fresh at dawn, half price by dusk

Problem statement

Post-harvest fish loss in small-scale fisheries is usually imagined as fish rotting and being discarded. FAO's field guide for loss assessment says the dominant form is different: "quality loss" — fish that "is sold for a lower price than that which would have been achieved if the fish were of 'best quality'. This is the most common PHFL in many areas." A trader without ice buys tilapia in the morning at about US$1 per kilogram and watches the price fall to under US$0.50 by evening and under US$0.20 for leftovers sold to processors, while "most customers wait until evening, when a fish seller is desperate for buyers as the quality is degrading fast." The problem is that this loss is invisible in tonnage statistics and cannot be measured or rewarded objectively: FAO's own methods concede that "determining quality can be subjective and is best done based on the operators' experience and understanding of quality," so neither fishers nor buyers nor loss-assessment teams have a shared, cheap, defensible measure of the quality that is draining away.

Why this matters

A March 2026 synthesis in Communications Sustainability estimates that only 54% of harvested fish is consumed directly by people and that roughly 35% of the global catch is lost or wasted along supply chains — "roughly 10–12 million tons annually" — with losses concentrated in small-scale fisheries "due to limited ice availability, delayed marketing, and rudimentary processing methods." In FAO's illustrative worked example of a questionnaire loss assessment (a respondent-agreement table, not a measured statistic), the authors conclude that discards at sea exceed 5 percent and at landing 2 percent, but that "the quality loss is very high with more than 40 percent of the fish volume being sold at less than 50 percent of the best price." That is income lost by some of the poorest producers in the food system, and it also blunts every intervention: a fisher who ices fish cannot prove it is better, buyers cannot pay for what they cannot verify, and — as FAO notes — in some communities "consumers think that fish that has been iced is not good quality and they are suspicious of such fish," so improved handling can even be penalized.

What’s been tried and why it hasn’t worked

FAO's guide provides three methods — the Informal Fish Loss Assessment Method, Load Tracking, and the Questionnaire Loss Assessment Method — and Load Tracking measures losses by weighing units before and after each stage. Weighing handles physical loss well: "Weighing in order to quantify physical loss is relatively straightforward. However, the situation becomes complicated when quantifying quality loss because one has to assess fish quality and then evaluate different qualities separately to calculate changes in quality." The guide falls back on operator judgment using signs such as belly burst, discoloration, mould, smell and breakage, and warns that "organoleptic or sensory assessments can be subjective." Ice and insulated boxes, the standard technical remedy, work where adopted — the 2026 synthesis cites an Indian icebox program that raised vendor income and notes that its discontinuation and reduced icebox access "caused vendor income to regress" — but adoption stalls where buyers do not pay for quality, consumers distrust iced fish, and no landing-site tool converts freshness into a price signal. Market-force losses compound this: FAO observes that "a market force loss may evolve over time into quality and/or physical losses" as unsold fish waits, and that "inadequate market information and barriers can prevent the producer from gaining access to the right market with the right product at the right time." The Nature-portfolio synthesis adds that adoption likelihood "is not a well-defined global parameter" and that socio-behavioural drivers of uptake "merit deeper investigation."

What would unlock progress

The missing piece is an objective, low-cost, landing-site freshness measure that both loss assessors and traders accept — the fish equivalent of a grain moisture meter — coupled to simple market mechanisms (grade-linked pricing, time-of-sale records) that let good handling be paid for. Adjacent precedents exist: sensory quality-index schemes and electronic freshness sensors are used in industrial fisheries and retail; smartphone-based colour and gill/eye image grading has been demonstrated for produce and meat; and grading-linked pricing transformed incentives in coffee and cocoa smallholder chains.

Entry points for student teams

The measurement half of this problem can be built anywhere fish is sold. A student team could construct a phone-camera freshness index for one common species from whole fish bought at a local market and held through a controlled warming series — photograph eyes, gills and skin at fixed intervals and temperatures, log hours since catch where the vendor can state it, and score each specimen against FAO's own organoleptic signs (belly burst, discoloration, mould, smell, breakage) with a small trained panel — which yields the index and its error bars without a landing site, a trader, or a permit. Testing whether a displayed grade moves the day's price is a live-market intervention with people whose income is at stake, so the semester deliverable is that trial's design rather than its result: the grade display, the price-and-hour recording sheet, the assignment scheme across vendors or days, and a pre-registered analysis plan, sized from the price decay in FAO Technical Paper 559 and handed to a fisheries extension service or NGO that already works the landing sites. An economics team can do the modeling arm now by reconstructing the "waiting until evening" dynamic from the price and loss figures in TP 559 and the published loss-assessment literature, estimating what a small ice subsidy or a grade-linked price would return before any field study is funded. A communication-design team could design and test buyer-facing messaging that reverses the perception that iced fish is inferior. Relevant skills: food science/sensory analysis, computer vision, development economics, field survey methods.