% paper/sections/literature.tex — UQO WP 27, "The Invisible Hand of Machines"
% Section: Related literature. Every citation key is in paper/references.bib and every claim is
% backed by the corresponding row of docs/literature_ledger.md.
\section{Related literature}
\label{sec:literature}

The paper draws on six literatures: agent-based models of financial markets; double-auction
microstructure and the measurement of price discovery; information aggregation with correlated
signals, herding and feedback; liquidity spirals, fire sales and contagion; algorithmic trading,
monoculture and algorithmic collusion; and, closest to our question, LLM agents in economic experiments
and market simulations. We close with what this paper adds and where it overlaps with existing work.

\subsection{Agent-based and heterogeneous-agent models}
\label{sec:literature:abm}

Our rule-based population descends from agent-based computational finance, in which interacting
heterogeneous rules reproduce fat tails, volatility clustering and bubbles \citep{lebaron2006}. In the
Santa Fe artificial stock market the speed of belief updating governs whether prices track fundamentals
or display excess volatility and technical-trading regimes \citep{lebaron1999}; in adaptive belief
systems, switching between predictors produces increasingly complex dynamics as the intensity of choice
rises \citep{brock1997}, and the share of trend followers determines whether the fundamental price is
stable \citep{brock1998}. Noise traders switching between optimistic, pessimistic and fundamentalist
states generate scaling laws and clustered volatility endogenously \citep{lux1999,lux2000}, contagious
optimism yields bubbles and crashes \citep{lux1995}, and recruitment dynamics generate persistent
herding without any informational reason \citep{kirman1993}. These are the channels through which a
correlated LLM order flow can be amplified by a classical population. Closer to our exchange,
\citet{chiarella2002} reproduce order-book regularities with fundamentalist, chartist and noise traders
in a double auction, and \citet{chiarella2009} show that heterogeneous rules in a limit-order book
generate fat tails, volatility clustering and order-flow autocorrelation, with the chartist share
driving excess volatility. Leverage in an agent-based model produces fat tails and clustered volatility
through synchronized margin-driven selling \citep{thurner2012}, the closest antecedent of our leverage
treatment, and multi-agent reinforcement-learning markets reproduce the same facts
\citep{lussange2021}. Following the surveys of \citet{hommes2006} and \citet{hommes2021}, we treat an
LLM policy as one more expectation type; following \citet{farmer2009} and \citet{axtell2025}, we use
the agent-based method for counterfactuals that cannot be solved analytically and address its open
frontiers---validation, calibration, scale---with matched seeds and a power analysis. Validation
targets are the moment set of \citet{franke2012} and the stylized facts of heavy tails
\citep{mandelbrot1963,cont2001}, conditional heteroskedasticity \citep{engle1982,bollerslev1986} and the
long memory of absolute returns \citep{ding1993}.

\subsection{Double auctions, zero intelligence and price-discovery measurement}
\label{sec:literature:lob}

\citet{smith1962} showed that double auctions with induced values converge to competitive equilibrium
with few traders, and \citet{smith1988} that experimental asset markets with known dividend streams
nevertheless produce bubbles and crashes; our dividend-paying asset and bubble metrics follow that
design. Zero-intelligence traders constrained by budgets reach near-full allocative efficiency, so
market structure rather than rationality drives efficiency \citep{gode1993,ladley2012}, and a
zero-intelligence model of the double auction predicts spread and price impact from order-flow rates
alone \citep{farmer2005,smith2003}. This fixes the benchmark against which any effect of
``intelligence'' must be read: changing the composition of a population can move liquidity merely by
changing order-flow rates, which is why we report realized order and volume shares alongside the
assigned treatment. The book is parametrized from the empirical regularities in \citet{bouchaud2002}, \citet{cont2010},
\citet{cont2014}, \citet{gould2013} and \citet{bouchaud2018}; the event-driven exchange with per-agent
latency follows ABIDES \citep{byrd2020} and the realism checklist is that of \citet{vyetrenko2020}.

The outcomes are standard measures of price discovery. The price-impact coefficient
$\lambda=\Cov(v,Y)/\Var(Y)$ of \citet{kyle1985} and the adverse-selection spread of \citet{glosten1985}
tie liquidity to the information in order flow; \citet{hasbrouck1991} measures that information by the
permanent price impact in a VAR of quotes and trades, \citet{hasbrouck1993} decomposes the transaction
price into an efficient random-walk component and a pricing error, and \citet{hasbrouck1995}
attributes price discovery through information shares, which we adapt to LLM versus rule-based flow.
Variance ratios test the random walk \citep{lo1988}, the effective spread follows from serial
covariance \citep{roll1984}, and illiquidity is absolute return per unit of volume \citep{amihud2002}.
\citet{bai2016} define informativeness by how well prices forecast future fundamentals,
\citet{bond2012} argue that it is the welfare-relevant object because real decisions feed on prices,
and \citet{hendershott2014} separate transitory price pressure from permanent information effects,
which justifies separating the pricing error from the efficient-price variance. Because the simulator
observes the latent value, these statistics are computed against the truth rather than a proxy, which
is how we operationalize the efficiency notions of \citet{fama1970}.

