(3/9) Sex, Lies and Birth Statistics by Manuel Bagues, @carmenvillaecon.bsky.social
Evidence of inconsistencies in Spanish birth data sheds light on missing women in official statistics.
www.rfberlin.com/wp-content/u...
También hemos escrito un blog en español, que podéis leer aquí. Mil gracias al equipo de @nadaesgratis.bsky.social nadaesgratis.es/bagues/el-mi...
The working paper can be found here: cepr.org/publications...
@cepr.org @warwickecon.bsky.social @econ.uzh.ch
Our findings highlight the responsibility of statistical agencies to cross-validate and, where necessary, flag known data errors to prevent misinterpretation by researchers. The Spanish women were never missing, but are an artefact of processing errors at INE 5/n
Provisional birth statistics for the same period show ratios within plausible values. Additional evidence in support of data errors explaining the high ratio come from missing values in the micro-data, which suggest the corruption of the sex field affected other variables 4/n
The registry also shows implausible variation in sex ratios at the province and month level. The outliers are not normally distributed but consistent with one-directional miscoding of females as males 3/n
The official birth registry data show very large sex ratios in Spain from 1975 until approximately 2000 (normal rates are 105-106). Evidence of the registry being wrong comes from comparing it to the census. The differences remain after account for mortality/migration 2/n
According to official birth registry data, in 1981 Spain had the highest sex ratio in the world (109 boys per 100 girls). Prior work attributed this anomaly to demographic/behavioural patterns. In a new WP, we show the high ratio is due to data errors 🧵with Manuel Bagues 1/n
Thank you Francesca!
Mixed-income developments and income diversity may help prevent negative social interactions, including the spatial concentration of gangs. Full WP here:cep.lse.ac.uk/_NEW/PUBLICATIONS/abstra...
Why high-rises? They concentrated deprived populations in specific areas. The Right to Buy policy (1980) made this worse—houses were more likely to be bought off than flats, leaving towers increasingly isolated and disadvantaged.
Areas with gangs have dramatically higher rates of: knife crime (+72%), drug offences (+55%), violence (+49%), and crimes involving children suspects (+46%), as measured in administrative data from the London Metropolitan Police in 2010-2019.
Bombed areas are: 18% more likely to have high-rise postwar housing, and 12% more likely to host gangs in later decades. This is robust to controlling for pre-WWII income levels, not driven by Inner London only, and not just due to population density effects.
Is this relationship causal? We use the 1940-41 Blitz as a shock to urban redevelopment. Since height regulations were relaxed in bombed areas, and bombing was random at small scales, this gives us plausibly exogenous variation in where postwar high-rises were built.
We combine novel spatial data on 440 gang territories with detailed council housing records and building attributes. Compared to areas with no council housing, high-rise post-war council housing areas are 3.3x more likely to host gangs.
🚨📰New WP 🚨📰
How does social housing design affect neighborhoods decades later? We study London gangs to show that postwar urban planning—specifically high-rise public housing construction—had lasting effects on gang formation
@cep-lse.bsky.social
Is this useful? Probably not. Productive? Definitely not. But hopefully entertaining!
Wishing you all a happy Christmas with loved ones. May 2026 bring joy both at and away from work.❤️
Reverse causality doesn't seem to be at play (lag productivity does not predict taking time off conditional on day of week, month, and year FE).
Does taking time off help or hurt? There seem to be some returns to "getting in the zone" for a few days, but they fade fast. The long breaks (9+ days off) are deadline-driven: job market, thesis defence, submissions etc.
There is no "Friday I'm in Love" effect on the intensive margin either, but when I do work weekends, I perceive myself as less productive—probably shorter days, distractions, missing that peer effect energy.
Can I predict productive days beyond major life events? Not really. Year fixed effects explain just 1.5% of variation, month FE only 2%. In the intensive margin, there is no "deficit of vitamin D" observed fall in winter, no significant "working from holiday" falls in summer.
Three things stand out from the aggregate trends:
a) my standards might be getting stricter over time
b) there are clear bursts when away from home (e.g. during my visiting at Chicago)
c) there is a post-job market crash, explained by changes in the ext. margin (long holiday)
A few years ago I started a productivity diary to stay motivated when progress felt invisible. Every day I log productivity (on a 1-4 scale), or whether I took the day off. Here's what 3 years of data reveal about my work patterns. 🧵
x.com/carmenvillae...
Eine neue Studie von @carmenvillaecon.bsky.social zeigt, dass die Anhebung des gesetzlichen Mindestalters für den Alkoholkonsum von 16 auf 18 Jahre, die schulischen Leistungen und die psychische Gesundheit von Jugendlichen deutlich verbessert.👇
www.tagesanzeiger.ch/alkohol-bei-...
/end The students induced to stay in school didn't gain qualifications despite more time in education. Financial incentives pushing students toward academic tracks may displace them from vocational or work-based training where they might thrive and build valuable skills.
6/ Despite some beneficial effects on crime, our estimate is that the Marginal Value of Public Funds was just 0.74 - meaning every £1 spent generated only 74p in societal returns. Even though this was a direct cash transfer, it failed to provide good value for money.
5/ Low achievers: stayed in education longer but saw no qualification gains, became more economically inactive at 18, had lower earnings throughout their 20s. GOOD NEWS: The EMA reduced criminal convictions among this group and this effect persisted into their late 20s.
4/ Effects varied by prior attainment. High achievers: more likely to attend university (but not graduate), lower part-time work while studying, lower earnings in early 20s that never recovered. 📉