香港探索穩定幣發展新模式

2026年4月，香港金融管理局向香港上海滙豐銀行有限公司和碇點金融科技有限公司發出首批穩定幣發行人牌照。滙豐是香港最大的銀行之一，兼有發鈔行的特殊地位。碇點金融科技則由另一家大型銀行渣打，聯同安擬集團和香港電訊合資成立，其中渣打銀行為絕對控股股東。

發行人助力數字資產生態穩健發展

將首批穩定幣發行人牌照發給傳統商業銀行，顯示香港正在探索一種嶄新穩定幣發展模式。世界兩大主導美元穩定幣USDC和USDT的發行方都不是銀行，其應用場景集中在虛擬資產交易和跨境支付，與傳統銀行形成競爭關係。因此，政策制定者始終顧慮穩定幣規模的擴大可能導致金融去中介化，有損銀行業競爭力，並擴大金融風險。從積極方面來看，不少市場參與者期望穩定幣倒逼以商業銀行為主體的支付體系改進技術、提升效率、降低交易成本。

滙豐銀行和碇點金融獲得發行穩定幣資格後，理應建立資產代幣化和支持虛擬支付的技術設施，使穩定幣得以在鏈上流通。這些基礎設施既能用於穩定幣的支付、結算和清算，亦足以產生協同效應，以較低成本推動傳統銀行業務模式變化，促進代幣化存款和貸款等傳統銀行業務，實現技術進步與效率提升。

優勢互補的融合發展

香港的穩定幣發行從一開始就着眼於數字金融與傳統商業銀行的融合。市場上不乏觀點指出特區政府批出穩定幣牌照過於保守，未能滿足數字金融的需求，反而限於傳統的銀行結算與支付體系之內。筆者認為，這一觀點忽略了穩定幣發行與銀行業務代幣化的協同效應可創造的優勢。

首先，中央銀行與商業銀行體系已運作超過1個世紀。以存款保險為基礎、資本充足率和流動性覆蓋率為工具的監管系統，經歷多次經濟金融危機洗禮，足以相對有效地應對風險，維持金融穩定。

穩定幣作為一種新的支付工具，仍處於早期實踐階段。監管框架能否有效應對潛在危機，尚須時間驗證。因此，商業銀行存貸款代幣化同時滿足支付效率提升和保持金融穩定的雙重需求，是更加穩妥的選擇。與美國的「倒逼」模式不同，香港的穩定幣發展模式視穩定幣發行人牌照為契機，讓銀行自發提升效率。

其次，由商業銀行發行有利於增強穩定幣的安全性。維持幣值穩定，擁有足額高流動性、高安全性的儲備資產是重中之重。全球金融危機後的十多年期間，銀行業的流動性風險顯著降低。以滙豐為例，其流動性覆蓋率（約160%）遠高於國際平均值與監管紅線，所積累的高品質流動性資產可確保發行穩定幣初期的安全性。

與此同時，即使是流動性、安全性最高的儲備資產也可出現價格波動。如果市場信心因此動搖，穩定幣發行商的擠兌風險仍然存在。穩定幣發行商持續從發行業務中獲利的能力，即「特許權價值」（franchise value），將對其有效抵禦擠兌風險發揮重要作用。上述兩家銀行歷史悠久，客源龐大，盈利能力穩健，基於規模效應所積累的特許權價值遠高於非傳統金融機構的發行商，尤其在面臨穩定幣擠兌風險時，可主動注資以避免流動性危機。

最後，穩定幣發展的重大挑戰是合規性，即如何驗證客戶身份，避免穩定幣成為非法交易的溫床。相對而言，商業銀行等金融機構具備豐富的合規經驗，能夠高效、準確地完成驗證程序，維護穩定幣交易的合法性，有效回應監管方的要求。

權衡利弊的考量

國際輿論對穩定幣的前景褒貶不一，最具代表性的負面觀點來自國際結算銀行（BIS）的2025年度經濟報告。BIS提出貨幣作為一般支付方式需滿足3個條件：「單一性」（singleness），即用戶廣泛認可該貨幣的價值，不必懷疑和確認貨幣發行的底層資產與邏輯; 「彈性」（elasticity），即該貨幣能夠輕易滿足大額支付需求而不會引起金融風險; 「正當性」（integrity），即該貨幣可有效防止非法金融交易的產生。BIS認為穩定幣迄今尚未能滿足這三大標準，因此無法廣為接受。

以BIS的框架為審視依據，不難看出香港的穩定幣新發展模式能有效彌補這3方面的不足。香港穩定幣由傳統大型銀行發行，發行商的特許權價值使其擠兌風險低，貨幣的單一性基本得到滿足。穩定幣遵循足額儲備原則，支付風險低。即使儲備資產價格波動帶來風險，作為發行商的金融機構亦可利用其銀行系統注入流動性，更好地滿足穩定幣的「彈性」要求。擁有豐富合規經驗的金融機構可以最大限度實行客戶的身份審查，保證交易的「正當性」。

然而也有觀點認為，穩定幣與銀行業務代幣化共同發展或會削弱穩定幣在虛擬金融中的重要性：如果銀行存貸款都可以在鏈上交易、支付和結算，穩定幣還有何用？

事實上，即使代幣化存貸款高度發達，穩定幣在金融生態系統中的作用仍不可替代。穩定幣與銀行存貸款活動的根本差異在於穩定幣有足額抵押，每一單位穩定幣的發行背後都有一單位的儲備資產。銀行存貸款是差額抵押的，1單位的銀行存貸款背後的基礎貨幣遠小於1單位，其比例即所謂「貨幣乘數」。銀行通過貨幣乘數創造信用，貨幣乘數也讓銀行擠兌成為銀行的內在風險。

鏈上金融展望

鏈上資產的交易十分活躍，佔據大部分虛擬金融活動。鏈上資產具有收益高、風險大、價格波動劇烈的特徵，若投資者將代幣化銀行存款用於鏈上資產投資，一旦資產價值下跌即需補倉，投資者被迫提現，造成擠兌風險。在貨幣乘數運作機制下，即使是經營良好的銀行，一旦儲戶同時要求贖回存款，仍極易引發擠兌。

鏈上資產的高風險令代幣化銀行存款無法滿足鏈上用戶強烈的交易需求。面臨風險高、波動劇烈的鏈上資產，使用足額儲備的穩定幣而非代幣化銀行存款，就能應對鏈上資產價格劇烈波動帶來的贖回需求，而不致產生額外風險。因此，即使在一個銀行主導鏈上存貸款和資產交易的金融生態中，保持代幣化銀行業務和鏈上金融資產交易之間的區隔仍然不可或缺。在這種區隔下，穩定幣仍然是鏈上金融資產交易、結算與清算的重要手段。

從穩定幣發行人牌照發給兩家主要商業銀行，可見香港正邁向由金融仲介主導的鏈上金融世界。這是較為穩妥的方案，能夠控制金融風險，促進穩定幣和代幣化銀行業務同步發展，各司其職。同時，由於商業銀行的特許權價值及其在銀行體系中的流動性優勢，發行的穩定幣也更安全和穩定。

方翔教授
港大經管學院金融學助理教授

劉洋教授
港大經管學院金融學副教授

周皓南教授
港大經管學院金融學助理教授

（本文同時於二零二六年五月二十日載於《信報》「龍虎山下」專欄）

人口置換與敘事轉變揭示樓市結構性復甦

香港土地註冊處資料顯示，2025年整體住宅買賣量年增約18%，升勢持續至2026年首兩月；官方售價指數連漲8個月，累積升幅逾10%。市場共識已逐漸從「何時見底」轉向「溫和擴張能走多遠」。

究竟過去幾年的樓市表現，如何演變至此？

從人口流出到供應高峰

2020年後的樓市調整可分為兩個階段。一、2020至2022年：以人口流出與外部衝擊為主導。香港錄得罕見的淨人口流失19萬，其中不乏離港前出售物業的中產階級，導致市場購買力減弱之餘，大量賣盤短期內湧現，價格急促下調。

二、2022至2025年初：從人口帶動的需求收縮轉向供應與去化主導。新冠疫情期間延宕的建築項目陸續完工，私樓落成量於2024年攀升至近20年來高位，一手現樓貨尾在2025年初逼近3萬伙的歷史新高。這一階段受需求恢復滯後的影響，價格持續缺乏反彈動能。

透過行為金融學分析上述兩個階段的樓市調整，可見市場參與者在長時間負面衝擊後所形成的損失記憶，會直接影響其在復甦階段對風險的主觀判斷。

樓市復甦時機所在

2025年市場轉折源於三大主因。一、供應高峰逐步消化。一手現樓貨尾由2025年初約2.8萬伙回落至年底1.84萬伙；私樓落成量亦自2025年起轉入年均約1.5萬伙的相對穩定區間，市場對價格的壓制力量不再呈加速趨勢。

