This year, the federal government requires that more than 10% of diesel used in the US be biofuel. To meet this requirement, Americans will have to use almost 7 billion gallons of biomass-based diesel (BBD), up from 4.3 billion gallons in 2025. Most BBD is made from vegetable oils such as soybean or palm. The BBD mandate stems from the Renewable Fuel Standard, a 2007 law that requires some gasoline and diesel to be replaced with biofuels.

As I wrote last November, using vegetable oils to make biofuel has generated more carbon emissions than fossil diesel because it caused deforestation. In this article, I propose a way to mandate biofuels without causing deforestation.

Subscribe now

Carbon Emissions from Biofuel

When we use a fossil fuel, we take carbon that was stored underground and put it into the atmosphere.

When we use a biofuel like BBD, we take carbon from the atmosphere and put it back into the atmosphere. As soybeans grow, they suck carbon dioxide out of the air through photosynthesis. Burning biofuel made from soybeans sends that carbon dioxide back into the atmosphere.

Now, the biofuel system is not completely circular because it causes additional emissions through things like operating a tractor on the farm, making fertilizer, and processing soybeans into fuel. Still, based on a comparison of photosynthesis to fossil fuel extraction, biofuels appear great for the climate; they make fuel from carbon that is already in the atmosphere rather than adding new carbon from underground.

Some of you are yelling at your screen right now because I excluded land from the above narrative. Growing crops for fuel also requires land, and land contains carbon. For example, when farmers clear forests to grow crops, the carbon in the trees and soil ends up in the atmosphere.

Accounting for land changes the narrative: biofuels transfer carbon from forests and soils into the atmosphere, whereas fossil fuels transfer carbon from under ground into the atmosphere. That sounds bad for biofuels. My research with Tzu-Hui Chen and Rich Sexton shows that, after accounting for induced deforestation in Indonesia and Malaysia, BBD from soybeans is worse for the climate than fossil diesel. Thus, policies in the United States can cause climate change by altering economic activity in other parts of the world.

Can US policy makers prevent people in other countries from deforesting in response to US biofuel policy?

The theoretical answer is yes, but only if biofuel policy does not increase vegetable oil prices. No increase in prices means no incentive to clear land to produce more vegetable oil.

Of course, long-time observers understand that biofuels are agricultural policy not climate policy. Their purpose is to raise crop prices for the benefit of corn and soybean farmers. This creates a dilemma. The political backing for biofuels relies on increasing crop prices, but avoiding deforestation requires that crop prices not increase.

Here is a solution to the dilemma: (i) tax the use of vegetable oils in food, and (ii) send the proceeds of the tax to farmers. The tax needs to be large enough to reduce food use by the same amount as the increased biofuel use so that total demand does not change and therefore prices do not change.

The graph below illustrates this idea using supply and demand curves. A biofuel mandate shifts the demand for vegetable oil to the right. A food tax shifts the demand for vegetable oil down.

Successfully enacting this policy would require knowledge of the elasticity of demand for vegetable oils. If demand were elastic, then a small tax would cause food consumers to cut back substantially. If demand were inelastic, then a large tax would be required.

How large would the tax need to be?

In a recent working paper, Tzu-Hui Chen, Rich Sexton, Andrew Swanson and I estimate a demand elasticity of -0.1 for US soybean oil. This estimate is in line with recent global estimates, and somewhat lower than older estimates for the United States. A lot of vegetable oil is used as a minor ingredient in processed food, so it is reasonable that demand from those users could be quite inelastic.

In 2025, the US used 23.5 billion pounds of crop-based vegetable oil for food and 19 billion pounds for biofuel (labeled as industrial in the figure below). In 2026, USDA predicts an additional 5 billions being used for biofuel. To offset the effects of this increase, we would need to reduce demand from food consumers by 5 billion pounds (about 10% of total consumption). An elasticity of -0.1 means that a 10% demand reduction requires a 100% price increase. If the demand elasticity were larger, then the tax would be smaller.

The price of vegetable oil in 2025 was 50c per pound. A 50c per pound tax on food use would raise an expected $12 billion. This is the same amount that the Trump Administration recently proposed to send farmers to compensate for economic losses due to tariffs and other cost increases.

The graph below shows US consumption of the four major oils. Soybean oil is 75% of US vegetable oil consumption, but the tax would need to be levied on all oils to prevent food users from avoiding the tax by switching to an alternate oil.

Tongue firmly in cheek

I am not seriously suggesting a food oil tax, although Health Secretary Kennedy’s hostility to “seed oils” in food suggests that there would be a political constituency for it.

The tax would be costly to administer. For example, to make sure food processing firms didn’t move offshore to avoid the tax, you would need to tax the vegetable oil in imported processed foods. Much better would be to just stop mandating vegetable-oil based biofuels.

Biofuels only induce deforestation when they induce cropland expansion. Thus, we can avoid cropland expansion if the demand for crops to use in biofuel is met entirely by reductions in demand from other users, rather than by increases in supply.

In short, vegetable-oil based biofuels can benefit the climate, but only if we let them take food out of our mouths.

I made the soybean oil use figure with this R code and the help of Claude Code.

I published a lightly edited version of this article on the EI Blog on 8/17/26.

This article was written by UC Berkeley ARE PhD student Ilaria Dal Barco. It is the second of two excellent articles written by students in my ARE 242 class this past spring that I am publishing here.

Subscribe now

Beef prices in the United States are at historically high levels. Ground beef prices are now close to twice their 2010 level, even after adjusting for inflation, and the increase has been particularly pronounced since 2020. Over the same period, steak prices have risen by more than 30 percent. The figures below show the evolution of prices in levels and their cumulative change over time.

Although recent inflation has contributed to higher food prices more broadly, the main reason beef has become so expensive is more specific: the cattle sector is currently in a tight phase of the cattle cycle, with unusually low herd sizes constraining supply and pushing prices up. This dynamic is especially strong in cattle because production adjusts much more slowly than in other livestock sectors.

What is the cattle cycle?

The cattle cycle refers to a recurring pattern in which cattle inventory and cattle prices swing in opposite directions over time. When inventories are low, beef supply is limited and prices rise; when producers rebuild herds, supply increases and prices fall. This dynamic is particularly pronounced in cattle relative to other livestock because cattle production is slow. It can take close to two years to go from breeding to slaughter, whereas production cycles in sectors such as poultry are much shorter. As a result, supply responds only gradually to changes in prices, which makes the cycle slow and persistent.

The figure below illustrates this inverse relationship. Decisions taken today affect supply several years ahead. When herd sizes are reduced, it takes time to rebuild them. When herd sizes are expanded, it takes time before the additional supply reaches the market. These lags generate fluctuations that typically unfold over a period of around 8 to 10 years.

The current episode fits this broader pattern, but it has also been amplified by recent shocks. A severe drought that began in 2022 increased production costs for farmers. When pasture conditions deteriorate, cattle producers need to rely more heavily on purchased feed and, in some cases, transport water to maintain their herds. Both adjustments raise costs and make it more difficult to sustain large inventories of cattle. A similar episode around 2014 also reduced herd sizes and contributed to higher cattle prices.

That earlier episode helps explain why herd rebuilding has been slow this time. When prices rose after the 2014 drought, many producers invested in expanding their herds, expecting sustained high returns. But as more cattle later reached the market, supply increased and prices declined, reducing the profitability of those investments. Producers now face the same trade-off: they can retain heifers for breeding and expand future production, or they can sell animals for slaughter and benefit from current high prices. Rebuilding a herd delays revenue and requires confidence that prices will remain favorable by the time the additional cattle reach the market.

Partly because of that recent experience, many producers appear reluctant to expand quickly and are behaving more cautiously. Expanding a herd requires retaining heifers for breeding, which delays revenue and only increases beef supply with a substantial lag. USDA noted in early 2025 that it could take several years before the U.S. cattle herd expands meaningfully, because heifer retention remained weak and, biologically, even the offspring from heifers born in 2024 would not enter the beef cow herd until 2027. This means that even if producers begin rebuilding, the resulting increase in beef supply will arrive only gradually.

Because herd rebuilding is slow, low cattle inventories translate into tight beef supply for an extended period. The figure below shows this clearly: across different cuts, beef prices tend to be higher when cattle inventory is lower. Today’s high beef prices therefore do not look like a purely temporary spike. Instead, they reflect a predictable phase of the cattle cycle, one made more severe by drought and the slow pace of herd rebuilding.

Imports have not been large enough to offset the decline in domestic supply. Trade flows are constrained by a combination of policy, sanitary regulations, and conditions in other producing countries, which have also faced similar pressures. At the same time, demand has remained relatively stable: consumption has not fallen sharply, and wage growth has supported continued spending on food. With supply still tight and demand holding up, beef prices are likely to remain elevated for some time.

You can replicate the figures with this R code.

This article was written by UC Berkeley ARE PhD student Nshan Burns. It is the first of two excellent articles written by students in my ARE 242 class this past spring that I will publish here.

California accounts for approximately 80% of the world’s and 100% of the United States’ almond supply. Between 1995 and 2024, almond acreage more than tripled and production increased more than sevenfold, due in part to the availability of groundwater, which farmers could pump without any charge, for a crop whose production is notoriously water-intensive. Roughly 1.6–1.7 gallons of water are required to produce a single almond.

California’s 2014 Sustainable Groundwater Management Act (SGMA) was designed to ensure the sustainable use of groundwater by farms. However, SGMA’s gradual rollout gave farmers years of warning without changing their immediate incentives, and high prices and uncertain future regulatory conditions may have caused farmers to ignore this signal in favor of continuing to invest in almond orchards. Enforcement is now bearing down on an industry that overinvested at the wrong time. Counterintuitively, a more immediate and aggressive signal may have reduced the long-run transition costs for farmers.

SGMA requires 127 water basins, covering 96% of California groundwater, to achieve sustainability by 2042. “Sustainability” here means that water extraction cannot exceed natural replenishment. Local authorities were required to establish Groundwater Sustainability Agencies by 2017, which were required to adopt Groundwater Sustainability Plans (GSPs) by 2022. These GSPs typically involve caps on pumping, overage fees, or groundwater trading markets.

Aquifers are currently so overdrawn that reaching sustainability will likely require the fallowing of irrigated land. Before GSPs took effect, farmers could still pump freely, and they only faced an uncertain, remote signal of future regulation – they may not have even known how heavily their basin would be regulated or what compliance would cost them.

Subscribe now

Much of the almond planting since 1995 happened during (and likely because of) a period of high prices from 2012-2019. Once these trees started to bear fruit, large consecutive harvests, combined with the impact of the COVID-19 pandemic, drove the price of almonds down far below $2.10 per pound (estimated by one analyst as the 2024 break-even price) from 2020–2023. Prices recovered to $2.14 in 2024, as plantings leveled off, yields reduced, and demand continued to grow. However, farms have been left financially destabilized and hoping for a market rebound.

Perennial crops like almonds exhibit a lower short-run elasticity of supply than annual crops, due primarily to “asset fixity” (see Siegle et al., 2024). Orchards lock in capital for decades, and production does not start until years after planting, meaning that immediate supply responses to price fluctuations are more costly. This suggests that myopia regarding future regulatory or market conditions could be especially burdensome for perennial farmers. Those who borrowed heavily during high prices might not be able to exit in the future without crystallizing significant losses.

The Public Policy Institute of California estimates that compliance with SGMA will eventually require the fallowing of at least 500,000 acres of irrigated cropland in the San Joaquin Valley by 2040. Combined with the effects of climate change and other environmental regulations, this figure increases to 900,000 acres (in the worst-case scenario), representing more than half of cropland in some basins. Appraisers estimate that almond orchards in parts of the San Joaquin Valley lost 25–50% of their value from 2023–24, especially in groundwater-reliant parcels, therefore attributing the loss partly to looming SGMA compliance.

Unfortunately, the preliminary economic research suggests that SGMA’s long implementation window didn’t help farmers adjust. Work by Ellen Bruno and Nick Hagerty shows no difference in well construction and orchard planting rates in more-regulated versus less-regulated basins after SGMA’s passage (see the below figure from their article). Farmers in more-regulated areas did not change their behavior throughout the entire 2014–2019 window when the law was on the books but regulations were not yet in effect. Perennial crop acreage increased by nearly 50% after SGMA passed. High almond prices dominated future regulation as a signal, the latter of which remained years away and highly uncertain.

This points to an irony of SGMA’s design. The long transition period was intended to give farmers time to adjust, but instead, it gave them time to overinvest. Because farmers in more heavily regulated basins have not been reducing planting or drilling relative to their neighbors, the industry’s exposure to the 2042 compliance deadline is larger than it otherwise would be. Now the adjustment will arrive to an industry with millions of water-dependent trees it cannot easily remove, loans it cannot easily repay, and land whose value has declined, all in the context of precarious market conditions.

None of this means that SGMA’s goals are wrong; California’s aquifers are being depleted at an unsustainable rate. But maybe its implementation could have been better. A uniform, state-implemented groundwater pumping fee that escalated predictably from 2014 onward could have lowered the net present value of planting new orchards, in a salient manner, exactly when the boom was happening, and caused farmers to make more efficient investment decisions.

