Macro Roundup

Melanie Wallskog, “Entrepreneurial Spillovers Across Coworkers,” Stanford University, November 11, 2021,
http://wallskog.su.domains/files/wallskog_jmp.pdf

Firm stats, “…Beyond firm survival, I explore other measures of firm success, including size, in terms of employment, payroll, and revenue, and revenue productivity. As shown in Table9, the patterns are generally similar to those for firm survival: individuals who work with more entrepreneurs tend to start “worse" firms, unless their entrepreneurial coworkers themselves were successful. For example, in column 1 of Panel A, I find that a one standard deviation (14.7 percentage point) increase in the share of coworkers who were recently entrepreneurs predicts 5.5% lower entry year employment. Yet, as column 2 shows, this pattern is offset if the entrepreneurial coworkers ran large firms | i.e., if their entrepreneurial firms were in the top 10% of entry year log employment, amongst firms that entered in the same year and industry: conditional on general exposure to entrepreneurial coworkers, a future entrepreneur with a one standard deviation (3.8 percentage point) higher share of coworkers who were entrepreneurs to particularly large firms has 4.3% higher entry year employment.As Panel B shows, these patterns persist with the inclusion of entrepreneurial firm industry fixed effects: individuals' future entrepreneurial success is predicted by their coworkers' success, even within the industry in which they start their firm. Table 9 shows similar patterns for the likelihood of a future entrepreneur's firm being in the top 10% of entry year employment, payroll, and revenue, relative to firms that enter in the same year and industry (where I measure entrepreneurial coworkers' successes by analogous measures). The one exception is in terms of revenue productivity, where on its own, general exposure to entrepreneurs predicts a marginally higher probability of starting a particularly productive firm (column 9), but this appears to be driven by the particularly productive coworkers (column 10). Conditional on exposure to entrepreneurial coworkers, a future entrepreneur who works with a one standard deviation (4.5 percentage point) higher share of coworkers who were entrepreneurs at particular high-productivity firms is on average 0.6 percentage points more likely to run a particularly productive firm, a 9.9% increase relative to the mean outcome.57These patterns suggest that what entrepreneurial coworkers teach individuals depends on the experiences of the coworkers, with successful entrepreneurs having a greater capacity to improve future entrepreneurs' prospects. In the remainder of this section, I consider additional outcomes that complement these standard measures and provide additional insight into the mechanisms of these spillovers….”

By Sector, “…Sector of individual's establishment Because workplaces and entrepreneurship patterns vary by industry, it is plausible that spillovers may vary dramatically across sectors.45 In fact, most sectors have similar coefficients to the aggregate coefficient, with some exceptions. Figure5shows the extensive margin spillovers by the sector of the individual's current establishment, estimated in a single regression by interacting the share of coworkers who were recently entrepreneurs with indicators for each sector, while continuing to include industry fixed effects that control for baseline differences in future entrepreneurship rates. There appears to be few spillovers for workers in the agriculture, utilities, and health sectors (which likely have high entry costs due to regulation) but substantial spillovers in the management and accommodation and food services sectors.46The fact that spillovers exist in most sectors but are strongest in the accommodation and food service sector suggests two conclusions. First, these spillovers exist across the economy | these spillovers are commonplace, and are not driven by the culture or structure of a single sector. Second, because the spillovers are largest in the relatively low-technology accommodation and food services sectors, these spillovers are unlikely to be predominantly transmitting knowledge of complex technologies or promoting innovation..”