\subsection{Correlated signals, herding and feedback}
\label{sec:literature:info}

The theoretical core of the paper is the aggregation of \emph{correlated} private signals. Competitive
prices aggregate dispersed signals imperfectly because of noise supply \citep{hellwig1980} and cannot
be fully efficient when information is costly \citep{grossman1980}. With strategic informed traders,
the correlation structure of signals, not only their precision, governs trading intensity and
informativeness \citep{foster1996}, and more correlated information leads to faster revelation
\citep{back2000}. Information diversity across traders is complementary, so a loss of diversity
reduces informativeness about some components of value \citep{goldstein2015}; abundant cheap data can
lower informativeness by crowding out costly deep information \citep{dugast2018}, a close parallel to
shared-model crowding; public information can raise disagreement through higher-order beliefs, so
shared inputs do not trivially homogenize behaviour \citep{kondor2012}; and information heterogeneity
among speculators generates contagion while homogeneity reduces it \citep{pasquariello2007}. Our
stylized model (Section~\ref{sec:theory}) borrows this correlated-error structure and reinterprets the
common component as the signature of a shared foundation model. Herding supplies the behavioural
counterpart: sequential decision-makers who follow predecessors destroy information aggregation
\citep{banerjee1992}, cascades become more likely as signals become more correlated
\citep{bikhchandani1992}, reputational concerns make managers mimic one another
\citep{scharfstein1990}, and short-horizon speculators herd on the same information source even when it
is unrelated to fundamentals \citep{froot1992}; the buy-side imbalance statistic of
\citet{lakonishok1992} is the standard measure and we compute it for LLM cohorts. Feedback trading
closes the loop, since the share of trend followers governs the stability of the fundamental price
\citep{brock1998}; Proposition~\ref{prop:feedback} is a mechanical version of this amplification.

\subsection{Liquidity spirals, fire sales and contagion}
\label{sec:literature:spirals}

Hypotheses H3 and H6 draw on the literature on funding constraints. Market liquidity and funding
liquidity reinforce each other through margins, producing liquidity spirals \citep{brunnermeier2009};
our margin-call mechanism follows this logic. Liquidation values fall when potential buyers are
themselves constrained \citep{shleifer1992}, constrained arbitrageurs can amplify rather than correct
mispricing \citep{shleifer1997} and may under-provide liquidity \citep{gromb2002}, forced sales push
prices below fundamentals and propagate losses \citep{shleifer2011}, endogenous margins amplify the
leverage cycle with leverage caps as a policy lever \citep{geanakoplos2010}, and procyclical
intermediary leverage amplifies price moves \citep{adrian2010}. Across assets, losses propagate through reduced risk-bearing capacity \citep{kyle2001}, distressed
selling induces endogenous correlation \citep{cont2013}, fire-sale spillovers scale with common
exposures and leverage \citep{greenwood2015}, rebalancing transmits shocks between unrelated assets
\citep{kodres2002}, and dense networks are fragile to large shocks \citep{acemoglu2015}. For measurement we use the spillover index of \citet{diebold2009}, its
generalized directional form \citep{diebold2012} and network interpretation \citep{diebold2014}, and we
heed the warning of \citet{forbes2002} that correlation-based contagion tests must adjust for
heteroskedasticity.

\subsection{Algorithmic trading, monoculture and algorithmic collusion}
\label{sec:literature:algo}

The average evidence on algorithmic trading is favourable: it improves liquidity
\citep{hendershott2011}, high-frequency traders trade with permanent price changes and against
transitory errors \citep{brogaard2014}, and algorithmic strategies in foreign exchange improve price
efficiency even though they are more correlated with each other than human trades
\citep{chaboud2014}---the finding we take most seriously, since correlation among algorithms is a
documented fact whose consequences depend on what they are correlated \emph{on}. High-frequency
market makers keep tight inventory control \citep{menkveld2013}, a feature of our market-maker policy;
\citet{menkveld2016} reviews the speed and adverse-selection consequences and \citet{foucault2016} show
how fast traders reacting to common news affect informativeness and liquidity. Under stress the picture darkens: during the Flash Crash, high-frequency traders did not
trigger the crash but their synchronized inventory management amplified it \citep{kirilenko2017};
order-flow toxicity rose and liquidity withdrew when flow became toxic \citep{easley2011}; the August
2007 quant unwind showed common strategies deleveraging simultaneously \citep{khandani2011}; and
ultrafast extreme events proliferate in a machine ecology beyond human response time
\citep{johnson2013}. \citet{budish2015} show that continuous trading creates a latency arms race with
correlated sniping and propose frequent batch auctions, which motivates our matched-latency treatment
and call-auction robustness arm.