Policymakers can learn from the SGMA example. Arizona’s groundwater basins are rapidly depleting. The state produces a large quantity of pecans and dates (both water-intensive perennials), but it has not yet imposed an equivalent to SGMA. Unfortunately, its proposed Rural Groundwater Management Act (RGMA) is designed to allow local councils to set their own conservation targets on decade-long review cycles — exactly the kind of uncertain regulatory signal that experience suggests farmers will ignore.

RGMA died in Arizona’s 2025 legislative session without a committee hearing, and it will likely do so again if reintroduced in 2026. Regardless, groundwater constraints will arrive, whether in the form of regulation, a pricing mechanism, or catastrophically low water levels resulting from inaction. The best policy for farmers (and for the long-term stability of supply, which would benefit consumers as well) is an immediate, strong signal that causes farmers to adapt sooner rather than later.

There are hundreds of thousands of farmers in the United States, many of them producing the same product as their neighbor. They have ample information about global markets for their product and can start or stop farming whenever they like.

These characteristics describe a perfectly competitive market. As every economics student learns, no-one consistently makes large profits in a perfectly competitive market. They make enough to fairly compensate them for their time and investment and that is it.

So, is it true that American farmers make low profits?

Subscribe now

One place to look for the answer is the Farm Income and Wealth Statistics produced by the Economic Research Service of the USDA.

Let’s start with gross farm income. American farmers received gross revenue of about $600 billion in each of the last six years, which is 2% of GDP. Crops and animal products each account for 45% of this revenue, and government payments accounted for 4%.

Some gross income goes to pay for inputs like fertilizer and seed, and some of it goes to profit. On average, net profits are 21% of gross farm income. After subtracting government payments, net profit from market sales is 17% of gross revenue. These are high percentages. The average net profit margin in the US economy is 10%. Grocery stores, which are the prototypical low-margin industry, have a net profit margin of 1%.

If net profit margin is 17%, why do I say farming is a low margin business?

Farmers make money in two ways. They own land that goes up in value and they sell agricultural products. The second part is the business of farming and that is where the low margins are.

Land owned by the farmer is not free. If they didn’t farm on it, they could rent it to someone else. They are giving up that income when they choose to farm on it. So, we should add foregone land rent to production costs, which reduces the profit margin.

The same argument applies to the farmer’s time (and the time of family members and others who give unpaid labor to the farm). If they weren’t working on the farm, they could work elsewhere for a wage.

USDA publishes estimated costs and returns for producing various commodities. These data count opportunity costs of owned land and unpaid labor as costs to the farm business. They exclude government payments.

Over the last 50 years average net profits for soybeans were 7% of the value of production. Average returns for corn were -1.2%, and average returns for milk and hogs were both less than 5%. The business of farming is indeed a low-margin business.

In contrast to overall farm profit, commodity-specific net returns are very volatile. Corn grower losses were 40% of revenue in 2005 and their profits were 20% of revenue in 2011, as shown above. Overall farm profits are never negative and were only below 10% of revenue in one year (1983).

So, individual farmers face significant ups and downs from year to year, but fortunes average out across farmers. Corn and soybean growers lost money in 2024 and 2025, but cattle farmers made high profits in those years.

So, are farmers poor?

No. Most commercial farmers own a valuable asset (farmland) that tends to increase in value. USDA ERS reports that “households operating commercial farms (those with gross cash farm income of at least $350,000) had $3.9 million in total wealth at the median, substantially more than the households of residence or intermediate farms.”

In 2024, ERS reports that more than 97% of farm households had more wealth than the median household. Also in that year, 40% of farm households had below-median income. Farming is a low-margin volatile business.

Some of you may be thinking of farmers you know who have made large profits. Look deeper, and you will probably see conditions that diverge from the perfect competition model. For example, a farmer may be first to use a productive new technology. They can reap profits from being better than everyone else, but soon others catch up and we are back to competition and low profits.

For another example, many California farmers made large profits on almonds and pistachios in the last 20 years. Supply was constrained by the fact that it takes several years for nut trees to reach bearing age after they are planted. But, eventually farmers produce more of highly profitable crops, creating extra supply that reduces the price and eliminates the profits, as has happened in almonds.

In summary

Low profits are intrinsic to commodity farming, yet our politics are beset by the sentiment that farmers don’t make enough money. The problem is that, whenever farm profits appear, they disappear just as quickly into the value of farmland or are captured by seed companies, fertilizer companies, or other suppliers.

Trying to persistently increase farm profits is a fool’s errand that policy makers have been chasing for decades.

I made the graphs using this R code.

Nowadays, dystopically: Professor stands at the front of a dimly lit room and talks through slides projected onto a giant screen, occasionally writing on a whiteboard with pens that don’t work. A small number of students stare blankly at their laptops or phones. Most of the class didn’t show up. They told themselves they would download the slides and watch a recording of the lecture later, but they probably won’t because ChatGPT will do their homework anyway.

In reality, things aren’t quite that dystopian, and the old days weren’t quite that good. However, because of AI, college professors can no longer ignore the fact that our traditional modes of teaching don’t work very well. It has always been possible for students to zone out during class or get someone else to do their homework. AI just makes it a whole lot easier.

I believe we owe it to students, their families, and taxpayers to adapt. I am not writing this post because I have figured it out. Far from it. However, I wanted to share a few things I tried this semester and some thoughts about how well I think they worked.

Subscribe now

This semester I taught Agricultural and Environmental Policy to a class of 156 mostly juniors and seniors (syllabus). My goal was to help them learn to analyze big issues in agriculture and the environment through an economics lens. I wanted them to get excited about the topics. I wanted them to engage their brains and work their analytical muscles. I really enjoyed teaching the class.

Thinking and doing

If I only assign tasks that students can easily get an AI to do for them, then I am preparing them to be replaced by AI. Why would an employer pay them to do a set of tasks that AI can do? I believe the ability to think deeply about how to define and analyze important problems is a skill that transcends the robots and equips people to adapt as the world changes.

I assume most of the students use AI tools in my class. Rather than a negative goal (prevent them from using AI), I have positive goals for the students (help them get excited to think deeply and to analyze problems) .

To these ends, I gave two writing assignments: a 600-800 word blog and a 2000-2500 word paper. I was anxious about whether this would work. How do you help 156 people do substantive work on a topic of their choice?

In this article, I describe how I set up the paper assignment. The blog was similar (with fewer steps). You can read my two favorite blog submissions here and here.

To inspire students to think deeply about a topic and analyze it critically, I had two strategies: (i) get them to talk about their topic, and (ii) give critical feedback. I think we had some good ways of achieving (i). I’m trying to figure out better ways to do (ii) in a class of this size.

I set up multiple intermediate deadlines in the assignment. Without these deadlines, it would be easy to procrastinate; you can’t study any topic deeply if you are rushing to do it at the last minute.

I allowed the students to work with a co-author on their paper, but each student had to do the oral components of the assignment individually.

First, the students submitted a proposal in the form of a single presentation slide containing their research question, why it is important or interesting, and how they plan to answer it.

A few days later during class, we divided the students into groups of five and had them each present to the other members of the group. They were to talk for no more than one minute and their groupmates were to question them for an additional two minutes. I suggested some questions they could ask, such as “What data will you use?”, “Do you have the data already?”, “Can you give a specific example?”, and “Why should we care?”. I used a timer so that each round lasted the specified time.

I think the first two steps worked well to get them started.

The next major steps also involved talking about their work, this time in the form of (i) a 5-minute video presentation, and (ii) an oral exam.

The oral exams required surgical precision. We devoted the entire last week of class to them (both lecture and section). I assigned each student an exam time — one every two minutes.

On the exam days, the three examiners (me and my two amazing graduate student assistants, Shuo and Molly) set up in three separate corners of the classroom. Students would spend two minutes talking to the first examiner, then move to the second, and then the third. One minute and 40 seconds into every exam, a timer buzzed, signaling that it was time to move to the next station.

Each examiner had a designated topic to cover, as shown in the above screenshot. We graded the students on the spot, but didn’t reveal the grades until later. We had formulated student-specific questions based on their video presentation, but rarely got the opportunity to ask them before time ran out. We did share those questions and comments with each student later to help them improve their paper.

In spite of the timer, we inevitably fell behind. In several cases, the last couple of students were finishing their exams as the next class was filing in. Not ideal, but we got it done.

For me, the exams were a lot of fun. Before the exams, I had only spoken one-on-one with a small minority of students. It was great to meet them all. I really enjoyed hearing how excited many of them were about their research topic.

The students are now finishing up writing their papers, which are due at the end of this week. Here’s a word cloud of the topics they chose.

The most glaring deficiency in this assignment is the amount of critical feedback we provided. I would like to modify the oral exams to allow time for the examiners to give immediate feedback. Ideally, we would deliver the feedback in oral and written form (e.g., could an AI record what we say and email it to the students? ). We could make students accountable for responding to the feedback.

I think we also need to require the students to submit a draft paper that we provide feedback on. The 5-minute video presentation and 6-minute oral exam allowed us to push the students on their basic arguments, but refining the details requires more time.

I wonder whether it would be possible to train an AI to do this effectively — at least on the structure and quality of the arguments. And, yes, I get the irony of working hard to get students to invest time and effort into a project while I work hard to get AI to reduce my time and effort. This is the challenge of scaling labor-intensive teaching to large classes.

Finally, I’m sure there are ways to improve the grading rubric.

In the future, I may write about other aspects of the class, such as the required weekly reading assignments, the in-class trading games Claude helped me code, giving credit for class participation using iClicker and Google forms, the in-class written exam, and the econometric replication projects in discussion section.

I hope this post provides an interesting window into the classroom. Please add comments with ideas, reactions, and suggestions. This is an exciting time to be a college professor.

An engineer can build you a new mousetrap. Just give them the specs and they will design it and make sure it works.

An economist can tell you how many new mousetraps you will be able to sell.

Subscribe now

OK, that’s promising too much. Economists can’t tell you exactly how many you will sell, but they can explain how prices, government policy, and market structure will determine outcomes. For example,

1. If your new mousetraps cost more than old mousetraps, then you will only be able to sell them if people like them better than old mousetraps.

2. If your new mousetraps cost more than old mousetraps and work just as well, but the government gives you a subsidy so you can sell them at a competitive price, then people may buy your mousetraps instead of old mousetraps.

3. If your region imports mousetraps and then you start producing mousetraps locally, the price of mousetraps won’t materially change. Global supply and demand determine the price and your production is too small to materially affect that price.

Why am I analogizing with mousetraps? It’s my latest attempt to explain the potential effects of California’s proposed tax credit for biofuel in airplanes, known as sustainable aviation fuel (SAF).

In case you missed the previous editions, the California governor has proposed a subsidy for using SAF. My co-authors and I pointed out the likely unintended consequences, including a large loss of road-repair funds and an increase in fuel prices at the pump. We also showed that the environmental benefits would be small. Advocates for the policy claimed that we had made some errors. I explained that our analysis was sound.

The latest contribution to the discourse is a piece by Harold “Skip” York of the Baker Institute for Public Policy and the Sustainable Aviation Advisory Council. York describes how (i) biofuel refineries that produce SAF also produce by-products that could be used in gasoline, (ii) if crude oil refineries produced less jet fuel, they could produce more gasoline, and (iii) with more in-state SAF production, ships that had been carrying imported jet fuel could instead carry imported gasoline.

He projects that, if the Phillips 66 refinery in Rodeo California were to produce SAF, then these three factors could cause California gasoline prices to decline by 20-40 cents per gallon!

Assuming it’s correct, this engineering analysis demonstrates three things that could happen, but whether they will happen depends on the economics.

Here’s my economic analysis:

1. Biofuels such as renewable diesel (RD) and SAF are much more expensive than their fossil fuel counterparts. They are also essentially identical to their fossil counterparts. Some airlines may buy a little SAF for green marketing purposes, but generally people will not buy these products of their own volition (see mousetrap item #1 above).

2. If government subsidies bring RD and SAF prices to parity with their fossil counterparts, then people will buy them. Federal policy subsidizes a specified volume of RD+SAF (see the table below). If you make SAF relatively more attractive, then people will buy less RD (see mousetrap item #2 above).

3. California imports gasoline and jet fuel now. It would still import gasoline and jet fuel if Phillips 66 produced SAF at full capacity, so it would still be subject to world market prices. Producing a little more gasoline in the state would not reduce the price of the marginal gallon used in the state (see mousetrap item #3 above).

There are valid reasons to subsidize SAF. If we believe that, by producing SAF, the industry will figure out how to dramatically reduce costs, then there’s value to subsidizing SAF now. This is the story of solar and batteries.

I’m somewhat skeptical that this argument applies to vegetable-oil based SAF because the big cost driver is the vegetable oil. SAF costs are high because vegetable oil is expensive, rather than because refineries haven’t yet figured out how to cheaply turn vegetable oil into fuel. For example, the current price of soybean oil is 75c/lb and you need 9 pounds to make a gallon of SAF, so you’re up to $6.75 per gallon even before accounting for production costs. Jet fuel currently trades for $4, up from $2 before the Iran war.

