The hottest Econometrics Substack posts right now

Over the last 40+ years labor’s share of income has fallen while profits and capital’s share rose, and much of the stock-market boom is due to investors paying much higher valuations (P/E) rather than a big rise in earnings relative to GDP.
Bitcoin trading relies heavily on highly leveraged perpetual-futures contracts that can force margin calls and cause cascading liquidations, making the market prone to sharp crashes.
The income gap between the median family and the 80th percentile has widened a lot, so what counts as a “middle-class” lifestyle has shifted up and leaves median earners feeling poorer by comparison.

When you add up many positively correlated variables with positive weights, different composite scores tend to become very similar because shared covariance grows faster than unique variance, so the sums converge toward perfect correlation as components increase.
GDP will naturally correlate highly with lots of other measures since it aggregates overlapping components (and is sometimes included in other indexes), and aggregation reduces within-group noise which mechanically inflates between-group correlations.
Adding items to make a composite more reliable often makes it harder to distinguish from other composites, so improving reliability can reduce discriminant validity (for example, measures like grit can converge with conscientiousness).

Upcoming workshop in late February and early March on demand estimation by Ariel Pakes and Jeff Gortmaker.
The workshop will cover demand estimation methods in applied microeconomics and bridge between reduced-form and structural frameworks.
Attendees will get hands-on experience with practical exercises and training using a python package for estimating demand.

Because prices link firms, workers, and markets, difference-in-differences estimates pick up relative changes across units (the slope) but miss common, economy-wide level shifts that get absorbed by fixed effects — the “missing intercept.”
Treatment spillovers mean control groups are almost never untouched, so naively scaling a micro DiD coefficient up to an aggregate shock can be very misleading; the true aggregate effect could be much smaller or much larger than the naive calculation.
To learn the aggregate or policy-relevant effect you need economic structure or extra identifying assumptions; techniques like synthetic DiD fix pre-trends but cannot recover common, market-wide shocks without a structural model.

Get a weekly roundup of the best Substack posts, by hacker news affinity:

Triple difference design can be used with continuous treatment by defining the parameters based on dosage levels.
When treatment is continuous, the target parameter shifts from average treatment effect to average causal response function.
Continuous treatments require careful definition of parameters to compare different dosages along a treatment curve.

Our risk and time preferences are influenced by a mix of genetics and environment, with the unique environment playing a significant role.
Despite genetic influences, the most crucial factor driving our preferences and decisions is our unique life experiences and choices.
We have the power to shape our destiny regardless of genetic predispositions or shared environments. What we make of our lives is ultimately up to us.

Econometrics helps to reveal truths in small-scale matters, but applying them to large societal issues requires many assumptions. Democratizing knowledge generation can be a social game-changer by putting economic tools in the hands of more people.
Academics often focus on big questions about human nature, but the implementation of policies by governmental and social organizations could benefit from more hands-on and practical application of econometric tools to measure the impact of these policies.
Government and social institutions should embrace a more incremental approach, like carpenters, making gradual improvements as opposed to sweeping changes. There's potential for significant social change when institutions start questioning and measuring the effectiveness of their own operations.

You can adjust cash flows for risk in two main ways: estimating expected cash flows across scenarios and using certainty equivalent cash flows. Both methods aim to accurately reflect investment risk.
Certainty equivalent cash flows account for risk by using a safer value an investor would accept instead of the expected cash flow. This helps to quantify how risk-averse someone is when valuing their investment.
Risk adjusting cash flows isn't necessarily easier than adjusting discount rates. It's important to know when to apply simple methods, like focusing on safe cash flows or dividends, but also to recognize their limitations.