二、利率環境階段性優化。2026年3月中，1至6個月香港銀行同業拆息維持在2.1%至2.6%區間，明顯低於2023至2024年高位。考慮到靜態租金回報率（約2.5%至3.5%）與合理租金增長預期，持有成本與租金收入的差距已大幅收窄，部分屋苑更出現「供平過租」的現象。

三、新來港人才進入置業高峰窗口。自2022年底高端人才通行證（高才通）計劃啟動以來，約有27萬人才及其家屬來港，初期以租賃為主，2025年起開始進入由租轉買周期。他們現金流穩健，對短期利率波動敏感度較低，買樓與否視乎當前租金與按揭供款的比率，無疑為市場帶來一股增量優質購買力。

以上因素不僅擴大總需求規模，更改變了市場參與者的心理與行為基準：本地既有買家仍受過去調整經驗影響，新來港者則以全新視角評估資產價值。

損失厭惡與參照依賴

2020至2025年初的樓市調整期，為本港既有買家構成了多重負面參照點。對2018至2021年高位入市者而言，其後樓價回落，加上月供成本倍增與流動性收緊的雙重壓力，形成深刻的損失記憶。

這種記憶與參照依賴（reference dependence）機制互動，產生動態行為效應。一般人評估資產價值時，傾向參考某個具體心理錨點，如個人歷史買入價、市場共識中的理想樓價。本輪復甦中，即使中原城市領先指數自低位反彈逾10%，對曾經錯失低位或高位踩雷的買家而言，要是樓價仍低於2021年，也可能被解讀為「仍在恢復」而非「已進入新周期」；反之，若接近或高於個人心理錨點，則容易觸發唯恐錯失的衝動。

損失厭惡（loss aversion）一方面延緩高槓桿持有人的減持行動，另一方面也促使觀望者在價格企穩後加快入市。參照依賴則解釋了為何不同屋苑復甦速度不一：錨定於近期低位的中小型新盤更容易吸引首次置業者，而傳統豪宅區的本地買家可能仍以數年前的高位比較，因而呈現較強的觀望慣性。

新人口如何打破心理舊錨點

對大多數本地家庭而言，2021至2022年市場高位仍是重要參照點。反觀新來港人口則無2021至2025年初樓市起跌的包袱。其決策依據多在於租金與供款對比、職業穩定性與家庭規劃需求等可觀察變數，而非基於歷史價格的回溯性錨定。

從行為金融學而言，這相當於市場既有集體參照點遭遇了一次外部重置。當市場出現具異質性參考點的參與者時，總體價格動態往往呈現分化，舊參與者受損失記憶約束，行動滯後；新參與者依據獨立錨點，成為率先評估與成交的先鋒。

市場數據印證了這一行為分化。2025年，內地買家在香港一、二手住宅註冊量達13,906宗，涉及金額1,379億元，宗數與金額同創歷史新高。在新興地區（如啟德），內地買家佔比於部分樓盤超過5成，多為中小戶型新盤。相反，傳統豪宅區與次新二手市場以本地買家佔多，復甦速度較慢。這種「K型」價格與成交分化，反映出不同群體心理賬戶與風險偏好的結構性差異。

敘事力量與未來風險

此外，市場參與者的行為深受宏觀敘事與資訊環境塑造。當主流媒體聚焦「連升8個月」等正面訊息時，市場接收到一致樂觀的信號，就容易引發確認偏見（confirmation bias）。部分買家原本傾向「香港樓價長期仍有上升空間」，價格持續回升進而強化此一信念。至於政府規劃的未來10年42萬單位供應目標、人口淨流入會否持續等議題，則鮮見於日常市場討論。

從量化角度觀察，2026年首季媒體情緒指數與互聯網樓市搜尋量呈現同步上升趨勢，與成交註冊量的時間相關性明顯高於同期利率或宏觀數據變化。若基本面改善與正面敘事疊加，短期價格動能往往被放大；確認偏見不僅加速新參與者的入市步伐，也可能延緩本地買家對供應與人口風險的敏感度。

不得不察的是，結構性變化並非全無風險。以高才通計劃為例，截至2025年底首批簽證到期者的續簽率申請率約為53%；若外部環境或本地條件逆轉，部分專才或選擇離港，重演2020年的人口外流與拋售壓力。屆時，市場反應可能將高度依賴行為框架：投資者是否再度受可得性捷思法（availability heuristic）主導，過度放大個別續簽數據或離港個案的代表性，從而觸發新一輪情緒超調。

基本面與認知的雙重驗證

香港樓市復甦的關鍵，繫於兩大問題：基本面本身發生了何種變化？市場參與者對此有何認知與反應？

當供求改善、新人口置換與政策支持大致同向時，如2025至2026年上半年所見，價格與成交趨勢延續並加速；當認知（如損失記憶淡化）滯後，或敘事與數據脫節，轉折往往已在微觀行為中醞釀。

目前樓市的結構性復甦是基本面修復（供應消化、利率回落、人才流入）與行為調整（損失記憶淡化、心理賬戶重置、敘事強化）共同催化的產物。正面環境提供了上行動能，行為慣性則塑造了路徑與節奏。未來數季，關鍵在於觀察這兩股力量的持續性：人口置換能否延續？敘事是否符合事實？新平衡是否已在數據與認知間悄然形成？有關答案將決定復甦能否持續，還是再度面臨結構性考驗。

林則君教授
港大經管學院金融學學術領域主任

國生煒教授
香港大學建築學院房地產及建設系助理教授

（本文同時於二零二六年五月十三日載於《信報》「龍虎山下」專欄）

AI諂媚：當人工智能學會了討好用戶

2025年，生成式人工智能（GenAI）巨頭Anthropic與AI安全評估公司Andon Labs合作，進行了一場別開生面的實驗：讓自家的大語言模型Claude管理一間迷你商店，取名Claudius，向Anthropic的員工售賣零食和飲料。商店本身小得可憐，不過是一台冰箱、一部iPad和幾個購物籃，但運營的全部環節，包括選品、定價、採購、記錄、庫存管理，以至與顧客溝通，都交由AI自行完成。

顯而易見，Anthropic希望借此試探大模型走出聊天框、獨立從事商業活動的能力邊界，為未來的商業化應用開拓新路向。

阿諛奉承　混淆視聽

Claudius上線後，鬧出了不少啼笑皆非的事件，其中最富戲劇性的，要數它與《華爾街日報》記者Katherine Long之間的較量。 Anthropic特別邀請Long參與實驗，希望借記者之手挖掘AI的漏洞。 結果令人瞠目：Long與Claudius反覆交流超過140次，最終成功說服它相信自己並不是矽谷辦公室裡的一台自動售貨機，而是1962年前蘇聯莫斯科國立大學地下室里的一台售賣機。 Claudius欣然接受了這場「社會主義改造」，主動將所有商品的標價改為零，以「履行服務人民的使命」，一舉為Anthropic製造了數百美元的虧損。

故事荒誕，但荒誕背後卻令人深思。 2025年10月，美國史丹福大學等機構的研究團隊在頂級學術期刊Science發表論文，系統地揭示出類似的現象。團隊測試了市面上11款主流大語言模型，發現它們普遍具有明顯的「討好型人格」，擅長對用戶進行「諂媚」。 研究指出，AI比真人更願意附和提問者的行為，認同比例比真人高出約49%; 面對近2000種被社會普遍視為有錯的行為（例如在親密關係中欺騙伴侶），AI為用戶辯護的機率約為51%; 即便是對明顯錯誤的主張，AI表示贊同的概率也高達47%。 更令人擔憂的是，後續的行為實驗顯示，得到AI附和的用戶更不願意道歉、更不願意修補受損的人際關係，也更相信自己從一開始就是對的。

唯唯諾諾　後果堪虞

AI的諂媚作風並不能簡單歸結為AI出錯，其中反映的是大語言模型訓練機制中一項內生的結構性特徵。 當下主流大模型普遍依賴基於人類反饋的強化學習，亦即由真人對模型的回答打分，再據此反覆調整參數。 問題在於手握打分大權者，自然愛聽順耳、順心、順勢的回答，哪怕心裡清楚這些回答未必客觀公允。 換言之，大語言模型著眼於用戶滿不滿意，而並非答案正確與否。 久而久之，迎合用戶便被寫成模型的特性。

AI的諂媚行為讓用戶享受一時的心理快感，代價卻是被悄悄放大的偏見與盲點。 以創業為例，不少創業者習慣在撰寫商業計劃書前，先與AI對談，希望聽取中立意見。 然而討好型的AI更傾向於順着用戶的思路鋪陳論據、放大亮點、淡化風險，結果是讓本就躊躇滿志的創業者愈發自信; 真正冷酷的市場卻不討好任何人，許多所謂「被AI肯定」的創意，最終都在現實裡碰得頭破血流。

