January 6, 2024: Research and Development

Happy New Year. Today we will look at research and development in general, with a particular look at agriculture.

It is good to start with the big picture.

World GDP from AD 1 to 2015, from Our World in Data. GDP figures are estimated on a purchasing power parity basis and are adjusted for inflation. Data is drawn from the World Bank and the Maddison Project Database. Obviously, there is much room for dispute about GDP reconstructions for many centuries in the past, but there there is little reason to believe the long-tern economic growth picture differs substantially from what is shown here.

Why this fantastic growth has happened, and how to insure it continues into the future, is a broad question and subject for lively discussion, but it is clear that scientific advancement is a major reason for this. And consequently, governments invest in research and development across a wide range of areas with the goal of advancing scientific prowess and economic growth (among other goals). However, evaluating the effectiveness of R&D spending is tricky.

One place that we do have some precise figures is this paper from Mark Rosegrant and others, so we will discuss this paper in some detail. It was published just a few months ago, and it conducts a cost-benefit analysis of agricultural R&D in the Global South.

Before we get to the main results, let’s discuss the term for the region that the paper describes. In development literature, “Global South” is a term that describes Central and South America, Africa, the Middle East, most of Asia (but not Japan or South Korea), and much of Oceania, though not Australia and New Zealand. The Global North comprises all countries that are not part of the Global South, and much like that frustrating term “The West”, there are geographic incongruities. The North/South distinction is similar, but not exactly the same, to the now-archaic First and Third World distinction and the starting-to-become-archaic developed/developing country distinction. I don’t care much for the term, but it’s what we have.

Anyway, the paper models an increase in agricultural R&D in the Global South over the next 35 years. They find that the additional spending would cost $61 billion over that time frame and yield $2.1 trillion in benefit. All figures are net present value measured at an 8% discount rate. The model is based on what is called the IMPACT model developed by the International Food Policy Research Institute, and the amount of spending is chosen to reduce the global prevalence of hunger below 5% (last year, World Vision reported that this rate is just above 10%). In more tangible terms, the paper finds,

After 35 years, the increased funding is estimated to increase agricultural output by 10%, reduce the prevalence of hunger by 35%, reduce food prices by 16%, and increase per capita incomes by 4% relative to a counterfactual where funding continues to rise on historical trends.

The increase in agricultural output means that less land is needed to feed the population, which reduces deforestation. The amount of deforestation reduced is on the order of one million hectares per year. Using the ecosystem service valuation figures of Costanza et al., the paper estimates that the value of the reduced deforestation would be 20-40% of the total $2.1 trillion benefit. However, because the Costanza et al. figures are so unreliable, Rosegrant et al. don’t include them in the calculations. They also find that over 35 years, annual greenhouse gas savings would be about 21 million tons CO2, which is about 0.5% of world total emissions. It would also save on water use, nitrogen runoff, and phosphorous pollution.

These numbers look very good. I haven’t looked into IMPACT, though, so the calculations are rather opaque to me. Also, agricultural R&D is a very broad term. The paper envisions it to encompass the full gamut of R&D ventures, which includes breeding and genetic engineering, extension services, better machinery, breakthrough ideas such as synthetic meat, and more.

There are some other analyses that show favorable returns to agricultural R&D. Here’s a study that finds a 20:1 return in the United States. Here’s one that gives a social rate of return of 20-60%. That means that from a societal point of view, the discount rate would have to be as high as 20-60%—far higher than any plausible appropriate value—for agricultural R&D not to be a good investment. More analysis indicates that agricultural R&D would save on greenhouse gas emissions and land use. Despite all that, public agricultural R&D spending has been declining in the United States.

After all this, we might think that public spending on agricultural R&D (I strongly suspect that similar numbers can be derived for R&D in other areas) is a good investment, even accounting for the possibility that some numbers are mismeasured and that government spending tends to be inefficient. But here I want to put on the free market cap and think about it a bit more critically.

