Dealing with Preharvest Food Loss
Today, I am starting a three part series about food loss: preharvest losses, which is crop that never grows to harvest; postharvest losses, which occur during or after harvesting and before reaching the consumer; and food waste, which is food that reaches the intended consumer but it not eaten. Today’s post will be on preharvest losses. Next week will be on postharvest losses, and the week after will be on food waste. This taxonomy is not used universally—harvest losses are often considered separately from postharvest losses—but it is good enough.
We will see that preharvest loss is a large but poorly understood problem, and while it is tied up with worries about the impact of climate change, preharvest loss presents many solutions. Food loss is an excellent illustration of the importance of economic development in bringing about good environmental outcomes.
Also below is a correction and expounding on cost figures relating to floating platforms and land reclamation, a subject of last week’s post.
In Genesis 41, Joseph, with God’s guidance, interprets the Egyptian pharaoh’s dreams to mean that Egypt is about to experience seven years of food surplus, during which they are to store food in preparation for the following seven years of famine. The story is a timeless reminder that, when we enjoy bounty today, we must not squander the fruits of that bounty or take it for granted, for hardship could come tomorrow.
Ancient people seldom had the gift of prophecy and were very much at the mercy of forces beyond their control when it came to crop yields. Recent research suggests that a severe drought around 1198-1196 BC may have been a major cause of the collapse of the Hittite Empire and a contributor to the broader Bronze Age Collapse that occurred around that time.
There is also evidence of aridity and crop failures contributing to the collapse of the Western Roman Empire, though as I’ve noted before, there were other factors in play. In particular, Rome was facing long-term depopulation, likely caused by chronic endemic disease. The collapse of the Western Roman Empire is perhaps the most studied ancient event, and like other major events, is overdetermined in its causality.
The problem may have been opposite with the Great Famine of 1315, one of the most calamitous disasters in European history. Evidence is that the famine years were some of the wettest in European history, leading to widespread crop failures as a result of flooding. Like other major events, it is hard to reduce the Great Famine to a single cause, and one can point to a population made less resilient by poverty and warfare prior to 1315. The Great Famine may be an illustration of a Malthusian catastrophe: the population of Western Europe had grown throughout the High Middle Ages to a point where the continent became vulnerable to the kind of disruption that occurred in the 14th century.
Various agripests—insects, weeds, and diseases—are also major causes of preharvest losses. For example, an outbreak of the brown planthopper devstated rice production in Japan in 1733, causing a famine that killed 12,000 people. A swarm of locusts was the eighth plague to strike Egypt, as described in Exodus 10:1-20. Pests remain a serious problem, with an average of 15% of crops worldwide lost each year to insects.
Especially notorious is the potato blight, caused by the pathogen phytophthora infestans, that struck Ireland in the 1840s, leading to the death in the Great Famine of a million people and mass emigration. Ireland’s vulnerability was exacerbated by dependence on a single crop—potatoes—for sustenance, and various political and economic circumstances.
In 1981, the economist Amartya Sen published Poverty and Famines: An Essay on Entitlement and Deprivation, which developed the now-common wisdom that famines in the modern world are not caused by a lack of food, but rather by political decisions. From 1930-1933, the Soviet Union suffered a famine that killed 5.7-8.7 million people. The famine resulted from the forced collectivization of agriculture from the first Five-Year Plan, and it entailed the Holodomor and the Kazakh Famine of 1930-1933, both believed to have been genocidal acts. The deadliest famine in history was the Chinese Great Famine of 1959-1961, which resulted from Mao Zedong’s deeply misguided Great Leap Forward.
World rates of malnutrition are decreasing over the long run, despite a worrying spike since 2017. Today, far more people are overweight than go hungry. Still, the 800 million people who are hungry today is 800 million too many, and there is no guarantee that the progress we have seen in the past will continue into the future.
Preharvest losses remain a serious problem today. This paper estimates that just over a billion tons of four major grain crops worldwide—wheat, maize (corn), rice, and soybeans—are lost before harvest, representing about 35% of possible biological yield. A 1991 report puts the figure at 36% of potential yield.
Looking at a few studies of country-crop combinations, this report finds that the two major causes of preharvest losses are pests (disease and animal, but not weeds) and a lack of rain. Freezes are a big problem for potatoes in Peru, especially at high altitudes, and weeds are a big problem for teff (a staple grain crop in the Horn of Africa) in Ethiopia. Flooding is a significant problem for most crops, and a few other causes of preharvest losses are given. I haven’t found any reliable breakdowns by country, but I would have to imagine that preharvest losses are generally much greater in low-income countries, where farmers have less access to pesticides and irrigation.