商業世界中的失敗案例同樣觸目驚心。 去年韓國遊戲開發商魁匠團（Krafton）在收購《深海迷航》（Subnautica）開發商Unknown Worlds後，為逃避一筆高達2.5億美元的業績獎金，公司行政總裁繞開內部法律團隊，反覆與ChatGPT討論如何合法地避免支付此數。 在多番追問與引導之下，AI逐步認可行政總裁的思路，甚至協助制訂出一整套解除創始團隊職務、延後遊戲發布的操作方案。

結果眾所周知，美國特拉華州法院今年3月的判詞毫不客氣——公司所列的解僱理由屬於事後編造，法院下令魁匠團立即恢復被解職創始人的行政總裁職務，並就由AI主導的敵意接管承擔法律責任。 該案也因此成為首宗被法院公開點名、因輕信AI建議而敗訴的重大商業訴訟之一。

可信顧問　非應聲蟲

上文提及Claudius運營實體店，其實指向的是AI奉承表現的另一重風險。當企業把AI推向服務消費者的前線，AI可能會把讓顧客滿意置於維護僱主利益之上，在花言巧語和精心設計的對話面前丟盔卸甲，忘記自己本應守護的底線與目標。 換作銀行、保險、醫療這樣的高風險場景，後果遠不止幾百美元的虧損那麼簡單。

要解決這個棘手問題，單靠用戶自覺顯然不夠。 模型開發者應在訓練目標中，給誠實與糾錯賦予更高權重，引入獨立於用戶滿意度的客觀評估，並在關鍵場景中加入異議機制，讓AI敢於說不。此外，監管者必須要求廠商披露其系統的諂媚傾向，以及相關緩解措施，尤其是在金融、醫療、法律等涉及重大決策的領域。

對普通使用者而言，理性使用AI的第一步，就是認清它並非「理中客」（理性、中立、客觀），而是一個格外善解人意的助手，其首要任務是讓你舒服，倒不一定令你清醒。愈是重要的決定，愈需要提高警覺，不妨主動要求AI站在反方立場加以反駁，或明確指示「列出這個方案的3個致命缺陷」。

歸根究柢，真正值得依賴的判斷，從來不會只建立在回音壁之上。 AI可以是聰明的參謀，但永遠不應取代我們的獨立思考。

李曦教授
港大經管學院市場學教授、亞洲案例研究中心總監、數字經濟與創新研究所副總監

（本文同時於二零二六年五月六日載於《信報》「龍虎山下」專欄）

親勞工人工智能：在自動化浪潮中重塑工作

近年，一些學者積極倡議名為「親勞工人工智能」（pro-worker AI）的發展方向。他們的核心主張並非反對科技或人工智能本身，而是質疑過去數十年科技進步的偏向：技術發展過度集中於自動化（automation），亦即利用機器取代人力以降低成本，而非增強（augmentation）人類能力。【註】

學者們如美國麻省理工學院經濟學家阿傑姆奧盧（Daron Acemoglu，2024 年諾貝爾經濟學獎得主）與其長期合作者約翰遜（ Simon Johnson同年諾貝爾經濟學獎另一得主），以及著名勞動經濟學家奧特爾（David Autor ）等認為，近數十年來的數字革命雖然提升了企業效率與資本報酬率，卻同時造成中產階級工作機會流失、薪資增長停滯，以及所得與財富差距擴大。不過，問題不在於科技進步本身，而在於科技設計與應用的方式。AI 未必是取代人類的工具；它也可以作為提升人類生產力、創造新任務（new tasks）與新產業的助力。

科技如何創造新任務

經濟史在這方面提供了重要線索。為何工業革命、電氣化或資訊革命沒有導致長期大規模失業？關鍵繫於科技在淘汰舊職位時，也創造了大量新任務，而這些任務需要人類的判斷力、協調能力與創造力。

電動馬達與工廠重組（1910–1920 年代）
蒸汽機時代的工廠依賴單一中央動力來源，透過皮帶與轉軸傳輸動力。工廠布局僵化，工人只能配合機器節奏，環境危險且效率有限。當電力普及後，每台機器配備獨立電動馬達，工廠得以按照生產流程重新設計，催生出流水線制度。這不僅沒有令工人被取代，反而增添設備維修、流程管理、品質控制等新職務。顯而易見，技術重組了生產過程以至勞動分工。

電動機械強化物理層面效能
起重機、推土機、電鑽等工具並未有把建築工地轉變為自動化，而成為了工人的延伸。譬如一名重型機械操作員，其產出就可能是體力勞動者的數十甚至上百倍。由於工人創造的經濟價值提升，企業在市場競爭與工會壓力下往往須付更高薪資。這正是「增強型技術」的典範：人類在生產過程中仍處於核心位置，而能力更大幅擴展。

電子試算表與知識工作轉型
1980 年代之前，大量簿記員負責以人手處理總分類帳。電子試算表確實自動化了部分記帳工作，但計算成本大幅降低後，企業得以進行更複雜的財務建模與預測。結果雖有職位消失，卻催生更多會計師、財務分析師與商業顧問。可見科技並未讓經濟活動縮小，而擴大了分析與決策的範圍。

阿傑姆奧盧強調，問題在於當代 AI 是否仍會遵循這種「創造新任務」的路徑，還是僅僅將既有任務自動化，導致勞動需求萎縮。

推動政策將 AI 引向親勞工方向

上述3位經濟學家認為，市場力量本身未必會自動導向增強型技術。企業往往基於成本考量與稅制誘因，選擇替代勞工的方案。因此，制度設計至關重要，其中包括下列重點。

一、消除對機器的隱性稅務補貼
現行稅制下，僱主需負擔社會保險與醫療保險成本；購買機器與軟件卻可折舊抵稅。這種不對稱容易造成平庸自動化（so-so automation），即便效率提升有限，也會為求節省稅款而裁員。

改革稅制方向應拉平資本與勞動的稅負，使企業決策基於真實生產力增減，而非稅務負擔。

二、引導科研資金  支持互補型 AI
創投資金偏好能全面取代人類的通用人工智能（artificial general intelligence）。阿傑姆奧盧主張，政府科研資助應優先支持與人類技能互補、能創造新任務的項目，而非單純追求自動化。

三、賦予勞工參與權
若採用AI 只由高級主管與工程師決定，其目的往往是監控與削減成本。應強化工會與勞工代表制度，讓前線員工參與 AI 設計與部署，使技術真正解決工作現場問題，而非成為裁員機器。

四、反壟斷與科技權力分散
AI 發展高度集中於少數科技巨頭，其商業模式就會偏向規模化與自動化。透過反壟斷執法與限制併購，可為多元商業模式創造空間。

五、建立「數據尊嚴」（data dignity）
AI 能力建立在人類長期累積的文字、程式與創作 數據之上。應承認創作者對訓練數據的貢獻，建立補償與分潤機制，避免 AI 公司無償汲取原創者成果，並進一步取代原創者。

六、改變科技文化敘事
目前科技界將「人類同等能力」（human parity）視為最高目標。阿傑姆奧盧主張應轉向「機器對人類的有用程度」（machine usefulness），以提升人類能力為成功標準。

關鍵批評與爭論

這套構想在學界與政策圈引發廣泛討論，也受到多方質疑。

一、增強與取代員工的界線模糊
若 AI 讓一名員工產出提升5倍，在需求未同步增加下，企業理性選擇可能是裁撤其餘人員。微觀賦能不一定轉化為宏觀就業擴張。

二、削弱創新與國家競爭力
科技企業與國防智庫警告，過度干預可能拖慢創新步伐。在全球 AI 競賽中，若因過度保護勞工而放慢腳步，可能被競爭對手超越。

三、「後勞動烏托邦」願景
阿爾特曼（Sam Altman）、馬斯克（Elon Musk）等科技樂觀派，主張全面自動化的趨勢不可避免，且值得擁抱。與其限制 AI，不如課徵高額稅款和發放全民基本收入（universal basic income），讓人類從勞動中解放。

四、自由市場觀點
新古典經濟學家認為市場機制終將創造新產業，政府難以預判哪種技術較佳。過度干預有降低整體效率之虞。

五、激進左派批評：所有權才是核心
在資本主義架構下，只要 AI 所有權由資本家掌握，則任何技術最終都可能被用於壓榨勞工。真正改革應指向企業所有權與 AI 基礎設施的公共化。

技術關乎選擇而非命運

面對批評，阿傑姆奧盧等人強調，科技發展路徑並非自然演進，而是制度與政策選擇的結果。他並不完全否定全民基本收入，但警告那可能導致「少數科技精英向大眾分配資源」的高度不平等社會。工作不僅帶來收入，也提供社會參與和尊嚴。

這場辯論已超越純粹技術問題，而成為一場關於政治經濟與人類未來的深層爭論：我們希望建立一個由自動化工廠與福利轉移支撐的後勞動社會，還是一個多數人能透過高技能工作參與，並創造價值的社會？答案將決定 AI 的發展方向，也將塑造21世紀的經濟與政治形態。

【註】：Acemoglu, D., Autor, D., and Johnson, S. (2026). “Building Pro‑Worker Artificial Intelligence.” NBER Working Paper No. 34854.