Societies can also choose to subsidize a company or a production facility to help the workers or owners of that facility. If drivers and taxpayers in the state of California want to pay a biofuel refinery to produce expensive airplane fuel, then they can do that. But, I think it’s important to be clear-eyed about the costs, benefits, and potential unintended consequences of any such policy.

Luciné is a senior studying Political Economy with a concentration in Global Development and the Environment. During her time at UC Berkeley, she served as President of Berkeley International Consulting and worked as a Lead Coach at The Practice Space. After graduation, she looks forward to joining the tech world in San Francisco and continuing to explore connections between development and business.

Tucked away in my hometown of Montrose, California is the historical Rockhaven Sanitarium, a feminist sanitarium that operated from 1923 to the early 2000s. Stories of actresses escaping the chaos of city life for the lush foothills of the San Gabriel Mountains are a defining part of Rockhaven’s history. Undeniably, the beautiful citrus groves that once sprawled here contributed to the serenity offered by Rockhaven.

As a child, my parents often brought me to benefit events for the Rockhaven Historical Society, where I heard stories about the citrus orchards that once surrounded the area and offered their supposed healing benefits. I could easily imagine these landscapes, but I rarely saw anything like them in present-day Los Angeles. Yet these orchards were more than scenery – they played a central role in shaping the city’s economy, culture, and identity.

Subscribe now

Los Angeles’s citrus industry was born in 1841 when pioneer William Wolfskill planted the county’s first commercial orange grove in what is now present day Downtown Los Angeles. While the California Gold Rush took off, Wolfskill made his way in “citrus gold.”

Between 1909 and 1950, Los Angeles was the nation’s leading farm county, with citrus fruits growing alongside a number of agricultural products. In the San Fernando Valley, fruit production – especially citrus – was the leading industry, and land used for orange and lemon groves could sell for up to eight times the price of other land. The citrus industry was undeniably defining agriculture in Los Angeles County.

In the early 1900s, there were about a million orange trees and a quarter million lemon trees in Los Angeles County, as shown in the figure below. California Board of Equalization reports indicate a large but brief jump around 1907 followed by a decline a few years later. It is unclear what caused these changes. What is clear is that there are now only about 76 acres of citrus in Los Angeles County, or about 15,000 trees.

Beyond its economic importance, citrus also shaped the cultural identity of Los Angeles. Promotional graphics of endless citrus groves helped construct the image of “Sunny California” and the “Golden State.” Furthermore, health became associated with the golden fruits grown in the sunshine of Los Angeles, attracting consumers to purchase citrus products and move to the dreamy land of sun.

In the case of my local Rockhaven and other sanitariums alike, the bliss of citrus and agricultural fields offered a serene backdrop for healing and wellness.

However, the post-World War II era brought dramatic changes to Los Angeles’s landscape. Increasing urbanization put pressure on farmers, and acres of groves were replaced by acres of homes, a hallmark of Los Angeles’s now famous suburban sprawl. As residential areas grew, Los Angeles’s agriculture was relocated to the Central Valley, where three of its counties lead in acreage dedicated to citrus plants as of the latest available data in 2022.

Presently, Los Angeles County ranks as 31st among the list of California counties and their acreage of citrus bearing and non-bearing trees, a far fall from its glory days as the agricultural capital of the U.S. However, its history is not completely lost.

Some public institutions alongside private partners have fought to preserve the legacy of citrus production in Los Angeles. Located in the San Fernando Valley, the California State University of Northridge (CSUN) maintains a five acre orange grove with over 400 navel trees. Open to the public, the grove is an enjoyable area for picnics, scenery, and recreation. Partnering with Fruit Forward, a nonprofit dedicated to redistributing surplus produce to those facing food insecurity, CSUN harvests their oranges for consumption.

Crop Swap Los Angeles is an urban agriculture initiative that grows and sells fresh produce throughout the county. At the Asanate Microfarm, Crop Swap harvest lemons, tangerines, and oranges. Initiatives like these preserve the legacy of citrus production in Los Angeles while also addressing food deserts and gaps in fresh-food access.

While Los Angeles may never experience another agricultural “gold rush,” the legacy of its citrus industry still shapes the region’s identity – from the “Golden State” imagery to the growing movement for urban agriculture. Expanding these efforts could preserve historical knowledge while improving food access, ensuring that future generations experience Los Angeles’s citrus legacy firsthand rather than only hearing about it.

Mattea is a sophomore studying Environmental Economics and Policy with a minor in Theater, as she is passionate about using storytelling to drive effective climate solutions. She has spent the past year interning at a climate technology venture capital firm, and she plans to continue in climate VC, focusing on innovations that can revolutionize and decarbonize the agricultural sector.

In middle school, following the example of my emerging-environmentalist older sister, I swore off beef. I was told cows had the worst greenhouse gas emissions among livestock, and though I didn’t want to go vegetarian (I loved chicken empanadas too much), I wanted to take action to help our planet. My sister and I have continued not eating beef to this day, and we even roped my mom into our diet shift.

For years, I’ve been proud that my lifetime emissions were reduced, and I’ve loved adoring the cute cows dotting California fields without guilt because I wasn’t the cause of their slaughter. Only a month ago, I realized I may have been complicit in their murder all along.

Subscribe now

One of my favorite sweet treats is a chocolate collagen powder that I mix in my non-dairy yogurt bowl every morning. It’s high in protein, low-carb, and delicious! I hooked many friends on it and even my grandpa, and we all agreed it was almost too good to be true. Because it was. I recently discovered that the powder has one simple ingredient: bovine hide collagen peptides. I’ve been eating cow skin every morning for years.

Horrified, I tried to research what percent of the lifetime emissions of cattle went to collagen to understand how much of their methane-riddled emissions I was inadvertently supporting with my collagen consumption. I found conflicting conclusions, revealing a currently debated subject among agricultural economists around how to measure the carbon footprints of livestock products. The emissions of one steer with 341 kg carcass weight are around 6,250 kg of CO₂e. This is equivalent to the emissions from driving 15,916 miles in an average gasoline-powered car.

Cattle produce multiple outputs—such as beef, hides, bones, and collagen—so determining the lifetime emissions of each product requires allocating cattle’s emissions to each product. However, methods of allocation vary, and each produces widely different results.

Collagen, as well as other non-meat carcass products like leather, are sometimes deemed byproducts, meaning they’re a sustainable utilization of otherwise wasted “leftovers” of meat and dairy production. I like that answer! However, these byproducts—encompassed in the non-beef categories of the figure above—account for about 5% of the total economic value per steer head. Given the small profit margins of the ranching competitive market, it’s likely that the revenue from these additional products keeps ranchers in business. Additionally, this recent study found that “neglecting co-product allocation leads to the systematic overestimation of the environmental impact of meat.” Thus, these non-meat products are more accurately co-products and should receive a share of cattle emissions.

But how should emissions be allocated? Using economic allocation based on relative market value, under the downstream assumption that 20% of the hide market is tied to collagen extraction, collagen accounts for over 5% of non-meat value, or 0.26% of total steer value. Allocation based on mass attributes 8% of live weight to hides (far greater than the 5% allocated to all coproducts under economic allocation), so the collagen share is around 1.6%. If you allocate based on protein share, collagen comprises 30% of the total protein mass in mammals.

Depending on the chosen allocation method, the share of beef steer lifecycle emissions attributable to collagen ranges from less than a percent to nearly a third of the total. To paint myself in the best light, I could simply decide that collagen is a byproduct, and my consumption sustainably reduces waste.

Organizations like the International Dairy Federation and OECD are trying to formalize an allocation framework that outlines which allocation method to use for different products. They conclude that revenue-generating co-products, like collagen, should use economic allocation, meaning collagen is only responsible for 0.26% of cattle emissions, or 16.25 CO₂e per cow, equivalent to the emissions from driving 41 miles.

But what about per serving of collagen? Each morning, I consume 20g of collagen, and each 700 kg live weight cow produces 11.2 kg of collagen under my previous assumptions. This leads to only 29 grams of CO₂e per serving attributable to cattle. Over a year, my daily habit results in 10.59 kg CO₂e annually from cattle, which is less than 0.17% of one cow’s lifecycle emissions and equivalent to driving 27 miles per year.

It’s important to note that these figures only include the emissions from cattle. Research on hydrolyzed bovine collagen shows that only 20% of the carbon footprint of collagen comes from agriculture, so the total carbon footprint of collagen is much larger. However, when focusing on the cattle emissions attributable to my collagen consumption, the answer appears to be negligible.

I spent hours researching different allocation frameworks and finding arguments for both my culpability and sainthood. The most disquieting fact is that, during the extensive time required to filter through the barrage of opinions, I grew overwhelmed by choice overload, and I kept eating my collagen. Easily understanding the carbon footprint of consumer products is not currently viable.

In an ideal, sustainable world, farmers would have efficient tools to measure their emissions. Accurate and clear allocation systems would allow every item in a grocery store to have an emissions label, akin to a price sticker. Then, customers could make educated choices, and there could be tax/subsidy systems to support lower-emission products. This would be a critical step in lowering global emissions. But that idyllic world is far from our reality.

Two weeks ago, my co-authors and I posted an article about a new tax credit that California Governor Gavin Newsom has proposed as a trailer bill to the 2026-27 state budget. It would allow oil companies that pay diesel taxes to reduce their tax bill if they also produce sustainable aviation fuel.

We argued that the proposed tax credit would reduce road funding, raise gasoline and diesel prices, and deliver small and expensive carbon emissions reductions. The Sustainable Aviation Advisory Council posted a response to our article entitled “Six Faulty Assumptions and Facts about California’s Proposed SAF Credit”. They included the above graphic to illustrate their arguments.

Here, I explain that the council’s points are incorrect. I list their six claims and follow each one with my response.

Subscribe now

First Claim

Assumption #1: “The SAF credit will significantly raise gasoline prices.”

Fact: This is inaccurate. This conclusion relies on a model that assumes renewable fuel producers are maxed out to full capacity, and if a facility were to produce more SAF, they would produce less renewable diesel. This is not correct. In reality, U.S. renewable fuel producers are running 40% under capacity. They have room to increase production of renewable fuels, which in turn will put downward pressure on Low Carbon Fuel Standard (LCFS) prices. Additionally, displacing fossil aviation fuel with SAF frees up refinery and market capacity to increase gasoline supplies, putting downward pressure on gasoline prices.

This claim is incorrect. We do not assume renewable fuel producers are “maxed out to full capacity”.

Let me explain. Producing more SAF leads to less renewable diesel because of two facts.

Fact 1: both renewable diesel and SAF are $2-$5 per gallon more expensive to produce than their fossil counterparts. Therefore, they need substantial subsidies to be cost competitive, otherwise they will not be produced.

Fact 2: the federal renewable fuel standard (RFS) requires that a minimum quantity of biomass-based diesel fuels be used in the United States each year. SAF counts as biomass-based diesel in the RFS. If the industry produces more biomass-based diesel than required, then the RFS subsidies will disappear. It has no incentive to go above the mandate because of the first fact. Therefore, more SAF means less renewable diesel in aggregate.

Imagine a person who has to drink at least a gallon of a medicine she doesn’t like. There are two types of medicine, and she can drink any amount of either type as long as she drinks at least a gallon in total. If you pay her to drink some of type A, then she will drink less of type B because she dislikes them both and will not drink more than she has to. In the fuels context, a tax credit that incentivizes more SAF will lead to less renewable diesel.

Second Claim

Assumption #2: “There is not enough feedstock to support both renewable diesel and SAF production.”

Fact: This is incorrect. This is the same argument that was used in 2022, before several new renewable fuel production facilities came online (mostly in California) and almost doubled US production capacity. According to the Advanced Biofuels Association, the global supply of eligible feedstock is currently over 5.9 million barrels per day (90 billion gallons) – almost 23 times the diesel demand of California. The global collections rates of used cooking oil are less than 10%, so the availability of renewable feedstocks will likely continue to increase as long as economic incentives remain stable.

We do not assume that “there is not enough feedstock”. In a market-based system, whether there is “enough” of something depends on the price. If there is not enough of something, then the price increases to encourage companies to produce more and consumers to use less. We assume that that if a refiner wants to use more vegetable oils or animal fats, then it has to offer a higher price than alternative users of the feedstock. We place no constraint on the amount of available feedstock.

It is true that enough feedstock exists in the world to produce approximately 23 times the diesel demand of California, but most of this feedstock is destined for the food system. About 20% of the calories consumed by Americans come from vegetable oils and fats. If we want to use more vegetable oils and fats for biofuel, then we need to convince consumers to eat less and farmers to produce more, and the only way to do that in a market system is for the price to increase.

I hope it is true that used cooking oil collections can increase without raising prices. And if this were to happen, using these additional feedstocks for renewable diesel rather than SAF would be less expensive and have a similar climate benefit.

