Category: Data Source

GAVI’s Ill-Advised Venture Into African Industrial Policy

GAVI, the Vaccine Alliance has saved millions of lives by delivering vaccines to the world’s poorest children at remarkably low cost. It’s frankly grotesque that RFK Jr. cites “safety” as a reason to cut funding—when the result of such cuts will be more children dying from preventable diseases. Own it.

You can find plenty of RFK Jr. criticism elsewhere, however, and GAVI is not above criticism. Thus, precisely because GAVI’s mission is important, I want to focus on a GAVI project that I think is ill-motivated and ill-advised, GAVI’s African Vaccine Manufacturing Accelerator (AVMA).

The motivation behind the AVMA is to “accelerate the expansion of commercially viable vaccine manufacturing in Africa” to overcome “vaccine inequity” as illustrated during the COVID crisis. The problem with this motivation is that most of Africa’s delay in receiving COVID vaccines was driven by funding issues and demand rather than supply. Working with Michael Kremer and others, I spent a lot of time encouraging countries to order vaccines and order early not just to save lives but to save GDP. We were advisors to the World Bank and encouraged them to offer loans but even after the World Bank offered billions in loans there was reluctance to spend big sums. There were supply shortages in 2021 in Africa, as there were elsewhere, but these quickly gave way to demand issues. Doshi et al. (2024) offer an accurate summary:

Several reasons likely account for low coverage with COVID-19 vaccines, including limited political commitment, logistical challenges, low perceived risk of COVID-19 illness, and variation in vaccine confidence and demand (3). Country immunization program capacity varies widely across the African Region. Challenges include weak public health infrastructure, limited number of trained personnel, and lack of sustainable funding to implement vaccination programs, exacerbated by competing priorities, including other disease outbreaks and endemic diseases as well as economic and political instability.

Thus, lack of domestic vaccine production wasn’t the real problem—remember, most developed countries had little or no domestic production either but they did get vaccines relatively quickly. The second flaw in the rationale for the AVMA is its pan-African framing. Africa is a continent, not a country. Why would manufacturing capacity in Senegal serve Kenya better than production in India or Belgium? There’s a peculiar assumption of pan-African solidarity, as if African countries operate with shared interests that go beyond those observed in other countries that share a continent.

Both problems with the rationale for AVMA are illustrated by South Africa’s Aspen pharmaceuticals. Aspen made a deal to manufacture the J&J vaccine in South Africa but then exported doses to Europe. After outrage ensued it was agreed that 90% of the doses would be kept in Africa but Aspen didn’t receive a single order from an African government. Not one.

Now to the more difficult issue of capacity. Africa produces less than .1% of the world’s vaccines today. The African Union has what it acknowledges is an “ambitious goal” to produce over 60 percent of the vaccines needed for Africa’s population locally by 2040. To evaluate the plausibility of this goal do note that this would require multiple Serum‑of‑India‑sized plants.

More generally, vaccines are complex products requiring big up-front investments and long lead times:

Vaccine manufacturing is one of the most demanding in industry. First, it requires setting up production facilities, and acquiring equipment, raw materials, and intellectual property rights. Then, the manufacturer will implement robust manufacturing processes and manage products portfolio during the life cycle. Therefore, manufacturers should dispose of an experienced workforce. Manufacturing a vaccine is costly and takes seven years on average. For instance, it took about 5–10 years to India, China, and Brazil to establish a fully integrated vaccine facility. A longer establishment time can be expected for African countries lacking dedicated expertise and finance. Manufacturing a vaccine can costs several dozens to hundreds of million USD in capital invested depending on the vaccine type and disease indication.

All countries in Africa rank low on the economic complexity index, a measure of whether a country can produce sophisticated and complex products (based on the diversity and complexity of their export basket). But let us suppose that domestic production is stood up. We must still ask, at what price? If domestic manufacturing ends up being more expensive than buying abroad (as GAVI acknowledges is a possibility even with GAVI’s subsidies), will African countries buy “locally” and pay more or will solidarity go out the window?