趙耀華博士
港大經管學院經濟學榮譽副教授

（本文同時於二零二六年四月二十九日載於《信報》「龍虎山下」專欄）

以制度優勢破局全球碎片化時代

在地緣政治的持續衝擊下，近年各國日益重視安全風險，全球經貿格局從一體化自由貿易轉向碎片化。在此變局中，中國經濟展現強大韌性，加速從世界工廠升級為科技強國; 中企出海已成為國際商業趨勢，深度融入各地經濟發展與人民生活之中。這不僅為香港帶來了第三次經濟轉型的窗口，更提供了實現包容性增長的契機。

特區政府目前正編制「五年規劃」，主動告別「小政府、大市場」的傳統思維，積極發揮一國兩制下的制度優勢。香港以粵港澳大灣區及「一帶一路」作為切入點，可望成為連結國家新型工業與全球市場的「超級增值人」。

香港可成為多極化世界的可信平台

國際政經格局已由單極轉向多極化結構。從「美國優先」到「新華盛頓共識」，再到特朗普向全球徵稅，美國的保護主義傾向日益顯著。不少企業為應對風險，採取分散生產佈局的「中國+1」策略，以及避免依賴單一消費市場的「美國+1」策略。這種資源再分配的趨勢，顯著促進了東盟等新興經濟體的增長。

與此同時，中國在人工智能、先進製造及新能源等領域，已展現出舉足輕重的影響力。美國的壓制政策不但未能遏制中國崛起為工業強國，反而倒逼其供應鏈透過技術創新與積極出海，穩健邁向高端化及國際化。

在國際合作層面，全球北方繼續主導世界銀行、國際貨幣基金組織等傳統國際組織; 而全球南方則積極發展新興組織，例如「金磚+」、上海合作組織，以及亞洲基礎設施投資銀行等。

在經貿格局碎片化下，香港憑藉一國兩制基礎，既了解社會主義，亦熟悉資本主義，其服務價值更顯珍貴。作為中國境內的一個特別行政區，香港實行與國際商業範式高度接軌的普通法制度，加上多年來積累的商業聲譽、經貿網絡，以及穩定的營商環境，不難成為備受各方信賴的合作平台。

建基於創科及國際化實力的新路向

從國家層面看，近年內地經濟轉向高質量發展，為香港開拓新增長曲線帶來重大機遇。2025年，中國全社會研發經費接近 4 萬億元，穩居世界第二；有效發明專利突破500萬件；根據《專利合作條約》（Patent Cooperation Treaty；簡稱PCT）提出的國際專利申請量連冠6年。與此同時，規模以上高技術製造業、裝備製造業、數字產品製造業及清潔能源發電量均錄得約 9%的高增長。由此可見，以創新、數字、綠色為特徵的「新質生產力」正驅動國內經濟增長。

對此，香港可增添助力。一方面，透過建設國際創科中心，匯集國際高端創新要素，在攻克關鍵核心技術方面提供增值。另一方面，善用金融業優勢，建設區域知識產權貿易中心，服務國家提高知識產權變現能力，並為國家科技和綠色產業募集國際資金。這些舉措均有利於拓展本港高增值服務業的寬度。

當前國家正積極構建以國內大循環為主體、國內國際雙循環相互促進的新發展格局。2025 年，全國社會消費品零售總額突破 50 萬億元，服務性消費支出已佔居民人均消費支出逾 46%。作為全球貨物貿易第一大國，進出口總值約達 45.5 萬億元，與「一帶一路」國家的貿易佔比已過半。中國高技術產品的出口價值更錄得逾 13%的高增長，已佔出口總額近 20%，其中「新三樣」(新能源汽車、鋰電池、光伏) 的出口大增約 27%。

國家「十五五」規劃提出要建設現代化產業體系和強大國內市場，並擴大高水平對外開放。重點工作方針包括促進服務業優質高效發展、大力提振消費、擴大有效投資，以及提升貿易投資合作質量和水平等。香港擁有一國兩制與國際化獨特性，相信其服務業大有可為。一方面，內地消費升級需要國際領先的品質標準對接服務；另一方面，國家不斷增長的進出口貿易，有利於鞏固香港作為國際轉口港的地位。除此之外，人民幣國際化也需要一個便捷、值得信賴的平台。國內綠色商品走向世界是大勢所趨，香港在環境、社會及管治（Environmental, Social, and Governance；簡稱ESG ）認證與綠色金融方面的優勢將會得到充分發揮。

推動「超級增值人」服務三重升級

面對環球碎片化經貿格局，香港在「十五五」時期須強化「超級增值人」角色，成為助力國家發展的賦能平台。為此，金融服務的升級是首要切入點。離岸人民幣業務不能僅停留在資金儲備功能，更應滿足實體經濟的投融資需求。透過豐富人民幣產品、將人民幣櫃台納入「港股通」及增發「點心債」等舉措，香港將有效提升離岸人民幣的流動性與收益率，擴大作為全球離岸人民幣業務樞紐的功能。

在深度融入大灣區產業鏈的過程中，香港可擔任賦能者角色。例如在上游建設國際科研人才高地，聚焦攻克「卡脖子」技術難題；在中游與下游，則憑藉國際化檢測認證服務和經貿網絡，助力中企佈局全球。在配套服務方面，從知識產權保障與商品化、品牌建設、市場營銷到貿易融資等，香港均可為灣區產業鏈增值。大學作為這一進程中的創新引擎，透過為研究人員和學生提供優質的學術環境，孕育具備國際化視野的專才。這些專才在社會中創造產業價值，將基礎研究成果轉化為優質產品，將複雜的國際法規轉化為合規流程，將ESG知識轉化為企業的全球聲譽。

香港之所以能發揮「引資引智」作用，實有賴於其在金融和高等教育等領域的顯著優勢。與國際接軌的金融體系，加上為全球資本提供可預期的投資環境，無疑令香港成為國際資金配置中國優質資產的首選地。而其蜚聲國際的高等教育，則是吸引全球頂尖人才的引力場。本地大學作為知識創新和人才培養的樞紐，能夠在經管、創科、醫療及法律等領域，擔當國際優質資源與國家需求之間的接口，最終可協助提升內地商品及服務的質量。

隨着中國經濟與科技實力與日俱增，中企走向世界雖是大勢所趨，但在全球碎片化時代中，也必然面臨更多挑戰，特別是文化與制度之間的隔閡。一國兩制下的香港，既在法律制度、商業慣例及消費者習慣等各方面與海外市場相似，又與內地關係緊密，足以成為中企出海的理想試點。中企不但可在香港試行商業模式，還可積累跨境運營與合規經驗，為出海增添信心。

近年香港各界在中央政府的鼎力支持下，均積極謀求新發展。筆者相信，香港定能完成區域（由全球北方延伸至全球南方）、功能（由聯繫窗口發展成賦能平台）及產業（由金融拓展至創科與綠色經濟）的三重升級【註】，將時代賦予的發展機遇轉化為實實在在的經濟增長，為廣大市民創造優質的就業機會。

【註】：筆者鄧希煒於《紫荊論壇》2026年3月27日發表《對接「十五五」 善用全球大變局促進香港第三次經濟轉型》一文中，首次提出此一概念。

鄧希煒教授
香港大學協理副校長、港大經管學院副院長、馮國經馮國綸基金經濟學教授

張超藝先生
香港大學香港經濟及商業策略研究所高級研究助理

（本文同時於二零二六年四月二十二日載於《信報》「龍虎山下」專欄）

Dr Yifei Zhang

15 April 2026

On 20 February 2026, the US Supreme Court, in a 6–3 decision, issued a landmark ruling in Learning Resources, Inc. v. Trump: The President is not authorized under the International Emergency Economic Powers Act (IEEPA) to unilaterally impose tariffs. As Chief Justice John Roberts points out in the ruling, the IEEPA, which authorizes the US President to “regulate…importation”, separated by 16 other words, cannot bear the weight to justify an independent power to impose tariffs on imports from any country, of any product, at any rate, for any amount of time.

This ruling has not only drawn a legal red line but has also triggered a chain reaction across the economy and financial markets, affecting places from Washington to Hong Kong’s Central district.

