AI资管时代：AI会让市场更有效还是更低效

Gen Al in Asset Management: Is Al making markets more or lessefficient? we get from investors. In this note, we show evidence to present both sides of the argumentand find that overall Al is likely to improve micro-efficiency -faster information processing,betterexecution,broaderdatacoverage:butcanworsenmacro/market-structureinefficiency through crowding,concentration,and sharper narrative-driven reversals.Webelieve Al would create lower average inefficiency but highertail inefficiency. +6563267641rupal.agarwal@bernsteinsg.com +85221232636cheng.zhang@bernsteinsg.com rests on its abilitytoprocessvastamountsofinformationfaster,moreconsistently,andat lowercostthan humans.Early evidenceis broadlysupportive:Italy's 2023temporaryChatGPTbanreducedinformationproductionandwidenedbid-askspreads,whilesinceGenAl's launch,EPS forecast dispersion amongmanylarge-cap/well-covered stocks hasfallen,suggesting lower information asymmetry.Alis also accelerating price discovery:IMF shows post-LLM equity price reactions to Fed minutes aligning more quickly withlonger-termpricemoves,whileS&P500 earningssurpriseshave steadilydeclined asanalystforecasts converge fastertoward actualoutcomes.Aldriven productivity gain(c.40% time save) is allowing broader coverage which is likely to compress the traditional‘neglectpremium'.Inthelast1yrtherehasbeen28%/23% jump in coverageofEMsmall-caps/ndiamid-caps.Gen Alagents (democratized access vs.onceavailableonlytoquantinvestors)are expected tomitigate behavioral inefficiencies as shown in arecentpaper-Alagents made rational decisions 61%-97%of the time vs.46%-51%for humans. Case for Al increasing inefficiency: Al could make markets more inefficient throughherding behavior, systemic fragility,synthetic information shocks and reduced competition.More Gen Alusage raises the risk of signal homogenization and consensus positioningThe August 2024 Nikkei sell-off, driven in part by the unwinding of crowded carry tradesReport also warned that Al could heighten systemic vulnerabilities through synchronizedtrading,modelconcentration,and cyber risks.Al-generatedmisinformation introducesthattemporarilymoved US equities in2023.Recent academic workfrom HEC Parisandother studies suggest Al trading agents can learn collusive-behavior without explicitcommunication,potentially reducing competition,liquidity and price discovery. The Reflexivity problem-ls Al itself the inefficiency?: As investors increasingly relysubsequently reinforces future model signals.Rather than simply reducing mispricing,Alcan create self-reinforcing feedbackloops that amplifymomentum,crowding and valuationextremes. This mirrors George Soros'theory of reflexivity,where perceptions influencereality and reality subsequently reinforces perceptions. This is so evident in markets today-more Al is being used by market participants while Alis also the dominant narrative. Thishasledtounprecedentedmarket concentration andcorrelation.Momentumtrade hasreached extreme valuations and crowding, seeing record high bullish analyst sentiment andproducing 25yr high factor dispersion. (see here, here, here). earnings surprises as forecasts convergemore quickly toward outcomes. At the same time, Al-driven productivity gains-savingroughly 40% of analyst time and expanding coverage of under-researched segments such as EM small-caps and India mid-caps-could compressthetraditionalneglectpremium.Finally,Alagents appearlesspronetobehavioural biasesthanhumaninvestors,withrecentresearchfindingsubstantiallyhigherratesofrationaldecision-making,potentiallyreducingbehaviouralinefficiencies anddemocratizing capabilities onceavailableprimarilytoquantitative investors.Overthelast12months,wehavehadmultipleconversationswithCiOsandPMsonhowGenAl is,andwouldimpactasset managementindustry.Wehavecoveredmultipleaspectsofthistopic inourpreviousworkfocusingonkeyuse-casesforinvestors,Altoolsforinvestors,bestpractices emergingat largeassetmanagers,keyconcerns andrisks,howAl-nativeassetmanagerswouldemergeby2O3O,expectedAlspendbyassetmanagersandtheimpactofAlontheindustryefficiencygains/margin improvement.Wehavealsotoucheduponsomelong-term impactofAlonthebroadermarketstructureandmarketparticipants. In this report, we try to address a key debate-"ls Al going to make markets more efficient or less efficient?" informationprocessing,betterexecution,broaderdatacoverage; butcanworsenmacro/market-structureinefficiencythroughcrowding,concentration,and sharpernarrative-drivenreversals.Webelieve Alwouldcreateloweraverageinefficiencybut highertail inefficiency. In a normal market regime, Al is likely to improve liquidity,tighten reaction time, reduce obviousmispricings and compress simple alpha. Howeverin a bubble regime, Al would amplifynarratives because similarmodelsidentify similar winners vs. losers while results in increased correlation,turnover, de-risking speed and liquidity gaps duringstress.Wediscussthisindetail below. 20 May 2026 -Gen Al in Asset Management: Key takeaways from o