TeX 53.6%
Python 46.2%
1\section{Conclusion}2\label{sec:conclusion}34Three decades of minute-level data on the full U.S.\ equity cross-section5allow the title's question to be answered in its three parts. Faster: common6information that in the mid-1990s took tens of minutes to appear in7cross-stock comovement now appears within the minute, whether because8incorporation accelerated or because the mechanical attenuation of9nonsynchronous trading receded, and most likely both. Wider: no. Isolated10idiosyncratic extremes---six-sigma one-minute moves in the largest five11hundred securities---raise market-wide tail activity by at most a few percent12of its base rate within two hours, in every decade and for either sign; what13transmission exists is concentrated among the shocked stock's twenty closest14lagged-correlation neighbors, and that concentrated channel became roughly15three times larger between the floor-trading and algorithmic eras, so16transmission became sharper rather than broader. More fragile: only in a17bounded sense. Within the post-decimalization era, ordinary extreme minutes18became an order of magnitude rarer while minutes in which a tenth of the19universe moves together are roughly ten times more frequent in 2008 and 202020than in calm years; across the decimalization boundary the question cannot be21answered with threshold-based measures, because price discreteness22manufactures synchronization in the 1990s.2324The evidence is deliberately associational where identification runs out,25and the paper reports the diagnostics that qualify its own headline numbers.26Real events arrive in minutes of elevated ambient activity, so the27market-wide response is bracketed rather than pinned at zero; the network28differential is a difference between two non-zero contrasts, and its secular29increase combines a change in transmission with a change in how precisely30one-minute correlations identify economic neighbors; and the co-jump series31conflates common news with endogenous synchronization. OHLCV bars do not32observe quotes, depth, order flow, or security type, so liquidity operates33through proxies and ETFs cannot be isolated; tickers are not permanent34identities, so long-run statements run within screened spells. The35extensions that would most sharpen the conclusions are, in order, neighbor36sets estimated on coarser-sampled or transaction-time correlations, the37pre-event network contrast, lower source and higher receiver thresholds on38wider universes, and the attribution of co-jump minutes to news using39timestamped announcement data.4041For market design and risk monitoring the practical implication is direct.42Surveillance aimed at single-name events monitors the part of the system that43modern markets absorb, or transmit only to a handful of linked names; the44binding constraint for market-wide stability is the market-wide minute, and45the margin that moved over three decades is the sharpness of transmission46along statistical linkages, not its breadth.47