spb/uqo-research-db

1% ============================================================================2\section{Results}3\label{sec:results}4% ============================================================================56\subsection{Portfolio sorts: a weak unconditional relation}7Table~\ref{tab:sorts} reports the quintile sorts over the full sample. The relation between AI8disclosure and future returns is positive and broadly monotone but economically modest. The9equal-weighted Q5$-$Q1 spread is $+4.6$ percentage points per year for AI adoption \tstat{1.83},10$+3.9$ for AI infrastructure \tstat{1.58}, $+5.3$ for AI disruption risk \tstat{2.27}, and $+4.9$ for11the composite \tstat{1.88}. Value-weighting attenuates the adoption and composite spreads, indicating12the effect is stronger among smaller index members. The left panel of Figure~\ref{fig:quint} displays13the full-sample pattern: returns rise gently from the lowest to the highest AI quintile. On its own,14this is suggestive but far from decisive evidence---and, as we now show, it averages over two very15different regimes.1617\input{tables/tab_sorts}1819\subsection{Does AI disclosure earn risk-adjusted alpha unconditionally? (H1)}20Table~\ref{tab:alpha} reports CAPM and FF6 alphas of the long--short portfolios. The AI-opportunity21composite earns a raw spread of about $+2.3\%$ per year, of which the FF6 alpha is $+1.9\%$---but22statistically insignificant \tstat{1.24}. The AI-adoption and AI-infrastructure long--shorts are23similarly insignificant after risk adjustment. The one exception is the AI-disruption-risk spread,24whose FF6 alpha of $+2.8\%$ per year is marginally significant \tstat{2.10}, hinting at a small risk25premium for firms that flag AI as a competitive threat. The factor loadings (not the focus here) show26the AI long--short is roughly market-neutral with modest negative profitability (RMW) and investment27(CMA) tilts, consistent with a growth-oriented long leg. The verdict on H1 is clear:28\textbf{unconditionally, AI disclosure is at best weakly priced.} This is not the end of the story but29the beginning of it.3031\input{tables/tab_alpha}3233\subsection{Panel and Fama--MacBeth estimates}34Table~\ref{tab:panel} juxtaposes the two-way fixed-effects panel with the Fama--MacBeth estimator. The35contrast is instructive. In the Fama--MacBeth (cross-sectional) regressions, the AI-adoption and36AI-infrastructure premia are positive---$+1.3$ and $+1.6$ percentage points per standard37deviation---with \textit{t}-statistics near 1.6--1.7, indicating a weak but positive cross-sectional38premium on average. In the two-way fixed-effects panel, however, every AI coefficient collapses to39essentially zero. Because firm fixed effects remove all permanent cross-firm differences, this tells40us the signal lives \emph{across} firms, not \emph{within} them: a given firm writing relatively more41about AI from one year to the next does not predict its own subsequent return (anticipating H4).42Throughout, size carries the familiar negative premium ($t\approx-8$) and volatility a positive one,43consistent with the broader cross-sectional literature in this large-cap universe.4445\input{tables/tab_panel}4647\subsection{The structural break: AI pricing switched on after ChatGPT (H2)}48The full-sample averages mask a sharp regime change that is the paper's central result. Splitting the49sample at the November 2022 ChatGPT launch, the center and right panels of Figure~\ref{fig:quint}50reveal a striking transformation: the quintile pattern is nearly flat before 2022 and steeply51increasing afterward. The high-minus-low composite spread is $+2.4$ percentage points per year52pre-ChatGPT but $+14.6$ points per year post-ChatGPT---a more than sixfold increase.5354\begin{figure}[htbp]\centering55\includegraphics[width=\textwidth]{fig1_quintile_returns.png}56\caption{Mean forward twelve-month return by AI-disclosure composite quintile, full sample and by57sub-period. The high-minus-low spread is $+2.4$ percentage points before ChatGPT and $+14.6$ points58after.}\label{fig:quint}59\end{figure}6061Table~\ref{tab:break} quantifies the break two ways. In sub-period sorts, the AI-adoption Q5$-$Q162spread rises from $+2.2\%$ per year \tstat{2.04} pre-2022 to $+14.3\%$ \tstat{2.15} post-2022; the63composite rises from $+2.5\%$ to $+14.6\%$; AI infrastructure from $+1.4\%$ to $+13.6\%$. In the64pooled interaction regressions, the post-2022 incremental slope is positive for all three signals and65statistically significant for AI infrastructure \tstat{2.63}---the facet most directly tied to the66chip, cloud, data-center, and power firms that led the 2023--2025 rally. Figure~\ref{fig:spread} shows67the annual spreads: small and mixed-sign through 2021, uniformly large and positive for every68post-2022 cohort.6970\input{tables/tab_break}7172\begin{figure}[htbp]\centering73\includegraphics[width=0.9\textwidth]{fig3_annual_spread.png}74\caption{Annual high-minus-low (Q5$-$Q1) forward-return spread on the AI-disclosure composite, by7510-K fiscal year. Post-ChatGPT cohorts (red) are uniformly and substantially positive.