Finally, even if complex vaccines are produced at a competitive price, we still haven’t solved the demand problem. GAVI again has a rather strange acknowledgment of this issue:

Secondly, adequate country demand is another critical enabler. For AVMA to be successful, African countries will need to buy the vaccines once they appear on the Gavi menu. The Secretariat is committed to ongoing work with the AU and Member States on demand solidarity under Pillar 3 of Gavi’s Manufacturing Strategy.

So to address vaccine inequity, GAVI is investing in local production….but the need to manufacture “demand solidarity” among African governments reveals both the flaw in the premise and the weakness of the plan.

Keep in mind that the WHO only recognizes South Africa and Egypt as capable of regulating the domestic production of vaccines (and Nigeria as capable of regulating vaccine imports). In other words, most African governments do not have regulatory systems capable of evaluating vaccine imports let alone domestic production.

GAVI wants to sell the AVMA as if were an AMC (Advance Market Commitment) but it isn’t. It’s industrial policy. An AMC would offer volume‑and‑price guarantees open to any manufacturer in the world. An AMC with local production constraints is a weighted down AMC, less likely to succeed.

None of this is to imply that GAVI has no role to play. In addition to a true AMC, GAVI could arrange contracts to pay existing global suppliers to maintain idle capacity that can pivot to African‑priority antigens within 100 days. GAVI could possibly also help with regulatory convergence. There is an African Medicines Agency which aims to operate like the EMA but it has only just begun. If the AMA can be geared up, it might speed up vaccine approval through mutual recognition pacts.

The bottom line is that the $1.2 billion committed to AVMA would likely better more lives if it was directed toward GAVI’s traditional strengths in pooled procurement and distribution, mechanisms that have proven successful over the past two decades. Instead, AVMA drags GAVI into African industrial policy. A poor gamble.

Markets, Culture, and Cooperation in 1850-1920 U.S.

From a very recent working paper draft by Max Posch and Itzchak Tzachi Raz:

We study how rising market integration shaped cooperative culture and behavior in the 1850–1920 United States. Leveraging plausibly exogenous changes in county-level market access driven by rail-road expansion and population growth, we show that increased market access fostered universalism, tolerance, and generalized trust—traits supporting cooperation with strangers—and shifted coopera-tion away from kin-based ties toward more generalized forms. Individual-level analyses of migrantsreveal rapid cultural adaptation after moving to more market-integrated places, especially among those exposed to commerce. These effects are unlikely to be explained by changes in population diversity,economic development, access to information, or legal institutions.

Here is the link.

Emotions and Policy Views

I would call this a story of negative emotional contagion:

This paper investigates the growing role of emotions in shaping policy views. Analyzing social citizens’ media postings and political party messaging over a large variety of policy issues from 2013 to 2024, we document a sharp rise in negative emotions, particularly anger. Content generating anger drives significantly more engagement. We then conduct two nationwide online experiments in the U.S, exposing participants to video treatments that induce positive or negative emotions to measure their causal effects on policy views. The results show that negative emotions increase support for protectionism, restrictive immigration policies, redistribution, and climate policies but do not reinforce populist attitudes. In contrast, positive emotions have little effect on policy preferences but reduce populist inclinations. Finally, distinguishing between fear and anger, we find that anger exerts a much stronger influence on citizens’ policy views, in line with its growing presence in the political rhetoric.

That is from a new paper by , and .

The Impact of Dating Apps on Young Adults: Evidence From Tinder

Online dating apps have transformed the dating market, yet their broader effects remain unclear. We study Tinder’s impact on college students using its initial marketing focus on Greek organizations for identification. We show that the full-scale launch of Tinder led to a sharp, persistent increase in sexual activity, but with little corresponding impact on the formation of long-term relationships or relationship quality. Dating outcome inequality, especially among men, rose, alongside rates of sexual assault and STDs. However, despite these changes, Tinder’s introduction did not worsen students’ mental health, on average, and may have even led to improvements for female students.

