The ever-brilliant Cory Doctorow calls our attention to the fact that, in the US of A, three companies control the market for school lunch payments, and they hit poor families with junk fees that go up to 60% of what families pay.
Now, some of you might have seen newer economics papers refer to something called “mode of provisioning”. The concept is not very intuitive, but – I believe – the story of school lunch payments is great to explain it.
Some of these papers come up with dramatic – and often very encouraging – results. As a recent example, consider How much growth is required to achieve good lives for all? Insights from needs-based analysis by Jason Hickel and Dylan Sullivan
https://doi.org/10.1016/j.wdp.2024.100612
This paper comes up with a terrific punchline: you could provide everyone on the planet with “decent living standard” (I will define that in a minute) with as little as 30% of the energy and materials that we consume now. This is surprising, especially given that now we are nowhere near those levels of global welfare: Hickel and Sullivan cite an upcoming paper by Hoffman et al. that find over 95% of residents of low- and middle-income countries are deprived over at least one dimension of the decent living standard! So how is it that we could technically vastly improve our welfare with one third of the resources we are spending now? There must be a gigantic source of inefficiency lying around, but what is it?
To answer this question, I need to backtrack a little and introduce needs-based analysis. This approach grew out of the dissatisfaction with the dominant method for measuring poverty, associated with the World Bank. This consists of choosing a “poverty line”, which is a level of per capita income. People who earn less than that level are labeled as poor. The World Bank defines extreme poverty as the conditions of the people whose income is lower than 1.9 dollars a day, measured in purchasing power parity (PPP).
PPPs were invented to normalize monetary measures to the different cost of living in different countries. Intuitively, a salary that barely allows a single individual to survive in Tokyo might support comfortably a family of four in Bucharest. But PPPs come with all sorts of statistical pitfalls. The one that interests us here is this: they are calculated on the basis of a basket of goods that attempts to account for a broad range of goods and services, including those that are irrelevant to measuring poverty – like jewelry, luxury restaurants, commercial airfare. In this situation, if the prices of essential goods rise faster than the growth of PPP income, poverty might increase.
To overcome this problem, needs-based analysis takes a different approach to measuring poverty. It starts by defining poverty as the inability to procure a chosen basket of physical goods and services, not income: for example, Allen’s basic needs poverty line is defined as 2,100 calories per day, plus certain quantities of protein, fat and mineral, plus some clothing and heating, plus 3 square meters of housing. One person is poor according to this definition if their income does not permit them to purchase this basket in full. To calculate how many people live in poverty in a given country, economists can now compare household income data with the price of this basket.
https://doi.org/10.1146/annurev-economics-091819-014652
This is where the mode of provision comes in. Consider, for example, China. Extreme poverty – as measured by the share of the population which did not have access to Allen’s basic needs basket – was, in 1981, 5.6%, much lower than in countries like Brazil or Venezuela – which stood at around 30%, despite their per capita incomes being around five times that of China. China’s low rate of extreme poverty was due to socialist policies that ensured basic necessities like food and housing were widespread and affordable. The capitalist reforms of the 1990s, while contributing to developing China’s manufacturing base, coincided with a large increase rate in the rate of extreme poverty, which peaked at 68% (!) before finally returning, in the 2010s, to the level of the 1980s. The paper by Hickel and Sullivan mentions other cases in which extreme poverty increased even as PPP per capita income was growing: Brazil, Indonesia, Kyrgyzstan.
The paper then goes on to discuss the relationship between growth and poverty, when poverty is measured in PPP income. I, instead, want to reflect on the “socialist policies” mentioned by the authors in passing. What might they be? 1980s China was a planned economy. One can easily imagine meetings of the Central Committee of the Chinese Communist Party deliberating to allocate resources like construction materials and workforce to construct X buildings in province Y, or to direct their vehicle factories to build more tractors and fewer cars for private transportation so that agricultural output would increase. But it is just as easy to imagine policies that obtain similar results by deploying appropriate incentives, such as taxes and subsidies, while leaving production decisions to private businesses. For example, the Italian government decided that its post-COVID recovery stimulus would target retrofitting the housing stock for better insulation, in the interest of reducing household consumption of fossil fuels. Households could claim a tax credit of up to 110% of the money spent on insulation. This process was no shining example and had plenty of problems, but it did lead to a robust rise in overall employment, even in presence of a limited economy-wide growth. This means that the (fully capitalist) Italian economy redirected part of its capacity to retrofitting the country’s housing stock for better insulation. In the bargain, it also got more jobs (construction is labour-intensive) and less precarity.
Having established that “socialist policies” can be enacted also in non-socialist economies (as long as they have robust governance of the economy), it’s time to go back to the mode of provisioning, and to those American companies taking a 60% cut on the lunch money of poor schoolchildren.
Suppose you are the minister of education, and your problem is to make sure that all children eat healthy food at school. You already made sure that school cafeterias serve healthy menus. However, there’s a problem: some of those children come from poor families, and those families might be tempted to reduce their expenses by giving children, instead, cheap processed food containing only empty calories. You have an idea: you could give each poor child an electronic wallet, containing electronic cash financed by your ministry, which can be only spent at the school’s cafeterias. Now that you have come up with that, you need to choose a way to make this service available: a “mode of provision”.
One solution is direct provisioning. Your ministry takes it on itself to produce the simplest electronic wallet possible, ideally starting from an existing open source solution that you would then contribute to maintaining. This solution, like any other, will still need to cover its running costs by taking a cut from the transactions involved; but, given economies of scale, you can imagine these costs to be of the order of magnitude of 1% of the resources your ministry credits to these wallets.
Another solution is market provisioning. Your ministry allocates the money for school lunches, but it leaves it to fintech companies to provide wallets, and cover their own costs, plus profit, by taking a cut. At this point, neoclassical economics usually invokes the Coase theorem to conclude that, in a world of perfect information and perfect competition, the costs of making something in house are the same as those of buying it on the market. But we do not live in that world: we live in the world of Myschoolbucks, Schoolcafé, and Linq Connect. In this world, a transaction comes down at 8% for the relatively affluent parents who are paying for school lunches out of pocket, and at up to 60% for the poorer ones, for which it is your ministry, hence the taxpayer, who is footing the bill.
So, with direct provisioning, buying one euro of school lunches will cost you, say EUR 1.01. With market provisioning, it will cost you up to EUR 2.50 (coincidentally, these numbers are not so far from those reported in Hickel and Sullivan’s paper). Where does the extra EUR 1.49 go? If we imagine both solutions use the same tech, which costs the same EUR 0.01, it goes into shareholder profit, where it is, presumably, used to finance the consumption of things like luxury cars, commercial air travel and so on. That consumption implies that, with market provisioning, school lunches need a lot more energy and materials to land on children’s plates; those extra energy and materials are needed to manufacture and deliver those luxury consumption goods. Without the incentives provided by the latter, private sector companies will have no reason to provide children with electronic wallets with which to pay for school lunches.
This explains the impressive efficiency gains associated to changes in the mode of provisioning estimated in papers by New Economics authors such as Millward, Hickel, Steinberger and others. It gives me a lot of hope, because it means we can massively increase our ability to satisfy human needs while staying within planetary boundaries at the relatively modest price of curbing the luxury consumption of the 1%.
Photo credit: Sustainable Economies Law Center on flickr.com, CC-BY