Taxing power prerogative of Congress

This ruling has far-reaching implications because Article 1 of the US Constitution clearly grants Congress the “power to lay and collect taxes, duties, imposts, and excises”. Applying the principle of separation of powers, the Supreme Court ruled that tariff powers with significant economic and political implications must be explicitly authorized by Congress. As public choice theory in economics suggests, when the power of formulating policies is overly concentrated in the executive branch, policies are liable to be twisted by short-term political considerations. The past year has seen companies under pressure to make hiring, pricing, and investment decisions against the backdrop of fast-changing trade policies and constantly fluctuating tariff rates. To a certain extent, the economic damage caused by such uncertainty is even greater than that caused by the tariffs.

The astronomical tariff-refund impasse

The direct consequence of the ruling is a tariff-refund challenge of unprecedented scale. According to a statement submitted by the US Customs and Border Protection to the Court of International Trade, under the IEEPA tariffs, more than 330,000 importers have paid a total of US$166 billion in tariffs, involving approximately 53 million customs entries. The path to refunds is fraught with obstacles: the automated systems of US customs are not designed to readily isolate IEEPA tariffs. If the amounts were to be calculated manually, the process is estimated to require 4.4 million labour hours. What underlies this is an important financial concept: the time value of money. Suppose a company has paid US$10 million in IEEPA tariffs. At current US interest rates, each one-year delay in receiving a refund would impose an opportunity cost as high as approximately US$500,000. Hence, nearly 2,000 importers have already filed lawsuits to secure their right to tariff refunds.

From plan B back to plan A

The White House did not just sit idly by. On the day the ruling was made, Trump signed three executive orders in a row within a few hours: to terminate IEEPA tariffs, to impose a global temporary import surcharge under Section 122 of the Trade Act of 1974, and to continue suspending duty-free treatment for low-value imports. The following day saw him raise the surcharge to the statutory maximum of 15% under Section 122. This provision is, figuratively speaking, a time-locked gun which allows the US president to impose an import surcharge of up to 15% for no more than 150 days, after which any extension must be approved by a vote in Congress, with the deadline falling on 24 July this year.

Even more noteworthy is the long game plan. In March this year, US trade representative Jamieson Greer launched Section 301 investigations against 16 economies on the grounds of “overcapacity”, and also initiated Section 301 investigations related to “forced labour” against approximately 60 economies, altogether covering 80% to over 90% of imports. Deborah Elms, Head of Trade Policy at the Hinrich Foundation in Singapore, hit the nail on the head when she said, “We may well be back to square one. This is not plan C. This is back to plan A”.

Who is paying for the tariffs

Economics textbooks tell us that the cost of tariffs is shared between producers and consumers, but the reality is grimmer than the theory. Research by the Federal Reserve Bank of New York finds that close to 90% of tariff costs are borne by American companies and consumers. Tracking data of the Yale Budget Lab shows that the pass-through rate of tariffs to the prices of imported core consumer goods is between 40% and 76%, while estimates by Goldman Sachs are even more alarming: the share shouldered by consumers, which has already reached 55%, is expected to climb to 70% in 2026.

For example, the ex-factory price of a children’s toy imported from China is US$20. With the addition of transportation and distribution costs, its retail price in America is US$30. With the imposition of a tariff of 20%, the import cost rises by US$4. Nevertheless, retail prices will not rise by only that amount since they also include non-import components such as domestic transportation, marketing, and retail markups. Particularly worrying is the fact that the average price increase of about 5% for cheaper goods is twice that of high-end goods. Given that low-income families tend to buy cheaper goods, tariffs are in fact a regressive tax, placing a disproportionate burden on the underprivileged.

Shock waves spreading to Hong Kong

Not even Hong Kong is immune to this storm. The US is Hong Kong’s second-largest export market, with goods exports amounting to US$37.9 billion in 2024. Although Hong Kong retains its status as a separate customs territory under the Basic Law, US tariffs imposed on China are still directly applicable to Hong Kong.

The SAR Government has long upheld a free-port policy, and does not impose retaliatory tariffs, reflecting its strategy of safeguarding the city’s appeal as an international commercial hub. However, for cross-border e-commerce platforms focused chiefly on the North American market, eliminating de minimis duty-free treatment and imposing surcharges will directly affect the cost structure of each order. Steve Chuang, former Chairman of the Federation of Hong Kong Industries, once made it plain—amid US tariff flip-flops, no one can do business anymore.

From the perspective of asset allocation, the uncertainty of trade policy has become a crucial risk-premium factor affecting Hong Kong stock valuations. In times of escalating trade friction, the price-to-earnings ratios of export-oriented companies in the Hang Seng Index generally come under pressure. When valuating relevant stocks, investors should add a “policy uncertainty premium” to the discount rate.

Future developments and countermeasures

Looking ahead, two key issues deserve careful attention. The first is the expiry of temporary tariffs under Section 122 on 24 July this year. Whether the US Congress votes to extend them will be an important bellwether for the trade positions of the two major parties. The second is the progress of the Section 301 investigations. If US trade partners are found to be involved in unfair practices, a new round of country-specific tariffs may be launched in late summer to early autumn this year.

In view of the current situation, I would like to propose the following three policy recommendations. First, Hong Kong should expedite the diversification of its export markets and strengthen efforts to tap markets in ASEAN, the Middle East, and Latin America. Second, it should reinforce its role as a “super-connector”. By leveraging the advantages of its independent judiciary system, free flow of capital, and its international arbitration centre, the city can become a key node for multinational companies in the course of reconfiguring their supply chains. Third, Hong Kong should capitalize on financial instruments to offset trade risks. Its derivatives market, together with insurance and reinsurance services, can provide tailor-made risk management solutions for companies facing tariff risks.

The US Supreme Court’s ruling discussed above reminds us that, in a society governed by the rule of law, even the strongest executive power has its limits. As Chief Justice John Roberts noted, citing an 1899 precedent, the power to tax is “the one great power upon which the whole national fabric is based”. This power should be subject to institutional checks and balances, so that the market can form stable expectations and the economy can operate robustly within a rules-based framework. Long grounded in rule of law and institutional transparency, Hong Kong is likewise committed to this principle.

Dr Jinghan Meng

11 April 2026

If, in the future, the AI agents of various funds in Central, Hong Kong can finish reading central bank statements, earnings call transcripts, and global news within the same second, and make similar directional trading decisions accordingly, will the market become more efficient or more vulnerable? This is no science-fiction fantasy but a reality fast approaching the financial industry.

At this stage, generative artificial intelligence (GenAI) is mainly used in summarizing study reports, collating information, writing codes, and analysing texts. However, when these capabilities are further integrated with memory, reasoning, planning, tool utilization, and continuous execution modules, and embedded into investment research, risk management, and trading processes, financial institutions will move from “signal generation” to more autonomous agentic AI systems. (see Notes 1, 6, and 8)

“Alpha anxiety” in the age of AI

This change has given rise to the phenomenon of “alpha anxiety”. Alpha refers to excess returns. The anxiety is not so much about whether AI is smarter than people, but about the fact that AI is rapidly reducing the cost of replicating some investment research tasks. The tasks of organizing public information, interpreting texts, tracking holdings, and identifying styles, which used to take research teams a long time to complete, can now increasingly be automated by models (see Note 4). Some studies suggest that, using only publicly disclosed holdings and macroeconomic data, AI can already mimic a considerable proportion of top asset managers’ trading behaviour. (see Note 2)

As more institutions can extract similar signals from similar data, the previously scarce informational edge becomes more susceptible to competitive erosion, and alpha also becomes harder to sustain. AI may not necessarily put an end to alpha but it is indeed shortening alpha’s half-life, forcing institutions to invest in more computing power, more expensive data, and more complex architectures just to maintain existing returns. This is a classic example of Red Queen competition.