“…Table 3 presents the point estimates from model (4) as controls are gradually added. As the table shows, individuals who work with proportionally entrepreneurial coworkers are more likely to become entrepreneurs in the future, regardless of the inclusion of controls. As more controls are added, this relationship decreases marginally but remains relatively stable. In the full speciation (column 8), the coefficient on the share of coworkers with entrepreneurial experience is 0.025: this predicts that an individual whose entire set of coworkers have entrepreneurial experience is 2.5 percentage points more likely to become an entrepreneur themself, compared to an individual who works with no entrepreneurial coworkers.27Recall that only 3.1% of the sample become entrepreneurs subsequently, such that 2.5 percentage points is very large relative to 3.1%, suggesting an 80% increase relative to the mean. However, this interpretation may be misleading, since very few individuals work with all former entrepreneurs. Instead, consider an increase in one standard deviation: the estimated model predicts that individuals who work a one standard deviation (9.5 percentage points) higher share of entrepreneurial coworkers are 0.236 percentage points more likely to become entrepreneurs in the next five years. This gap is still large: a 0.236 percentage point increase in the predicted future entrepreneurship maps into a 7.6% increase relative to the mean;28 this increase is comparable to the prior findings in the literature. To provide a simple evaluation of the size of this estimate, I conduct a back-of-the-envelope calculation to approximate how much the spillovers boost aggregate entrepreneurship. I predict the number of “additional" future entrepreneurs that start firms in the presence of spillovers by multiplying the coefficient on the share with the mean share of coworkers with entrepreneurship experience (0.03356) and the number of individuals (46.68 million). This calculation yields a predicted additional 39,000 future entrepreneurs, which amounts to a 2.75% increase. While this back-of-the-envelope calculation is inherently simple, ignoring any general equilibrium forces or dynamics of how spillovers aggregate over time, it does suggest a meaningful role for spillovers: spillovers boost aggregate entrepreneurship by 3%...”

Key finding“…I find evidence of positive extensive margin spillovers: individuals who work with one standard deviation (10 percentage points) higher share of entrepreneurial coworkers are 2.5 percentage points more likely to become entrepreneurs themselves within the next five years, an 8% increase relative to the average likelihood…”

The evidence

“…Entrepreneurial coworkers It is additionally important to note that coworkers with recent entrepreneurial experience were not necessarily the most productive entrepreneurs: indeed, many of their firms no longer exist, and they are now primarily employed at a different firm. As discussed below,this likely shapes the lessons that these coworkers can teach. Table2documents summary statistics on the firm outcomes for coworkers and individuals who were entrepreneurs within the previous five years (1999-2003).The table shows the outcomes both in terms of the average coworkers (columns 1 and 2, i.e., average outcomes for coworkers who were previous entrepreneurs) and at the individual level (columns 3 and 4, i.e., average outcomes for individuals who were previous entrepreneurs). For this paper, the past outcomes of the average set of entrepreneurial coworkers (columns 1 and 2) describe the average “treatment" that individuals face in the workforce. The average past outcomes of previous entrepreneurs in general (columns 3 and 4, many of whom started their current firm) provide benchmarks for the success of the average set of entrepreneurial coworkers. Table 2 highlights that entrepreneurial coworkers generally started relatively unsuccessful (i.e., shortersurviving, smaller, and less productive) firms, although some were relatively successful. Less than half of individuals' entrepreneurial coworkers started _rms that survive to age 5, on average, while over 60% of all recent entrepreneurs' firm ssurvive that long. Alongside these higher exit rates, many entrepreneurial coworkers have since returned to being a standard worker: while around 52% of the recent entrepreneurs in general started their current firms, less than 9% of an individual's entrepreneurial coworkers are entrepreneurs of the current firms, on average. This means that many of these entrepreneurial coworkers have left the firm they started and now work at someone else's firm. Yet, some entrepreneurial coworkers started relatively successful firms; for instance, on average 15% of entrepreneurial coworkers started firms that were in the top 10% of entry year log employment, amongst firms that started in the same year and industry as them. Taken together, these summary statistics suggest that exposure to entrepreneurial coworkers is likely quite different from, e.g., mentorship through start-up accelerator programs: the average entrepreneurial coworker did not start a hyper-productive firms.However, the fact that there is heterogeneity in these entrepreneurial coworkers' outcomes | some individuals get “lucky" and work with successful former entrepreneurs | allows me to explore the roles of both exposure to more entrepreneurs and to more successful entrepreneurs in SectionsIIIandIVbelow….”