First off, R&D can be public or private. This review of agricultural R&D shows that from around 2008-2013, public spending in the United States decreased by about $2 billion. Over that same time, private spending increased by about the same amount. Could it be that public spending was crowding out private spending and thus not making much contribution to the total amount of R&D spending? That could be true, or it could be there is no causal relationship between these two trends despite occurring at the same time. It is also unclear to me what the difference is between public and private spending in terms of impact. A 2022 paper by Fuglie and Nelson suggests that there is a difference that is relevant to the impact on agricultural productivity. They imply that a decline in total factor productivity in agriculture of ~6% in the 2010s can be explained by falling public R&D. I’m not very convinced, though.

A paper that I have come back to many times, Are Ideas Getting Harder to Find? by Nicholas Bloom et al., is relevant here. Crop yields are one of the case studies they look at, and they document a fairly dramatic fall in gain in crop yields per dollar spent on R&D. It is not clear why this is happening. It could be that crop yields are approaching theoretical limits, and so there are inherent diminishing returns to research. It could be that we are seeing sharply diminishing returns as a result of the amount of spending. It could be that research focus is shifting toward other priorities, such as reducing greenhouse gas emissions. Or it could be all of these things.

The main rationale for public R&D spending is the public good argument: there are significant spillover benefits to R&D spending that are not captured by a private actor, and so there is a case for public spending, even if it means tolerating some of the inevitable inefficiency that comes with any government program. But I wonder if there is a better way. The idea of patenting goes back to 15th century Venice glass making, with some antecedents going back farther than that. A patent is a legal monopoly that is granted to the discoverer of a technology for a period of time, allowing the discoverer to profit from the invention and thus internalizing the benefit of discovery, incentivizing private research. It’s not a perfect system, but patents are a valuable tool in advancing technology. What other tools could incentivize private actors to invest in R&D at a socially optimal level? I would be curious to hear what readers have to think on this question.

With all this said, maybe the best policy is a certain level of public R&D spending on agriculture and other areas. It certainly isn’t the worst use of public money that I can think of.

Quick Hits

Brian Potter of Construction Physics has written a good history of United States Steel. They were once the largest company in the United States, but after decades of rent-seeking, overly bureaucratized management, and a lack of innovation, they are a shadow of their former selves and have been recently acquired by Nippon Steel, pending regulatory approval.

Peter Hague, in There Is No Business Case For Civilization, considers the folly of treating every social decision as though it were a VC pitch.

Mickey Mouse and Steamboat Willie have reached the end of their 95 year copyright protection and are in the public domain as of January 1, 2024. The short can be seen here. The copyright term was extended by the Copyright Term Extension Act, known derisively also as the Mickey Mouse Protection Act; Steamboat Willie would have been in the public domain in 2004 without that. Without that at the Copyright Act of 1976, Steamboat Willie should have been in the public domain in 1985.

Nigel Jaquiss has an interview with Rob Justus, a prolific developer of low-income housing in Portland, Oregon, who has recently stepped aside from the industry. Justus identifies—correctly in my opinion—an overly complex funding structure as a driver of low-income housing costs and a barrier to their construction.

Forest City, a new ecological city project in Malaysia that is part of China’s Belt and Road Initiative, is so far a ghost city. Scott Beyer examines why this has happened. Beyer is more bullish on privately run charter cities, as he describes in this profile of Próspera, a charter city in Honduras.

Jack Watling in Foreign Affairs analyzes the apparent (and what he believes is a mischaracterization) stalemate in the war in Ukraine. One notable takeaway from his piece is the argument that the United States and Western Europe has squandered the last two years in building a military-industrial capacity. As I’ve noted before, since the first Persian Gulf War, Americans have become accustomed to wars that are either short or fought at a low intensity, and thus they have not seen great urgency in maintaining a strong defense industry. The last two years in Eastern Europe should be a wake-up call that this approach is inadequate.