Original analyses of the US credit union industry. Each piece tests a single hypothesis against real call-report data and surfaces a takeaway you can act on.
Pure NCUA data. Single hypothesis per piece, single chart, single takeaway. New ones land each quarter.
Hypothesis: Top-quartile efficiency CUs grow faster than bottom-quartile CUs.
Refutes the "you have to spend to grow" intuition. The efficiency ratio is one of the cheapest single metrics to track and one of the highest-signal.
Hypothesis: There is an inverted-U between L/S ratio and ROA — peak is somewhere around 85-95%, both extremes underperform.
Most CU CFO boardroom debates are some flavor of "should we lend more aggressively?" Quantifies where the curve turns.
Hypothesis: The conventional view says yes; the 2025 data shows fast growers are at least as well-capitalized as slow growers.
Counterintuitive. Useful for capital-planning conversations and for CUs that have been holding back growth out of capital caution.
Hypothesis: Efficiency improves dramatically from <$50M to $1B; flattens above $1B.
Drives the "should we merge?" board conversation. Quantifies the tier at which the math materially changes.
Hypothesis: CUs growing loans aggressively trade credit standards for volume and book higher delinquency.
Risk officers and examiners pay attention to lead indicators. The 2025 data inverts the hypothesis — fast growers have cleaner books, not dirtier ones.
Hypothesis: Diminishing returns; a "minimum viable" floor below which CUs lose members, an upper bound above which spend doesn't buy growth.
The CMO board question, finally answered with data instead of vendor case studies.
Hypothesis: CUs that grew members every single one of the last 4 quarters outperform CUs that grew on average.
Tests whether smoothness of execution matters beyond raw averages — relevant to board-level KPI discussions.
Hypothesis: CUs with higher members-per-branch (a digital-density proxy) earn higher ROA, holding asset tier constant.
The branch-vs-digital strategy debate has been narrative-led for a decade. Quantifies the actual operational trade-off.
Joins NCUA call-report data to Tudovu's rate-puller scrape. No other vendor can publish these because they don't own both sides of the data. The live cross-section is available today; the time-series analyses unlock as the daily rate-scrape history accumulates (started May 2026 — first multi-quarter pieces ship Q3-Q4 2026).
Hypothesis: Live cross-section. The top APYs across CDs, IRA CDs, savings, and money market — joined to NCUA credit-union metadata.
Doesn't need time-series history; runs off the daily rate scrape. Evergreen consumer-search SEO ("highest CD rates") + first cross-dataset proof point.
Hypothesis: CUs that priced CDs above the national median saw faster share growth the next quarter.
Available Q3 2026 — needs 90+ days of accumulated daily rate snapshots. The rate_puller pipeline went live May 2026.
Hypothesis: CUs that adjust CD rates within 30 days of a Fed move grow deposits faster than CUs that lag 90+ days.
Available Q4 2026 — needs an observed Fed rate move within the accumulated window. Directly actionable for ALCO / treasury teams; strong industry-press angle.
Hypothesis: Every 25bps of premium APY costs ~X bps of NIM but buys ~Y bps of deposit growth.
Available Q4 2026 — needs to pair the rate-puller time series with the NCUA quarterly NIM trajectory. Board-level "should we raise CD rates?" question, resolved with numbers.
Hypothesis: CUs with low rate-volatility retain members at a higher rate than CUs that chase the market.
Available 2027 — needs at least one full year of rate history to define "stability" credibly. Tests whether trust/consistency beats chasing yield.
Each piece picks a single question, defines a single hypothesis, segments US credit unions into buckets along a single dimension, and reports a single outcome metric. No black-box scoring, no composite indices — just clean splits you can verify against the source data.
We refresh each piece quarterly when new NCUA data lands (typically ~45 days after quarter end). The narrative usually holds; the exact numbers move. If you cite a stat from here in a board deck or article, the link will keep working — the page rewrites itself with current data.
Have a question you'd like us to answer with the data? Reach out at david@tudovu.com.