Bad Milk Economics
Do free markets make milk unaffordable and diluted?
The subject of today’s post is a meme, shown below. Unserious as it is, it is popular, and the garden of industrial organization economics must be watered with the blood of bad takes like this one.
The actual trajectory of milk production in the US has been very good. In 1975, a gallon of milk was $1.57, and even a federal minimum wage earner would be making $2.10 an hour. Today, a gallon of milk is $4, and those earning the federal minimum (an increasingly tiny fraction of the labor force) make $7.25 an hour, meaning the cost of milk is lower today as a fraction of earnings, even for the lowest of earners. The median personal income went up by nearly 700% in nominal terms from 1975 to 2024. Much success! Lots of milk to go around.
In the US context, gallon and half-gallon canisters have remained standard for packaging and selling milk for a long time. Just look up milk on Walmart’s website, and you’ll find them. They haven’t shrunk the packaging; that was true even during the spike of inflation that occurred after the COVID-19 pandemic. Liters are used in the image, so maybe they’re referencing shrinkflation that occurred in the UK or some other country that uses the metric system, but the subject is “market competition in practice,” so I feel comfortable focusing on any form of it. The broader point they’re making is that greedy capitalists covertly raise the per-unit price, but obviously that’s not what we can observe.
Product adultery
Do milk producers dilute their product with water? Well, it’s illegal in the US and every developed country to do that, so no, they don’t. It’s not clear if they would do it if they were allowed to, since we don’t really have any useful counterfactual developed countries where the only force that might prevent milk dilution with water is market competition.
One user provided some helpful examples of a lack of regulation leading to milk adulteration, with the primary one being the swill milk scandal in 1850s New York. Milk producers, seeking cheap grain, used the byproducts of fermentation and distillation at distilleries (swill) to feed their cows, and the result was thousands of babies dying after consuming the milk. They also attempted to mask the adulteration by adding other products like water and eggs. A similar scandal occurred in China in 2008.
I am not convinced that these other situations analogize to the modern US, though I think it’s a possibility. In 1850s New York, milk wasn’t sold under big, recognizable brands like Horizon, Fairlife, or Walmart’s Great Value.1 Americans today also have access to more substitute products like almond and oat milk, and have become more literate and educated since the 1850s. (At the time, they were under the false impression that infants should be fed cow’s milk, hence the outbreak of infant deaths.) All of this is to say that it seems unlikely that an established brand like Horizon would deliberately adulterate their milk, since they have a reputation to uphold and face relatively smart consumers who can discipline them.
One could argue that the only reason why brands are able to function as indicators of quality is that the government protects consumers from fraudulent rebranding of milk. If it were legal to pass off any milk as Fairlife milk, it wouldn’t be a useful signal of quality. But I think this kind of policy should fall neatly under standard private property protections considered core to the functioning of a free market, so it’s fine to consider this sort of dynamic a form of free market success.2
Do Americans really react to poor food quality by punishing companies with fewer sales? I’m sure it depends on how severe the problem is, but to provide an example, Chipotle was hit with declining sales after an outbreak of E. coli at its restaurants in late 2015.
The company suffered from an immediate crash, losing about a third of its sales, and had not recovered fully from this outbreak even two years later. It’s easy for consumers to form associations between a brand and the bad outcomes customers experience, and that’s what allows the market to discipline companies and discourage them from selling bad products. Similarly, the 2008 Chinese milk scandal resulted in a consumer panic and a huge drop in sales.
We should think more clearly about how this works. Here’s a graph:
Companies will tend to invest in product safety so long as it’s profitable. Since there is some risk of sales issues if you don’t make sure your product is safe, investing in safety allows a company to make more money. It is costly to do this, and the marginal benefit of safety investment diminishes, so they invest a finite amount of money in it and tolerate some degree of risk. Governments can raise the marginal benefit of safety investment by introducing new risks like government lawsuits and criminal charges. That raises total investment in safety.