Algorithmic monoculture names the homogenization mechanism: when many decision-makers rely on the same
algorithm, outcomes can be worse even if it is individually more accurate \citep{kleinberg2021};
foundation models carry a homogenization risk because downstream systems inherit the same failure modes
\citep{bommasani2021}, and shared models mis-serve the same individuals across deployers
\citep{bommasani2022}. Generative AI raises individual creativity but reduces collective diversity
\citep{doshi2024}, standard LLM outputs show diminished diversity of thought \citep{park2024brm}, and
LLM assistance homogenizes human ideation \citep{anderson2024}; conversely, ensembles of twelve LLMs
match human crowd forecasting accuracy \citep{schoenegger2024}, the diversity benefit that H4 tests in a
market. Learning algorithms also coordinate without being told to: Q-learning pricing algorithms learn
tacit collusion \citep{calvano2020,klein2021}, algorithmic pricing raised retail gasoline margins
\citep{assad2024}, asymmetric adoption softens competition \citep{brown2023}, LLM pricing agents reach
supracompetitive prices with a sensitivity to prompt wording \citep{fish2024}, and LLM sellers in double
auctions collude more when they can communicate \citep{agrawal2025}. In securities markets, Q-learning
market makers quote markups that shrink with the number of algorithms \citep{colliard2026}, and
reinforcement-learning speculators in a Kyle-type market sustain supracompetitive profits through
price-trigger strategies or homogenized learning biases, lowering informativeness and liquidity
\citep{dou2025}. \citet{dou2025} is the nearest theoretical neighbour of our mechanism: their channel is
strategic, ours is a shared-error structure in signals, and they have no language models. The policy
literature has drawn the systemic conclusion: deep learning increases uniformity because models built
on the same data generate correlated predictions \citep{gensler2020}; common AI models and data sources
could raise correlations in trading, lending and pricing \citep{fsb2024}; AI amplifies existing
vulnerabilities and makes crises faster \citep{danielsson2025}; monoculture and correlation are among
the stability concerns of an intelligent financial system \citep{aldasoro2025}; and the AI adoption
share and signal correlation raise systemic coupling convexly, with 13F evidence of tail-loss
amplification \citep{meng2026}. None of these provides a controlled experiment.

\subsection{LLM agents in economics and markets}
\label{sec:literature:llm}

\paragraph{Simulated subjects.}
\citet{horton2023} proposed LLMs as simulated economic agents for replicating behavioural experiments
at low cost; \citet{aher2023} showed that they can simulate multiple humans with fidelity varying by
model and task, and \citet{argyle2023} that demographic backstories yield silicon samples matching
sub-population responses. GPT choices score higher on rationality than humans and are stable across
demographic prompts \citep{chen2023pnas}, chatbot choices in behavioural games fall within the human
distribution \citep{mei2024}, LLMs cooperate well but coordinate poorly in repeated games
\citep{akata2025}, and LLM-elicited expectations are extrapolative \citep{bybee2023}. Generative agents
with memory and reflection produce believable emergent behaviour \citep{park2023,gao2023s3}, EconAgent
produces realistic macroeconomic dynamics \citep{li2024econagent}, and \citet{immorlica2024} treat
generative agents as economic agents with prompts as primitives \citep[see also][]{manning2024}.
\citet{anthis2025} conclude that LLM social simulations are promising provided that homogeneity,
sycophancy and drift are addressed; homogeneity is the concern this paper turns into a treatment.

\paragraph{Single-agent trading and backtests.}
ChatGPT-derived headline sentiment predicts next-day returns \citep{lopezlira2023chatgpt}. FinMem
\citep{yu2023finmem}, TradingGPT \citep{li2023tradinggpt} and FinAgent \citep{zhang2024finagent}
evaluate one agent, or a small team with layered memory, by backtesting on historical prices. The
market does not react to the agent, so nothing is learned about equilibrium prices, and such backtests
are contaminated by memorization of the historical period \citep{lopezlira2025memorization}---two
reasons for a synthetic economy with an endogenous price and no history to memorize.