“..Despite the fact that future entrepreneurs are different from workers in general, they work and become entrepreneurs across the economy. Figure2shows the sectoral composition of all individuals and future entrepreneurs, plotting the shares of all individuals compared to the shares of future entrepreneurs employed in each 2004 primary firm sector (based on NAICS codes), along with the share of future entrepreneurs' firms' entry year sectors. As the figure shows, future entrepreneurs work in all industries in 2004 and start firms in all industries, though they disproportionately work and start firms in construction, professional/ scientific/technical services (e.g., R&D and law and accounting services), and accommodation and food services, and less often appear in manufacturing and health, compared to the general workforce. Nearly half of future entrepreneurs start firms in the same sector as their 2004 establishment….”

Workers vs. entrepreneurs, “…Entrepreneurs are quite different from the general labor force. Yet, entrepreneurs work and start firms in all sectors of the economy, making the potential scope of entrepreneurial spillovers large. For the 2004 sample, in Table1 I present entrepreneurial, demographic, job, and establishment characteristics of all individuals in 2004 and of those who become entrepreneurs between 2005 and 2009. Relative to the general population, future entrepreneurs tend to be young, male, educated, White and Asian, born outside the U.S., higher earning, and working at smaller, younger firms.They also tend to work with more entrepreneurial coworkers, which I explore more systematically in the remainder of the paper. While entrepreneurs tend to be younger than workers, there is a distinct inverse-U relationship between age and entrepreneurship, as shown in Figure 1. Entrepreneurship rates by age peak around age 35, with the share of individuals aged 35 who are entrepreneurs being nearly double that of individuals aged 20 or 70….”

Summery statistics, useful factoids

Core findings, “…In my extensive margin analysis, I estimate regression models of whether individuals who work with more former entrepreneurs in 2004 are more likely to become entrepreneurs themselves subsequently, between 2005 and 2009. I find evidence of positive spillovers: individuals who work with one standard deviation (about 10 percentage points) higher share of coworkers who were entrepreneurs in the past five years are 8% more likely to become entrepreneurs themselves in the next five years, relative to the average likelihood. Through a back of- the-envelope calculation, this pattern suggests that spillovers generate an additional 3% of entrepreneurs in the aggregate. While the average extensive margin spillover is large (8%, relative of the mean),there is substantial heterogeneity. The spillovers tend to be amplified by exposure to relatively more successful entrepreneurial coworkers (i.e., entrepreneurs whose firms were relatively large or productive). While spillovers exist in most sectors, they are largest in the accommodation and food service sector, suggesting that the aggregate evidence of spillovers are not driven by the “high tech" sector that previous literature has studied….Additionally, these spillovers are stronger for younger individuals and do not exist for near-retirement individuals, with individuals learning" the most from their relatively older coworkers. The spillovers are strongest across coworkers who may be more likely to regularly interact in the workplace or to form mentorship relationships, namely coworkers who earn similar wages or who are of the same sex or immigration status. Finally, spillovers generate new entrepreneurs: individuals who themselves have recent entrepreneurial experience are not more likely to become entrepreneurs after working with more entrepreneurial coworkers, consistent with these individuals having had their own experiences and thus little to learn from entrepreneurial coworkers….”

Her basic story, “…I study the entrepreneurial outcomes of the individuals who become entrepreneurs between 2005 and 2009 and find evidence of both spillovers of institutional knowledge and entrepreneurial skill. I estimate regressions models of whether an individual's future entrepreneurial firm's characteristics vary if they were exposed to more entrepreneurial coworkers in 2004. I find that individuals who work with more entrepreneurial coworkers tend to start firms that are smaller in both employment and sales and are less likely to survive, consistent with a net pattern of individuals on average simply being inspired or learning the institutional knowledge needed to start a firm, as this leads to less productive individuals choosing to become entrepreneurs. However, if the individuals' entrepreneurial coworkers ran larger or longer-surviving firms, the individuals are more likely to start firms that are larger and more likely to survive. These results suggest scope for some true productivity gains via entrepreneurial skill spillovers, if the spillovers are from particularly successful entrepreneurs…”