So, clearly there’s some reason for companies to invest in safety, even in lieu of government intervention. The question is whether the natural equilibrium is optimal for society. I think not: if consumers were fully informed, they would be more punishing towards companies that make these mistakes, and the marginal benefit of safety investment would be greater. Governments can fix this problem by introducing new risks associated with selling an unsafe product. All of this reasoning analogizes to adulterated products in general, since consumers likewise will tend to punish companies for covertly making their products lower-quality, such as by diluting milk with water.3
Is cheap milk just government intervention?
There’s another potential problem: the US government subsidizes the dairy industry! Case closed, pack it up. Declining milk prices aren’t a success of free markets. This is just government intervention in action.
I don’t think subsidies explain the decline in milk prices. For one, even in countries like Australia, where the dairy industry is unprotected, milk has become more affordable with time. More importantly, the subsidies just aren’t big enough to explain why milk prices have risen more slowly than inflation and incomes.
To put an upper bound on the effect of dairy subsidies on the price of a gallon of milk, we can assume that all of the subsidies pass through to consumers in the form of lower prices and that milk was unsubsidized in the 1970s. (Both of those statements are false!) There are two subsidy programs to focus on for our napkin math:
The Dairy Margin Coverage program: This program provides payments to enrolled farmers (which include most American dairy farms) when the difference between the national price of milk and the price of feed falls below certain levels. We can assume full pass-through on the $2.7 billion in payments made from 2019 to 2024 and assume they’re made all in the same year.
Dairy Revenue Protection: A subsidized insurance scheme. We can pretend this costs nothing and just say this is $1.26 billion a year, which it is not.
US milk production in 2024 was 226 billion pounds, or 26.28 billion gallons, and these typically retail for about $4 a gallon. A very high upper bound on the effect of these subsidies on the price is ($2.7 + $1.26)/26.28 = $0.15 per gallon of milk, suggesting the subsidy-free price would be $4.15. That’s still far cheaper than in the 1970s.
If you’re concerned that milk was unusually expensive in the 1970s due to a shortage, you’re kind of right, but the picture doesn’t change much. Below, you can see how milk prices have fallen relative to inflation. Milk was indeed unusually expensive come 1975, so perhaps it would have been better to use the 1970 price, but this was not deliberate on my part. (I have a list of prices for various products in 1975 and 2024 hanging behind my desk.)
So clearly, milk prices are not low because of government subsidies. What’s the real reason?
Milk appears to have become more affordable over time because the cost of producing milk has gone down, and that decline in costs has been passed through to consumers. One way we can see this change in the cost of production is through the rising quantity of milk produced per cow, shown in orange:
A nice side effect of this increase in efficiency is that there are fewer cows experiencing the evil manmade hellscape that is factory farming. We have become more morally efficient as well as economically efficient.
But all of this is irrelevant to the question of whether free market competition is occurring. Even a monopoly would seek profits by cutting costs, and would still pass through some of these reductions in costs in the form of lower prices (though not to the same degree as a competitive market). There are exceptions to this rule, but they’re rare. That brings us to our next section:
Is the dairy industry a competitive market or a cartel?
Whether the dairy industry counts as an example of free market competition is not easy to say. One reply suggested that milk companies engage in price leadership, a form of implicit collusion where… okay, it actually wasn’t clear whether or not this person understood what price leadership is or how any of this works, so let me give you exactly what they said:
Uh huh. Well, as we’ve established by this point, government subsidies are pretty small compared to the size of the American dairy industry, so I don’t think that causes milk overproduction. And if dairy companies were implicitly colluding, they would be underproducing milk in order to raise prices and profits. That’s how collusion works.
I tried to squeeze a more coherent picture out of this person, but when I kept pressing them on what they meant by industries setting an “acceptable price” rather than the competitive price, they just screamed at me:
We all have bad days sometimes. And in case you didn’t notice, this explanation does not add any new information. Of course they set prices to make money—but are they implicitly colluding by following a price leader and producing a smaller quantity? They insisted that these companies aren’t colluding, not even implicitly, but at the same time, they aren’t competing on price. I tried giving them an out by suggesting the possibility of imperfect, oligopolistic competition, but they insisted on this strange contradiction of terms.
This means that I am resigned to arguing and explaining on behalf of my opponents. Picking on weak arguments would only mislead you into thinking the dairy industry is more competitive than it really is.