That is from a new paper by Berkeren Büyükeren, Alexey Makarin, and Heyu Xiong.

New York facts of the day

It’s truly astonishing how fiscally irresponsible New York is. The state budget proposal calls for $254 billion in spending, which is 8.3 percent higher than last year. That comes despite New York’s population having peaked in 2020. It’s a spending increase far in excess of the rate of inflation to provide government services for fewer people.

Ditch compares the New York state budget to the Florida state budget, a sensible comparison since both are big states with major urban and rural areas and high levels of demographic and economic diversity. He finds:

  • New York’s spending per capita was 30 percent higher than Florida’s in 2000. It was 133 percent higher last year.
  • New York’s Medicaid spending per capita was 112 percent higher than Florida’s in 2000. It was 208 percent higher last year. Florida has not expanded Medicaid under Obamacare, while New York has expanded it more aggressively than any other state. “For perspective, in 2024 New York spent nearly as much per capita on Medicaid ($4,551) as Florida did for its entire state budget ($5,076).”
  • New York’s education spending per student is highest in the country, at about $35,000. Florida spends about $13,000 per student. Florida fourth-graders rank third in the country in reading and fourth in math. New York fourth-graders rank 36th and 46th.
  • Florida has surpassed New York in population and continues to boom.

Here is more from Dominic Pino.

A consumption basket approach to measuring AI progress

Many AI evaluations go out of their way to find hard problems.  That makes sense because you can track progress over time, and furthermore many of the world’s important problems are hard problems, such as building out advances in the biosciences.  One common approach, for instance, is to track the performance of current AI models on say International Math Olympiad problems.

I am all for those efforts, and I do not wish to cut back on them.

Still, they introduce biases in our estimates of progress. Many of those measures show that the AIs still are not solving most of the core problems, and sometimes they are not coming close.

In contrast, actual human users typically deploy AIs to help them with relatively easy problems.  They use AIs for (standard) legal advice, to help with the homework, to plot travel plans, to help modify a recipe, as a therapist or advisor, and so on.  You could say that is the actual consumption basket for LLM use, circa 2025.

It would be interesting to chart the rate of LLM progress, weighted by how people actually use them.  The simplest form of weighting would be “time spent with the LLM,” though probably a better form of weighting would be “willingness to pay for each LLM use.”

I strongly suspect we would find the following:

1. Progress over the last few years has been staggeringly high, much higher than is measured by many of the other evaluations  For everyday practical uses, current models are much better and more reliable and more versatile than what we had in late 2022, regardless of their defects in Math Olympiad problems.

2. Future progress will be much lower than expected.  A lot of the answers are so good already that they just can’t get that much better, or they will do so at a slow pace.  (If you do not think this is true now, it will be true very soon.  But in fact it is true now for the best models.)  For instance, once a correct answer has been generated, legal advice cannot improve very much, no matter how potent the LLM.

As in standard economics, consumption baskets change over time, and that can lead to different measures of progress (or in the economics context, different estimates of advances in living standards, depending on whether the ex ante or ex post bundle weights are used).  Researchers could attempt the more speculative endeavor of estimating how LLMs will be used five years from now in everyday life (which will differ from the status quo), and then track progress on that metric, using those value weights.  “How rapidly are we improving these systems on their future uses?”

This alternate consumption basket approach gives you a very different perspective on progress in AI.

Note also that the difference between the “Math Olympiad measurements of AI progress” and the “consumption basket measurements of AI progress” may iincrease over time, especiallly if the basket of everyday uses does not change radically.  The everyday uses will peak out near maximum levels of performance, but there will always be a new series of very hard problems to stump the AIs.  It will become increasingly unclear exactly how much AI progress we really are making.

