作者： 本週美股股利增減動態
掌握本週美國上市公司股利變動趨勢，解析配息調升與削減原因，並辨識股票拆分陷阱。
在美國市場的大多數週次中，增加股利的次數多於削減股利的次數。在最近一個完整的週一至週日公告週期中，18 家美國上市公司調升了定期現金股利，而 3 家則調降了股利。下方的第一個表格涵蓋不同的時間區間，即過去七天，並列出其中 12 家規模最大的公司。本頁面上的每一項變動，都是將公司新公布的定期配息與其先前相同週期的定期配息進行比較，且在計算前已剔除特別股利與股票拆分。
本週宣布的股利增長
本列表列出過去七天內，市值最大的 12 家企業。更多符合條件的公司請參閱下方的週增長面板。前一欄為該公司先前的定期配息金額，新的一欄則是董事會剛宣布的金額。
每個數據背後的精確 SQL 語法
WITH feed AS (
SELECT max(declaration_date) AS latest
FROM global_markets.stocks_dividends
WHERE declaration_date <= today()
),
universe AS (
SELECT ticker, argMax(market_cap, date) AS mcap
FROM global_markets.stocks_ratios
WHERE date >= today() - 10 AND market_cap > 0
GROUP BY ticker
),
regular AS (
SELECT ticker,
ex_dividend_date,
max(declaration_date) AS declared_on,
argMax(cash_amount, declaration_date) AS amt,
argMax(frequency, declaration_date) AS freq
FROM global_markets.stocks_dividends
WHERE distribution_type = 'recurring'
AND cash_amount > 0
AND currency = 'USD'
AND frequency IN (1, 2, 4, 12)
AND ex_dividend_date >= today() - 800
AND ticker NOT IN ('SPCX')
GROUP BY ticker, ex_dividend_date
),
seq AS (
SELECT ticker, ex_dividend_date, declared_on, amt, freq,
lagInFrame(amt) OVER w AS prev_amt,
lagInFrame(freq) OVER w AS prev_freq,
lagInFrame(ex_dividend_date) OVER w AS prev_ex
FROM regular
WINDOW w AS (PARTITION BY ticker ORDER BY ex_dividend_date ASC ROWS BETWEEN 1 PRECEDING AND CURRENT ROW)
),
splits AS (
SELECT ticker, groupArray(execution_date) AS split_dates
FROM global_markets.stocks_splits
WHERE execution_date >= today() - 900
GROUP BY ticker
)
SELECT s.ticker AS ticker,
round(s.prev_amt, 4) AS prior_usd,
round(s.amt, 4) AS new_usd,
round(100 * (s.amt / s.prev_amt - 1), 1) AS pct_change,
toString(s.declared_on) AS declared_date,
toString(s.ex_dividend_date) AS ex_date,
round(u.mcap / 1e9, 1) AS company_size_bn,
toString(f.latest) AS declarations_through
FROM seq s
INNER JOIN universe u ON u.ticker = s.ticker
LEFT JOIN splits sp ON sp.ticker = s.ticker
CROSS JOIN feed f
WHERE s.declared_on > today() - 8
AND s.declared_on <= today()
AND s.prev_amt > 0
AND s.freq = s.prev_freq
AND arrayCount(x -> x > s.prev_ex, sp.split_dates) = 0
AND dateDiff('day', s.prev_ex, s.ex_dividend_date) BETWEEN intDiv(240, s.freq) AND intDiv(520, s.freq)
AND s.amt > s.prev_amt * 1.001
ORDER BY u.mcap DESC, s.ticker
LIMIT 12名單中規模最大的公司是 C，市值約 221.2 十億美元。該公司將每股定期配息從 0.6 美元提高至 0.67 美元，增幅為 11.7%。新金額將適用於該列對應的除息日；投資人必須在除息日前持有股票才能領取該筆股利。名單中規模最小的公司為 EHC，其配息變動了 10.5%。表中的 12 家公司增幅差異很大，這屬於正常現象。董事會可以僅微調一美分的配息，或是直接大幅調整。
股利調升屬於前瞻性承諾。新的每股金額將成為下一季計算的基準，且傾向於持續調升的董事會通常會維持此趨勢。擁有最長連續增長紀錄的公司，請參閱我們的 二十年連續增長股利 頁面；而 即將到來的除息日 日曆則列出了下一批除息的公司。
過去 90 天內宣布的股利削減
股利削減並不常見。通常一週內最多僅有幾家公司會進行調整，因此下方的篩選範圍擴大至整個季度。此列表列出該季度內 14 次最大幅度的削減，並包含該期間內符合條件的其他公司。每一列的宣布日期顯示了該事件的近期程度。
每個數據背後的精確 SQL 語法
WITH universe AS (
SELECT ticker, argMax(market_cap, date) AS mcap
FROM global_markets.stocks_ratios
WHERE date >= today() - 10 AND market_cap > 0
GROUP BY ticker
),
regular AS (
SELECT ticker,
ex_dividend_date,
max(declaration_date) AS declared_on,
argMax(cash_amount, declaration_date) AS amt,
argMax(frequency, declaration_date) AS freq
FROM global_markets.stocks_dividends
WHERE distribution_type = 'recurring'
AND cash_amount > 0
AND currency = 'USD'
AND frequency IN (1, 2, 4, 12)
AND ex_dividend_date >= today() - 800