Determining the degree of competition in any industry is difficult. To demonstrate this, imagine if all of the coffee shops in America were bought out by one company. Despite owning every shop, they continue to sell coffee at the competitive price, say $5 for a medium latte. They might do this because they recognize that barriers to entry into the business are low. On paper, they have a “monopoly position,” but it’s unprotected—if they started charging the most profitable price, they might attract new entrants who wish to undercut them and get some of those profits.
We also might have an apparently competitive industry that actually operates like a monopoly. If there are 10 companies producing coffee, but they all collude to set the monopoly price and share in the profits, the situation is worse than the “monopoly” I just described.
Could we determine whether there’s a problem by simply looking at the profit margin? Life isn’t that easy, either. Perfect competition does not eliminate the accounting profits we can observe by subtracting expenses from revenue. There are unobserved opportunity costs of working in one industry rather than another, so we’d expect some visible markup to remain in the long run.
Standard measures of revenues and costs also ignore the long-term returns of things like branding and research. Suppose you observe high profits at some company, with revenues at $200k and expenses at $100k for some year. Unobserved is the investment they made last year: $100k spent on surveying consumers and asking them for their thoughts on the company’s product, all to improve it and sell more later while advertising to them. They’re underwater for a moment, but the investment pays off the next year, and that turns out to look like “high profits.”
Nevertheless, we can get some semblance of an idea of the degree of competition by looking at how concentrated the industry is. A 2010 Congressional Research Service report describes the degree of concentration in the dairy industry in 2008. The top 10 dairy cooperatives produced 48% of all milk volume, while the top 50 produced 79%. The largest producer, the Dairy Farmers of America, clocked in at just 19.9%. DFA has grown to 30% market share, so there’s been some more consolidation since then, but presumably the overall picture looks similar: a big industry leader and lots of smaller companies. This is not a guarantee of competition, but it does tell us that the coordination costs required to engage in monopoly pricing are higher than they would be in a more concentrated industry.
One way to make coordination easier would be price leadership, mentioned earlier. The Dairy Farmers of America, seeking to engage in monopoly pricing, could pick the monopoly price and then hope everyone else follows along. (If they explicitly asked other producers to do this, they would be breaking the law and attract the attention of the FTC.) The problem with this strategy, like every other anticompetitive strategy, is that every member has an incentive to cheat. You only need one of the dozens of dairy producers to break the tacit agreement for the scheme to fall apart and profits to begin collapsing as the undercutting producer eats the others. The more extreme the price leadership, the greater the reward for defection as consumers switch to the undercutting brand.
Pricing in the dairy industry is complicated. Farmers, cooperatives, processors, retailers, and the federal government all play a role. Dairy farms are generally members of cooperatives like the Dairy Farmers of America, who sell their raw milk to processors, who then sell milk to retailers. Processors must pay at least the minimum price set by federal milk marketing orders (FMMOs), which is one potential coordination mechanism for engaging in monopoly pricing. If the federal government is setting a minimum price, you don’t have to worry about anyone defecting from the cartel and undercutting everyone else. The key question, then, is whether processors generally pay the minimum price or some price above it—that is, whether the price floor is binding.
This is very unclear to me. I can’t find any data describing whether the actual price paid is regularly above the minimum price or if it’s usually equal to it. The federal government constantly changes the minimums, so we can’t simply observe the price charged over time to check if it’s binding. (If it were fixed far above the competitive price, the price would be flat over time.) Regardless, if you want it, here it is below. Note that cwt stands for hundredweight, meaning 100 lbs, the standard unit of measurement for bulk milk sales.
There have been some cases of antitrust litigation against dairy companies in the US, like this one:
The lawsuit, filed in the United States District Court for the Northern District of California on Sept. 26, 2011, alleged that between 2003 and 2010, more than 500,000 cows were slaughtered prematurely under CWT’s dairy herd retirement program in a concerted effort to reduce the supply of milk and inflate its price nationally. According to the complaint, the increased price allowed CWT members to earn more than $9 billion in additional revenue.