Compared to postharvest losses, our understanding of preharvest losses is spotty, and this is a major gap in our understanding of the issue.
I hope that the above historical discussion is more than enough motivation as to why reducing preharvest losses is important. But let us also look at the environmental aspects.
Our World in Data reports that 45% of the world’s habitable land, excluding glaciers, deserts, salt flats, beaches, and dunes, is used for agriculture. This compares to about 1% for cities and infrastructure, such as roads and power lines (this exact figure is disputed, but it is clearly much less than for agriculture). Of the agricultural land, 80% is for livestock, which is mostly grazing land, but also cropland for feed; 16% is for crops for food; and 4% is for non-food crops, such as biofuels and cotton.
The International Union for Conversation of Nature maintains its Red List, which tracks threatened and endangered species and lists the risk factor(s) to them. The biggest risk factor by far is land for agriculture, which includes both cropland and grazing land. Due to intensity, cropland is a greater threat to species than grazing land, despite the latter being a larger area. Reducing the land required for agriculture is the most important species conservation action available.
Food loss is one of the most obvious places to start looking. As we saw above, a reasonable guess is that preharvest losses reduce agricultural yield by a third of what would be biologically possible; this does not take into account additional losses at the postharvest or consumer stages. Cutting preharvest losses in half would reduce the cropland for food from 16% to under 13% of the world’s total, saving three times as much land as all the cities and infrastructure in the world.
It is estimated that the greenhouse gas emissions embodied in postharvest losses and wasted food is the equivalent of 4.4 billion tons of carbon dioxide, just over 10% of the world total. I haven’t found comparable estimates for preharvest losses, but I find it plausible that they would be a similar order of magnitude. Thus protecting crop yields is an important climate change solution.
Pesticides are indispensable to modern agriculture and one of the tools of the Green Revolution. It has been estimated that without pesticides, average yields of fruits would fall by 78%, vegetables by 54%, and cereal by 32%. Pesticides are controversial from an environmental standpoint, but turning back the clock on them is not a viable option.
From 1990 to 2021, world pesticide use increased by 97%. Over that same time, pesticide usage per area of crop land increased by 85%, while the usage per value of agricultural output is up just 2%.
Ideally, pesticides would target only the particular pest(s) of concern and not harm other organisms, but this doesn’t happen in practice. The modern environmental movement was launched with Rachel Carson’s 1962 Silent Spring, which documents a loss of birds—hence the title—as a result of the pesticide DDT. The book helped push public opinion toward a ban on DDT. Harmful impacts of modern pesticides include poisoning of farm workers; contamination of food via pesticide residue; poisoning of non-target species; and pollution of surface water, groundwater, and soil. The external costs of pesticides have been estimated at $4-19 per kilogram of active ingredient, which based on FAOSTAT’s figures, cause $14-67 billion of external damage worldwide each year. By comparison, using mainstream estimates of the social cost of carbon of $50-100 per ton, world damages from CO2 emissions are around $2-4 trillion per year.
Fortunately, there are many solutions to preharvest losses, all of which are already in use today and are a major reason why malnutrition is receding.
Solutions are needed that perform pest control and are more environmentally sensitive than chemical pesticides. One such solution is integrated pest management, a broad term that encompasses a strategy of mechanical, biological, and chemical pest control and focuses on reducing pests to economically tolerable levels rather than eradicating them completely. IPM has been a popular phrase in ecology since the 1970s, but the main challenge is that it requires a higher up-front investment in training and higher ongoing labor costs. I wonder if this is a case where greater automation, as we are seeing with precision agriculture, will enable wider use of IPM.
Biotechnology will play an important role. Several drought-resistant crops are under development: grasses; oats; barley; MON 87460, Monsanto’s drought-resistant maize; NXI-4T, a drought-resistant sugarcane in Indonesia; and others. Since the 1990s, insect resistant transgenic crops incorporating genes from the bacterium Bacillus thuringiensis (Bt), such as maize, soybeans, and cotton, have come into usage.
Supply and management of water are important solutions. For example, drip irrigation supplies water via pipes to a plant’s root system, either just above or below the surface. Drip irrigation was invented in Germany in the 1860s and came into widespread use in Israel a century later. Water efficiency and more efficient desalination have allowed Saudi Arabia and Australia, two of the most arid countries in the world, to supply desalinated water to much of their cropland.
Greenhouses offer substantially higher yields than open-air farming, a 50-90% water savings per unit crop, and protection against pests and disease. Greenhouses are more energy- and labor-intensive than open field farming, but unlike vertical farming, the economics of greenhouses are favorable enough that they are in widespread use worldwide. Greenhouse technology continues to develop, such as with sensors to detect diseases.