Massive Rent-Seeking in India’s Government Job Examination System

In India, government jobs pay far more than equivalent jobs in the private sector–so much so that the entire labor market and educational system have become grossly distorted by rent seeking to obtain these jobs. Teachers in the public sector, for example, are paid at least five times more than in the private sector. It’s not just the salary. When accounting for lifetime tenure, generous perks, and potentially remunerative possibilities for corruption, a government job’s total value can be up to 10 times that of an equivalent private sector job. (See also here).

As a result, it’s not uncommon for thousands of people to apply for every government job–a ratio far higher than in the private sector In one famous example, 2.3 million people submitted applications for 368 “office boy” positions in Uttar Pradesh. 

The consequences of this intense competition for government jobs are severe. First, as Karthik Muralildharan argues, the Indian government can’t afford to pay for all the workers it needs. India has all the laws of say the United States but about 1/5 th the number of government workers per capita leading to low state capacity. But there is a second problem which may be even more serious. Competition to obtain government jobs wastes tremendous amounts of resources and distorts the labor and educational market.

If jobs were allocated randomly, applications would be like lottery tickets with few social costs. Government jobs, however, are often allocated by exam performance. Thus, obtaining a government job requires an “investment” in exam preparation. Many young people spend years out of the workforce studying for exams that, for nearly all of them, will yield nothing. In Tamil Nadu alone, between one to two million people apply annually for government jobs, but far less than 1% are hired. Despite the long odds, the rewards are so large that applicants leave the workforce to compete. Kunal Mangal estimates that around 80% of the unemployed in Tamil Nadu are studying for government exams.

Classical rent-seeking logic predicts full dissipation: if a prize is worth a certain amount, rational individuals will collectively spend resources up to that amount attempting to win it. When the prize is a government job, the ‘spending’ is not cash, but years of a young person’s productive life. Mangal calculates that the total opportunity cost (time out of the workforce) that job applicants “spend” in Tamil Nadu is worth more than the combined lifetime salaries of the available jobs (recall jobs are worth more than salaries so this is consistent with theory). Simply put, for every ₹100 the government spends on salaries, Indian society burns ₹168 in a collective effort of rent-seeking just to decide who gets them. The winners are happy but the loss to Indian society of unemployed young, educated workers who do nothing but study for government exams is in the billions. Indeed, India spends about 3.86% of GDP on state salaries (27% of state revenues times 14.3% of GDP). If we take Mangal’s numbers from Tamil Nadu, a conservative (multiplier of 1 instead of 1.68) back of the envelope number suggests that India could be wasting on the order of 1.4% of GDP annually on rent seeking. (Multiply 3.86% of GDP by 15 (30 years at 5% discount) to get lifetime value and take .025 as annual worker turnover.) Take this with a grain of salt but regardless the number is large.

India’s most educated young people—precisely those it needs in the workforce—are devoting years of their life cramming for government exams instead of working productively. These exams cultivate no real-world skills; they are pure sorting mechanisms, not tools of human capital development. But beyond the staggering economic waste, there is a deeper, more corrosive human cost. As Rajagopalan and I have argued, India suffers from premature imitation: In this case, India is producing Western-educated youth without the economic structure to employ them. In one survey, 88% of grade 12 students preferred a government job to a private sector job. But these jobs do not and cannot exist. The result is disillusioned cohorts trained to expect a middle-class, white-collar lifestyle, convinced that only a government job can deliver it. India is thus creating large numbers of educated young people who are inevitably disillusioned–that is not a sustainable equilibrium.

Mangal valiantly proposes redesigning the exams to reduce waste, but this skirts the core issue: India’s wildly skewed public wage structure. Government salaries far exceed what is justified by GDP per capita or job requirements, distorting education, employment, and unemployment throughout the entire economy in deeply wasteful ways. The only real solution is to bring public sector pay back in line with economic fundamentals.