Technical point: We assume in our model that the elasticity of net supply of waste oils and fats equals 2, i.e., a 1% increase in price causes a 2% increase in supply. Increasing this elasticity a little would moderate the gasoline and diesel increases stemming from the tax credit because it would be less expensive to acquire the additional feedstock needed to meet the requirements of the RFS and LCFS while also taking advantage of the SAF tax credit.

Increasing the elasticity a lot would flip the balance for multiple different fuels that operate in the tangled web of state and federal biofuel policies. For example, we ran our model with an elasticity of 10 instead of 2. Without the proposed California SAF tax credit, this higher elasticity causes relatively more renewable diesel from waste oils and fats to be used in California (which would benefit the Phillips 66 refinery in California that currently produces renewable diesel mostly from waste oils and fats). If the proposed SAF tax credit were implemented, then even more renewable diesel would switch to SAF than in our baseline model, i.e., the state would lose even more diesel excise tax receipts than in our baseline model. Details at the end of this post.

Third Claim

Assumption #3: “Renewable diesel refineries can easily switch to producing SAF.”

Fact: This is false. Producing jet-quality renewable fuels requires tighter process control and additional refinery investment. Converting a facility optimized for renewable diesel into SAF production is neither automatic nor inexpensive, which limits the large-scale diversion assumed in the Haas and LAO analyses.

We do not assume that “renewable diesel refineries can easily switch to producing SAF.” At least six refineries in the United States are currently equipped to produce SAF from vegetable oils or animal fats. In May 2025, EIA reported that these refineries could produce 460 million gallons of SAF per year. Farmdocdaily published a larger number, estimating 834 million gallons of existing SAF capacity in 2025 and 2026. Our model of the proposed SAF tax credit projects between 200 and 370 million gallons of SAF from vegetable oils and animal fats, which could be produced using current capacity.

In addition, we project 90-570 million gallons of SAF from alcohol-to-jet (ATJ), which involves refining ethanol or isobutanol made from corn or sugarcane into jet fuel. Although cost estimates vary, this production process is widely understood to be more expensive and nascent relative to HEFA SAF. LanzaJet owns the only ATJ plant currently operating in the United States.

Fourth Claim

Assumption #4: “SAF policies don’t meaningfully reduce emissions.”

Fact: This is incorrect. In fact, SAF has the highest ROI for addressing climate goals in the aviation sector by immediately reducing lifecycle carbon emissions, particulate matter, and other harmful greenhouse gases, without requiring new infrastructure. It benefits local air quality leading to health benefits for vulnerable communities. Additionally, there are no good options to decarbonize aviation. Batteries are too heavy for long-haul aircraft and hydrogen faces major infrastructure challenges. Sustainable aviation fuel is the only scalable pathway to reduce aviation emissions over the coming decades.

Our recent research paper finds that the average cost of reducing carbon emissions with current biofuel policies is around $461 per ton of CO₂, which is more than twice the social cost of carbon. Our analysis of the proposed California SAF tax credit finds that the incremental emissions reductions would cost $997-$2768 per ton of CO₂, depending on the viability of making SAF from ethanol.

This tax credit proposal gives low bang for the buck because it mostly incentivizes switching from renewable diesel to SAF. We are already getting the carbon gains from using renewable diesel, so switching to SAF doesn’t help much.

We only have so many resources to spend on climate change mitigation — choosing expensive policies to reduce emissions means we are choosing to reduce emissions less than we could with more efficient policies.

Fifth Claim

Assumption #5: “The SAF credit would cost the state more than $1 billion per year.”

Fact: The estimate from the LAO report assumes far more SAF production capacity (834 million gallons) than currently exists in the U.S. (460 million gallons) and further assumes rapid conversion of RD refineries (see Assumption #3). When realistic production and logistics constraints are considered, the potential fiscal exposure to California’s budget appears substantially lower.

See my response to the third claim above.

Sixth Claim

Assumption #6: “California can’t reduce global GHG emissions by acting alone.”

Fact: Nearly every climate policy implemented at the state level faces some degree of market and emissions “leakage”. California’s clean fuel policies – like LCFS and cap-and-trade – operate within a global energy system and yet still have successfully driven the expansion of overall low-carbon fuel supply. Despite its small contributions to reducing global GHGs, California’s leadership on decarbonization has already created real-world impacts on markets and technologies at a global scale.

I agree that decarbonizing aviation is important in the long term, but I am skeptical that incentivizing SAF in California will meaningfully accelerate that process. SAF produced from vegetable oils and animal fats is a relatively mature technology with little scope for large cost reductions.

SAF producers need to buy 9 pounds of vegetable oils or animal fats at 50-70 cents per pound to make a gallon of SAF. On top of the feedstock cost, our model assumes an additional $1.63 of production costs. Even if process efficiency gains were to reduce production costs by a dollar, SAF would still be twice the price of fossil jet fuel.

Conclusion

I stand by our three projected outcomes:

1. The proposed tax credit will incentivize much more SAF than projected by the administration.

2. That increase in SAF comes at a cost: diesel excise tax receipts decrease by at least 20% and as much as 75% within a few years, and the prices of gasoline and diesel increase by 10-15 cents because of interactions with existing policies (or more or less with different elasticity assumptions).

3. Because the SAF largely would come from diverting biofuels from surface transport to aviation, the net reductions in carbon emissions would be small and expensive.

The company best placed to immediately benefit from the proposed tax credit is Phillips 66, a Texas oil company that converted its oil refinery in Rodeo California to make renewable diesel and SAF mostly from waste oils and fats. If the objective of the proposed tax credit is to provide financial assistance to Phillips 66, then I encourage the legislature to explore ways to subsidize that company directly, rather than risking potentially large unintended consequences for California drivers and taxpayers.

Disclosure: In the last couple of weeks, I have had two zoom meetings with representatives of Phillips 66 and the Sustainable Aviation Advisory Council in which I listened to their arguments and shared my responses. As of the time of writing, they have not corrected their “six faulty assumptions and facts” post. If they make a correction in the future, then I will note that fact here.

Addendum: Model Results Using a More Elastic Supply of Waste Oils and Fats

In a recent working paper, we developed a detailed model of US state and federal biofuel policies. A couple of weeks ago, we added the proposed CA tax credit to the model. Last week, we made one additional change to our model and re-ran it: we increased the supply elasticity of waste oil and fat feedstocks from 2 to 10 to illustrate what would happen if these feedstocks were to be much more available than we assumed.

The model parameters are calibrated to match 2018-2023 data and the model projects outcomes in 2030. When there is no SAF tax credit, a high elasticity means more waste oils and fats used in renewable diesel production in 2030. These feedstocks make almost three quarters of California renewable diesel up from one quarter in the baseline scenario. This outcome would benefit companies such as Phillips 66 that produce renewable diesel mostly from waste oils and fats. There is no SAF used in California in this scenario because it is too expensive relative to renewable diesel.

When the proposed California SAF tax credit is introduced, the model projects 1.8 billion gallons of SAF, enough to eliminate all diesel tax receipts. There is a correspondingly large reduction in renewable diesel. This quantity of SAF exceeds current production capacity and so would require more SAF capacity to be built, which may not happen in reality as quickly as it does in our model. However, the point remains that the proposed SAF tax credit, when coupled with an elastic supply of waste oils and fats, creates a huge incentive for SAF expansion.

A more elastic waste feedstock supply lowers biofuel costs in the absence of the SAF tax credit. The wholesale price of gasoline (i.e., the price before adding excise taxes and retail markups) is $2.40 compared to $2.54 in our baseline. Adding the SAF tax credit increases the gasoline price to $2.69. When the tax credit is in place, the price of gasoline is similar across the two elasticity scenarios. The price pattern is broadly similar for diesel.

Because of the complexity of state and federal biofuel policy, different elasticity assumptions will change the magnitude of the price effects. However, the main point does not change: the proposed SAF tax credit would cause substitution from renewable diesel to SAF and increase consumer fuel prices.

This blog is co-authored with Kristen McCormack, William Scott, James Stock, Mengying Wu, and Jingran Zhang. It is also cross-posted at the Energy Institute blog.

In California, road construction, maintenance, and repairs are funded primarily through excise taxes on gasoline and diesel. The Governor’s latest budget includes a proposal that would diminish this dedicated source of revenue. It would allow fuel suppliers to reduce their diesel excise tax liability by selling sustainable aviation fuel (SAF).

Deciphering the implications of this proposal is tricky because SAF is closely related to renewable diesel (RD), the main biofuel substitute for petroleum diesel that now comprises 70% of diesel in California. Moreover, the proposed tax credit would exist in a complex web of regulations and tax credits at the Federal level, in California, and in other states.

The Newsom Administration estimates the SAF tax credit would reduce diesel excise tax receipts by about 10% ($165m). We have analyzed this proposal in the broader context of national and state biofuel policies.

Based on our analysis, we project three outcomes.

1. The tax credit will incentivize much more SAF than projected by the administration.

2. That increase in SAF comes at a cost: diesel excise tax receipts decrease by at least 20% and as much as 75% within a few years, and the prices of gasoline and diesel increase by 10-15 cents because of interactions with existing policies.

3. Because the SAF largely would come from diverting biofuels from surface transport to aviation, the net reductions in carbon emissions would be small and expensive.

   Subscribe now

Where would SAF come from?

The easiest and most immediate way to scale up SAF production involves using the hydroprocessed esters and fatty acids (HEFA) process to convert vegetable oils and waste fats into fuel. Refiners in the United States currently produce three billion gallons of RD per year using this process. At least six of these refineries are already equipped to produce SAF alongside RD if given a sufficient financial incentive, including the Phillips 66 refinery in Rodeo, CA (see the map below).

As we wrote last fall, current policies will yield very little SAF because it costs at least three times as much to produce as petroleum jet fuel. Although SAF currently earns substantial incentives, including credits in the federal renewable fuel standard, low carbon fuel standard (LCFS) credits from California, and federal tax credits, it does not receive enough to make up for its higher production cost relative to fossil jet fuel. At the same time, RD costs somewhat less to produce than SAF, benefits from a more favorable subsidy stack, and faces a smaller price gap with petroleum diesel, while also competing with SAF for feedstocks and production capacity.

SAF produced at a carbon intensity at least 50% below that of jet fuel would be eligible for California’s proposed tax credit, which is $1 per gallon of SAF, plus 2 cents per gallon for each additional percentage point of reduction beyond this 50% threshold, up to a maximum $2 per gallon. Based on CARB’s existing carbon intensity models, the tax credit would apply to SAF produced from waste oils and fats such as used cooking oil, tallow, and distiller’s corn oil. Crop-based oils such as soybean or canola oil would not qualify, which is a good thing given that increasing demand for these oils causes substantial carbon losses through tropical deforestation.

The other viable way to produce SAF is alcohol-to-jet (ATJ), which involves refining ethanol or isobutanol made from corn or sugarcane into jet fuel. Although cost estimates vary, this production process is widely understood to be more expensive and nascent relative to HEFA SAF. LanzaJet owns the only ATJ plant currently operating in the United States. In addition, CARB’s existing carbon intensity models imply that ATJ would face an additional hurdle to claiming the tax credit because it would likely need to integrate carbon capture and storage (CCS) to meet the 50% reduction threshold.

If the six current HEFA refineries that are equipped for SAF were to produce at maximum capacity, they would crank out 834 million gallons per year. At $1.25 per gallon, that is $1.04b in tax credits, or 65% of diesel excise tax receipts. Adding ATJ SAF would generate even more tax credits, as would more RD facilities retooling to produce SAF.

What is the likely effect of the California SAF tax credit?

This question cannot be answered without accounting for all the complex interactions created by existing fuel policies. In a recent working paper, we developed a detailed model to do just that. Last week, we added the proposed CA tax credit to the model.

We consider two policy scenarios, reflecting different potential costs of producing ATJ. The lower-cost scenario uses estimated production costs provided by industry analysts, and it implies that ATJ with CCS is competitive with HEFA for SAF production. However, given that ATJ technology has yet to prove itself viable at scale, we also present a higher-cost scenario in which the marginal cost of ATJ production is 25% higher. We view the higher-cost scenario as a conservative projection for 2030.

Under the lower-cost scenario, we find an increase in both HEFA SAF and ATJ SAF, totaling close to 1 billion gallons annually and split between the two technologies. Some, but not all, of the SAF production comes out of renewable diesel, which decreases by 700 million gallons. The higher-cost scenario generates a quarter of a billion gallons of SAF annually, almost all of which is from HEFA. We plot these outcomes in the chart below.

The two SAF technologies (HEFA with waste oils and ATJ with CCS) generate tax credits of about $1.25 per gallon. Thus, our two scenarios imply potential tax credits between $300m and $1.2b annually, or between 20% and 75% of diesel excise tax receipts.

How would the SAF proposal change fuel prices?

As a result of this policy, we estimate that both gasoline and diesel prices would rise in California. Gasoline prices would increase by 11-14c and diesel prices by 12c. Californians use about 13 billion gallons of gasoline and 4 billion gallons of diesel per year, so these price increases imply that annual costs to consumers would rise by between $1.9 and $2.3 billion.

