Social and environmental analysis of food waste abatement via the peer-to-peer sharing economy

The study addresses whether peer-to-peer (P2P) sharing platforms can effectively and sustainably reduce post-retail and consumer-level food waste in high-income countries, where waste is substantial and often still edible. It situates the problem within global food loss (1.3 billion t annually) and associated environmental burdens (notably greenhouse gas emissions, water, and land use). The authors highlight drivers of retail and household food waste (aesthetic and safety standards, over-purchasing, confusing date labels, and poor household food management) and the coexisting issue of food insecurity in high-income contexts. Given that much discarded food is still edible, the research explores P2P redistribution as a potential win–win solution. The introduction frames key uncertainties: whether sufficient supply and demand exist for unwanted food on P2P platforms, potential environmental trade-offs from added transport and induced consumption, and whether such platforms serve populations experiencing food insecurity given known participation biases in the sharing economy. The paper investigates these questions using data from OLIO, a large food-sharing app.

The paper reviews evidence that post-retail and consumer-level food waste in high-income countries is large, costly, and environmentally impactful, and that much of it remains edible at disposal. It discusses the rise of the digital sharing economy, its ability to reduce transaction costs and match underutilized assets, and its potential scalability and suitability to perishable food exchange. However, literature also questions environmental benefits of sharing platforms due to stimulated demand (e.g., for transport, tourism, durable goods) and potential for rebound effects. Psychological barriers such as disgust toward non-new items and skepticism about consuming resources produced via recycling are noted as potential demand constraints. Social literature indicates that higher-income and especially highly educated individuals may disproportionately benefit from sharing platforms, potentially limiting benefits for those experiencing food insecurity. Overall, prior work motivates empirical assessment of P2P food-sharing’s uptake, environmental performance, and social distribution of benefits.

Design: Mixed-methods empirical analysis of OLIO platform data (April 2017–October 2018), covering >30,000 active users, with most activity in the UK and Channel Islands. The study examines item types, quantities, and values; life-cycle GHG impacts; and socioeconomic patterns of exchanges. Classification of listings: All listings were classified into 13 food categories (and nonfood categories) using a supervised deep-learning LSTM text classifier trained on 53,000 manually tagged listings. The classifier achieved ~0.9 accuracy. Only the 170,000 unique edible food listings and their users were analyzed. Collection assessment: A listing with at least one successful collection was treated as fully collected. Listings and collection rates were analyzed by food category and by user type: “food waste heroes” (volunteers who collect surplus from businesses and list it) vs. regular users. Weights and value estimation: Because most listings lacked exact weights, the authors conducted Monte Carlo simulations using an empirical sample of 3,300 listings where weights and monetary values were manually estimated from images and descriptions. Weights-per-listing were estimated separately for heroes and regular users across food categories. When sample sizes were <50 per group, weights were drawn from fitted log-normal distributions. Simulations were repeated to yield total exchanged mass and value ranges. Environmental assessment (Greater London focus): Benefits (avoided GHG) were computed as the product of total exchanged mass and a UK-specific average full life-cycle emissions factor for avoided food waste (normal distribution, mean 4.3 t CO2eq per t food waste, coefficient of variation 10%). Sensitivity analysis used a lower EU-wide factor (2.1 t CO2eq per t). Added transport burdens were estimated for six travel scenarios crossing mode (car vs. London bus) and purpose (two-way dedicated vs. one-way/combined trips; walking for trips <1.6 km). Road distances and times from collectors’ notification locations to pickup points were computed using STATA Georoute (HERE API). Exchanges with one-way travel time >30 minutes were excluded, leaving 62,570 exchanges. Travel emissions were calculated using full life-cycle per passenger-km factors for cars and London buses. Net GHG benefits equal avoided food waste emissions minus travel emissions. Social/socio-demographic analysis: Exchanges were mapped between providers’ and collectors’ home-area characteristics (income and education percentiles) using geographic census proxies for users’ home addresses, enabling a matrix of exchange frequencies by decile for income and education. Network analysis: Participation roles (giver, collector, both) and interaction structure were characterized, identifying mega-providers (often food waste heroes) and recurring exchanges.