}76\label{fig:spread}77\end{figure}7879The same regime change is visible in two further lenses. Figure~\ref{fig:cum} plots the cumulative80value of one dollar invested in the equal-weighted, annually rebalanced, self-financing81high-minus-low portfolio: it drifts sideways-to-up for a decade and then inflects upward after the82dashed ChatGPT line. Figure~\ref{fig:fmyear} plots the year-by-year Fama--MacBeth slope on the AI83composite (with controls): it hovers near zero for the entire 2010s and then turns sharply positive,84peaking in the 2024 cohort. The latter figure also delivers an important nuance: the size-controlled85premium is largest in the most recent cohort, consistent with the AI \emph{infrastructure}86super-cycle of 2024--2025 (the build-out of data centers and accelerated compute). We return to this87concentration when discussing limitations.8889\begin{figure}[htbp]\centering90\includegraphics[width=0.86\textwidth]{fig2_cumulative_ls.png}91\caption{Cumulative value of \$1 invested in the equal-weighted, annually rebalanced, self-financing92high-minus-low AI-disclosure portfolio (AI opportunity composite and AI disruption-risk).}93\label{fig:cum}94\end{figure}9596\begin{figure}[htbp]\centering97\includegraphics[width=0.9\textwidth]{fig6_fm_byyear.png}98\caption{Year-by-year Fama--MacBeth slope on the AI-disclosure composite (forward return per99cross-sectional standard deviation), controlling for size, momentum, and volatility. Bars are $\pm1$100standard error.}\label{fig:fmyear}101\end{figure}102103Finally, Figure~\ref{fig:bin} confirms the pattern is not an artifact of a few influential104observations: the binned relation between the AI composite and forward returns is essentially flat in105the pre-ChatGPT sub-period and distinctly upward-sloping afterward, across the entire distribution of106AI disclosure rather than only at the extreme quintiles.107108\begin{figure}[htbp]\centering109\includegraphics[width=\textwidth]{fig5_binscatter.png}110\caption{Binned scatter of forward twelve-month return against the AI-disclosure composite, by111sub-period. Each point is a portfolio of firms in a bin of the AI signal; the line is an OLS fit. The112conditional slope steepens markedly after ChatGPT.}\label{fig:bin}113\end{figure}114115\subsection{The anatomy of the premium: size and sector}116Where in the cross-section does the post-2022 premium concentrate? Table~\ref{tab:doublesort} and117Figure~\ref{fig:double} report an AI~$\times$~size double sort. Before 2022 the AI gradient is flat118within both size halves. After 2022 it is steep, and dramatically so among smaller firms: small,119high-AI firms earned a mean forward return of 34.6\%, versus 19.5\% for small, low-AI firms---a12015-point gap---while the corresponding gap among large firms is 7 points. The premium is thus121\emph{not} merely a mega-cap phenomenon (a point we reinforce in Section~\ref{sec:robustness}); if122anything it is strongest among the smaller, less-followed names, consistent with an attention-based123mechanism in which information diffuses more slowly into smaller stocks.124125\input{tables/tab_doublesort}126127\begin{figure}[htbp]\centering128\includegraphics[width=\textwidth]{fig9_double_sort.png}129\caption{AI-disclosure $\times$ firm-size double sort: mean forward twelve-month return by AI tercile130within small and large firms, pre- and post-ChatGPT.}\label{fig:double}131\end{figure}132133\subsection{Economic significance}134The statistical break is matched by an economic one. The Sharpe ratio of the equal-weighted135AI-composite long--short portfolio---a direct gauge of risk-adjusted attractiveness---rises from 0.15136in the pre-ChatGPT decade to 1.33 afterward; for the AI-disruption-risk spread it rises from 0.33 to1371.15. A Sharpe ratio above one is exceptional for a single-characteristic, self-financing equity138strategy, and its emergence precisely at the ChatGPT boundary is hard to reconcile with a constant139risk premium. The point estimates in the quintile sorts are equally vivid: a 14-point annual spread140between the top and bottom AI quintiles is roughly three times the unconditional equity risk premium141and an order of magnitude larger than the pre-2022 spread. Even acknowledging that a portion of this142reflects the higher realized volatility of AI-exposed firms (documented in Section~\ref{sec:robustness})143and the broad technology rally, the magnitude of the post-2022 re-rating is among the largest144single-theme cross-sectional effects in the recent asset-pricing record.145