Finally, trade is essential as a means of mitigating risk. If agricultural production falters in one part of the world, either because of climate, or a pest outbreak, or a geopolitical problem, or for whatever other reason, they can make up the difference with imports, rather than going without.
Based on some of the papers cited above on the historical perspective on crop failures, I would caution against analogizing too much between ancient events and modern concerns about climate change. Yes, climate change is a risk to agriculture. It is a highly manageable risk, and we have surveyed only a few of the solutions above. The far greater risk is bad policy. Ill-conceived fears about GMO crops, desalination, and opposition to trade will turn the highly manageable problem of climate change into an unmanageable problem.
Costs of Floating Platforms and Land Reclamation
I was sloppy in reviewing cost figures in last week’s analysis of floating cities, and so I want to revisit that subject. I cited a few figures for costs of recent land reclamation projects, which range, in CPI-adjusted terms, to around $35-350 per square foot. That contrasts with a cost of $978/sqft that was estimated by the Seasteading Institute for the capital cost of residential and commercial space on floating platforms. However, these figures are not directly comparable for a few reasons.
First, the Seasteading Institute, if I read their figures correctly, is looking at finished buildings, while the land reclamation figures are just looking at creating the “land” without any buildings on it. For the cost of buildings, here is a blog post from a company that works with construction. There are a lot of numbers here, but maybe the most relevant are $562/sqft for the average U.S. cost of a midrise commercial office building and $660/sqft for highrise commercial office buildings. The most expensive class of buildings is museums and performing arts centers at $892/sqft. Looking at it this way, floating platforms to extend the area of expensive coastal downtowns looks doable.
Second, the numbers compare two different quantities: usable floor space in the case of seasteading, and land area in the case of land reclamation. To properly compare them, we would have to know the ratio of the two quantities for a proposed project. In the case of a single lot, this ratio is known as the floor-area ratio. This ratio is typically 6-7 in downtown Washington, D.C. and 19-20 in midtown Manhattan. Since these FAR values are for lots, and not a whole city neighborhood, which would include roads, then the more expansive version of FAR would be less. Still, I would guess that this consideration reduces the land reclamation portion of the cost for new floor space to no more than $10-100/sqft.
Then we come to environmental costs, which are not factored into the above prices. Here I will quote a couple of papers, both analyzing land reclamation projects in China. This paper analyzes reclamation around Jiaozhou Bay and finds a net present value of around $2.10/sqft at a 7% discount rate, though it varies substantially over time. This paper looks at the city of Xiemen and finds a cost of $18-28/sqft for various project scenarios. Both papers use ecosystem service valuation, a technique into which I place low confidence, but that’s what I have. These costs are probably not enough so that, if they were incorporated into the internal financial calculus of a developer, they would substantially alter decisions.
I would like to do a more rigorous evaluation someday, but for now, I stand by my judgment that it will probably be very difficult to put together a financial case for floating platforms instead of land reclamation, even taking environmental costs into account.
Quick Hits
Jessica Weinkle has written an important article for the Breakthrough Journal about the Planetary Boundaries framework. Planetary Boundaries is a concept developed by Johan Rockström and coauthors in a 2009 article and since expanded on, and it has a home at the Stockholm Resilience Center. Planetary Boundaries holds that there are nine major environmental challenges that are defined as “planetary boundaries”, such that transgressing them poses an existential risk to civilization. As of 2023, six boundaries were assessed to be transgressed. As Weinkle documents, Planetary Boundaries is a spiritual successor of the Limits to Growth model, in that both are broad models, both disguise political propositions in scientific jargon, and both have a Malthusian orientation that reaches immediately to population control and degrowth as solutions to environmental challenges. This week’s piece is one of dozens I could write about how actual solutions to environmental challenges are based on ingenuity rather than austerity. What is particularly disturbing, though, is the extent that, as Weinkle documents, Planetary Boundaries has become an operative framework for wide swathes of academia, and even other fields including health and finance.
There is a new paper on the ArXiv introducing Husky, a language agent that is able to break natural language problems into atomic pieces and apply external tools to solve them and assemble a full solution. Here the external tools are modules for executing code, basic math, search queries, and common sense reasoning. What I find especially interesting is how the authors analogize to a classical (1971) planning system called STRIPS. It is common wisdom that deep learning has proven so superior to all other machine learning strategies that all but deep learning are now obsolete, but I expect that in the next few years, we will see fruitful research into ways that deep learning can revitalize older approaches that seemed to hit roadblocks, even the expert systems that seemed to go out the window in the first AI winter.