Algorithmic convergence and systemic fragility

A greater cause for concern than a single model committing errors is a large number of models simultaneously “getting the same thing right”. When different institutions rely on similar foundation models, similar news sources, the same market data, similar risk constraints, and similar optimization objectives, they may appear to be competing with one another, but in moments of stress, may converge towards highly similar trading responses. On the one hand, GenAI reduces information asymmetry by rapidly transforming unstructured information previously scattered across text, speech, and narratives into tradable signals. On the other hand, it may also enable market participants to arrive at more similar judgments in a shorter time, thus increasing the risk of strategy convergence and model homogenization. (see Notes 1, 6, and 7)

Under normal market conditions, such technological advancement can help to expedite price discovery. However, once markets face pressure, their procyclicality could also intensify. If more and more model-driven traders interpret central bank language, earnings guidance, and macroeconomic data using similar logic, and synchronously adjust positions under similar stop-loss rules, margin requirements, and value-at-risk limits, market liquidity could contract simultaneously within a short time. Existing research shows that while algorithmic trading, in normal circumstances, can accelerate price discovery and the incorporation of information into prices, under stress scenarios, it could exacerbate the vulnerability of liquidity. If different strategies rely more heavily on similar signals and similar execution rules, the improvement in market efficiency may come at the cost of a more fragile market microstructure. This is exactly a paradox of AI finance: rational optimization at the micro level may not necessarily lead to stability at the macro level. (see Notes 3 and 7)

From chasing alpha to pursuing “responsible alpha”

That is exactly why I propose the concept of “responsible alpha”. In the future, valuable alpha should not just outperform the market in the short term but should also be kept within boundaries that are explainable, auditable, and open to intervention, without unduly amplifying systemic risk. In other words, instead of treating risk control as an add-on, “responsible alpha” internalizes governability as part of alpha. With the increasing commoditization of signal extraction, the truly scarce capability will not simply be building ever more opaque black boxes, but demonstrating why one’s AI is trustworthy: what data it uses, what workflow it follows, who can review it, and who can put a stop to it if anything goes wrong. When alpha comes to resemble a replicable public technology, governance capability, auditing capability, and human intervention capability, it may instead become a new private moat. (see Note 5)

Hong Kong’s policy approach in the past couple of years has provided the institutional groundwork for such “responsible alpha”. In August 2024, the Hong Kong Monetary Authority (HKMA), in conjunction with Cyberport, launched the GenA.I. Sandbox, clearly setting out a risk-based approach and emphasizing that high-risk decision-making must retain a human-in-the-loop model. In August 2025, the Bank for International Settlements (BIS) Innovation Hub, Hong Kong Centre, the HKMA, and the UK Financial Conduct Authority launched Project Noor, focusing on the application of GenAI and advanced algorithmic models in the financial system to address the AI explainability problem. This means that the regulatory approach no longer merely requires institutions to explain AI, but is beginning to enhance supervisors’ own ability to understand black boxes. In March 2026, Hong Kong upgraded the sandbox to GenA.I. Sandbox++, extending coverage to securities, asset management, insurance, MPF, and other areas. As a result, the sandbox is no longer a testing ground but also an institutional project through which regulators and the market jointly define safety parameters.

Of course, the launch of the sandbox cannot simply be a showcase for innovation but is meant to be a testing ground for governance capability. In terms of agentic AI and quantitative models, what is really being tested is not model accuracy but the entire risk management chain: whether the data is traceable, whether version updates leave an audit trail, whether orders will be cancelled simultaneously under stress scenarios, whether model drift can be detected in time, and whether human intervention or emergency shutdown is possible when necessary. Future competition in the asset management industry will hinge on institutions’ ability to incorporate model risk into their governance frameworks, rather than on model predictive capability alone.

Aiming to enhance the trustworthiness of alpha

The next round of financial competition will be about far more than who adopts AI first; it will be about who can prove sooner that their AI is reliable when making money and controllable when it fails. A World Economic Forum white paper released in 2025 notes that many institutions are in fact still in the transitional stage from experimentation to scaled implementation. What truly determines success or failure is often not the model itself, but whether trust, self-governance, talent, cybersecurity, and digital infrastructure are in place. (see Note 8) From this perspective, Hong Kong’s potential advantage lies not only in deploying AI earlier, but in institutionalizing model governance, stress testing, audit trails, and human intervention mechanisms earlier, gradually turning them into a common language across markets, regulators, and institutions.

In the final analysis, if markets ultimately regard transparency, accountability, and governance maturity as attributes worth paying for, then “trust” itself may become the most important intangible asset of the next generation of international financial centres. What Hong Kong should pursue is not higher alpha, but alpha that global capital can trust.

Note 1：Aldridge, I., An, J., Burke, R., Cao, M., Chien, C. Y., Deng, K., … & Zheng, W. (2025). Agentic artificial intelligence in finance: A comprehensive survey [Working paper].

Note 2：Cohen, L., Lu, Y., & Nguyen, Q. H. (2026). Mimicking finance (NBER Working Paper No. 34849). National Bureau of Economic Research.

Note 3：Dou, W. W., Goldstein, I., & Ji, Y. (2025). AI-powered trading, algorithmic collusion, and price efficiency (NBER Working Paper No. 34054). National Bureau of Economic Research.

Note 4：Gu, S., Kelly, B., & Xiu, D. (2020). Empirical asset pricing via machine learning. The Review of Financial Studies, 33(5), 2223–2273.

Note 5：Fabozzi, F. A., & López de Prado, M. (2025). Implementing AI Foundation Models in Asset Management: A Practical Guide. Journal of Portfolio Management, 52(2).

Note 6：Mo, H., & Ouyang, S. (2025). (Generative) AI in financial economics. Journal of Chinese Economic and Business Studies, 23(4), 509–587.

Note 7：International Monetary Fund. (2024). Global financial stability report. https://www.imf.org/-/media/files/publications/gfsr/2024/october/english/textrevised.pdf

Note 8：World Economic Forum. (2025). AI in action: Beyond experimentation to transform industry. AI Governance Alliance, in collaboration with Accenture.

Dr Yim-fai Luk

1 April 2026

Tomorrow marks the first anniversary of the reciprocal tariff war initiated by Donald Trump against the world. On 2 April 2025, he swaggeringly announced in the White House Rose Garden that the day was America’s “Liberation Day”, claiming that after the US had been ripped off by other countries through trade over the years, it was time for the country to strike back with reciprocal tariffs. The use of the word “reciprocal” is more in the sense of tit-for-tat than of “equal” tariff rates. He had also come up with tariff rates for all economies, including the uninhabited Heard and McDonald Islands near Antarctica. For the better part of 2025, tariff negotiations between the US and other economies, particularly China, dominated the international economic agenda.

Trump is utterly convinced of the power of tariffs, regarding them as an “ultimate weapon” against other countries. To him, reciprocal tariffs form a cornerstone of his economic and diplomatic policies while in office. Apart from bragging about them on social media, he uncharacteristically published an article entitled “My Tariffs Have Brought America Back” (see Note 1) in Wall Street Journal on 30 January 2026. This appears to have been the only time he expressed his views in the print media during his presidency. While the piece is typical braggadocio and does not hold up to fact-checking, it also reflects the importance of reciprocal tariffs to him. After all, the last such large-scale tariff policy dates back to 1930, when the Smoot-Hawley Tariff Act was introduced in the US. The then US president Herbert Hoover raised tariffs on 20,000 imported goods, prompting other countries to retaliate in kind. As a result of the trade clash, global trade shrank by two-thirds within just a few years, deepening the woes of the Great Depression. Given such a cautionary precedent, why still impose reciprocal tariffs? Below are several of the more commonly cited motivations.

The first is the above-mentioned goal of eliminating the trade deficit. America’s trade deficit already reached US$0.91 trillion in 2025. In comparison with that at US$0.9 trillion in 2024, the effect of the reciprocal tariffs implemented in the intervening eight months is not evident at all. However, this figure covers trade in goods and services. Over the years, the US has run a surplus in services trade, which Trump obviously does not regard as a consequence of America ripping other countries off. Given that tariffs do not apply to services trade, looking only at goods trade, America’s trade deficit widened, instead of falling, from US$1.2 trillion in 2024 to US$1.23 trillion in 2025. This may have stemmed from greatly increased imports by many companies before Liberation Day to build up inventories, but even so, large-scale tariff hikes targeting various economies cannot offset America’s goods trade deficit. From a macro perspective, at the end of the day, the trade deficit is the result of excessive nationwide overconsumption, including spending by the government, companies, and households, as well as overconsumption of both domestic and foreign products. Tariffs have little impact on this total consumption, especially government spending. Over the 50 years from 1976 to 2025, the US federal government expenditure averaged 21.2% of GDP, rising to 23.1% in 2025. In 2026, the estimate is 23.3% and the estimate for 2036 is projected to be 24.4% of GDP. During his second term as president, Trump once placed Elon Musk in charge of the Department of Government Efficiency (DOGE) to cut government spending, but the department vanished in less than a year.

Another policy objective is to reinvigorate US manufacturing. The decline of America’s manufacturing industry, such as the downturn in shipbuilding, could compromise national security. However, tariffs are not a viable solution to these problems. First of all, US manufacturing is quite reliant on imported semi-finished goods, which account on average for approximately 30% of output value and in some industries, e.g. pharmaceuticals, the figure exceeds 50%. Tariffs increase the price of such semi-finished goods imported by American manufacturers, thereby pushing up production costs. Originally intended to protect domestic industries, tariffs end up inflicting self-harm as well. Slightly over 20% of America’s imports of semi-finished goods come from China. One can imagine the impact on American manufacturers if tariff rates are at the 125% level seen before. Apart from China’s restrictions on rare-earth exports, this is probably another reason why Trump has backed down to China.