• Platform activity and diversion: Of ~170,000 food listings, 60% were collected. Over 19 months, 91 ± 1 t of food (5th–95th MC percentiles) with an estimated retail value of £720–750 thousand were diverted from disposal across OLIO’s network.
• Item types and collection rates: Most commonly listed categories were baked goods (29%), kitchen/pantry staples (17%), fresh produce (16%), and prepared food (13%). Collection rates were highest for mixed (71%), sandwiches (70%), prepared food (66%), and fresh produce (65%); most categories were ≥53% except baby food (29%).
• Roles and supply structure: Among ~22,000 users who engaged in at least one exchange, 12% both gave and received, 26% only gave, and 62% only collected. A small number of mega-providers supplied most items, often food waste heroes (volunteers collecting from businesses). Heroes provided 71% of all listings (collected and uncollected) and predominantly supplied ready-to-eat items (sandwiches, baked goods), whereas regular users more often listed pantry/frozen items. Hero listings had higher collection rates (66%) than regular users (47%) across all categories. Heroes interacted with more users on average (27 vs. 2.5) and had more recurring exchanges (79% vs. 31%).
• Environmental performance (Greater London): Approximately 41 t of food were exchanged, with 226–451 thousand added vehicle/bus km depending on travel scenario. Across all modeled scenarios, avoided food waste emissions exceeded added transport emissions. Net GHG benefits ranged from 87 t CO2eq (72–102) in Scenario #1 (two-way dedicated car trips) to 156 t CO2eq (150–162) in Scenario #6 (mix allowing short trips on foot, akin to average London transport behavior). Sensitivity analyses generally preserved net benefits except in the most carbon-intensive travel assumption. A displacement rate as low as 12–50% (depending on scenario) would still yield net benefits.
• Social patterns: Exchanges predominantly occurred among users from areas associated with relatively lower income and higher education levels. Exchanges tended to occur between users with similar income and education profiles.
• Scale of impact: For London users, the average net GHG benefit corresponds to roughly 0.6% of per-capita annual food-related emissions (assuming 2.7 t CO2eq per capita-year for food consumption).

The findings demonstrate that P2P food sharing can substantially increase utilization of edible post-retail and household food by matching unwanted items with secondary consumers, including short-shelf-life items often unsuitable for centralized redistribution. Collection rates above 60% and up to 70% for several categories indicate robust demand and platform viability. Environmentally, avoided cradle-to-grave emissions from displaced food production and waste management outweigh added travel burdens under realistic urban transport behaviors; only extreme, car-only, dedicated round trips risk negating benefits. Considering carbon opportunity costs of foregone land use would further amplify net benefits. Socially, while exchanges reach lower-income areas, participation skews toward users with higher education levels, suggesting access and cultural capital requirements that may limit direct benefits to the most food-insecure populations. Additionally, if shared food displaces new purchases, monetary savings could induce rebound consumption that partially offsets environmental gains. Overall, P2P sharing offers a promising, scalable complement to existing waste-reduction strategies but requires attention to transport behaviors, substitution rates, and equitable access to maximize societal benefits.

P2P food-sharing platforms like OLIO can effectively divert edible food from disposal, achieving high collection rates and yielding net GHG reductions even when accounting for added transport, particularly in urban contexts. The platform’s current scale delivered about 90 t of food reuse and meaningful environmental benefits, with strong performance for ready-to-eat and perishable items. However, to realize larger system-level impacts, further scaling is needed alongside efforts to ensure exchanges displace new purchases, minimize dedicated car travel, and broaden participation among populations facing food insecurity. Future research should: (1) characterize item-level composition to optimize exchanges of high-impact foods; (2) measure actual substitution rates and rebound effects; (3) assess non-urban contexts and transport patterns; (4) test platform design features (e.g., timing, location constraints, item provenance, images/text) that increase successful, low-impact exchanges; and (5) refine social targeting to improve inclusivity and benefits for food-insecure households.

• Listing-level resolution: Multi-item listings were assumed fully collected if any collection occurred; anecdotal data suggest ~90% full collection, but item-level confirmation was unavailable.
• Location uncertainty: Collectors’ default notification locations were used as origin proxies due to privacy constraints, introducing uncertainty in travel distance/time estimates.
• Emissions factor generalization: A single UK-average full life-cycle emissions factor for food waste was applied due to mixed-item compositions, potentially masking variation by food type (e.g., meat vs. plant-based); sensitivity tested with a lower EU factor.
• Substitution assumption: Main analysis assumes 1:1 displacement of new purchases by shared food; in practice, substitution may be lower, affecting net GHG benefits. Minimal displacement rates of 12–50% (scenario-dependent) are needed for net benefit.
• Social inference: Socioeconomic attributes were inferred from area-level census data, which may not reflect individual users (ecological fallacy risk).
• Context specificity: Environmental results reflect Greater London’s transport and infrastructure; rural/suburban settings with higher car dependence and longer distances may reduce or negate net benefits.
• Exclusion of long trips: Exchanges with one-way travel time >30 minutes were excluded, potentially biasing against longer-distance collections.