Controls that address your concern, “… Coworkers are not randomly assigned to individuals, creating identification concerns, despite the rich set of controls included in the baseline specifications. For instance, entrepreneurship-prone individuals may cluster at firms or establishments for a variety or reasons not captured by the observable controls, such that the spillovers I measure may not be evidence of learning from coworkers. Instead, these spillovers may reflect firm or establishment effects, consistent with the literature of entrepreneurs “spawning" from their employers. Or, these spillovers may reflect pure spurious correlation, if these entrepreneurship-prone individuals work at firms for reasons entirely unrelated to entrepreneurship, for instance if they are also more educated. Additionally, individuals living and working in particular locations or industries may experience common business shocks that make entrepreneurship more or less attractive. I address these identification concerns through several robustness analyses. First, I show that these spillovers do not reflect entrepreneurial individuals simply clustering at certain firms or establishments. Instead, individuals disproportionately appear to learn from their true coworkers other workers at their establishment as opposed to other workers at the same firm but at other establishments or workers who work at the same establishment either before they joined or after they leave.Second, I show that these spillovers cannot be fully accounted for by selection into having specific coworkers; the results are not driven by individuals seeking out entrepreneurial coworkers. Instead, spillovers are also apparent from coworkers who joined an individual's establishment after they joined, whom the individual should not have been able to select on when choosing to join the firm. Finally, spillovers are not driven by local or industry common shocks, as the patterns persist with the inclusion of additional location and industry fixed effects. To further bolster a causal interpretation of the spillovers, I present direct survey evidence of entrepreneurial spillovers across coworkers. By linking individuals who become entrepreneurs to firm owners in a Census survey, I analyze reported motivations for these individuals' entrepreneurship. I find that the individuals who previously worked with more former entrepreneurs are more likely to report that they had an entrepreneurial role model who led them to start a firm. In other words, the individuals who I predict become entrepreneurs because of their exposure to entrepreneurial coworkers are more likely to report having an entrepreneurial role model, consistent with these spillovers actually existing…”

She doesn’t explicitly address that question but she does implicitly address it (in her words her goal)“…I find that if an individual interacts more with former entrepreneurs, then they are more likely to become an entrepreneur subsequently. I compare the future entrepreneurship of individuals who work with more coworkers who were recently entrepreneurs to those who work with proportionally fewer entrepreneurial coworkers. Effectively, I want to compare individuals who are very similar, both in terms of their own demographics and entrepreneurship experience and their current firms, but who have different exposure to entrepreneurial experience….”

"...All else equal, bigger companies with greater market share have more pricing power than smaller ones. A survey of American cfos conducted by Duke University and the Federal Reserve Banks of Richmond and Atlanta found that85% of large firms reported passing on cost increases to customers, compared with 72% of small firms. A “pricing-power score” for companies in the s&p 1500 compiled by ubs is based on four indicators: mark-up, market share, and the volatility and skew of profit margins. The bank found that firms providing consumer staples, communication services and it have the most pricing power and that energy, financial and materials companies have the least (see chart 1). When ubs compared the financial performance of companies with strong and weak pricing power, they found that the former have delivered more profit growth since 2010 and generated better stock returns, particularly during periods of high inflation (see chart 2).Firms that score well on this index have lagged in the past year, notes ubs. This may be explained by cyclical factors. When profit margins are expanding, the argument goes, firms with pricing power tend to generate relatively low returns; when margins are shrinking, they produce high returns. At the moment, profits are still healthy. For now, demand is robust and consumers seem relatively insensitive to price changes. But companies are planning more price increases...."

Economist Staff, "Pricing power is highly prized on Wall Street,"The Economist, November 6, 2021, https://www.economist.com/business/2021/11/06/pricing-power-is-highly-prized-on-wall-street

Pricing power is highly prized on Wall Street

Mcdonald’s has employed a “barbell” pricing strategy for decades, luring customers with low-cost items in the hope that they will then splurge on pricier fare. This balancing act is now at risk. On October 27th the fast-food giant said that, due to rising costs, prices at its American restaurants will increase by 6% this year compared with 2020. The burger chain says labour expenses have risen by 10% at its franchised restaurants and 15% at its company-owned locations. Add the rising cost of ingredients and the result is higher prices for burgers and fries. For now, it seems, customers can stomach it. Chris Kempczinski, McDonald’s boss, said the increase “has been pretty well received”. After digesting the news, investors have sent shares in the fast-food firm up by 6%.