AND ticker NOT IN ('SPCX')
GROUP BY ticker, ex_dividend_date
),
seq AS (
SELECT ticker, ex_dividend_date, declared_on, amt, freq,
lagInFrame(amt) OVER w AS prev_amt,
lagInFrame(freq) OVER w AS prev_freq,
lagInFrame(ex_dividend_date) OVER w AS prev_ex
FROM regular
WINDOW w AS (PARTITION BY ticker ORDER BY ex_dividend_date ASC ROWS BETWEEN 1 PRECEDING AND CURRENT ROW)
),
splits AS (
SELECT ticker, groupArray(execution_date) AS split_dates
FROM global_markets.stocks_splits
WHERE execution_date >= today() - 900
GROUP BY ticker
)
SELECT s.ticker AS ticker,
round(s.prev_amt, 4) AS prior_usd,
round(s.amt, 4) AS new_usd,
round(100 * (s.amt / s.prev_amt - 1), 1) AS pct_change,
toString(s.declared_on) AS declared_date,
toString(s.ex_dividend_date) AS ex_date
FROM seq s
INNER JOIN universe u ON u.ticker = s.ticker
LEFT JOIN splits sp ON sp.ticker = s.ticker
WHERE s.declared_on > today() - 91
AND s.declared_on <= today()
AND s.prev_amt > 0
AND s.freq = s.prev_freq
AND arrayCount(x -> x > s.prev_ex, sp.split_dates) = 0
AND dateDiff('day', s.prev_ex, s.ex_dividend_date) BETWEEN intDiv(240, s.freq) AND intDiv(520, s.freq)
AND s.amt < s.prev_amt * 0.995
ORDER BY pct_change ASC, ticker
LIMIT 14紀錄中最嚴重的削減幅度為 EMBC，其每股定期配息從 $0.15 降至 $0.01，變動幅度為 -93.3%。在同一表格中，變動幅度最小的為 GNK，變動幅度為 -30%。所有 14 列皆為每股宣布金額的實質削減，且皆是與該公司先前相同週期的配息金額進行比較。
此類列表通常由兩類公司主導。第一類是營運型企業，因現金流無法支應配息而重新調整配息額度，這是典型的股利削減，且通常發生在股價已經下跌之後。第二類是設計上採變動配息模式的轉嫁型工具：權益信託（royalty trusts）會配發底層油井的收益，而業務發展公司（BDCs）則配發其貸款組合產生的收益。這類公司的配息變動方向較為頻繁。若要查看美國大型企業股利削減的歷史概況（包含 2020 年的浪潮），請參閱 哪些公司削減了股利。
每週增減趨勢對比
單一公告本身參考價值有限。透過整個季節的名稱統計，才能呈現週期規律。下表列出過去 26 個完整申報週，並統計所有符合條件的變動，且不設上限。
每個數據背後的精確 SQL 語法
WITH universe AS (
SELECT ticker, argMax(market_cap, date) AS mcap
FROM global_markets.stocks_ratios
WHERE date >= today() - 10 AND market_cap > 0
GROUP BY ticker
),
regular AS (
SELECT ticker,
ex_dividend_date,
max(declaration_date) AS declared_on,
argMax(cash_amount, declaration_date) AS amt,
argMax(frequency, declaration_date) AS freq
FROM global_markets.stocks_dividends
WHERE distribution_type = 'recurring'
AND cash_amount > 0
AND currency = 'USD'
AND frequency IN (1, 2, 4, 12)
AND ex_dividend_date >= today() - 1200
AND ticker NOT IN ('SPCX')
GROUP BY ticker, ex_dividend_date
),
seq AS (
SELECT ticker, ex_dividend_date, declared_on, amt, freq,
lagInFrame(amt) OVER w AS prev_amt,
lagInFrame(freq) OVER w AS prev_freq,
lagInFrame(ex_dividend_date) OVER w AS prev_ex
FROM regular
WINDOW w AS (PARTITION BY ticker ORDER BY ex_dividend_date ASC ROWS BETWEEN 1 PRECEDING AND CURRENT ROW)
),
splits AS (
SELECT ticker, groupArray(execution_date) AS split_dates
FROM global_markets.stocks_splits
WHERE execution_date >= today() - 1300
GROUP BY ticker
)
SELECT toString(toMonday(s.declared_on)) AS week,
formatDateTime(toMonday(s.declared_on), '%b %e') AS week_label,
countIf(s.amt > s.prev_amt * 1.001) AS increases,
countIf(s.amt < s.prev_amt * 0.995) AS cuts
FROM seq s
INNER JOIN universe u ON u.ticker = s.ticker
LEFT JOIN splits sp ON sp.ticker = s.ticker
WHERE s.prev_amt > 0
AND s.freq = s.prev_freq
AND arrayCount(x -> x > s.prev_ex, sp.split_dates) = 0
AND dateDiff('day', s.prev_ex, s.ex_dividend_date) BETWEEN intDiv(240, s.freq) AND intDiv(520, s.freq)
AND s.declared_on >= toMonday(today()) - 182
