What I have been reading: mid-October update
Capitalism promotes positive-sum thinking and reciprocity. You can see this in the historical account of attitudes to interest rates. Zero-sum thinking prevailed under Malthusian conditions. This further solidifies the notion that low growth fosters zero-sum mindsets; corrosive to our institutions.
Tentative progress in anti-ageing research. Yes we need more trials to establish whether these results hold up for humans, yet I see no reason why the underlying mechanisms cannot apply to us. What are the odds that, by the end of this century, life expectancy in developed countries will exceed 100? More likely than not is my prior.
Derek Thompson on the decline of partying. Although parties and house gatherings are far from the only means to socialise, this fits in with the wider pattern of a general increase in loneliness throughout the same period. I do wonder to what extent the pandemic distorts this picture though. Also, if socialisation is a normal good, then the sluggish recovery from the 2008 financial crisis may also play a role. Much commentary has focused on social media as a substitute to in-person gatherings, however whilst social media perhaps diminishes the importance of focal points, it also reduces the search costs to meeting like-minded peers. This is consistent with a shift towards fragmented socialisation involving smaller gatherings, with the overall level of socialisation remaining the same. Nonetheless, given our looming mental health crisis, further inquiry into this topic is important. Note too that alcohol abuse amongst adolescents has fallen; an unequivocal positive, yet if alcohol reduces the transaction costs to socialising (in this case, acute social anxiety), which likely has an evolutionary explanation, then will a more sober world (all other things equal) be a lonelier one? This does not necessarily have to be the case (all socialisation arguments defending alcohol could equally have been made to smoking), yet perhaps changes in attitudes to socialisation lag changes in attitudes to alcohol? Again, an important hypothesis to resolve.
Mandating disclosure of clinical trial results may slow down the pace of innovation. I also propose an alternative mechanism for these results: using Aghion’s and Howitt’s model of creative destruction (for which they won the Nobel for this week), greater aggregate investment into research and development raises the probability that a firm's invention will be outcompeted by a superior substitute. Perhaps greater disclosure speeds up the perception of this process. Firms perceive more innovation overall, but this reduces expected profits, leading to a slower pace of investment here.
A good primer on the state of the literature so far regarding the effects of AI on employment practices. Still early stages, yet it seems that if there is any conclusion to draw so far, that AI may be reducing hiring for AI-exposed entry-level occupations, yet this is far from inevitable and will likely change once businesses fully adjust. Note that there is yet no evidence for employment losses for all AI-exposed roles.
The 250 Greatest Songs of the 21st Century list skews towards the first decade. I always thought that pop music is becoming worse - perhaps this is even data to support that claim?
On the economics Nobel this year, I have seen many great summaries of the work of Mokyr, Aghion and Howitt. These two are my favourites. Also read Aghion on Chinese industrial policy and on the implications of AI for economic growth. The Nobel committee has an excellent overview of Mokyr’s theory. Here is the paper for the Aghion and Howitt model of creative destruction. Also add Mokyr to your reading list: his theory underpins the role of culture in fostering growth and discovery - a culture that we are unfortunately losing today.
On early-onset cancer. Incidence is up (likely due to obesity), yet mortality rates have substantially fallen (although hitting diminishing returns in recent years?), adjusting for population ageing. Better detection likely explains the increase in the incidence. Treatment is also much better. These results replicate across different countries.
Conditional convergence, or a damning indictment of economic stagnation in Britain?
GPT-5 and Gemini 2.5 Pro achieved gold in the IOAA. These are also the best models to use for math problems. I think most of the debate surrounding whether AI progress is stalling depends upon its application. Perhaps for general tasks, yet certainly not for specialised fields of inquiry, where the main LLMs are now at PhD level.
Dallas-Fed predictions.
Immigration is reducing the income gap between blacks and whites in America. As I continuously emphasise, heterogeneity in outcomes within racial categorisations is far more salient than group differences between them, no matter what your view is on the causes of group disparities.
A few notable examples of selection bias. We need to teach basic statistical literacy in high school as a mandatory subject.
Is the Eurozone now a fiscal union? Perhaps more importantly, did those nations signing up to the common currency sign up to these implicit fiscal transfers?
It takes a model to beat a model. Causal identification is central to the economics profession today, yet one requires a robust theoretical framework to interpret your findings. Autor's work does not explain the full story.
Although violent crime seems uncorrelated with economic conditions, there is a case to be made for a link with property crime. Worth noting that drug abuse falls too - adding to the body of evidence suggesting that rewards conditional on abstinence work. These results are consistent with rational addiction, but not with disease theories (e.g. the AA model) that implies an unstoppable escalation until "rock bottom", which fits the overall pattern of the data here. Yes, addiction is characterised by increasing marginal utilities (the more you consume the more you want to consume), and rational addiction theory is a black box: the functional form of the utility function is perfectly consistent with the dopamine reward system hijacking the brain. Yet constraints, opportunity costs, superior substitutes, and the general proclivity towards deferring gratification with maturity, dampen the influence of our impulses.
Evidence in favour of the signalling hypothesis for university education.
High rates of polygyny do not lead to large numbers of unmarried men. I'm unconvinced whether the census data in some of those countries is of good quality, so by itself, not worth updating your priors over. However, I agree with Lyman's general claim: marriage markets do not attain equilibrium, so ceteris paribus, whilst polygyny would reduce the proportion who marry, with large frictions to matching, this isn't observed. Also polygyny is endogenous to the wider cultural variables (traditional and conservative values) that correlate with high marriage rates.
A lesson in unintended consequences: EITC increases increase the rate of high school dropouts.
Race is a social construct, contrary to what illiberals claim on either side of the political divide.
LLMs may beat the best superforecasters soon, albeit currently they still yield an edge.