A growing number of companies are raising prices as costs for labour and raw materials rise, often with no ill effects. This summer PepsiCo, an American food giant, lifted prices for its fizzy drinks and snacks to offset higher commodity and transport costs; it plans further increases early next year. Ramon Laguarta, the firm’s boss, suggested in an earnings call in October that customers do not seem bothered. “Across the world consumers seem to be looking at pricing a little bit differently than before,” he said. In September Procter & Gamble, a multinational consumer-goods giant, raised prices for many of its products. The effect on demand was minimal. “We have not seen any material reaction from consumers,” Andre Schulten, the firm’s chief financial officer (cfo), told analysts last month.

“Pricing power”, the ability to pass costs to customers without harming sales, has long been prized by investors. Warren Buffett has described it as “the single most important decision in evaluating a business”. It is easy to see why. When hit with an unexpected expense, firms without pricing power are forced to cut costs, boost productivity or simply absorb the costs through lower profit margins. Those with pricing power can push costs onto customers, keeping margins steady.

Today, firms are eager to flaunt their price-setting clout. “We can reprice our product every second of every day,” Christopher Nassetta, boss of Hilton Worldwide, a hotel operator, told investors last month. “We believe we’ve got pricing power really better than almost anybody if not everybody in the industry,” boasted John Hartung, cfo of Chipotle, a restaurant chain, in October. Companies such as Starbucks, Levi Strauss and GlaxoSmithKline make similar claims. “We are a luxury company, so we do have pricing power,” bragged Tracey Travis, cfo of Estée Lauder, a cosmetics firm, on November 2nd.

They are not alone. Of the s&p 500 companies that have reported third-quarter results, over three-quarters beat projections, according to Bank of America Merrill Lynch. “This earnings season there was a lot of angst on the part of investors that higher input costs would erode margins,” says Patrick Palfrey of Credit Suisse, a bank. “In fact, what we have seen is another spectacular quarter on behalf of corporations so far in spite of input cost pressures.” According to Savita Subramanian and Ohsung Kwon of Bank of America mentions of “price” or “pricing” in American earnings calls—a proxy measure for pricing power—increased by 79% in the third quarter from a year earlier. In the second quarter, such mentions were up by 52% year on year.

If costs spiral out of control, the power to raise prices will become ever more important. On November 2nd JPMorgan Chase’s global purchasing-managers index, a measure of manufacturing activity, showed that input prices in the sector increased in October at the highest rate in more than 13 years. But the prices of manufactured goods and services also rose at the fastest pace since records began in 2009. A gap between input and output price inflation is typically interpreted as a sign that firms are struggling to raise prices and that margins are being squeezed. That isn’t happening yet.

Identifying firms with pricing power is crucial for investors. Analysts tend to look for three things. The first is a big mark-up—the difference between the price of a good and its marginal cost—which only firms with market power can get away with. Big and steady profit margins are another sign of pricing power. “If you are a firm that is dominant in your market, you are much more resilient to shocks,” explains Jan Eeckhout, an economist and the author of “The Profit Paradox”, a book published earlier this year.

Size is another factor. All else equal, bigger companies with greater market share have more pricing power than smaller ones. A survey of American cfos conducted by Duke University and the Federal Reserve Banks of Richmond and Atlanta found that 85% of large firms reported passing on cost increases to customers, compared with 72% of small firms.

For now, demand is robust and consumers seem relatively insensitive to price changes. But companies are planning more price increases. A survey by America’s National Federation of Independent Business, a trade group, found that the margin of small-business owners planning to raise prices in the next three months over those planning to lower them grew to 46%, the biggest gap since October 1979. This is a concern for some central bankers such as James Bullard, president of the Federal Reserve Bank of St Louis. In October he noted that for years companies have worried that if they raised prices, they would lose market share. “That may be breaking down,” he says.