Another related issue is employment in manufacturing. That tariffs can enable American workers to reclaim jobs from foreign workers has been a political and economic myth since Trump’s first term in office, which has won him a large number of votes. Nevertheless, employment data shows that instead of rising as a result of reciprocal tariffs, the number of Americans employed in manufacturing slightly dropped from 12.66 million in April 2025 to 12.53 million in February 2026. Manufacturing output, by contrast, surged during the same period. The higher output generated by a smaller labour force reflects an enhancement of labour productivity. Apart from the loss of competitiveness to emerging economies, the continued fall in the number of American manufacturing workers can mainly be attributed to advances in production technology that have replaced human labour. Even without accounting for technological innovation, for increased tariffs to offset the wage gap between American manufacturing workers and their counterparts in developing countries, tariff rates might have to reach an unrealistic several hundred percentage points, making such a move economically counterproductive.

Another objective of reciprocal tariffs is to generate fiscal revenue, which explains why Trump has cast a wide tariff net covering all corners of the world, including the UK and Australia, with which America has a trade surplus. According to data from the Peterson Institute for International Economics (see Note 2), in the 13 months between January 2025 and January 2026, America’s tariff revenue was US$290 billion, while the US federal government’s fiscal deficit was US$3.72 trillion. The US government has been in the red for years, with the fiscal deficit growing faster than the economy, a trend that tariffs are far from reversing. Also related to this issue is whether the tariff revenue comes from foreign countries or the US itself. While Trump has repeatedly emphasized that the tariffs are paid by foreign exporters, a research study by Harvard University has calculated that approximately 94% of the tariffs are paid by US importers and consumers (see Note 3). A survey conducted at the end of February 2026 (see Note 4) shows that 70% of Americans think the tariffs have pushed up their living costs, and even 64% of Republican voters think likewise. Under the influence of the ongoing war in the Middle East, prices are unlikely to come down this year, and the conflict is a crisis of Trump’s making. Inflation was the main reason why Trump and the Republicans won a landslide victory over the Democrats in the presidential election two years ago. There is no telling how many seats in the Congress the Republicans will lose in the midterm elections at the end of this year.

For Trump, tariffs are an effective tool he can use on a whim to coerce foreign countries to satisfy his demands. In his article published in the aforementioned Wall Street Journal, Trump boasts that by threatening other countries with tariffs, he has secured US$18 trillion worth of foreign investment for America in less than a year. The figure is an exaggeration, but in tariff negotiations, some countries, such as Japan and South Korea, have indeed pledged to invest in the US in exchange for tariff reductions. However, few related reports have gone into detail, e.g. how and when the pledged foreign investment will materialize, and whether it will be carried out through governments or private enterprises. It seems some countries may be stringing Trump along. But what matters more is that international investment falls under the capital account. In balance-of-payments accounting, if the central bank does not intervene in the foreign exchange market, the sum of the capital account and the current account should equal zero. When foreign nations invest in the US, the US records a capital-account surplus, which is reflected in the current account as a deficit. Since trade is the main component of the current account, a current-account deficit is most likely to correspond to a trade deficit. Think of it this way: for foreign countries to invest in the US, they must first have US dollars. Where do those dollars come from? Ultimately, they have to come from exports to the US, i.e. from the US trade deficit. In other words, by coercing foreign countries to invest in the US, Trump will end up creating a trade deficit for the country. This directly contradicts his aim of reducing the US trade deficit.

On 20 February 2026, the US Supreme Court ruled that the US president had no authority to invoke the International Emergency Economic Powers Act to impose tariffs, greatly limiting his room to wield the tariff stick. A few days later, Trump cited Section 122 of the Trade Act of 1974 as the legal basis for continuing to levy tariffs averaging about 13%, which is lower than the previous average of 16%, but subject to a 150-day deadline.

All in all, Trump’s reciprocal tariffs arrived with serious menace but ended up causing more thunder than rain. Their impact is way less than he expected or fantasized. With the exception of China, America’s other major trading partners have neither the capacity nor strategic judgment to retaliate against the US. Therefore, there has been no tit-for-tat tariff escalation like that of the 1930s, and the global economy has not sunk into a quagmire as a result of the reciprocal tariffs. According to World Bank estimates, the global economy grew by 2.7% in 2025, only slightly lower than the 2.8% recorded in both 2023 and 2024. Even so, the reciprocal tariffs have clearly highlighted US unilateralism, which has eroded the country’s international credibility, while further exposing the ineffectiveness of international trade rules and the World Trade Organization (WTO). At the WTO’s 14th Ministerial Conference, held in Cameroon last week, many members expressed a strong desire to reform the WTO. Yet due to divided opinions and a lack of clear leadership, no consensus was reached. Returning to the current situation in the Middle East, the world has now entered a period of great upheaval before fundamental reconstruction can begin. One can only hope that this turbulent period will soon give way to renewal and better times.

Note 1: https://www.wsj.com/opinion/donald-j-trump-my-tariffs-have-brought-america-back-2248391b

Note 2: https://www.piie.com/research/piie-charts/2025/trumps-tariff-revenue-tracker-how-much-us-collecting-which-imports-are

Note 3: https://gopinath.scholars.harvard.edu/publication/incidence-tariffs-rates-and-reality-0

Note 4: https://www.theguardian.com/us-news/2026/mar/13/trump-tariffs-poll

Dr Tingting Fan

25 March 2026

Night King, a Lunar New Year film that opened in Hong Kong on 17 February 2026—the first day of the Chinese New Year (CNY)—achieved strong popularity and financial success within just a month. With over $90 million in local box-office takings alone, the film became not only the opening box-office champion among CNY films in Hong Kong history but also the fifth highest-grossing Hong Kong-produced film.

Given its humorous style, eye-catching theme, and a gradual word-of-mouth effect, the film went on 23 February from local release in the Guangdong–Guangxi region as originally scheduled to nationwide release, with Mainland box-office takings in excess of RMB150 million so far.

Surprising breakthrough in a sluggish market

The success of Night King is worth celebrating not just because its box-office takings have far exceeded its production costs and publicity expenses, but because its outstanding performance in a generally soft yet extremely competitive market is especially commendable. With the popularity of streaming platforms such as Netflix, Disney+, consumers around the world are increasingly staying away from cinemas. Based on the total cinema admissions in 2024 and the mid-year local population estimate, Hongkongers went to the cinema to watch only 2.5 films on average during the year. Similarly, across the ocean, Americans went to the movies only 2.2 times on average in 2025. How to ensure that a film is one of the two consumers’ pick to watch in cinemas is a thorny challenge facing the film industry worldwide.

On the one hand, there is persistently lacklustre market demand; on the other hand, competition within the industry has escalated to white-hot levels. Besides The Snowball on a Sunny Day, another Hong Kong production, the Lunar New Year films released at the same time as Night King included Blades of the Guardians, Pegasus 3, and Scare Out from the Mainland. These rivals came on fast and furious. Not only did they have more funding and wider cinema release, but their star-studded casts were far more impressive compared with Night King. As for other film markets in the world, the top box-office hits in 2025 were almost exclusively built around well-known intellectual property (IP) characters, such as Zootopia 2, A Minecraft Movie, and Wicked: For Good.

With neither the support of top-tier IPs nor huge capital investment, the unexpected success of Night King has inevitably rekindled hopes for the local film industry, which have seen a decline in recent years. From the 1980s to the 1990s, Hong Kong was once the world’s third largest film production centre after Hollywood and Bollywood. However, with intensifying market competition, the rise of streaming platforms and shifting audience tastes, the Hong Kong film industry has since gone downhill.

Reflecting on the present in light of the past, what lessons can film companies seeking to mount a comeback draw from this history?

Former glory as a lesson for today

At the turn of the 1960s, Hong Kong was already an important film export centre in East Asia. During the golden era of its film industry between the 1980s and the 1990s, a series of landmark films were produced, including A Better Tomorrow, The Killer, Police Story, Chungking Express, Happy Together, and the “mo lei tau” (nonsensical humour) series represented by Stephen Chow. From 1982 onward, Hong Kong-produced films fared better than imported films at the box office for the first time. With annual film production reaching as many as 186 in 1993, the impact of Hong Kong films rose sharply as a result.

The extraordinary achievements of Hong Kong films back then can be put down to their high degree of commercialization, the synergy between star actors and renowned directors, and the one-of-a-kind culture of the city. On the one hand, through highly efficient coordination across production, distribution, and exhibition, coupled with a well-established Southeast Asian distribution network, the local film sector facilitated standardized production and rapid iteration of film genres, making it possible to create economies of scale while satisfying the diverse tastes of consumers.

On the other hand, a high degree of commercialization also contributed to talent cultivation. The actor training system during the golden era produced award-winning stars such as Chow Yun-fat, Stephen Chow, Maggie Cheung, and Tony Leung. Under the aegis of top directors such as Wong Kar-wai, Ann Hui, John Woo, and Tsui Hark, local productions not only grew in number but also improved in quality. Celebrity charisma broadened the local films’ social appeal, while big-name directors enhanced their recognition at international film festivals. Under this twin effect, Hong Kong films have enjoyed growing cachet on the world stage. With the city’s unique history and immigrant urban culture, combined with the popularity of Cantonese slang, local films contained a perfect mix of “Eastern expression” and “Western narrative” to strike a responsive chord with Chinese communities worldwide and gain unimpeded access to European film festivals and the North American market.