Fuel prices rise because of complex interactions between the tax credit and existing policies. In particular, California’s low carbon fuel standard requires that every gallon of fossil gasoline and diesel be matched by some amount of cleaner (i.e., lower carbon intensity) fuel such as RD or SAF.

When California fuel suppliers reduce RD in favor of SAF, the lost RD has to be backfilled with fossil diesel to meet road transportation fuel demand. But, more fossil diesel means that more low-carbon fuel has to enter the system to meet the low carbon fuel standard goal. This is why the tax credits cause a smaller decline in RD than the increase in SAF, as shown above. The additional required low-carbon fuels raise costs.

How much emissions reduction and at what cost?

Carbon emissions decrease because the SAF is produced using relatively low-carbon processes compared to crop-based RD. In the lower-cost ATJ scenario, annual US carbon dioxide emissions fall by 2.1 million metric tons (MMT) relative to current policy, which is the equivalent of taking 1.5% of California’s gasoline-powered cars off the road. This net reduction, however, obscures substantial leakage: California emissions would decline by 3.6 MMT and emissions in the rest of the US would increase by 1.5 MMT as lower fuel prices in other states spur consumption there. The net reductions cost $997 per ton, which is about five times the social cost of carbon.

The higher-cost ATJ scenario sees smaller emissions reductions because less SAF enters the system. The emissions reduction is 0.5 MMT, or the equivalent of taking 0.4% of California’s gasoline-powered cars off the road. Again, there is substantial leakage, with California emissions declining by 2.0 MMT and rest-of-the-US emissions increasing by 1.5 MMT. The average abatement cost in this scenario is exorbitant at $2768 per ton.

Potential Complications

First, can all the tax credits be claimed? Firms that sell wholesale diesel pay the diesel excise tax. If those firms sell SAF, then they can claim credits against their excise tax liability (for example, Phillips 66 sells both SAF and diesel). If the firms selling SAF don’t have (much) excise tax liability, then they may not be able to claim all of the credit (e.g. Lanzajet). If enough money is on the line, we would expect firms to engage in some financial engineering to align tax liabilities with available credits, e.g., mergers or joint ventures.

Second, what about the state’s cap-and-invest program? As discussed in our recent paper, accounting methods and emissions covered by California’s LCFS and cap-and-invest program differ, so in some cases, the additional fossil diesel consumed in California might be offset by a decrease in emissions in non-transportation sectors. The magnitude of these effects is relatively small.

A Series of Unintended Consequences

Any expansion of biofuels into SAF will conflict with the regulations and tax credits that currently incentivize their use in surface transportation. These policy interactions make trying to do something that seems simple, like incentivizing SAF, hard to do without unintended consequences.

We predict that the proposed tax credit would increase SAF adoption more than expected and generate some additional emissions reductions, but at a high cost through lost road funding and higher fuel prices. If incentivizing SAF were to spur future cost reductions through new technologies, then it may be worth paying a high price, but there is limited evidence for significant cost declines in the biofuels sector so far.

Decarbonizing aviation will be challenging. Current batteries are too heavy and hydrogen takes up too much space, which leaves liquid fuels as the dominant energy source for the next few decades. However, promoting SAF only makes sense if it creates a clear path to future cost reductions and carefully manages the unintended consequences.

Update: I have two additional posts on this topic that respond to critiques. Read them below.

The scent comes first, when you sink your knife – or your teeth – into a feijoa’s skin. It is zingy, heady, a burst of bright perfumed flavour unlike any other. In New Zealand, the traditional method is to scoop out the creamy-clear insides with a teaspoon and discard the skins, though in other parts of the world people simply eat them whole. In the centre, the flesh is translucent and jelly-like where the tiny seeds hang in spiralled suspension. Closer to the skin, it is opaque and slightly gritty. Some have compared the taste to a mixture of pineapples and strawberries, but really, the flavour is something all its own. — Feijoa: A Story of Obsession and Belonging by Kate Evans.

When we redid our backyard in 2016, my wife suggested planting some pineapple guava trees. I agreed, and promptly forgot about the decision. They became part of the landscaping.

A couple of years later, on one of those early fall afternoons when the air suddenly feels cool, I caught a whiff of a familiar scent. I followed it to the pineapple guava trees and found the ground littered with oval green fruits, bruised from their fall.

“Those are feijoas!!”

Subscribe now

I had never thought deeply about feijoa. They were omnipresent in New Zealand, especially in the autumn when feijoas covered the ground in neighborhoods across the country. Yet I don’t recall ever seeing them for sale in the supermarket. They would show up occasionally as flavors in chocolate or beer (yuck), before disappearing just as quickly.

The major commercial fruit and vegetable crops in New Zealand are wine grapes, kiwifruit, and apples. Together these three crops occupy over 60,000 hectares. In 2024, there were just 116 hectares of commercial feijoa trees. That’s about 200 (American) football fields — so little you can barely see it on the graph below.

The feijoa originated in South America and spread around the world in the early 20th century. In Santa Barbara, California, a horticulturalist name Francesco Franceschi proclaimed feijoa the “coming fruit of the century”. Numerous plant breeders around the world attempted to develop it for commercial production.

However, it never became a major commercially produced fruit. Instead, the feijoa is a backyard staple in New Zealand, as well as in places like Davis, California. It’s an attractive evergreen tree that flowers in the spring and produces fruit in the fall.

Why did the kiwifruit take off while the feijoa did not?

Feijoa arrived in New Zealand in the 1920s around the same time as the kiwifruit, which was then known as the Chinese Gooseberry because of its origin and similar taste to the common gooseberry.

In the 1930s, a horticulturalist named Hayward Wright developed a variety of Chinese Gooseberry that produced large tasty fruit and had a long shelf life. Fruit packers started exporting it. In 1959, they renamed it the kiwifruit in honor of New Zealand’s national bird and it took off. If only they had trademarked the name!

The feijoa, by contrast, has poor shelf life, bruises easily, is difficult to propagate and pollinate, and there is no guaranteed way of telling if the fruit is ripe until you cut it open. These features do not make for a profitable crop.

I had none of these worries with the feijoas littering my yard. I picked up the best-looking 25 of them, scooped out the insides and used them to make jam. It was the first (and perhaps last) time I made jam, and it tasted great on my morning toast.

Our dog enjoyed eating many of the rest of our crop.

A recent book by Kate Evans provides a beautiful ode to the feijoa, covering its life, history and place in the culture. I’ll close with this sentence from her book:

In the United States, where feijoas are called pineapple guavas, a 1912 newspaper article declared, ‘he who drinks beer, thinks beer. But he who eats pineapple guava thinks of pineapple, raspberries, and banana, all at once.’

Merry Christmas and Happy Holidays!

The last 100 years have seen truly remarkable advancements in our agriculture and food system. The wonders of international trade bring fresh pineapples and grapes and corn and carrots to our grocery stores year round. Advances in plant breeding mean that supermarket brussels sprouts and apples taste so much better than they used to.

American farmers now produce almost five times the output they did in 1910, yet they use about the same quantity of inputs (similar amount of land, more machines, fewer workers). This increased productivity means that our food has gotten cheaper. In 1901, Americans spent 42.5% of their budget on food. Now, it’s only 10% — and that includes eating out.

Is the cheap-food era over?

My former UC Davis colleague (and current co-author) Rich Sexton just published an excellent book that sounds the alarm on the future of agriculture and food. The changing climate, growing population, rising incomes, biofuels, and environmental damage are making it harder to produce high-quality cheap food. Misguided policies are exacerbating these challenges.

Rich is an esteemed agricultural economist. He is a former editor of the flagship academic journal in the field (AJAE) and a former president of the premier professional association in the field (AAEA). He is a fellow of the AAEA and a multi-term chair of the UC Davis Agricultural and Resource Economics department. When he speaks, it is worth listening.

Subscribe now

The book begins by outlining the four factors that will drive demand for agricultural products in the coming decades.

1. Population. More people means more mouths to feed.

2. Income. As people rise out of poverty, they typically eat more meat. To produce a calorie of beef, farmers have to produce to 25 calories of food for the cow.

3. Eating habits. Our food preferences evolve over time, typically moving away from basic grains and towards things that require more resources to produce. An acre of corn or rice produces about 15 times the number of calories as an acre of broccoli or spinach.

4. Biofuel. About of quarter of the world’s corn is used to make ethanol for cars, and almost 10% of the world’s vegetable oil is used to make biomass-based diesel for trucks. Land used for biofuel is land not used for food.

Rich emphasizes the massive uncertainty in how these four demand factors will evolve, which is honest and refreshing. He walks through the arguments for and against large future increases in each factor and concludes that “forecasting models are likely underestimating demand for agricultural products in the twenty-first century.”

Of the four factors, my money is on income and biofuels being the two things that drive demand skyward. But, there are plenty of other arguments in the book to consider.

Can the food system meet future demand?

Now we get to the pessimistic part of the book.

Rich describes the many reasons to be worried about future food supply. For example, productivity growth appears to be slowing following a slowdown in public R&D spending, climate change causes more extreme heat that damages crops, and pushing people to buy local, organic, or GMO-free means less food produced with more resources. Many of the constraints on supply are driven by counterproductive government policies that seem to be expanding.

I think I’m a little more optimistic than Rich. I am optimistic that new technology will improve agricultural productivity and environmental outcomes. I am optimistic that economic forces will keep inefficient practices on the edges. I am optimistic that producers in the developing world will become massively more productive as they modernize……

OK, I feel seen. Optimism is worthless if no-one takes heed of the warnings and prioritizes solutions. Avoiding high food prices in the future requires much better policy and significant investment, as Rich describes in the book.

One thing I am not optimistic about is that we’ll stop using cropland to produce biofuels. Since that is the topic in the book I am closest to, perhaps I should take the lesson that my optimism in other areas is naive. Food for thought.

Buy Food Fight. Amazon can deliver it to your house tomorrow (or right now if you buy the kindle version). It’s a quick read, but densely packed with facts and thought-provoking insights, only a fraction of which I mentioned here.

Californians pay the highest gasoline prices in the United States, a fact that has grown more politically salient amid rising affordability concerns. On October 2 of 2025, Governor Newsom signed legislation that proponents claim will reduce gasoline prices by up to $0.20 per gallon and thereby save California drivers as much as $2.7 billion annually.

The new legislation allows gas stations to sell fuel that contains 15% ethanol and 85% refined petroleum, known in the industry as E15. At present, almost all gasoline in the United States is E10; it contains 10% ethanol and 90% petroleum.

Subscribe now

Why do we currently use E10 and not some other percentage?

The federal Renewable Fuel Standard mandates the use of ethanol. The reason we blend it at exactly 10% lies in a set of arcane air-quality rules. When gasoline evaporates, it releases smog-forming chemicals, and evaporation increases in summer heat. To limit this pollution, the EPA requires summer gasoline to be formulated so it evaporates less. However, a 1978 amendment to the Clean Air Act weakens this requirement for gasoline blended with exactly 10% ethanol in conventional gasoline markets (about half the US market). This effectively means gas stations can sell E10, but not E9, E11, or E15 in these markets.

Air quality requirements are not the only drag on E15. Many carmakers have approved it for use in their cars in the past decade, but some still recommend against it. EPA has approved it for sale in any 2001 or newer model of car, but prohibits it from use in motorcycles, boats, lawnmowers, snowmobiles and similar machines. Moreover, fuel pumps dispensing E15 are required to include a label stating these restrictions.

How would E15 be sold in California

California’s new legislation specifies that “blends of gasoline containing 10.5 percent to 15 percent ethanol by volume may be sold in the state“ until the state air resources board (CARB) completes an evaluation of these fuels. At that point, the state will either adopt a regulation establishing the specifications of these fuels or, if they are found to be incompatible with air quality regulations, cease the sale of them.

A study from scientists at UC Riverside found that increasing the ethanol blend rate to 15% would not exacerbate air pollution. These findings suggest it is likely that E15 will pass the evaluation.

CARB has two options to establish specifications for E15. First, it could redefine the specifications of standard California gasoline (known as CaRFG) to include E15. Under this option, gas stations could stock E15 in their regular pumps as long as they posted the EPA-mandated label. Switching to E15 everywhere would likely require oil refiners to coordinate to reconfigure their gasoline so that the blend meets technical and environmental specifications.

The second option is to develop a new specification for E15 and add it to the list of approved alternative fuels, a category that already includes hydrogen, natural gas, and E85. CARB staff indicated that they are leaning towards the second option. This option would require retailers to install pump infrastructure to dispense this alternate fuel in addition to standard gasoline.

E15 did not expand in other states when it was allowed

Classifying E15 as an alternative fuel has been tried at the federal level in conventional gasoline markets every summer since 2019. It is always allowed as an alternative fuel every winter in most of the country and throughout the year in reformulated gasoline markets, which are cities with high smog levels.

E15 is currently available in about 3,000 of the 145,000 gas stations in the United States, mainly in corn belt states. Gas stations have sold very little of it.The figure below shows that (i) the ethanol blend rate is the same in the winter when E15 is allowed as in the summer when it was not allowed, and (ii) the overall ethanol blend rate remains close to 10%.