High fixed costs push artistic innovation towards popular culture, whilst low fixed costs facilitate more avant-garde and niche pursuits. The internet has clearly pushed culture towards the latter, via both reduced search and transaction costs and fixed costs. AI will likely exacerbate this.
Yes, if you see a Starbucks or Costa opening in your neighborhood, it is a good sign for that area.
This prediction turned out to be correct: we see a slight improvement in the Conservatives’ polling since conference (yes polling will fluctuate again, yet given availability bias, the rise now is probably linked to current events). Focus on the bread and butter of living standards, not in outcompeting Reform in a nativist match you cannot win. This is not to say that immigration does not matter, and I have argued before that the median voter is underrepresented on this issue, yet the median voter position is not fixed. Why not make the case for open borders, and run the experiment?
Share buybacks do not reduce investment. My view goes further: An EMH-like argument emphasising asymmetric info. The firm has insider information regarding its future profitability. This trade encodes that into prices. Attempts to discourage share buybacks (e.g. via taxes) hence harm price discovery.
LLMs can develop pathological behaviours. Are these behaviours fundamental to all intelligent beings, or are these the results of training on human-created data, thereby reflecting our biases?
Quality not quantity of your exercise matters. There are no excuses! I am a prolific walker though (an average of an hour, five miles, each day) so I'm writing from a position where no major adjustment needs to be made, but my advice if you're starting to exercise is to start small with 100% effort.
Heart disease, the greatest killer of mankind today, could be a relic of the past within my lifetime. In particular, note that these drugs target genetic pathways. Genes matter, as well as environment and lifestyle. I see no contradiction here.
On the Nobel prize for physics, relating to experiments demonstrating quantum tunnelling at a macroscopic level.
Convenience yields are positively correlated with the magnitude of fiscal surpluses. In a monetary union, as risk-free rates do not adjust at the country-level, this implies a large fiscal cost to running deficits. Luckily for the periphery and France, the Eurozone is now a fiscal union, so alas no major structural reforms will be implemented anytime soon.
Policy recommendations for the development of AI in Britain.
Economists back an abundance agenda for Britain, alongside welfare and labour-market reform. Abundance-related items are prioritised over EU relations, devolution, public services, closing regional gaps, the third-runway, and pension nationalism. It seems like there is an inverse correlation between what we should prioritise vs what garners the most press attention…
However we account for income misreporting, this paper just shows that Piketty and Saez's estimates of income (or wealth) inequality are not accurate. A problem that the left needs to confront.
"The UK not only developed the science of three-person babies, but it also became the first country in the world to introduce laws to allow their creation after a vote in Parliament in 2015.” There is still a bullish case for Britain; we just need the right institutions to reach our potential.
Freedom of navigation in the skies is important and underrated in facilitating trade. So Yglesias may even be understating the success of increased competition in airlines. Yes, privatisation in most cases works.
Whatever your views on the CCP are, you have to admire China's technological feats. Reject abundance, and we lose ground to China. It's as simple as that!
We see a divergence between the West vs China on automating unskilled service-sector roles. Why? My theory is that hospitality is a lot more established as an industry in the West, and revolves around the human touch here. In China, less so, leaving more scope for radical change to established practices? Regulatory differences may also be a factor: only recently has Waymo been approved in London.
A good antidote to race realism. As I'm a novice on this subject, I have purposely refrained from giving strong opinions on the matter. It seems that the genetic data is mired in confounding, data issues (obviously as the UK biobank data was leaked without permission a replication hasn't been done), and all the other limitations to empirical science. So I take one's view on whether racial gaps in average IQs are a function of genetics to mainly reflect one's priors more than anything. As for Crémieux, from what I can gather regarding his true identity, I don't think he argues in good faith on this subject. He's on much stronger ground on other subjects (here's an example regarding climate change), so I will continue to use that account as a source of information.
I do wonder about the exact question wording/framing - those numbers in support of all-male groups seem low, especially when considering sports teams. It's as if Democrats don't watch as much sports as Republicans, and I find such a large gap hard to believe. People are viewing these results overwhelmingly through culture war lens, but what the results are probably endogenous to framing and what respondents were thinking of when considering all-male/female groups.
An inverse relationship between thinking time and accuracy in LLMs?
Marriage markets in the West: a tale of less matches but greater matching efficiency?
Human cloning is a lot closer than we may think. I too will donate white blood cells to maximise the probability that (at least some version of me) will outlast death. On a related note, Alvin Lucier is a musical genius. The volume of "Still and Moving Lines in Families of Hyperbolas" follows an almost perfect sine wave.
A thread on how DEI and ESG is destroying the British stockmarket.
Life on Saturn’s moons? Any reasonable observer has to conclude at this point that extraterrestrial life more likely than not existed in our solar-system's history.
On differences between male and female brains. Yes, biological sex is real, but unlike radical feminists (feminists that accept the biological reality of sex, as opposed to intersectional feminists), I take this statement to its logical conclusion, that there are stark biological and psychological differences between the sexes, and therefore from division of labour and comparative advantage, in equilibrium we would expect differential gender roles in society to be prevalent. Indeed this has been the case in virtually all of human history, across almost all human societies (albeit a few exceptions in some tribal societies where matriarchy or full egalitarianism prevails). My attitude on this, as with most sociocultural topics, is Burkean.
Similar to the latest findings on the Cambrian revolution. Is our knowledge of key events in prehistory out by a million years or so?
Researchers created egg cells from adult somatic cells via 'mitomeiosis'; transferring a nucleus from a skin cell into a donor egg then inducing it to become haploid. Interestingly, this happens without recombination, which would increase within-family genetic variance and efficacy of embryo selection. However, only a small number had the correct number of chromosomes, so mutations may be an issue.