Event study, 640 dividend cuts, ten years

Finance research poster example

A finance research poster built on an event study leads with its table, because that is where the result lives. This A0 landscape board measures the share price reaction to 640 dividend cuts over ten years: the three day abnormal return falls from 1.1% for small cuts to 6.8% for cuts over half.

Create a poster with OneCraftA0 Landscape, printed at 1189 by 841 millimetres

The whole board

The poster at full size, exactly as it prints. Every number, citation and caption on it was written for this example, so the layout is being judged on real content.

HVU

What does a dividend cut cost a shareholder? Abnormal returns around 640 dividend cut announcements, 2016 to 2025

K. Hollis ¹, B. Adebayo ¹

1 Department of Finance, Harwick Vale University

Abstract

We measured the share price reaction to 640 dividend cuts by listed firms over ten years. The three day abnormal return averages 2.9% down and grows with the size of the cut, from 1.1% for cuts under 10% to 6.8% for cuts over half.

METHODS

Events. 640 announced cuts to a regular dividend, 2016 to 2025, by firms with daily prices for a year before the announcement.

Design. Event study. Day 0 is the announcement day, or the next trading day when the news came out after the close.

Model. Market model fitted on days 250 to 30 before the event, giving each firm an expected return for each event day.

Window. Cumulative abnormal return over days -1 to +1, with days -10 to -2 and +2 to +10 either side.

Groups. Four bands of cut size, measured as the fall in dividend per share against the previous payment.

Three day return by cut size

Size of the cut	Events	Three day abnormal return	Share of events
Under 10%	214	-1.1%	33%
10 to 25%	196	-2.4%	31%
25 to 50%	142	-4.0%	22%
Over 50%	88	-6.8%	14%
All cuts	640	-2.9%	100%

Why it matters

Boards often cut a dividend in one step rather than two because they expect the market to punish any cut by the same amount. If the penalty grows with the size of the cut, that reasoning is wrong, and a smaller cut now costs shareholders less.

REFERENCES

1.Marsh, D. and Iqbal, R. (2023) Dividend signals in small markets. Journal of Corporate Payouts, 11(2), 88 to 110.

2.Teale, F. (2024) Market model choice in short window event studies. Finance Methods Review, 6(1), 33 to 51.

3.Adebayo, B. (2025) Announcement timing and day zero. Harwick Vale Finance Paper 2, 2025.

4.Quint, L. and Rowe, S. (2024) Cut size and payout policy. Applied Finance Letters, 19(4), 301 to 315.

Figure 1. Abnormal return, per cent. Bars left to right in each group: days -10 to -2, days -1 to +1, days +2 to +10, all 21 days.

Source: daily prices, 640 events, 2016 to 2025.

Results

The loss is concentrated on days -1 to +1 and grows with each band of cut size. After day +1 the drift is small: about 0.5 points for all cuts and 0.7 points for cuts over half, so the market does not reverse the first reaction within ten days.

6.8%

lost in three days on a cut over 50%

Cumulative abnormal return, days -1 to +1, 88 events

Key finding

The penalty grows with the size of the cut, so two small cuts can cost less than one big one.

All 640 cuts: 2.9% abnormal loss over three days

Cuts under 10%: 1.1%; cuts over half: 6.8%

No reversal of the first reaction by day +10

Most of the loss lands on the announcement day

Block by block

What each block on the board is for, in the order a reader walks it.

Title band and two authors: The title asks what a dividend cut costs a shareholder and names the 640 announcements and the ten years.
Abstract and methods: The abstract states the average and the range. Five entries cover the events, how day 0 is set, the market model and its estimation window, the event window and the four cut bands.
Three day return by cut size: The results table: size of cut, events, three day abnormal return and share of events, with an all cuts row. The design clips a table caption, so the window is stated in the methods instead.
Why it matters and references: Why boards cut in one step and why a size dependent penalty changes that reasoning, followed by four numbered references.
Figure 1 and results: Bars for days -10 to -2, -1 to +1, +2 to +10 and the full 21 days, for all cuts and cuts over half. Results note the drift after day +1 is 0.5 and 0.7 points.
Key number, key finding and points: The stat block holds 6.8% over three days. The callout says two small cuts can cost less than one large one, and four checked points close the board.

What makes this board work

The table does the arguing

Four bands of cut size, their event counts and their three day returns, in a table that takes the widest block. The counts, 214, 196, 142 and 88, add to 640, and the all cuts row of -2.9% is their weighted mean.

The figure covers all 21 days

Figure 1 splits days -10 to +10 into before, during and after the three day window, for all cuts and for cuts over half. The before and after bars are small, which is the evidence that the market reacts once.

The headline sits in the table too

The stat block repeats 6.8% for cuts over 50%, a number the reader can find in the table row above, with its 88 events and window in the detail line.

Questions people ask

What should a finance research poster include for an event study?

The event definition and count, how day 0 is set, the expected return model and its estimation window, the event window, abnormal returns with their groups, and evidence on what happens outside the window. Finance audiences check the estimation window first, so this board states days 250 to 30 before the event.

Why lead with a table rather than a chart?

Because finance readers compare exact returns across groups, and a table gives them the figure to two decimal places with its event count beside it. A bar chart of four returns adds little. The chart on this board is kept for the time pattern, which a table shows badly.

How should negative returns appear on a poster?

With a minus sign and a consistent unit, and a heading that says the direction. The table shows -1.1% to -6.8%, while the stat block and abstract say lost and down so a reader skimming the large type does not misread the sign. Bars below zero make the direction obvious.

Why does the table on this board have no caption?

This table led design places the caption line below the table in a fixed height block, and the caption gets clipped. Rather than ship a half visible line, the board leaves it blank and puts what a caption would say, the window and the model, in the methods entries.

How many references fit on this finance layout?

Four, in the numbered block at the foot of the left column. The block on this design is capped at four entries, so a list of six from a working paper has to be cut to the four a viewer is most likely to look up: the method, the data and the closest prior result.

Could I show the full day by day path instead?

You could, but 21 bars per group in a block this size crowd the axis, and tick labels start colliding with the legend. Grouping the days into before, during and after keeps every number readable and still answers the question a reader asks: did the price drift or reverse later.

Why split the cuts into four size bands?

Because the claim being tested is that the penalty grows with the size of the cut, and four bands are enough to show a gradient without making any group too small. The smallest band here still holds 88 events. More bands would thin the over 50% group and widen its uncertainty.

Build your own in about a minute

The button below opens the generator with this use case already described. Change the wording to match your own, generate, then edit anything you like.

Make one like this

Sources

Written and checked by the OneCraft team. Last checked 4 October 2026.