequently disagree, sometimes dramatically, on the ESG score of the same firm. This divergence is not a measurement error to be corrected; it is a structural feature of the ratings landscape with the potential to systematically distort buyer decisions in ways that are invisible to most supply chain participants. Using large-sample archival data on buyer-supplier relationships and causal identification strategies, I examine how divergence across raters affects buyer strategies for supplier selection and, through them, the resilience of the supply chains buyers build. The core finding is that divergence creates a form of strategic ambiguity that rational buyers resolve through risk-hedging behaviour — a response that can paradoxically reduce supply chain resilience by inducing over-diversification and undermining the deep relational investments that make supply chains robust to disruption. A related paper examines how ESG signals function as persistent quality indicators in supply chain finance, studying whether and under what market conditions ESG-based credit pricing is efficient rather than merely regulatory-compliant.

A parallel thread applies game theory and mechanism design to FinTech-bank competition in supply chain finance markets. Under open banking — where firms can share financial data across institutions — I model the strategic interaction between a FinTech lender and an incumbent bank as a Bayesian game with asymmetric information and capital constraints, characterising the mixed-strategy Nash equilibrium in which neither full FinTech dominance nor full bank retention is stable. The equilibrium structure shows that open banking does not straightforwardly favour FinTech entrants: the advantage depends critically on the level of information asymmetry between the firm, the FinTech, and the bank, and on the bank's capital friction relative to the FinTech's cost of funds — a finding with direct implications for how open banking regulations should be designed. A connected paper uses mechanism design to examine when regulatory requirements for algorithmic transparency in platform supply chains help or harm efficiency, identifying the conditions under which mandating explainability produces better supply chain outcomes and when it merely enables strategically sophisticated actors to game the disclosed rules.

A third thread, more empirically grounded, examines blockchain-enabled transparency and forced labour governance in agrifood supply chains. Using fieldwork, archival records, and theoretical frameworks drawn from institutional and organisation theory, I examine how the technical design of blockchain workflows — the structure of records, the assignment of write permissions, the conditions under which entries are validated — determines whether blockchain actually improves labour governance or produces a verifiable record that obscures as much as it reveals. The key finding is that observability is not a property of the technology but of the governance architecture surrounding it: the same infrastructure can enable or neutralise accountability depending on the incentives participants have to populate records truthfully. A companion paper applies Extreme Value Theory, borrowed from financial risk modelling, to treat modern slavery as a tail risk in ESG-vulnerable sectors — showing how ESG frameworks can be redesigned to price and hedge this exposure in the same way that catastrophic loss events are priced in actuarial and risk management practice.

At the heart of every supply chain is a relationship governed — explicitly or implicitly — by contracts: wholesale price agreements, two-part tariffs, buyback provisions, revenue-sharing clauses. My work in this stream asks how those contracts should be designed, what strategic behaviour they induce, and how information asymmetry, financial constraints, and uncertainty change the picture. The methods are primarily game-theoretic: I develop analytical models of strategic interaction between supply chain parties, characterise equilibrium outcomes, and identify the contract structures that achieve coordination or approximate centralised efficiency in decentralised systems.

In a series of analytical papers, I have developed frameworks for contract design in Industry 4.0 settings. One paper examines how a manufacturer should decide whether to introduce a smart, connected product line alongside a traditional one — characterising the critical market potential threshold at which the innovative product overtakes the traditional one in sales, even when priced at a premium. The key result is counter-intuitive: under both wholesale price and two-part tariff contracts, the cessation points and market potential thresholds move in opposite directions. This means that the contract type a supply chain uses changes not just the distribution of profit but the conditions under which investment in innovation is rational — a practically important finding for firms navigating the transition to smart manufacturing, where pricing and contract structure are often treated as separate decisions.

A second line examines multi-echelon supply chain coordination — how to sequence contracts and design cut-off policies across multiple tiers so that the whole system achieves near-optimal performance without requiring centralised information sharing. The central challenge is that bilateral coordination mechanisms can fail when applied sequentially across three or more tiers, because each contract changes the information structure and incentive environment that the next contract faces. I characterise the contract sequences and cut-off thresholds that sustain coordination, and show when a decentralised system with optimally designed sequential contracts can replicate the performance of a centrally managed one.

More recent work examines partner selection under commitment uncertainty — a problem that arises whenever a firm must choose a supplier before observing whether that supplier will honour its commitments. I develop a game-theoretic model comparing ex-ante incentive structures — bonuses conditioned on anticipated performance — with ex-post penalty structures — charges imposed on observed non-performance. The analysis shows that the optimal mechanism depends on the cost structure of commitment, the verifiability of outcomes, and the supplier's private information about its own reliability. A connected thread examines platform financing and product returns: cash-constrained sellers on major e-commerce platforms face a structural tension between accepting platform credit, which eases their liquidity constraint, and the platform's return policies, which create contingent liabilities that can erode the financial benefit of borrowing — a trade-off that platform designers have largely overlooked in the design of integrated logistics-finance systems.

My approach to digital technologies in supply chains is deliberately analytical rather than aspirational: I ask not whether a technology promises to transform operations, but under what conditions it actually does so — for whom, at what cost to other parties, and with what unintended consequences. This produces work that is more precise, and sometimes more cautious, than the general literature on digital adoption.

A recent paper accepted at the Journal of Operations Management addresses a structural problem in how our field has studied digital transformation. Operations management and information systems scholars have examined the same phenomenon — how organisations integrate digital technologies into their processes — through largely disconnected theoretical frameworks, separate empirical traditions, and different research communities. I identify the mechanisms through which this fragmentation has developed, characterise what each discipline has produced independently and what it cannot produce alone, and develop an integrative framework connecting OM's emphasis on process efficiency, coordination, and operational performance with IS's emphasis on sociotechnical adoption dynamics and information system design. The contribution is not a synthesis but an interface: a conceptual structure that allows findings from each tradition to inform the other without collapsing their distinctive analytical commitments.

My work on blockchain technology spans empirical and formal modelling approaches. An empirical study using structured survey and interview data with supply chain managers in third-party logistics documented the practical barriers that prevent blockchain from delivering its theoretical promise: interoperability failures with existing ERP systems, inconsistent data quality across supply chain partners, the absence of industry-wide standards, and free-rider problems in consortia governance where no single firm has an incentive to invest in infrastructure that benefits all. These findings were used to develop a decision framework that firms can apply to evaluate whether blockchain adoption is appropriate for their specific context and operational maturity. Subsequent theoretical work develops a dynamic Stackelberg differential game to model how blockchain-enabled transparency affects investment incentives in low-carbon supply chains. A fundamental tension emerges: transparency reduces information asymmetry and should, in theory, encourage green investment; but it also changes the strategic dynamics of the investment game in ways that can reduce firms' incentives to participate, because the transparency that makes cooperation visible also makes the cost of green investment visible in ways that invite competitive disadvantage.

A newer thread examines generative AI and algorithmic management in platform supply chains. One paper develops a tripartite evolutionary game to formally model how Gen-AI capabilities alter the economics of community group buying — a distribution model growing rapidly across Asia — showing that AI-driven demand aggregation changes the stable coalition structures among buyers, intermediaries, and suppliers in ways that are both efficiency-enhancing and potentially exclusionary. A parallel paper uses mechanism design to study when platform operators' regulatory obligations to explain algorithmic pricing or allocation decisions produce better supply chain outcomes, and when they inadvertently enable strategically sophisticated participants to game the disclosed rules. The finding is that transparency mandates are not uniformly beneficial: their welfare effects depend on the information structure of the market, the degree of strategic sophistication among participants, and the verifiability of the platform's explanations.