The anti-alcohol campaign in the USSR

Although alcohol consumption remains high in many countries, causal evidence on its effects at the societal level is limited because sustained, society-wide reductions in alcohol consumption rarely occur. We take advantage of a country-wide 1985-1990 anti-alcohol campaign in the Soviet Union that resulted in immediate, substantial and sustained reductions in alcohol consumption. We exploit regional differences in precampaign alcohol related mortality in the Russian republic and show immediate declines in male and female adult mortality in urban and rural areas across the entire age distribution, which translate into a rise in life expectancy. The campaign led to a substantial decline in deaths that are both directly (alcohol poisoning, homicides and suicides) and indirectly linked to alcohol consumption (respiratory and infectious). We find a decline in infant mortality rates among boys and girls due to causes most affected by post-natal parental behavior (choking and respiratory). Finally, both divorce and fertility rates rose, while abortions and maternal mortality due to abortions declined. This study provides novel evidence that alcohol consumption not only directly affects the mortality of drinkers but can have spillover effects on family outcomes.

That is from a recent paper by Elizabeth Brainerd and Olga Malkova.

Some European countries have mastered a happiness trick?

Using Eurobarometer data for 21 Western European countries since 1973 we show the U-shape in life satisfaction by age, present for so long, has now vanished. In 13 northern European countries – Belgium, Denmark, Finland, Germany, Iceland, Ireland, Luxembourg, the Netherlands, Norway, Sweden, Switzerland, Turkey and the UK – the U-shape has been replaced by life satisfaction rising in age. We confirm these findings with evidence from the European Social Surveys, the Global Flourishing Survey and Global Minds. Evidence of change in the U-shape is mixed for Austria and France. In six southern European countries – Cyprus, Greece, Italy, Malta, Spain and Portugal – the U-shape was replaced by life satisfaction declining in age. In these southern European countries, life satisfaction of the young has been rising since around 2015. A contributory factor is the rapid decline in youth unemployment from its 2015 peak.

Here is the full NBER paper by David G. Blanchflower and Alex Bryson.

The objectivity of Community Notes?

We use crowd-sourced assessments from X’s Community Notes program to examine whether there are partisan differences in the sharing of misleading information. Unlike previous studies, misleadingness here is determined by agreement across a diverse community of platform users, rather than by fact-checkers. We find that 2.3 times more posts by Republicans are flagged as misleading compared to posts by Democrats. These results are not base rate artifacts, as we find no meaningful overrepresentation of Republicans among X users. Our findings provide strong evidence of a partisan asymmetry in misinformation sharing which cannot be attributed to political bias on the part of raters, and indicate that Republicans will be sanctioned more than Democrats even if platforms transition from professional fact-checking to Community Notes.

Here is the full paper.  I guess it agrees with Richard Hanania…

Are cultural products getting longer?

Ted Gioia argues that cultural products are getting longer:

Some video creators have already figured this out. That’s why the number of videos longer than 20 minutes uploaded on YouTube grew from 1.3 million to 8.5 million in just two years

Songs are also getting longer. The top ten hits on Billboard actually increased twenty seconds in duration last year. Five top ten hits ran for more than five minutes…

I’ve charted the duration of [Taylor] Swift’s studio albums over the last two decades, and it tells the same story. She has gradually learned that her audience prefers longer musical experiences…

I  calculated the average length of the current fiction bestsellers, and they are longer than in any of the previous measurement periods.

Movies are getting longer too.  Of course this is the exact opposite of what the “smart phones are ruining our brains” theorists have been telling us.  I think I would sooner say that the variance of our attention spans is going up?  In any case, here is part of Ted’s theory:

  1. The dopamine boosts from endlessly scrolling short videos eventually produce anhedonia—the complete absence of enjoyment in an experience supposedly pursued for pleasure. (I write about that here.) So even addicts grow dissatisfied with their addiction.
  2. More and more people are now rebelling against these manipulative digital interfaces. A sizable portion of the population simply refuses to become addicts. This has always been true with booze and drugs, and it’s now true with digital entertainment.
  3. Short form clickbait gets digested easily, and spreads quickly. But this doesn’t generate longterm loyalty. Short form is like a meme—spreading easily and then disappearing. Whereas long immersive experiences reach deeper into the hearts and souls of the audience. This creates a much stronger bond than any 15-second video or melody will ever match.