AND s.declared_on < toMonday(today())
GROUP BY week, week_label
ORDER BY weekJan 19 當週以 31 次增加對比 4 次減少開場，並於 Jul 13 當週以 18 次增加與 3 次減少結束。增加趨勢線呈現波動。委員會依其各自的季度節奏進行申報，變動最劇烈的週次通常集中在每季結束後的幾週，而兩季之間的間隔則較為平緩。週次數據較少屬於日曆效應。減少趨勢線全年皆維持在低位。
Who cuts, and who does not
Increases and cuts sit differently across the size ladder. Every regular-dividend declaration recorded so far this calendar year, split into four bands by what the company is worth today:
每個數據背後的精確 SQL 語法
WITH universe AS (
SELECT ticker, argMax(market_cap, date) AS mcap
FROM global_markets.stocks_ratios
WHERE date >= today() - 10 AND market_cap > 0
GROUP BY ticker
),
regular AS (
SELECT ticker,
ex_dividend_date,
max(declaration_date) AS declared_on,
argMax(cash_amount, declaration_date) AS amt,
argMax(frequency, declaration_date) AS freq
FROM global_markets.stocks_dividends
WHERE distribution_type = 'recurring'
AND cash_amount > 0
AND currency = 'USD'
AND frequency IN (1, 2, 4, 12)
AND ex_dividend_date >= today() - 1200
AND ticker NOT IN ('SPCX')
GROUP BY ticker, ex_dividend_date
),
seq AS (
SELECT ticker, ex_dividend_date, declared_on, amt, freq,
lagInFrame(amt) OVER w AS prev_amt,
lagInFrame(freq) OVER w AS prev_freq,
lagInFrame(ex_dividend_date) OVER w AS prev_ex
FROM regular
WINDOW w AS (PARTITION BY ticker ORDER BY ex_dividend_date ASC ROWS BETWEEN 1 PRECEDING AND CURRENT ROW)
),
splits AS (
SELECT ticker, groupArray(execution_date) AS split_dates
FROM global_markets.stocks_splits
WHERE execution_date >= today() - 1300
GROUP BY ticker
),
changes AS (
SELECT s.ticker AS ticker,
s.declared_on AS declared_on,
s.amt / s.prev_amt - 1 AS chg,
u.mcap AS mcap
FROM seq s
INNER JOIN universe u ON u.ticker = s.ticker
LEFT JOIN splits sp ON sp.ticker = s.ticker
WHERE s.declared_on >= toStartOfYear(today())
AND s.declared_on <= today()
AND s.prev_amt > 0
AND s.freq = s.prev_freq
AND arrayCount(x -> x > s.prev_ex, sp.split_dates) = 0
AND dateDiff('day', s.prev_ex, s.ex_dividend_date) BETWEEN intDiv(240, s.freq) AND intDiv(520, s.freq)
)
SELECT multiIf(mcap >= 1e11, 'Mega cap ($100B and up)',
mcap >= 1e10, 'Large cap ($10B to $100B)',
mcap >= 2e9, 'Mid cap ($2B to $10B)',
'Small cap (under $2B)') AS size_band,
count() AS declarations,
countIf(chg > 0.001) AS increases,
countIf(chg < -0.005) AS cuts,
round(100 * countIf(chg < -0.005) / count(), 1) AS cut_share_pct,
round(100 * quantileDeterministicIf(0.5)(chg, cityHash64(ticker, declared_on), chg > 0.001), 1) AS median_raise_pct
FROM changes
GROUP BY size_band
HAVING countIf(chg > 0.001) > 0
ORDER BY max(mcap) DESCIn the top band, Mega cap ($100B and up), 53 of 210 declarations raised the payment, against a cut count of 0, a cut share of 0%. The picture inverts at the bottom of the market: the Small cap (under $2B) band carried a cut count of 66 out of 1333 declarations, a cut share of 5%, well above the rate in each of the three larger bands. The typical raise is a similar size everywhere, near 7% in the Large cap ($10B to $100B) band and 8.9% among the small caps. The bands differ in durability, not in the size of the typical raise. One caveat sits under the panel: the bands use each company's market value today, so a payer that cut in the spring and shrank afterwards is banded by the smaller value it carries now.