Dairy industry revenues were $31.8 billion in 2007, and the $9 billion figure for added revenue was over the span of 7 years, so this doesn’t seem like a case of the price being far above the competitive level.
I can understand the concern with saying the dairy industry has market competition, but even an oligopolistic industry with occasional antitrust scandals has some degree of competition. More importantly, it doesn’t exhibit the rising prices and product adulteration described in the post, which is the main focus here; whether it really counts as market competition in the first place is just an interesting question to ruminate on, and more of a semantic one in the end.4
One way we can escape this conundrum is by simply focusing on milk alternatives that don’t face price floors. Oat and almond milk, unlike cow’s milk, do not face federal minimum prices at any stage of production, so they might give us an idea of whether market competition disciplines companies to sell at lower prices.
At Walmart, you can get oat milk at a price of about 6.5 cents per fluid ounce, compared to 2.9 cents per fluid ounce of cow’s milk. Almond milk is a bit cheaper, being about 4 cents per fluid ounce, comparable to some more premium cow’s milk brands. Price data is not as easy to come by, but this paper (PDF warning) provides a small window into pricing over time for a few alternative milks:
Of their own volition, an unregulated competitive alternative milk industry chose to cut prices on almond milk from 2014 to 2020, all while the general price level in the economy was rising. This measure isn’t biased by shrinkflation, since it keeps the unit of weight constant, but we might wonder if almond milk producers have increased the ratio of water to almonds. That would be visible in a changing nutrition facts label, perhaps not immediately obvious to consumers. (There’s also a much simpler method: regularly drink alternative milks. I had them on occasion over the period shown above, when I was a teenager, and didn’t notice any watering down.)
A 2012 blog post reports the nutrition facts for Almond Breeze sweetened vanilla almond milk: 90 calories, 2.5g of fat, 150mg of sodium, 60mg of potassium, 16g of carbs, 1g of dietary fiber, 15g of sugar, and 1g of protein. We can compare this to the current nutrition facts, which are identical with the following exceptions: it has 10 fewer calories, an additional 110mg of potassium, and 14g total carbs rather than 16g. It appears that Almond Breeze has simply dropped some of their sweetener without watering the product down, which would be visible in a fall in fat or protein content. We know this because there aren’t any significant added forms of fat or protein in the current ingredients list. Sunflower lecithin is included, which does have some fat, but only a negligible amount.
So, confusing competition issues aside, the original points of the meme about free enterprises covertly raising per-unit costs and watering their products down appear to be mostly made up when it comes to the milk industry. There have been product adulteration incidents in the past, but milk has become more affordable with time in the US, not less. Insofar as government intervention has influenced the price, it has potentially raised it through FMMOs rather than lowering it, and it is not clear whether this has really occurred through binding price floors for raw milk sold to processors.
There are a lot of other replies, some sillier than others, that need to be addressed before I can end the post.
Don’t we need to adjust 1975 incomes for inflation?
A user I won’t be linking (because they ultimately learned from our exchange and decided their initial impression was wrong) thought that my use of nominal income was incorrect, and we need to adjust for purchasing power first. Unfortunately, that is not correct. By comparing the incomes people were paid on paper at the time to the actual prices they paid, we’re getting exactly what we need: the fraction of their income they needed to spend on milk. We can then compare that to the fraction today. If we adjust their income for inflation without adjusting the prices they paid for inflation, we’re comparing super-high incomes they didn’t actually receive to the actual (lower) prices they paid, making milk seem more affordable than it really was.
100% of the food is diluted with soybean oil, canola oil, and Palm oil
I do not think the milk we buy has oil in it. That would taste gross. I would like to buy some oil-free peanut butter, though, so I understand how you feel. This just isn’t relevant to the topic of discussion (milk, famous for not containing palm oil).
Milk would be much more expensive in a free market
One worry I’ve discussed is that the milk industry isn’t really a free market, which is why milk is cheap. I think the information on subsidies I provided is enough to show that isn’t true. This is essentially the same claim, but focusing on the other direction, and I don’t think it’s true either. It’s worth discussing some more.