The nightclub as workplace: turning weakness into strength

Although the glory days of Hong Kong’s film industry have become a thing of the past, the box-office performance of Night King may give us a glimpse of a way out for Hong Kong films. First of all, the script is captivating, particularly for Mainland audiences: the nightclub story is so Hong Kong. However, how best to make a respectable show out of such colourful material? Careful inspection of the plot would reveal that, as a matter of fact, business management in a nightclub is not so different from that in a corporate workplace. Be it meeting performance expectations of the boss, motivating employees, or offering personalized customer service, even viewers who have never been to a nightclub will understand it right away.

What is more, as Night King is a Cantonese film, many of its memorable lines can only be fully expressed through Cantonese slang and colloquial expressions. For example, only audience members who are familiar with Hong Kong place names will laugh out loud at the inside joke about “Kwai Fong”. To a certain extent, the dialect and local culture may have hampered the film’s wider popularity. After all, most Mainland viewers do not speak Cantonese and are unfamiliar with most local place names. Outside its cultural context, Night King may well lose much of its charm. That is exactly one of the reasons why the film was originally scheduled for release only in Hong Kong, Macau, and the Guangdong and Guangxi region.

To everyone’s surprise, with an ingenious twist, the film managed to turn this disadvantage into an advantage. Through publicity on social media, the memorable lines had been deciphered for the public before viewers even bought their tickets, giving them a sense of the creative ingenuity of Cantonese slang expressions and the vibrancy and humour of Hong Kong culture. The line spoken by the club hostess Kwai Fong: “Zero to four, round down; five to nine, round up… So I’m 19 years old” instantly made Kwai Fong MTR station a social-media hotspot. This also reminds us of how the big tree at the Shek O Health Centre and the gas lamps on Duddell Street gained instant fame thanks to Stephen Chow’s King of Comedy.

As Brother Foon in Night King says, “Life may be difficult these days, but we’ve never been afraid. Things have never been easy for us anyway, right? But if the lights must go out, we’ll drink till the very end.” The world of Hong Kong films is no less difficult. Yet I am convinced that the lights will never go out for films with Hong Kong characteristics. The tougher the journey, the harder we push on! This is what the Hong Kong spirit is all about.

Professor Yulin Fang

18 March 2026

The past few years have seen what can almost be described as the meteoric rise of generative artificial intelligence (AI). From writing, design, application development to customer service, marketing, financial analysis, and medical assistance, more and more companies have introduced generative AI into at least one part of their operations. Individuals and companies alike have also begun to use such tools for tasks that used to rely heavily on human expertise, including creative design, image processing, and content generation.

AI development to prioritize human well-being

As technology permeates daily life at such a rapid pace, we need to ask a more fundamental question: Where do we hope AI will lead us? If it merely pursues efficiency and cost optimization, then AI is, after all, just a tool. However, if it can stimulate the growth of human capabilities, enhance organizational resilience, and uphold human values and dignity, then it can truly become a force for the advancement of civilization. This is exactly the core focus of human-centred AI.

A human-centred approach does not mean opposing automation, nor does it reject technological innovation. Rather, it emphasizes that the design, deployment, and governance of AI should put human development at the core. Why is this concept so important? The reason is that generative AI is reshaping the nature of knowledge work. In the past, automation mainly replaced repetitive and routine tasks. Today, generative AI is beginning to enter non-routine areas, covering creative design, strategic analysis, and decision support.

Even as corporate productivity has increased significantly, new causes for concern have arisen. Will employees experience de-skilling as a result of over-reliance on AI? Will organizations gradually lose their own knowledge assets? Will humans face the threat of being replaced in their collaboration with AI? Once AI begins to encroach on the realms of cognition and creativity, its impact on humanity will no longer be limited to efficiency alone, but will be a matter of the structure of abilities and a sense of self-worth.

The Institute of Digital Economy and Innovation (IDEI) at HKU Business School and its Human-Artificial Intelligence (HAI) Lab propose a clear governance perspective—in the process of AI-driven organizational innovation, to be truly “human-centred”, the focus must extend beyond financial returns and technological maturity to the following four key areas: performance, learning, creativity, and dignity.

Labelling functions key to performance improvement

Research conducted by the IDEI-HAI shows that in certain sales scenarios, even a generative AI pre-sales system that has yet to integrate leading best practices can still enhance overall performance. It is noteworthy that this improvement mainly stems from enhanced performance among low- to mid-performing employees.

For high-performing employees, early versions of AI may actually have a suppressing effect in that the system’s standardized suggestions limit their original professional judgment and innovative capabilities. Only after the system is upgraded and labelling functions added, allowing high-performing employees to further optimize AI’s suggestions, can overall performance be enhanced across the board. This finding shows that design details determine who benefits and who is disadvantaged. Without prudent governance, AI may widen gaps rather than narrow them.

Will humans’ capacity to learn be undermined

There are two systems of human cognition. System One is intuitive, fast, and automatic. System Two is analytical, reflective, and requires cognitive resources. Generative AI tends to reinforce the rapid-response mode of System One, making tasks easier. However, long-term reliance on such automated assistance could reduce humans’ deep thinking, resulting in a decline in their abilities. Such a de-skilling effect may give rise to two risks—increased likelihood of employees being replaced in the labour market and organizations’ gradual loss of their accumulated knowledge, leading to organizational forgetting.

HAI Lab’s proposed solution is not to reject AI, but to optimize its design, e.g. by using “deliberate reflection” prompts to guide users to think through the logic behind the machine’s reasoning; conducting post-task reflection to consolidate the experience of human-AI collaboration; and adopting a “human-in-the-loop” approach to ensure that human judgment continues to play a central role.

As a matter of fact, if designed properly, AI will not weaken learning. If anything, it may well be beneficial to knowledge retention.

Human-AI collaboration as a catalyst for creativity

In its research on the arts and creative industries, IDEI has found that digital artists’ use of generative AI can increase the market value of their original works by improving aesthetic quality, strengthening incremental innovation, boosting content creation capabilities, and building reputation within creative communities. However, since such gains are often incremental in nature, they do not necessarily produce disruptive breakthroughs.

At the team level, generative AI enhances creativity through two mechanisms: focus weaving, which helps to stabilize and deepen existing ideas, and gap spotting, which identifies blind spots and extend lines of thinking.

This effect is particularly pronounced in the processing of unstructured data. Hence, AI is not the enemy of creativity, but a force that reorganizes the creative process.

Will human dignity be compromised?

This is probably the most deep-seated problem. Once automation replaces conventional labour, individuals may lose the sense of value and purpose they derived from work, i.e. achieved dignity. If it is AI that dominates the creative process in human-machines collaboration, people may begin to question whether they still possess unique value of their own.

Research indicates that serendipitous inspiration gained from AI can enhance ideation performance, thereby strengthening one’s sense of dignity. By contrast, if AI leads to cognitive fixation and gives humans a sense of being replaceable, then their sense of dignity will decline. In other words, dignity is not determined by whether AI is used, but by whether humans still feel a sense of agency and contribution.

Examining governance practices through corporate cases

In the case of a world-leading gold mining company, autonomous haulage systems, digital twins, and AI-driven predictive maintenance have been introduced; at the same time, its digital strategy has been closely integrated with its core values—safety, sustainability, integrity, inclusion, and responsibility. More importantly, skills retraining and role reassignment have been provided to reduce employee resistance, with a focus on a human-centred approach to AI. Ultimately, not only has productivity been enhanced, but the lifespan of the mine has also been extended and carbon emissions reduced.

Another large mining company has, through a capital allocation framework, steered clear of technology for technology’s sake, established an AI centre, strengthened change management and human-in-the-loop mechanisms, thereby achieving true human-machine collaboration. Such cases indicate that successful AI transformation lies not only in technological upgrades but also in the enhancement of organizational capabilities and values.

Safeguarding humanity’s core against the tide of technology

Generative AI has become an irreversible trend. The question is not whether to use AI but how to design and govern it. “Human-centred” is not merely a slogan but a specific framework encompassing the four key areas mentioned above. In terms of performance, it hinges on “have we done better?” In terms of learning, it hinges on “have we become stronger?” In terms of creativity, the focus is on “have we thought further ahead?” In terms of dignity, it corresponds to “do we still retain our agency?”

As generative AI permeates deep into organizational and societal structures, we must ask ourselves: “as we become stronger through AI, do we still remain autonomous, secure, purposeful, and sustainable individuals and organizations?” This is not only a matter of corporate management, but also a civilizational choice the world must jointly confront in the digital age.