Scott Alexander reports on new genetics research that suggests some of the genes associated with mental illness may have compensatory benefits, "...But genes for some other conditions increase non-cognitive skills related to educational attainment. Bipolar disorder, fine…manic creativity hard work. OCD, fine, it’s probably on a spectrum with OCPD which is on a spectrum with regular perfectionism. Autism, fine, probably a lot of Math PhDs. But the big shock here is schizophrenia. As of last time I checked, the leading hypothesis was that schizophrenia genes were just really bad, evolutionary detritus that we hadn’t quite managed to weed out. And although they definitely decrease IQ, they seem to be good in other ways. Not with certainty: the correction for false discovery rate kills a lot of the effect (though this is the question I would have been most interested in before reading these results, so maybe I can ignore that?). But there’s at least a faint signal here….At some point, some geneticists just did the hard thing and found some actual genes for actual intelligence, separate from educational attainment. And if you have both the educational attainment genes and the intelligence genes, you can subtract the one from the other to find the non-intelligence-related genes that affect educational attainment in other ways. That's the thought process behind Investigating the genetic architecture of noncognitive skills using GWAS-by-subtraction, by Demange et al...They admit not doing a perfect job. Their genetic measure of non-cognitive skills (ie factors that affect educational attainment other than intelligence) was still correlated at r = 0.31 with IQ (higher than most things in social sciences ever correlate with anything). So take their results as directionally rather than exactly true. Still, what did they find? Unsurprisingly, both sets of skills contribute to educational attainment (that’s the point), and cognitive skills contribute more to IQ (that’s also the point). I was pretty surprised by how much more specifically “self-reported math ability” isolates cognitive ability compared to “highest math class taken”, but I guess people are good at self-assessment. But the real fun is Figure 4: This is correlation between the genes for cognitive/non-cognitive skills related to educational attainment, and the genes for something else (eg household income). Note that these are genetic correlations, so they’re not looking at your actual household income, they’re looking at the genes that would cause you to have a high household income. This is probably better since it avoids lots of possible confounders...."

Scott Alexander, "Non-Cognitive Skills For Educational Attainment Suggest Benefits Of Mental Illness Genes," Astral Codex Ten, November 3, 2021, https://astralcodexten.substack.com/p/non-cognitive-skills-for-educational

Non-Cognitive Skills For Educational Attainment Suggest Benefits Of Mental Illness Genes

Suppose you want to study the genetics of intelligence. You probably want a sample size in the six digits, and you can't make a six digit number of people sit down and take IQ tests. Also, whenever you say "genetics" and "intelligence" in the same sentence, an angry mob shows up at your door. One solution is to switch to a more popular / less stressful line of work, like Mafia snitch or al-Qaeda second-in-command. But another solution is to use educational attainment.

"Educational attainment" means how far you got in school - did you graduate high school? Get halfway through college? Go to grad school? It's a nice simple number that everyone already knows, and studies show it's closely correlated with intelligence. This is the most common way people investigate cognitive genetics, and if you hear someone talk offhanded about finding "genes for intelligence", they probably found genes for educational attainment.

Educational attainment is closely correlated with intelligence, but not perfectly correlated. How far you go in school depends not just on your IQ, but on other skills: how hard-working and motivated you are, how well you can cope with adversity - and arguably also less desirable qualities, like whether you so desperately seek societal approval that you're willing to throw away your entire twenties on a PhD with no job prospects at the end of it. "Genes for educational attainment" will be a combination of genes for intelligence, and genes for this other stuff.

At some point, some geneticists just did the hard thing and found some actual genes for actual intelligence, separate from educational attainment. And if you have both the educational attainment genes and the intelligence genes, you can subtract the one from the other to find the non-intelligence-related genes that affect educational attainment in other ways. That's the thought process behind Investigating the genetic architecture of noncognitive skills using GWAS-by-subtraction, by Demange et al (the "al" includes some researchers whose work has featured here before, like Paige Harden, Elliot Tucker-Drob, and Abdel Abdellaoiu).

They admit not doing a perfect job. Their genetic measure of non-cognitive skills (ie factors that affect educational attainment other than intelligence) was still correlated at r = 0.31 with IQ (higher than most things in social sciences ever correlate with anything). So take their results as directionally rather than exactly true. Still, what did they find?