Australia’s dairy industry is unsubsidized and doesn’t have the kind of price floor intervention seen in the US. Despite that, it appears to likewise produce affordable milk. We could naively compare the price of milk in Australia as a fraction of the median income there to the same fraction in the US, but part of this difference would be explained by the difference in productivity between the two countries, rather than the difference in their milk industries in particular. We can attempt to adjust for this by multiplying the typical Australian income (in Australian dollars) by the ratio of median income (PPP) in the US to that of Australia. That ratio is about 1.24 for 2020, the latest year for which Our World in Data provides data for Australia. Using this, and price and income data from 2020, we get:
United States: ($3.32 per gallon)/$35850 = 0.00009260808
Australia: (~4.914 AUD per gallon)/(54890 AUD * 1.24) = 0.00007219718
This suggests that milk is about as affordable in the deregulated Australian market—slightly more affordable, in fact—as it is in the US. It remains about as affordable if you drop the productivity adjustment.
Actually, Australian agriculture is like Soviet command economics, since the government invests in agricultural infrastructure
Are Uber and Lyft Soviet-style state enterprises, since the government pays for the roads? This is just really silly. Certainly, the Australian government has some involvement, but they don’t set quotas or anything.
Whatever this is
A view into the mind of a normal person can be fascinating. It seems this person not only believes the Fed wants the cost of living to be higher, but that relative prices are determined by a mysterious, unidentified force related to the money supply, rather than supply and demand. Well done.
They sort of diluted milk by switching to Holsteins (no they didn’t)
This doesn’t explain the falling price. By 1970, Holstein cows had already taken over from Jersey cows, at 84% of the population. They were 85% of the population by 2014.
It appears that Jersey cow milk, which has higher fat content, retails for $5.19-ish, probably depending on the brand. That’s still more affordable than milk in the past, probably explained by rising per-cow milk production even among Jerseys, which produce less milk volume during their lifetime than Holstein-Friesians. Even if we shorten the time horizon and measure from 1995 to today, milk would need to retail for $6.67 a gallon to be as expensive today as it was in 1995 (i.e., take up the same fraction of the median personal income). If we use the overall CPI-U-RS price level instead of income, it would need to retail for $5.30 a gallon to have inflated as much as everything else, yet even Jersey milk is cheaper.
I think fewer people should get their ideas about how the world works from memes about capitalism. As I’ve written before, a lot of internet arguments are built on non-specificity, this is one of those cases, the other side has some good points, etc. etc. I’m tired of this. Go read Waldman and Jensen to do your IO homework, and complain about the housing market or something instead; getting mad about the dairy industry is stupid if you aren’t Canadian.
Or so it seems. It’s very hard to find information on this subject. You can go look it up yourself to see what I mean.
Exactly what counts as a “free market” is a little vague. But I think we should avoid defining them out of existence, much like how we ought to count representative democracies like the US as a form of democracy rather than treating republics as a distinct concept. When people talk about democratic nations, they almost always mean republics like the US, France, Germany, and others.
“Aha!” someone might say, “So you admit it! What’s even the point of this post if you’re going to turn around and admit that government intervention is necessary?”
Read more carefully. I’m including this footnote as a warning. I agree only that government intervention can improve the natural state of affairs, not that it is absolutely necessary for any pressure towards product quality to exist at all.
Like, if you have a case of duopoly where the two companies are charging above the competitive price but not charging the monopoly price, does that count as market competition? There’s certainly some “competition” occurring, since they aren’t maximizing profits by charging the monopoly price. If you have one hundred firms with differentiated products, and they each have a little bit of pricing power, does that count as market competition? Consumers do have a choice, after all.
So it seems like much of the question here is over where we draw the line. Certainly, we shouldn’t decide that every market that exhibits any degree of pricing power on the part of suppliers is “not really market competition,” or that any amount of government intervention at all renders a market not really free.
> One user provided some helpful examples of a lack of regulation leading to milk adulteration, with the primary one being the swill milk scandal in 1850s New York.
I suspect milk adulteration was a fair bit more common in the 19th century, as there’s even a rather famous Thoreau quote about milk being adulterated with water:
> Some circumstantial evidence is very strong, as when you find a trout in the milk.