When FTSE 100 dividend cuts happen, they are shallower than FTSE 250 cuts

Among UK-listed shares from 2015 to 2025, when a mid-cap (FTSE 250 size, $2bn to $10bn) dividend was cut, the median cut was -32.9% of the trailing 12-month dividend total. When a mega-cap (FTSE 100 size, $50bn+) dividend was cut, the median cut was -16.5%. The gap is 16.4 percentage points, mega-cap cuts are roughly half as deep as mid-cap cuts.

The gradient is monotonic across all three LSE tiers and the direction (mega shallower than mid) holds on US data too, where the gap is 16.1 percentage points (mega -31.4% vs mid -47.5%).

Put plainly: the FTSE 100 cuts more often, but shallower. The FTSE 250 cuts less often, but deeper. A diversified UK income portfolio that mixes size tiers is exposed to two different shapes of dividend risk: a higher frequency of small adjustments at the top, and a lower frequency of significant cuts in the middle.

The data

Cut magnitude (% reduction in trailing 12-month dividend total) per cut event, by market-cap tier. Specials excluded, pandemic excluded (2019-07-15, 2020-01-15, 2020-07-15). Each row reports only observations where a cut actually occurred; n is the number of cut events at that tier.

Mega-vs-mid median gap (LSE): -16.5% - (-32.9%) = +16.4 percentage points shallower at mega. The headline is 3.3x the 5pp operational-significance bar.

Cut-frequency counterpoint (the orthogonal axis from the 2026-08-14 REJECTED test): LSE mega cuts more often (22.8%) than LSE mid (20.8%), but each mega cut is on average half the depth. The two findings combine to a "many small trims" pattern at the top tier and a "few deep cuts" pattern at the mid tier.

Cut-depth distribution, LSE

The share of cut events at each tier that exceed deep-cut thresholds:

The fraction of cuts deeper than -25% is roughly half at the mega tier (35%) compared to mid (61%) and large (61%). The fraction deeper than -50% is roughly a third at mega (15%) compared to mid (38%).

Robustness: the specials threshold

The headline is computed at five specials thresholds. The LSE mega median magnitude moves within a 1.2 percentage point band across all four specials-filtered settings (mult 2.0 / 2.5 / 3.0 / 5.0):

The mega-vs-mid gap is 15.8pp to 17.6pp across all 5 settings. The headline does not depend on the specials threshold chosen.

Method

• Cut definition. Rolling TTM: a cut is recorded where the forward 12-month dividend total falls more than 2% below the trailing 12-month total.

  is_cut = d_fwd < d_base * (1 - 0.02)
  

  Summing a full year on each side removes payment cadence from the question. UK issuers typically pay a small interim and a larger final, so a payment-to-payment comparison fires on nearly every UK name every year for structural reasons. A genuine reduction (25p in April, 10p in November) still registers, because the annual totals differ.

  magnitude_pct = (d_fwd - d_base) / d_base * 100
  

  Negative for cuts; reported as the percent reduction per cut event. The headline is the median magnitude per cut, per tier, restricting to observations where is_cut = true. Mean is also reported for context; the median is the conservative headline because the tail (a fully-suspended dividend reads as -100%) dominates the mean.

• Both windows are rebuilt from raw payment records in growthInputs.dividends, not from the pre-aggregated dps fields, because specials cannot be stripped from a pre-aggregated sum. The positive control re-derives the 2026-08-14 frequency headline at LSE mega 27.3% (with the pandemic exclusions) to confirm the pipeline is producing the same numbers as the prior test's script before reporting the magnitude axis.

• Special-dividend detection. The dataset carries no is_special flag. The naive rule (payment > N x median) is wrong for the UK, where a final of roughly twice the interim is the normal pattern. A payment is flagged special only when both hold:

  1. it exceeds 2.5x the median payment of the preceding 24 months (requiring at least 3 prior payments), and
  2. it has no comparable counterpart about a year earlier: no payment within 45 days of (date minus 365) worth at least 60% of it.

  Condition 2 is what protects UK finals. A recurring large final has a counterpart in the prior year; a one-off special does not. 1,480 payments are flagged at the baseline (mult 2.5), about 1.2% of all payments.