股票拆分會造成股利削減的假象嗎？
這是大多數股利篩選工具常會出錯的地方。若一家公司進行 10 比 1 的股票拆分，並在同一天將每股股利除以 10。實際上股利並未削減，股東持有十倍數量的股票，獲得的總現金金額保持不變。若篩選工具直接比較拆分日前後的每股金額，會顯示出驚人的虛假削減。9 大型發放者，各別以其自身的拆分數據衡量：
每個數據背後的精確 SQL 語法
WITH sp AS (
SELECT ticker,
max(execution_date) AS exec_date,
argMax(split_to / split_from, execution_date) AS ratio
FROM global_markets.stocks_splits
WHERE adjustment_type = 'forward_split'
AND split_from > 0
AND split_to >= split_from * 2
AND modulo(split_to, split_from) = 0
AND ticker IN ('NVDA', 'AVGO', 'WMT', 'LRCX', 'CTAS', 'ETR', 'FAST', 'ODFL', 'BKNG')
AND execution_date >= today() - 1500
AND execution_date <= today() - 90
GROUP BY ticker
),
divs AS (
SELECT ticker, ex_dividend_date, argMax(cash_amount, declaration_date) AS amt
FROM global_markets.stocks_dividends
WHERE distribution_type = 'recurring'
AND cash_amount > 0
AND currency = 'USD'
AND ticker IN ('NVDA', 'AVGO', 'WMT', 'LRCX', 'CTAS', 'ETR', 'FAST', 'ODFL', 'BKNG')
GROUP BY ticker, ex_dividend_date
)
SELECT sp.ticker AS ticker,
concat(toString(toUInt32(round(sp.ratio))), '-for-1') AS split_ratio,
toString(sp.exec_date) AS executed_date,
round(argMaxIf(d.amt, d.ex_dividend_date, d.ex_dividend_date < sp.exec_date), 4) AS before_split_usd,
round(argMinIf(d.amt, d.ex_dividend_date, d.ex_dividend_date >= sp.exec_date), 4) AS after_split_usd,
round(100 * (argMinIf(d.amt, d.ex_dividend_date, d.ex_dividend_date >= sp.exec_date)
/ argMaxIf(d.amt, d.ex_dividend_date, d.ex_dividend_date < sp.exec_date) - 1), 1) AS apparent_pct,
round(100 * (argMinIf(d.amt, d.ex_dividend_date, d.ex_dividend_date >= sp.exec_date) * sp.ratio
/ argMaxIf(d.amt, d.ex_dividend_date, d.ex_dividend_date < sp.exec_date) - 1), 1) AS adjusted_pct
FROM sp
INNER JOIN divs d ON d.ticker = sp.ticker
GROUP BY sp.ticker, sp.ratio, sp.exec_date
HAVING countIf(d.ex_dividend_date < sp.exec_date) > 0
AND countIf(d.ex_dividend_date >= sp.exec_date) > 0