E15 is not a lower cost fuel

The figure below shows the difference between the wholesale prices of ethanol and gasoline. It shows two lines, (i) the difference between the prices in cents per gallon (unadjusted) and (ii) the difference after accounting for the fact that gasoline contains 50% more energy per gallon than ethanol. This means that a gallon of E15 contains about 1.7% less energy than a gallon of E10. In road trials, cars running on E15 get 1% to 2% fewer miles per gallon than those burning E10.

The top graph shows that unadjusted ethanol prices have been below gasoline prices for much of the last 15 years. Yet, fuel suppliers did not voluntarily expand ethanol blends even in 2018 when ethanol was 70c cheaper than gasoline. The bottom energy-adjusted plot shows a major reason why. After adjusting for energy content, we see that ethanol is more expensive than gasoline for almost all of that time.

This analysis focuses only on ethanol’s energy content and ignores other key properties, especially octane. E10 became cost-effective partly because oil refiners could produce cheaper, lower-octane gasoline and rely on ethanol to raise the final fuel to required standards. Today, refiners blend 84-octane gasoline with 10% ethanol to make 87-octane fuel. If E15 became common, refiners could make 82-octane gasoline and still meet the 87-octane standard after blending with ethanol. However, it is unclear how much money refiners would actually save by switching from 84- to 82-octane gasoline.

Why is E15 priced lower than E10 in other states?

A recent study sponsored by the Renewable Fuels Association uses state-level data over a 20-year period to show that fuels with higher ethanol blends are priced lower than E10. This study has been used by ethanol advocates and California politicians to claim that allowing E15 will lower gas prices for Californians by about 20 cents per gallon. This same information can be seen in the map below from E15prices.com. It shows E15 prices 25 cents per gallon, or 8.2%, lower than E10.

The lower price of E15 does not reflect a lower cost of producing ethanol. In fact, even if the extra 5% (by volume) of ethanol in E15 were free to produce, the cost of production would drop by only 13 cents. The ethanol is not free, however. As shown above, it typically costs slightly more than gasoline to produce, after adjusting for the lower energy content.

Policy incentives help explain why E15 is slightly cheaper than E10, but not enough to justify a 25-cent discount. Under the federal Renewable Fuel Standard, each gallon of E15 earns about 5.7 cents more in net subsidies than E10, savings that can be passed on to consumers. In California, E15 would also avoid about 1.3 cents in cap-and-invest permit costs and earn less than one cent more in low carbon fuel standard credits. These state-level savings, however, would leave less cap-and-invest revenue to distribute and less subsidy for other clean fuel producers in the low carbon fuel standard.

So, if it is not supply, then it must be demand. When companies wish to sell a product that consumers are not enthusiastic about, then the price has to drop to clear the market. People don’t want to buy E15, and sellers appear to be taking a loss on the small amounts they sell, perhaps with the goal of expanding the market. If demand were to increase to the point that E15 became a viable fuel, then the price would rise.

Conclusion

Thanks to our work with the legislature, we have averted billions of dollars in higher costs at the pump by avoiding the kinds of severe gasoline price spikes we saw a few years ago.

— California Governor Gavin Newsom, October 2, 2025

The outcomes of California’s new E15 law will depend on how CARB chooses to regulate E15 under the new law. If it adds an alternative fuel specification for ethanol, as seems most likely, then I expect the results to be the same as in other states when the EPA allowed summer E15. Few gas stations will offer it and few drivers will buy it. Consumers don’t understand it and may be wary of it; gas stations will be averse to paying for the labeling and new fuel dispensers.

If CARB were to update the CaRFG specifications to include E15 – which it has indicated it will not do – then there would be more scope for consumer adoption because consumers would use it without knowing. Gas stations could provide it without installing new capital, but not at a substantially lower cost relative to E10.

Either way, the new law will not avert billions of dollars in higher costs at the pump.

This article is cross posted at the Energy Institute blog. I generated the two graphs using this R code.

Diesel made from vegetable oils has grown rapidly around the world as a substitute for petroleum. It used about 17% of global vegetable oil in 2023, up from 1% in 2000. In California, 70% of diesel is now bio-based, much of it produced from vegetable oils.

Biomass-based diesel costs about $2 per gallon more than petroleum diesel. Governments subsidize it on the belief that it has lower greenhouse gas emissions than petroleum. In a new research paper, Tzu-Hui Chen, Rich Sexton, and I show that biofuels derived from vegetable oils have likely increased, rather than reduced, global emissions.

Where is this vegetable oil coming from?

Most vegetable oils are extracted from soybeans, canola, sunflowers, or palm. We use them to fry food and as ingredients in baked goods and salad dressings.

Some of the vegetable oil used in biofuel comes from other users cutting back and some comes from increased production. The contribution of each source depends on elasticities, which are among the most useful numbers in economics. Elasticities tell us how willing consumers are to fry less food and how easily can farmers scale up production.

It really maters how much of the vegetable oil for biofuels comes from new production. Palm oil is the most widely used vegetable oil worldwide and is mostly produced in Indonesia and Malaysia. Expanding production in that region usually means clearing carbon-rich tropical forests. Converting forest to cropland causes the carbon in the forest to be lost to the atmosphere, offsetting the emissions benefit of reducing petroleum use.

Subscribe now

This figure below summarizes our paper. It is a simple story: more demand for vegetable oil pushes up the price, which incentivizes people to convert forests in Indonesia and Malaysia to palm plantations. Unlike most previous work on this topic, which is based on theoretical computer models, we analyze data on what has actually happened over the last couple of decades.

Let me walk you through the steps in our analysis.

First, this is not just a California or a United States thing. Canada’s clean fuel regulations are bringing vegetable oil diesel into the system. Biomass-based diesel (BBD) is also expanding quickly in Europe, South America, and parts of Asia.

Second, between 2002 and 2018 about 8.5% of the four major vegetable oils were used to make diesel. There were rapid increases between 2005 and 2010 and again in the last few years.

Third, and this is really important, the four major oils are pretty much interchangeable. Their prices move up and down together. This means that, if the US uses more soybean oil for biofuel (as we have done), then we will import palm and canola oils to use in the food system (as we have done.) Policy makers should treat the four oils as one product.

We estimate elasticities of demand and supply for vegetable oils. These elasticities imply that using 8.5% of vegetable oil for biofuel caused palm oil prices to increase by 20.8%.

Fourth, we use satellite data from 32,305 sites in Indonesia and Malaysia to measure deforestation. Each site measures 9km × 9km. We show that, when the price of palm goes up, more of these sites are converted from forest to palm. Specifically, the 20.8% increase in the palm oil price caused 1.7 million hectares to be converted from forest to palm between 2002 and 2018. This accounts for 20% of the deforestation that occurred in that region during that period.

Because we observe where the deforestation occurs, we can use existing carbon maps to calculate the amount of lost carbon from deforesting those 1.7 million hectares, accounting for above- and below-ground biomass carbon, peat decomposition, non-peat soil carbon, and the carbon sequestered by oil palm.

We find 1,075 Tg of CO₂ emissions from this deforestation, amortized over 25 years, which is comparable to adding about 9 million cars to the road for those 25 years.

To put this emissions number in context, we present it relative to the fuel energy produced from vegetable oil and compare it to the emissions from petroleum diesel.

Our study shows that each megajoule of energy from vegetable oil biofuel caused 83 grams of carbon dioxide emissions just from the induced deforestation in Indonesia and Malaysia. Add the emissions from cultivation, fertilizer use, processing, transport, and distribution and you get between 118 and 147 grams per megajoule. And we haven’t even considered emissions from land use change in other parts of the world.

For comparison, petroleum diesel emits 97 grams per megajoule over its complete life cycle. In short, biofuel from vegetable oils is worse than petroleum diesel for carbon emissions.

It is time to pump the brakes on biofuels from crop-based vegetable oils. The data show that these fuels make the climate worse, not better.

Citation and link to paper: Using Vegetable Oils for Biofuel Accelerates Tropical Deforestation and Increases Carbon Emissions. Chen, T. J., Sexton, R. J., & Smith, A. Working Paper. 2025.

Postscript: Two other stellar economists I have worked with are on the PhD job market this year. Shuo Yu is graduating from UC Berkeley and Mengying (Mandy) Wu is currently a postdoc at Harvard. If you’re hiring in agricultural or environmental economists, go look at their files right now. I’m sure you will be impressed.

The air is about 80% nitrogen, an element that promotes plant growth. You might think plants could absorb nitrogen from the air and grow like crazy. However, atmospheric nitrogen needs to be converted to a different form before plants can use it, and most plants can’t do the conversion.

Subscribe now

More than 100 years ago, German chemist Fritz Haber discovered how to do the conversion by reacting nitrogen with hydrogen under high heat and pressure. The company BASF bought Haber’s technology and assigned Carl Bosch to figure out how to scale it up, creating synthetic fertilizer. Both Haber and Bosch won Nobel Prizes for their work, which transformed agriculture and provided massive benefits to humanity.

Excess nitrogen fertilizer pollutes the water

But, there’s a dark side. Crops don’t use all of the nitrogen fertilizer that farmers spread on their fields. The unused nitrogen becomes a pollutant; it is either converted to a potent greenhouse gas (nitrous oxide) or washed into nearby waterways. Excessive nitrogen in water causes rapid algae growth, which reduces recreation value and can be toxic. When the algae die and decompose, they consume all the oxygen in the water, leaving nothing for fish.

Nitrogen pollution is a major global problem, especially in areas with intensive farming like the Mississippi River Basin. In the United States, only 32% of streams and rivers are in good condition for nitrogen. Near the mouth of the Mississippi River, a hypoxic “dead zone” the size of Connecticut has so little oxygen that almost nothing can live there.

In a recently published paper, my esteemed co-author Kostas Metaxoglou and I studied the data connecting fertilizer-intensive crops to water pollution. The US government measures nitrogen concentration in streams and rivers at thousands of locations in the corn belt. We linked those measurements to the crops that were grown upstream. Most previous studies rely on simulation models of plant growth and water flow to estimate this relationship, but we use actual observed data.

Our study finds 50 times less water pollution than commonly-used models

Here’s a map of the water monitors in waterways east of the 100th meridian, i.e., in places where farmers grow corn mostly without irrigation.

Water quality has not improved in the last 30 years (see below). Since 1990, nitrogen concentration has been about 2.4mg/L in the average river or stream. Despite a lot of attention, the problem isn’t getting better.

From each monitor, we follow all rivers and streams upstream for 50 miles. For example, the map below shows a monitor on the Mississippi River along the Iowa-Illinois border. The yellow triangle is the monitor. The little blue lines are all rivers and streams within 50 miles that flow towards the yellow triangle.

Our econometric model measures the additional nitrogen at the yellow triangle when another acre of corn is planted in the counties containing the little blue lines. We looked at other crops, but corn was the only one that showed a significant connection to water quality.

We find that the nitrogen load in small rivers and streams increases by about 0.4 lb per year for each additional acre of corn planted within 50 miles upstream. That number is 50 times smaller than what commonly-used mathematical simulation models say.

So, did we prove the models wrong? Is growing corn causing much less pollution than what we thought?

Sadly, the answer appears to be no. The models are good at calculating how much nitrogen a corn plant uses based on its growth, so they have a good idea of how much unused nitrogen is on corn fields. Our study shows that this excess does not enter rivers and streams within the same year.

So, what happens to the excess nitrogen?

A substantial portion gets stored in the subsurface soil and groundwater and will eventually find its way into rivers and streams. This is known as legacy nitrogen, and there are millions of tons of it (as illustrated below).

Kim Van Meter and co-authors looked at soil samples in Iowa and Illinois and found that 22-62lb of nitrogen per acre per year had accumulated below ground in recent decades. This number matches the amount of excess nitrogen on cornfields computed by the mathematical models. It also supports our finding that only a small fraction of that excess nitrogen reaches nearby rivers and streams in the same year. Most of it remains stored in the soil.

Potential options to deal with the nitrogen pollution problem include:

1. take land out of corn production

2. get farmers to use less fertilizer on corn

3. expand wetlands

We argue in our paper that option 1 is not a good idea for several reasons. The lost production value would exceed the value of avoided pollution damages, taking one corn field out of production would cause people to plant corn somewhere else thereby blunting the impact, and it does nothing to deal with the legacy nitrogen.

Option 2 sounds great in theory, but it’s hard in practice. Farmers don’t have the technology to apply nitrogen exactly where and when it’s needed and exactly in the right quantities. They spread it as precisely as they can and hope the plants find it and the rain doesn’t wash it away. I am optimistic that AI and robotics will help mitigate this challenge, as may new varieties of corn that convert their own nitrogen.

Economists sometimes suggest taxing fertilizer as a way to reduce use, but that’s difficult because we don’t want to tax all fertilizer (most of it is good), just the excess (which we don’t know). And, using less fertilizer does nothing to deal with the legacy nitrogen.