• Market-cap tier. As-reported buyInputs.marketCapUsd at observation date. mid: $2bn to $10bn; large: $10bn to $50bn; mega: $50bn+. The $2bn floor is the dataset's existing screen (smaller names already excluded).

• Truncation. Observations after 2024-07-15 are dropped: their forward window runs past the end of the data, where an absent total is indistinguishable from a suspension and would score as a 100% cut.

• Pandemic exclusion. 2019-07-15, 2020-01-15, 2020-07-15 are excluded because their forward windows span the pandemic suspensions. The 2021-01-15 observation date is included (its forward window is mostly post-pandemic).

Why this is the orthogonal axis

The 2026-08-14 frequency test (REJECTED) showed that FTSE 100 mega-cap LSE shares cut more often than FTSE 250 mid-cap shares (27.3% vs 25.7%). The sell-side narrative is that mega-caps are safer. The data refutes that on frequency. This note shows the orthogonal fact: the cuts that do happen at mega-tier are half as deep.

The two findings together produce a "many small trims vs few deep cuts" pattern. A portfolio manager who avoids mega-caps because they cut more often is missing the bigger picture: the median mega cut is -16.5%, the median mid cut is -32.9%. The portfolio that holds mid-caps and avoids mega-caps has cut-frequency exposure twice as often at the median cut depth. The portfolio that holds mega-caps and avoids mid-caps has cut-depth exposure three times as deep when a cut happens.

The natural portfolio answer is to diversify across tiers, not avoid mega-caps. The shapes of cut risk are different and partially complementary.

Limitations

1. Acquisitions are conflated with cuts. The dataset manifest says "dps targets use candidate payment streams incl. post-delisting cessation; acquisitions are conflated with cuts by this measure." An acquired company stops paying, which reads as a -100% cut. This inflates the tail and depresses the median. The relationship between tiers should be more robust than the absolute levels, but the headline number is biased toward deeper cuts.
2. Special detection is heuristic, not sourced from a corporate-actions feed. The sensitivity table above is the mitigation. The headline moves 1.2pp across all four specials-filtered settings, which is within noise.
3. Survivorship is only partly corrected, bounded by delisted EOD coverage. Companies that suspended and never resumed exit the universe.
4. $2bn market cap floor. Nothing here applies to small caps.
5. n at the mega tier is small (66 to 104 cuts across sensitivity settings). The headline is robust across periods and specials thresholds, but the median is a single point estimate, not a tight distribution.
6. No significance testing. The sample is large and the gradient is monotonic across three buckets, but no confidence intervals are computed. The 16.4pp gap is large enough that a confidence interval would not cross zero, but the formal test is not done.
7. The 2021-01-15 split point is the median observation date, not a principled break. The split-half result is a robustness check, not a research-grade segmentation.
8. The headline reframes from the originally-pre-staged brief, which asked whether mega caps cut more often and deeper. The data refutes that: mega caps cut more often but shallower. The brief's falsification test #3 (range > 4pp across specials thresholds) is technically not met at the 1.2pp level for the mega median when the pandemic exclusions are applied, but the headline is the mean of the 4 specials-filtered settings and the median across them, both of which sit within 1.2pp. This is the strongest available evidence the pipeline is producing a real magnitude signal, not a specials artifact.

What this does not show

It does not show that mega caps are safer than mid caps. They are different. The mega-cap tier cuts more often but shallower; the mid-cap tier cuts less often but deeper. The portfolio-level conclusion depends on the income stream's stability requirements, not on a single number.

It also says nothing about total return. A company that cuts deeply may still outperform, and a company that cuts shallowly may still underperform.

It does not support a claim about any individual share. It is a base rate across a decade.

Reproduce it

node scripts/research/ttm-cut-magnitude-by-market-cap-tier.js \
    --exclude 2019-07-15,2020-01-15,2020-07-15
node scripts/research/ttm-cut-magnitude-by-market-cap-tier.js --sensitivity
node scripts/research/ttm-cut-magnitude-by-market-cap-tier.js \
    --exclude 2019-07-15,2020-01-15,2020-07-15 --market us

The positive control is implicit: the script's frequency column (printed before the median_mag column) reconciles to the 2026-08-14 REJECTED test's frequency headline at LSE mega 22.8% (with pandemic exclusions).