ORDER BY adjusted_pct DESC, sp.ticker以 NVDA 為例。在 10-for-1 拆分前的最後一次配息為每股 $0.04，拆分後的首次配息為 $0.01，原始變動幅度為 -75%。若將拆分後的配息金額乘以拆分比例，同一事件會變成 150% 的增長。表格另一端的 ODFL 經歷了 2-for-1 拆分，其原始變動為 -50%，調整後變動為 0%：每投入一美元的收益完全沒有變動。
本篩選機制之衡量方式
規則必須完整說明。若篩選機制包含隱藏規則，其結果僅是數據，實則為個人觀點。
- 以宣告日而非除權息日為準。 變動日期以董事會宣告之日為準，這也是新聞發布的時間。除權息日欄位顯示新金額與股價脫鉤的時間。
- 僅限定期配息。 僅比較標記為「定期」的紀錄。一次性的特別股息或補貼股息不屬於此系列；若將下一次定期配息與特別股息進行比較，會產生虛假的減息紀錄。詳細機制請參閱 什麼是特別股息。
- 兩者頻率必須一致。 進行比較的兩筆配息必須具有相同的宣告頻率，且其除權息日之間的間隔必須符合該頻率。將季度配息與年度配息進行比較並不屬於減息。
- 剔除拆股，而非進行調整。 若任何公司在兩筆配息中較早者之日或之後進行股票拆股，該公司將從這些表格中剔除。我們的股利數據源包含供應商提供的拆股調整欄位，但抽查發現該欄位在公司的配息歷史中應用並不一致，因此本頁面比較的是宣告金額，並剔除受影響的公司。
- 以美元及美國上市為主。 僅包含由我們的基本面數據源（具備市值資料）所涵蓋、且以美元宣告配息的公司。這會剔除指數型 ETF（其分配金為轉嫁收入且每期波動），以及美元金額隨匯率變動的外國發行者。
- 兩份排名表格皆有上限。 增幅表格最多顯示過去七天內最大的十二筆增幅；減幅表格最多顯示過去 90 天內最深的十四筆減幅。由於每個時間窗口內符合條件的公司可能更多，因此兩者皆非完整列表。週報面板則不設上限。
- 規模區間以今日市值為準。 我們的基本面數據源提供當前市值快照，不提供歷史數據。因此，若公司在春季減息後市值下跌，將被歸入較小規模區間；若發放股息的公司自 1 月以來已被收購或下市，則因缺乏快照而從年初至今（YTD）面板中剔除。
- 重要性門檻。 增幅需超過 0.1% 且減幅需低於 0.5% 才會被計入，藉此過濾分分錢（fractional-cent）配息產生的捨入誤差。
- 時效性。 宣告資料紀錄至 2026-07-23。週報系列以最後一個完整的週為結尾，因此當前未完成的週不會被顯示為活動驟減。
常見問題
本週有哪些公司調升股利？
在最近一個完整的週一至週日宣告週期內，共有 18 家美股上市公司調升了定期現金股利。若以過去七天（時間區間略有不同）來看，調升幅度最大的是 C，其每股定期配息從 $0.6 提高到 $0.67；其次是 BX，從 $1.16 提高到 $1.29。上方表格依公司規模由大到小排列 12 家公司。
2026年有哪些股票削減股利？
在今年至今紀錄的所有定期股利宣告中，削減股利的情況絕大多數集中在小型股：Small cap (under $2B) 級別中有 66 家，而最大規模級別僅有 0 家。在過去 90 天內宣告的最深幅削減是 EMBC，其每股定期配息從 $0.15 降至 $0.01；其次是 SACH，從 $0.05 降至 $0.01。上方表格列出該季度最深的 14 次削減；在同一時間區間內還有其他符合條件的削減。
股票拆分算作削減股利嗎？
不計算。1 拆 4 的拆股會將一股變成四股，並將每股股利除以四，因此持有人收到的總現金金額保持不變。上方的拆股面板顯示了 9 個真實案例，其中一個案例的原始數據顯示為 -75%，但經拆股調整後的變動為增加 150%。
與調升相比，股利削減發生的頻率如何？
在正常年份，調升的頻率遠高於削減。今年以來，最大規模的配息者紀錄了 53 次調升，對比 0 次削減；即使是削減最常見的小型股級別，也紀錄了 216 次調升，對比 66 次削減。
各個代號的股利概況
大型配息股各具獨特的特性：包含當前殖利率、完整的配息紀錄、成長趨勢以及除息行為。本文涵蓋 JPMorgan、Morgan Stanley、Bank of America、Wells Fargo、Coca-Cola、Procter & Gamble、Johnson & Johnson、Exxon Mobil、Chevron、AT&T、Verizon 以及 Pfizer，並透過 什麼是股利殖利率 說明每股配息如何轉換為篩選器顯示的殖利率數值。
此處的每個表格皆為針對已公告股利紀錄所進行的儲存且具版本控制的查詢，每週重新建構一次。您可以展開任何面板下的 SQL 語法以審核單筆資料，或在 Strasmore 終端機上執行該篩選器。