Option 3 has some promise. Establishing wetlands in streams and rivers can help mitigate both legacy and new surplus nitrogen by catching the nitrogen in waterways and using it for plant growth. In a recent study, Amy Hansen and co-authors find that “fluvial wetland restoration was most effective for achieving long-standing policy goals for sediment and nitrate reduction.”

In ongoing work, Kostas and I are using satellite data on land use to get more granular estimates of how agricultural activity affects nitrogen concentration. We hope this will lead to more precise policy and practice recommendations.

In a different study to the one I cited above, Van Meter and co-authors estimate that, even if surplus nitrogen on cropland were to go to zero and remain there, it would take about 30 years for nitrogen loads in the Mississippi River Basin to reach their 1970 levels. Millions of tons of nitrogen are waiting to pollute waterways, a legacy of corn belt agriculture.

Citation to our paper: “Agriculture’s Nitrogen Legacy” Metaxogolou, K. and Smith, A. Journal of Environmental Economics and Management, 130:103132. 2025.

It’s been a tough year in the corn belt. Farmers are facing crop prices too low to cover their costs. The Trump administration is planning to send them checks to soften the blows, but payments have been delayed by the government shutdown.

Last month China stopped buying US soybeans. Many in the media have drawn a straight line from this change to the low farm profits. But can we really blame China for the tough times soybean farmers are going through?

Here are 3.5 facts that imply the answer is no and 1.5 facts that suggest yes.

Subscribe now

Fact 1: Soybean prices have been declining for 3 years

US soybean production has been flat for a decade. Over the same period, Brazilian production increased by 83%. Brazil now produces 40% of world soybeans and 50% more than the US. This large supply increase means more soybeans for global markets to absorb, so prices have declined.

Low and declining prices started way before last month when China stopped buying US soybeans.

Fact 2: US soybean farmers attained record high yields this year

US soybean farmers are projected to produce 53.5 bushels per acre this year, which is the highest ever. Earlier this year, Brazilian farmers obtained their second highest ever yields. When yields increase, prices have to drop to convince people to buy the extra product. Major crops have inelastic demand, which means that prices typically drop by more than production increases, i.e., farmers often make less money in years they produce a big crop.

Fact 3: Countries other than China also import soybeans

Countries other than China import 40% of the world’s soybeans. The US exports 25% of the world’s soybeans. If China won’t buy US soybeans, then the US can sell to other countries.

Opposing Fact: Brazil prices are now higher than US prices

The price of soybeans at US Gulf export terminals is typically about 8% higher than the price at the port of Paranaguá in Brazil. This difference reflects the lower cost of shipping to China from the Gulf. However, you may have noticed in the first price graph that Brazil prices have increased in 2025, whereas US prices have stayed flat. The US price is now about 5% below the Brazil price. This fact suggests that lack of demand from China suppressed US prices by about 13% through September.

The price gap should return to normal once markets figure out where else to send US soybeans, just as it did in the 2018 trade war. Brazil prices will not stay above US prices.

Bonus mixed fact: Soybean prices mostly did not react to tariff news, except for today

After Trump won the presidential election last year, I noted that “corn and soybean prices have barely moved, implying that agricultural commodity traders see little change for the major agricultural crops.” In April this year after Trump announced the liberation day tariffs, I wrote an article titled “Soybean Markets Yawn at the Tariff War” about the lack of soybean price reaction to what were historically high tariffs. In September, prices dropped about 3% as it became clear China was not buying this year’s crop, but prices reversed those losses in the first part of October.

However, this morning’s markets seem to have noticed the trade war. News reports over the weekend suggested that US and China officials had reached (another) trade agreement that will make soybean farmers “feel very good about what’s going on both for this season and the coming seasons for several years”. Look closely at the plot above and you’ll see prices are up 2.5% today. They are now back to where they were in June and February.

The three facts I presented above paint a big picture that lack of demand from China is not to blame for the struggles of soybean farmers. Nonetheless, the last two facts show that trade negotiations can cause price fluctuations, at least in the short run.

If I were in the soybean business, I would be much more worried about Brazil’s massively increasing supply than China’s demand.

I made the graphs with this R code.

Singapore Airlines offers the longest current commercial passenger flight. It travels from Singapore to New York and takes 18 hours and 40 minutes. Mercifully, they don’t offer economy seats — it’s premium economy or business only! During the flight, the plane emits about 2.7 tons of CO₂ per passenger, or about the same as driving a standard gasoline-fueled car from San Francisco to Boston and back.

Subscribe now

Batteries and hydrogen potentially could dramatically reduce CO₂ emissions from aviation, but their weight and/or volume make them infeasible at present. Current batteries are 50 times heavier than jet fuel per unit of energy, and they take up 20 times as much space. Compressed hydrogen takes up 6 times as much space as jet fuel. The fuel in a typical long-haul jet (including the Singapore-NYC plane) comprises up to 40% of its weight on takeoff. This leaves little capacity for batteries or hydrogen.

The low energy density of batteries and hydrogen implies that near-term decarbonization of long-haul jets relies on sustainable aviation fuel (SAF). SAFs are liquid aviation fuels derived from feedstocks such as waste oils and fats, crops such as corn, soy, and sugarcane, cellulosic biomass, municipal solid waste, or captured CO₂. SAFs are chemically very similar to fossil jet fuel. They are currently approved to be mixed with conventional jet fuel at levels up to 50%, making them compatible with existing aircraft and requiring no modifications to current fueling infrastructure.

A Grand Challenge

In 2011, the US Federal Aviation Administration set a goal for US airlines to use one billion gallons of SAF annually by 2018. Actual SAF production fell far short of this target; in 2018 only 1.8 million gallons were produced.

In 2021, the US Federal government launched the SAF Grand Challenge, which set a target of 3 billion gallons annual production by 2030, which would result in SAF comprising about 12% of US jet fuel consumption in that year. At the beginning of 2024, US annual production capacity for SAF was approximately 30.66 million gallons, representing just 0.1% of the current US jet fuel supply. In late 2024, two additional plants came online, generating a significant increase in capacity to about 450 million gallons per year.

What would it take to reach the SAF Grand Challenge goal?

In an NBER working paper out today, Mengying Wu, Kristen McCormack, Will Scott, Jingran Zhang, Jim Stock, and I answer that question.

In the paper, we develop a detailed model of the road and air transportation fuel sectors, including the 12 most common biofuel-feedstock combinations. We use this model to evaluate six alternative policy scenarios designed to achieve the 3 billion gallon SAF target and discuss the estimated emissions reductions, cost effectiveness, fuel-feedstock mix, and cost incidence of each policy.

Two SAF production technologies are currently operational in the US, both based on biofuels: (i) hydroprocessed esters and fatty acids (HEFA) using vegetable oils and waste fats, and (ii) alcohol-to-jet (ATJ) using ethanol made from corn or some other bio feedstock. The three most common road transportation biofuels are ethanol, biodiesel and renewable diesel. Biodiesel and renewable diesel are produced from the same feedstocks as HEFA SAF.

Because currently available SAFs are predominantly biofuels, SAF policy will interact with the tangled web of existing policies that promote biofuels for surface transportation. In the United States, these policies include the Renewable Fuel Standard (RFS), which mandates blending biofuels into the surface transportation fuel supply, federal tax credits (45Z), state-level Low Carbon Fuel Standards (LCFS), and state tax credits. These policies provide subsidies for producers and consumers of biofuels. Depending on the policy, the cost of these subsidies fall on drivers, surface freight shippers, or taxpayers (illustrated below).

Any proposed SAF policy must be developed and assessed in the context of existing biofuel policies. The mix of policies in place help determine the total quantity, as well as the composition by technology and feedstock, of biofuel produced and consumed for surface and aviation fuel.

We have five main findings

First, current policies are insufficient to achieve the SAF production target. Using SAF generates credits in the RFS and LCFS programs, but the value of these credits is insufficient to bring much SAF into the market. In our model, only 60 million gallons of SAF are produced in 2030 under current policy, and a large amount of biodiesel feedstock that could be used for SAF is converted instead to less-expensive renewable diesel, which is used to replace petroleum diesel to satisfy the California LCFS.

In short, the current US policy mix incentivizes the use of biofuels for surface transportation over aviation (as illustrated below).

Second, we find that three alternative policy options to reach the target would all result in similar total emissions and overall economic costs. But they differ in who would bear those costs. The three are

1. a regulatory change to the RFS that creates a SAF requirement, but does not obligate petroleum jet fuel producers to buy renewable fuel credits (we call this Nested D2)

2. a regulatory change to the RFS that creates a SAF requirement and obligates petroleum jet fuel producers to buy renewable fuel credits (we call this D2 with Aviation Obligation)

3. a clean aviation standard in addition to current policy (we call this Aviation Intensity Standard)

The average cost of reducing carbon emissions with any of these three policies is around $470 per ton of CO₂, which is more than twice the social cost of carbon. These three policy tools would result in slightly higher average economic costs and lower total emissions than the current policy mix (see figure below).

Our cost estimate is based on current models of greenhouse gas emissions from biofuels. These models adjust for the fact that increasing demand for biofuels induces more land to be converted into farmland, which can release large amounts of carbon stored in plants and soil. However, there’s ongoing debate about whether these adjustments are large enough. Such emissions are especially high when forests are cleared in places like Brazil for soybeans or Indonesia for palm oil. Accounting for emissions from higher induced land use would move the dots in the above figure up and to the left.

Our third finding is that replacing all current federal and state biofuel policies with a $15 per metric ton carbon tax on fossil transportation fuels and a $4.41 per gallon SAF tax credit could reduce emissions, reach the 3 billion gallon target, and substantially reduce total policy costs, all relative to the current policy mix. Almost all reductions are a result of reduced demand for fuels.

Fourth, additional policy costs of reaching the target are disproportionately borne by California, which experiences gasoline and diesel price increases of 30-40 cents per gallon under the three policies mentioned above. This occurs because increased demand for SAF outside of California, encouraged by state tax incentives that currently exist in Hawaii, Illinois, Iowa, Minnesota, Nebraska, and Washington, increases competition for bio-based diesel feedstocks and therefore makes compliance with California’s LCFS more expensive.

Finally, it follows from the first two findings that adding policies to achieve the target is more costly than current policy. The static costs reported in this paper, however, do not account for any technological innovation induced by a durable SAF policy that would spur investment; costs arguably could be driven down further by extensive federal R&D support for advanced low-carbon SAFs, such as SAF produced from captured CO₂, which has low carbon emissions if produced using clean electricity.

Conclusion

Open up an in-flight magazine or read press releases from airlines and you will see an industry that is conscious of the high greenhouse gas emissions from flying. Many companies, including Singapore Airlines, have experimented with adding SAF to their fuel supply. Such voluntary initiatives could potentially support some SAF, but volumes would likely remain small, in large part because SAFs are significantly more expensive than fossil jet fuel and somewhat more expensive than biofuels for surface transportation. For SAF to expand, it will need long-term policy support. Our findings show that such policies would mostly divert biofuels from surface transportation to aviation at somewhat higher cost.

Considering potential innovation is important, particularly because many SAF technologies are still relatively nascent. In surface transportation, recent research indicates that learning-by-doing that drives down costs has been substantial for ethanol but not for bio-based diesel. If using biofuel SAFs can accelerate the technological transition to deep decarbonization (e.g., batteries, hydrogen, or SAF from captured CO₂), and if the induced land-use change effects of biofuels are not too large, then incentivizing biofuel SAFs could perhaps be justified.

This article is cross posted at the Energy Institute blog.

Recently, I wrote about America’s weird government-run sugar industry, which saddles us with really high sugar prices. This explains why American food manufacturers use a lot of high fructose corn syrup as a sweetener instead of sugar.

What if this is good?

High quantities of sugar are bad for human health. Too much sugar leads to metabolic syndrome, obesity, type 2 diabetes, and other health challenges. When we have too much of a good, the standard economic remedy is to tax it. The Federal sugar program is a big sugar tax.

Before endorsing such a tax, we need to know how much sugar is too much.

When I go on a long bike ride, I mix powdered glucose and fructose into my water bottle (in a 2:1 ratio) and supplement with sugary gels. I typically consume about 1200 calories of sugar in a three-hour ride, which replaces less than half the calories I burn.

Professional endurance athletes take this to another level. During Stage 19 of last month’s Tour de France, pro cyclist Ben Healy burned 5065 calories (5031kJ) over 5 hours and 22 minutes. He probably took in about 2600 calories of sugars during the race and a bunch more immediately after the finish.

Athletes like Ben Healy are not consuming too much sugar; it helps give them energy during the race and aids post-race recovery. However, an average person would be in very poor health if they consumed thousands of calories of sugar every day. The American Heart Association recommends that a typical man should consume no more than 150 calories of added sugar per day, which is about half what the average person consumes.

So, how much sugar is too much depends on a person’s activity level.

Subscribe now

It also depends on who bears the burden of eating too much sugar. In a free society, people get to consume things that make them happy, as long as they aren’t affecting other people.