An important piece and useful corrective.

The Effects of Ranked Choice Voting on Substantive Representation

Ranked choice voting (RCV) is an increasingly popular electoral institution that has been posited by reformers and media outlets to produce transformative effects on electoral outcomes and representation. However, there is little social scientific evidence available that evaluates these claims. I test the effects of RCV on municipal fiscal outcomes and the ideological composition of city councils. I also estimate RCV’s effects on these outcomes relative to public opinion — in other words, whether RCV narrows the gap between outcomes and mass policy preferences. This article finds no empirical support for the proposition that RCV changed fiscal outcomes or the ideological composition of city councils — both on absolute terms and relative to mass opinion. Furthermore, the roll-call based ideal points of legislators serving before and after RCV did not change, and the relationship between city district opinion and city legislator ideology is unchanged post-adoption. Taken as a whole, this article does not find evidence that RCV has produced the types of transformative political effects that reformers have postulated.

Here is the full paper by Arjun Vishwanath.  Source.

How do declining fertility and climate change interact?

There are lots of assumptions behind these results, but still it is good to see someone working through some scenarios:

A smaller human population would emit less carbon, other things equal, but how large is the effect? Here we test the widely-shared view that an important benefit of the ongoing, global decline in fertility will be reductions in long-run temperatures. We contrast a baseline of global depopulation (the most likely future) with a counterfactual in which the world population continues to grow for two more centuries. Although the two population paths differ by billions of people in 2200, we find that the implied temperatures would differ by less than one tenth of a degree C—far too small to impact climate goals. Timing drives the result. Depopulation is coming within the 21st century, but not for decades. Fertility shifts take generations to meaningfully change population size, by which time per capita emissions are projected to have significantly declined, even under pessimistic policy assumptions. Meanwhile, a smaller population slows the non-rival innovation that powers improvements in long-run productivity and living standards, an effect we estimate to be quantitatively important. Once the possibility of large-scale net-negative emissions is accounted for, even the sign of the population-temperature link becomes ambiguous. Humans cause greenhouse gas emissions, but human depopulation, starting in a few decades, will not meet today’s climate challenges.

That is from a new NBER working paper by Mark Budolfson, Michael Geruso, Kevin J. Kuruc, Dean Spears & Sangita Vyas. You also can read this as an argument that declining fertility will not be a major problem for some long while.

The antitrust case against U.S. higher education

Thirty prestigious independent American institutions of higher education were at some time members of the 568 higher education group (often labeled a cartel). Seventeen of them were sued by the U.S. Government and representative students who alleged that their meetings and deliberations resulted in collusion that caused students to pay higher prices. Twelve of the seventeen institutions subsequently settled their cases and by 2024 collectively had paid $284 million to do so. However, an inspection of these institutions’ pricing reveals that the median 568 Group institution lowered its average real net annual cost to its undergraduate students by 19.07% between 2009 and 2022. Further, this reduction was 1.70 times larger than the average real price reduction granted during the same period by the median institution among a sample of 475 other accredited, non-profit, independent four-year institutions and 11.63 times larger than the median price reduction granted by 78 public flagship state universities. The 568 group’s real price reductions stretched across every one of the five household income categories commonly used by the Government. Thus, there is little empirical support for the allegations that the Government has levied against the representative 568 group institution, and thus multiple members of this group appear to have paid unmerited fines to the Government to settle claims against them.

That is from a new paper by James V. Koch.  Via the excellent Kevin Lewis.