\paragraph{LLM-populated market simulators.}
A second generation places many LLM agents on a matching engine. ASFM finds reactions to rate and
inflation shocks consistent with economic priors \citep{gao2024asfm}; StockAgent adds bulletin-board
communication and reports that GPT- and Gemini-driven agents differ systematically
\citep{zhang2024stockagent}; TwinMarket scales to a social-media-connected population, reproduces
bubbles and crashes, and reports that agents of one model family are more homogeneous
\citep{yang2025twinmarket}; FCLAgent reproduces path-dependent patterns absent from conventional
agent-based models \citep{hashimoto2025}; StockSim is an open order-level simulator with latency and
price--time matching \citep{papadakis2025}; and trading-style switching in a year-long simulation is only
partially consistent with behavioural-finance theory \citep{li2026switch}. \citet{lopezlira2025trade} is
closest in spirit: LLM agents in a persistent order book adhere to assigned strategies, generate price
discovery and bubbles, and are analysed with partial-dependence-style perturbations. These papers show
that LLM markets \emph{can} produce the patterns we study; they hold the population composition fixed,
do not equalize the information set of LLM agents with that of the agents they replace, and do not
separate the fundamental value, the efficient price and the transaction price.

\paragraph{Experimental-finance replications.}
In Smith--Suchanek--Williams-type markets, LLM agents price near fundamentals and form only muted
bubbles, with less variable strategy text than humans \citep{henning2025}; in a call auction they
exhibit disposition effects and extrapolative beliefs, and prompt interventions causally change bubble
size \citep{ouyang2026}; across fifteen models, behaviour ranges from fundamental pricing to human-like
bubbles, mixed populations are unstable, and advanced models adapt to exploit weaker ones
\citep{saxena2026}; replications of \citet{smith1962} converge more slowly or not at all, with
model-family differences \citep{struski2026}; and the institution alone moves the efficiency of
identical agents from 56\,\% to 89\,\% \citep{chupilkin2026}. These studies ask whether LLMs behave
like humans and their answers conflict. Our results do not require fidelity, only that errors be
correlated within a model.

\paragraph{Correlated behaviour from shared inputs.}
The one paper that measures correlated trading from shared inputs is \citet{song2026}: overlap in
retrieved documents among retrieval-augmented trading agents raises mean pairwise action correlation
from 0.04 to 0.41 and increases position volatility without raising profits. That is a backtest without
market feedback, so it cannot speak to informativeness or liquidity. \citet{cho2025} document peer
conformity in LLM multi-agent systems, and \citet{hansen2025} find in re-created herding experiments
that AI agents rely more on private signals than humans but can be prompted to herd when maximizing
profit. LLM service behaviour drifts over months \citep{chen2024hdsr}, which is why every response is
cached and replayed from a frozen library.

\subsection{What this paper adds}
\label{sec:literature:adds}

We should be candid about overlap. The hypothesis that shared models correlate trading is not ours: it
is stated by \citet{gensler2020}, the \citet{fsb2024} and \citet{danielsson2025}, modelled as
homogenized learning biases by \citet{dou2025}, and measured in a backtest by \citet{song2026}.
Model-family homogeneity inside LLM markets has been observed by \citet{yang2025twinmarket},
\citet{zhang2024stockagent} and \citet{saxena2026}, and open LLM order-book simulators exist
\citep{lopezlira2025trade,papadakis2025,gao2024asfm}. Our simulator is not the contribution; the design
is. First, the LLM penetration share is a randomized treatment assigned to trader slots by an
independent seed stream, so that matched pairs share every other stream---fundamentals, news,
endowments, latencies, policy assignment, activation, tie-breaks---and paired differences are causal
effects on the same market history (Section~\ref{sec:identification}). Second, we separate three
prices---the latent value known to the simulator only, the public value implied by announced dividends
and news, and the transaction price---so that an efficiency loss can be distinguished from transitory
pressure \citep{hasbrouck1993,hendershott2014}. Third, we derive the shared-error mechanism in a
Kyle--Hellwig setting with a model-specific common error and an effective number of models, then test
it by intervention---prompt diversity at equal share, a two-model mix, inherited information sets,
memory, communication cohorts---so that the effective number of models is manipulated rather than
observed. Fourth, we report the realized dose, the share of orders and volume actually originated by
LLM agents, alongside the assigned share, and run fairness designs with equal decision opportunities,
equal dollar budgets and matched latency, because naive comparisons confound being an LLM with being
slow and infrequent. Fifth, the leverage interaction and safeguard counterfactuals connect the LLM
literature to \citet{brunnermeier2009}, \citet{thurner2012} and \citet{kirilenko2017}, with thresholds,
estimators and seed counts fixed before the first confirmatory call.