There are two broad ways that excess consumption of added sugar affects other people. First, healthcare costs are borne by taxpayers (for government-funded care) and people who pay health-insurance premiums (for privately-funded care). Various studies estimate that drinking an ounce of a sugary drink imposes about a cent of costs on others through increased health-system costs from future disease.

The second way that excess sugars may affect “another person” is to make life worse for your future self. In a series of papers, Hunt Allcott, Benjamin Lockwood, and Dmitry Taubinsky call this an internality. An internality occurs when you do something today that you regret in the future. It is not an internality if you make a calculated decision to drink a Coke today knowing the potential future health consequences are worth it. It is an internality if you drink a Coke because you lack the self control to make the best long-term decision for yourself.

To measure the internality of drinking sugary beverages, Allcott and co-authors study the food purchase decisions of about 20,000 consumers in the Neilsen Homescan dataset. They supplement the data with a survey in which they asked the same people about (i) their nutrition knowledge, (ii) whether they think they drink more sugary beverages than they should, and (iii) how much they like sugary drinks.

Using an econometric model, the authors estimate that the average U.S. household would consume about 31% fewer sugary drinks “if they had the nutrition knowledge of dietitians and nutritionists and no self-control problems”. Then, based on a demand elasticity, they estimate that reducing consumption by 31% would require an a tax of about a cent per ounce.

So, externalities and internalities imply suggest a two cent per ounce tax on sugary drinks — or 24 cents on a can of Coke.

Lots of places have tried this, and the results have been surprisingly good. My colleague Sofia Villas-Boas has studied the effects of these taxes. In one study, Sofia and her co-authors looked at sugary drink taxes in five US cities. These taxes averaged 1.3 cents per ounce, which caused a 33% price increase. Purchases declined by a massive 33%, as shown below.

So, it seems sugary drink taxes may work well in improving health and welfare on average. They will be too harsh for endurance athletes and too mild for soda addicts, but about right on average. Does this mean that our wacky Federal sugar policy has been helping us all along?

Answer: not really.

The Federal sugar program effectively places a 100% tax on wholesale sugar prices, which sounds like a lot. But, the cost of sugar makes up a tiny percentage of the price of a sugary drink, so raising the cost of sugar by 100% doesn’t do much to the price of the drink in the store.

Specifically, the sugar program raises wholesale sugar prices by 20c per pound of sugar, which translates to about 1.7 cents per can of Coke (or 0.14c per ounce of Coke). This is about one tenth of the sugary drink taxes that have been implemented, so we would expect one tenth of the effect. That is, the sugar program may have reduced sugary drink consumption by the relatively small amount of 3.3%.

To look at it from a different angle, we saw above that externality and internality cost of consuming a can of Coke is about 24 cents. The world price of the sugar in that can is 1.7 cents. Put these numbers together and the appropriate sugar tax to account for the external and internal costs is 24/1.7 = 1400%. People love sugar so much that it needs a 1400% tax to compensate for the damage it causes, and yet we would still use a lot of it even with the tax.

Addendum: Here’s a NYT article about how much Ben Healy eats during the Tour de France: 6,252 calories and 1,245 grams of carbs: What it takes to fuel a winning ride at the Tour de France

Last week, President Trump announced that he had called the Coca Cola company to ask them to use cane sugar instead of high fructose corn syrup (HFCS) in their flagship beverage. He claimed they had agreed to do so, and Coca Cola later announced that it would add a cane-sugar version this fall. However, this reads like a pretty weak promise, given that you can already buy Coke made with sugar in the US; we call it “Mexican Coke” and it comes in a glass bottle.

Subscribe now

The president is right that Coke made with sugar is better. I know this because my daughter and I did a blind taste test, the results of which I posted in Ag Data News in 2021. Here’s an excerpt:

My daughter and I tasted four cola products from Coke and Pepsi. From each brand, we tasted one cola sweetened with sugar, one sweetened with HCFS, and two sweetened with aspartame. Both Coke and Pepsi have "diet" product and a "zero" product. It seems each developed their zero product in an attempt to improve their diet recipe. We also tasted a cola made by Empire and sweetened with sugar.

Our favorite was Coke sweetened with sugar, also known as Mexican Coke as it is typically imported from Mexico. I have always thought I liked it better. Apparently, I really do. My wife, whose palate is more refined than mine, observed that sugar seemed to mask some tangy flavors that HCFS did not. I also liked sugar better than HCFS in Pepsi, but my daughter disagreed.

We both liked Coke better than Pepsi.

By the way, we have done a few other taste tests, including (i) Which Chicken Sandwich is the Best? (ii) Which Milk Substitute is the Best?, and (iii) Which Fries are the Best?

The president may be right about the taste, but he is taking the wrong approach to engineering a sugar switch.

If sugar tastes better, why is American coke made with HFCS?

Answer: Because government policies make sugar prices twice as high as the rest of the world. If I were to describe these policies without context, you would think I was describing the Soviet Union in the 1960s; it is very much not free market capitalism.

Here’s how US sugar policy works.

1. To keep the price high, USDA guarantees a minimum price to sugar processors, which in turn pay high prices to farmers.

2. To prevent over-supply at these high guaranteed prices, USDA assigns to each sugar processor the maximum amount they are allowed to produce. It chooses these amounts to total 85% of predicted demand.

3. To prevent cheap imports from flooding the market, the US imposes a tariff rate quota, which permits a small quantity to be imported duty free and then imposes a large tariff on imports above the quota. Each country gets its own quota.

The sugar program has taken this form essentially since 1981. Coke and Pepsi switched from sugar to HCFS around the same time. HFCS consumption grew rapidly until 1999, after which it slowly declined on a per capita basis.

In 2024, the world price of sugar was 21c per pound, and the US price was 38c per pound. The world price is the price of raw cane sugar for delivery to a port within the country of origin of the sugar, sometimes known as No.11 Futures. This US price is the price of raw cane sugar at one of five US refinery ports, known as No.16 Futures. (I have no idea why there are no contracts numbered 1-10 or 12-15.)

For the last 20 years, US sugar producers have been unable to reach their allotted volume of 85% of consumption. The government is forcing consumers to pay double the price that consumers in the rest of the world pay, yet farmers still cannot produce their quota. Each year since 2007, the USDA has told each sugar processor how much domestic sugar it is allowed to process, and then later in the year it had to allow more imports to make up for the domestic shortfall.

This pattern suggests that sugar producers are not making exorbitant profits. If they were, then more farmers would enter the market. Instead, they are wasting resources by producing high-cost sugar.

Where is sugar produced?

Most sugar is made from sugar cane grown in the tropics, with Brazil and India the world’s largest producers. Sugar beets provide 10% of the world’s sugar and are grown mostly in Europe and North America. The US produces less than 3% of world sugar, according to UN FAO data.

US production is split evenly between beets and cane. Specifically, USDA requires that 54.35% of the domestic production allotment be distributed among beet processors and 45.65% be distributed among the sugarcane processors. Specific allotments are distributed to each processor according to formulas prescribed by the Agricultural Adjustment Act of 1938. I am not kidding — here are the official 2024 and 2025 allocations.

In 1960, Hawaii produced half of US domestic cane sugar. Its production declined steadily to zero until 2016 when the Maui Sugar Mill closed. It is much cheaper to produce sugar in places other than beautiful islands in the middle of the Pacific.

What about corn farmers?

Corn is America’s biggest crop, and it gets plenty of government support. About a third of US corn is used to make ethanol thanks to a federal policy that requires corn ethanol to be blended into our gasoline. Most of the rest of US corn is used to feed animals.

HFCS uses 3% of the corn crop. In my 2017 paper with Colin Carter and Gordon Rausser, we estimate that a 13% increase in corn demand from the ethanol sector caused a 30% price increase. This estimate implies that losing the HCFS market would reduce corn prices by 3*30/13 = 7%.

What about health?

There is little evidence that HFCS is worse for health than sugar. The HFCS used in soft drinks is 55% fructose and 45% sucrose; sugar is 50% fructose and 50% sucrose. So, despite what the MAHA movement claims, switching from HCFS to sugar is not a boon for health.

However, excess consumption caloric sweeteners is a leading contributor to diabetes, cardiovascular diseases, certain types of cancers, and overweight and obesity. These diseases impose costs on society that could justify a tax on sugar - perhaps even the 100% tax the US government imposes through its communist sugar policy. (Unlike a typical tax, the proceeds of the “sugar tax” go to sugar producers rather than to the government.)

Or maybe GLP-1 drugs will make the health issues moot.

Derek Thompson

If GLP-1 Drugs Are Good For Everything, Should We All Be on Them?

Several years ago, scientists took a close look at GLP-1 drugs, such as Ozempic, and learned that they were good at helping people lose weight. Then they took an even closer look and learned the drugs are also good at just about everything else…

Read more

a year ago · 234 likes · 49 comments · Derek Thompson

Would the price of a Coke go down if they used imported sugar?

Domestic sugar and HFCS are 20c per pound more expensive than imported sugar (assuming no tariffs). A can of coke contains 39g of sugar, so the cost of producing a can of Coke would go down by 1.7c if the sugar program went away. It’s not nothing.

Don’t hold your breath

In 1982, the New York Times wrote an article titled “A Shameful Sugar Policy”. Since that time, myriad agricultural economists have pilloried the program, all to no avail. It was expanded in the OBBBA.

I’m pessimistic that we’ll see the sugar program disappear any time soon, but eliminating it would be a surefire way to make sugar cheaper, which would cause soft drink companies to use it instead of HFCS.

I made the graphs in this article using this R code.

My family and I recently spent a day on the island of Lanai in Hawaii, also known as Pineapple Island. James Dole bought the island in 1922 and used it to build the world’s largest pineapple plantation. Lanai is now 98% owned by Larry Ellison, and no longer grows pineapples. Ellison has invested half a billion dollars (more than it cost him to buy the island) trying to grow leafy greens and tomatoes in greenhouses, so far unsuccessfully.

So, Lanai is no longer Pineapple Island. What about the rest of the state?

Hawaii agriculture is synonymous with pineapples and sugar. Sugarcane plantations boomed in the mid 1800s to supply northern US states that were cut off from the south during the Civil War. James Dole arrived in the islands in 1899 and started the Hawaiian Pineapple Company, which at its peak produced 75% of the world's pineapples.

In the last decade, media has reported the disappearance of these two commodities from Hawaiian agriculture. Dole and Del Monte moved most of their pineapple production out of Hawaii in the 1980s. The last sugar mill in Hawaii closed at the end of 2016.

Subscribe now

Overall, the real value of Hawaiian farm products has dropped in half in the last 40 years as the state has shifted its emphasis towards tourism and agricultural producers have moved to lower cost places. In contrast, real farm receipts increased in California and in the US as a whole over the same period.

So, agriculture has declined, and sugar production has disappeared completely, but does Hawaii still produce pineapples?

According to USDA, pineapples were not one of the top 20 commodities produced in the state in 2022. The state farm profile from the 2022 census does not mention pineapples.

In 2023, USDA's Hawaii tropical fruits report listed the value of Hawaiian production for 32 tropical fruits, ranging from a high of $1.8m (lychee) to low of $5000 (white sapote). But, no mention of pineapples.

Yet, when you visit the islands, as I was fortunate to do last week, you'll find tasty local pineapples.

According to United Nations FAO data, the United States produced $144m worth of pineapples in 2023. This is less than 1% of world production, but still large relative to the other commodities produced in Hawaii. The world's biggest pineapple producer is Costa Rica, followed by China and Nigeria. In 1961, the US produced more than a quarter of the world's pineapples, worth about $740m in 2025 dollars.

Essentially all US production occurs in the state of Hawaii; there is a tiny amount of pineapple production in Puerto Rico and American Samoa. FAO estimates that the $144m in 2023 production came from 170,000 metric tonnes of pineapples produced on 13,800 acres.

According to the 2023 Ag Census, the top crops in Hawaii by acreage were macadamia nuts (16,900) and coffee (9,400), but they don't mention pineapples.

In 2015 and 2020, the Hawaii Department of Agriculture constructed a detailed crop map for Hawaii. They estimated pineapple acreage at 4,396. They label small pineapple plantings that are mixed within smaller diversified farm operations as Diversified Crops, so actual pineapple plantings are higher than 4,396.

Their findings imply that pineapple acreage is less than half of the FAO number, but still more than other prominent crops such as banana and papaya.

Most of the pineapples are grown on Oahu.

So, in spite of what USDA data imply, the pineapple remains one of the larger crops in Hawaii, even though it has receded from its heyday. And just as well because I swear the local pineapples taste better, although that might be the vacation talking.

This article is an update of one I published in 2022. I made the plots in the article with this R code.

Related: Do They Still Grow Macadamia Nuts in Hawaii?

Addendum: In this year’s Farm Robotics Challenge, organized by UC Agriculture and Natural Resources and the AI Institute for Next Generation Food Systems, a team from the University of Hawaii at Manoa earned second place for their robot and drone system, which “identifies pineapple blooms and reconstructs their positions to optimize harvesting operations and improve labor efficiency.” You can watch a video presentation of their project here.