But the real fun is Figure 4:

Both cognitive and non-cognitive skills increase your income a lot, no surprise there. Both sets of skills improve your lifespan (probably more educated people are better at judging health advice - get your COVID vaccine!) and prevent you from making bad decisions like teenage pregnancy, smoking, or excessive drinking.

The Big 5 personality factors are yet more evidence for stereotype accuracy: the intelligent people are more introverted, more disagreeable, and less conscientious; the people who do well for reasons other than intelligence are more likely to be extraverted, agreeable, and diligent. Both cognitive and non-cognitive skills are negatively correlated with neuroticism, which is not what I would have expected.

And then there’s mental illness.

I have been saying for years that I think some of the genes for some mental illnesses must have compensatory benefits. Everyone else said that was dumb, they’re mostly selected against and decrease IQ. But here we get a pretty clear picture of where this is and isn’t true.

Depression is just bad. I strongly recommend not having it. Don’t even have any risk genes, if you can avoid it. All of you people trying to come up with clever evolutionary benefits for depression, I still think you’re wrong, and so does the genetic correlation with cognitive and non-cognitive aspects of educational attainment.

ADHD is also just bad here. This doesn’t entirely match my previous beliefs; I think it’s helpful with a certain kind of high-stress task switching. But we can’t blame this study for not picking that up; it’s just using the giant bucket “non-cognitive skills that affect your ability to stay in school”, of which ADHD is definitely one and definitely net bad. I suspect that there are different forms of the syndrome that are vs. aren’t associated with low intelligence (and maybe some associated with high intelligence), but this is total speculation and this study tells us nothing about this one way or the other, except that it’s correlated with lower intelligence on net.

But genes for some other conditions increase non-cognitive skills related to educational attainment. Bipolar disorder, fine, something something manic creativity hard work. OCD, fine, it’s probably on a spectrum with OCPD which is on a spectrum with regular perfectionism. Autism, fine, probably a lot of Math PhDs.

But the big shock here is schizophrenia. As of last time I checked, the leading hypothesis was that schizophrenia genes were just really bad, evolutionary detritus that we hadn’t quite managed to weed out. And although they definitely decrease IQ, they seem to be good in other ways. Not with certainty: the correction for false discovery rate kills a lot of the effect (though this is the question I would have been most interested in before reading these results, so maybe I can ignore that?). But there’s at least a faint signal here.

I may write some future posts going down the very weird rabbit holes this result opens up.

"...Does manufacturing still offer “good wages”? The short answer to this question is yes, but not as broadly as in the past. It depends on a worker’s occupation within the sector. Approximately three out of every five manufacturing workers are nonproduction workers, who typically earn an hourly wage that is 14 percent higher than they would receive if they worked in another sector. The wage premium for production workers, on the other hand, is smaller. While this premium was 10 percent or higher throughout the 1980s, it has been no more than 4 percent since 2003. The fact that the manufacturing wage premium is small for production workers, including those with no more than a high school diploma, undermines one of the reasons governments provide special assistance to manufacturers...A manufacturing wage premium is the average percentage difference between the wage a worker earns in manufacturing and the wage similar workers earn in industries other than manufacturing. Using standard wage regressions, we find that, between 1979 and 2018, the manufacturing wage premium declined much more for production workers (such as machine operators) than for nonproduction workers (such as managers or administrative assistants). As a result, the production-workers’ wage premium was 4 percent during 2015 to 2018, while the nonproduction-workers’ wage premium was 14 percent. The decline in the production-workers’ wage premium is broad based, both demographically and geographically. We argue that this decline was most likely caused by the loss of manufacturing production jobs, which put downward pressure on wages for this work. The decline in manufacturing wage premiums has important implications for the difficulties associated with recruiting manufacturing workers and for public investment in training for manufacturing jobs...."

Joel Elvery and Julianne Dunn, "Manufacturing Wage Premiums Have Diverged between Production and Nonproduction Workers," Federal Reserve Bank Of Cleveland, November 9, 2021, https://www.clevelandfed.org/newsroom-and-events/publications/regional-policy-reports/rpr-20211109-manufacturing-wage-premiums-have-diverged-between-production-and-nonproduction-workers.aspx