1% paper/tables/ledger.tex — key rows of docs/parameter_ledger.csv (fetched 2026-10-27)2% Status: calibrated = computed from downloaded data; literature = verified publication; regulation; illustrative.3\begin{table}[htbp]4\centering5\scriptsize\singlespacing6\setlength{\tabcolsep}{4pt}7\begin{threeparttable}8\caption{Parameter ledger: key rows}9\label{tab:ledger}10\begin{tabular}{@{}>{\raggedright\arraybackslash}p{3.9cm}l>{\raggedright\arraybackslash}p{2.6cm}>{\raggedright\arraybackslash}p{5.7cm}>{\raggedright\arraybackslash}p{1.6cm}@{}}11\toprule12Parameter & Symbol & Value (range) & Source (short) & Status\\13\midrule14\multicolumn{5}{@{}l}{\textit{Preferences and technology}}\\15Discount factor (annual), Model A / B & $\beta$ & 0.96 / 0.95 (ledger 0.985--0.995) & real bill rate 0.57\%, TIPS 0.97\% (FRED) & calibrated\tnote{a}\\16Relative risk aversion & $\gamma$ & 2 (1--10) & Mehra--Prescott (1985) admissible range & literature\\17Frisch elasticity, intensive / extensive & $\nu$ & 0.5 / 0.25 & Chetty et al.\ (2011) & literature\\18Capital share & $\alpha$ & 0.33 (0.33--0.38) & Fernald annual $\alpha$, 1984--2025: 0.354 & calibrated\tnote{a}\\19Labour share (level) & $1-\alpha$ & 0.618 (0.546--0.64) & BLS index rescaled by PWT 2017 (FRED) & calibrated\\20Depreciation, fixed assets & $\delta$ & 0.06 (0.05--0.10) & BEA fixed assets, 1984--2024: 0.055 & calibrated\tnote{a}\\21TFP persistence (annual) & $\rho_z$ & 0.9 (0.83--0.93) & Fernald TFP, linear detrend: 0.876 & calibrated\tnote{a}\\22TFP innovation s.d. & $\sigma_\varepsilon$ & 0.013 (0.0074--0.0145) & Fernald TFP, AR(1) residual: 0.0126 & calibrated\tnote{a}\\23Capital--labour substitution & $\sigma$ & 0.7 (0.5--1.25) & Oberfield--Raval (2021); Karabarbounis--Neiman (2014) & literature\\24\addlinespace25\multicolumn{5}{@{}l}{\textit{Earnings and wealth}}\\26Earnings persistence (annual) & $\rho_e$ & 0.96 (0.91--1.0) & Flod\'en--Lind\'e 0.9136; Storesletten et al.\ 0.98--1.0 & literature\tnote{a}\\27Earnings innovation s.d. & $\sigma_e$ & 0.17 (FL: 0.206) & Flod\'en--Lind\'e $\sigma_\eta^2=0.0426$ & literature\tnote{a}\\28Borrowing limit & $\underline a$ & 0 ($-1$ to 0) & Aiyagari (1994) convention & illustrative\\29Top 1\% / top 10\% wealth share & $S_{1}$, $S_{10}$ & 32.5\% / 68.9\% & Fed DFA, 2026:Q2 & calibrated\\30Bottom 50\% / top 0.1\% wealth share & $S_{50}$, $S_{0.1}$ & 2.3\% / 15.0\% & Fed DFA, 2026:Q2 & calibrated\\31Wealth Gini (lower bound) & $G_w$ & 0.751 (0.75--0.87) & DFA grouped shares; SCF Gini not obtained & calibrated\\32\addlinespace33\multicolumn{5}{@{}l}{\textit{Financial}}\\34Real risk-free rate & $r_f$ & 0.57\% (0.56--0.97) & FRED TB3MS $-$ CPI, 1984--2025 & calibrated\\35Mean real equity return & $\E[r_e]$ & 8.64\% (6.98--9.9) & Shiller data, 1872--2025 & calibrated\\36Equity return s.d.\ (Model A $\sigma_R$) & $\sigma_e$, $\sigma_R$ & 17.9\% (15.7--18.9); 0.16 & Shiller data, 1872--2025; 1984--2025 & calibrated\\37Equity premium over bills & $\E[r_e-r_f]$ & 9.17\% (6.18--9.7) & Shiller + FRED TB3MS, 1934--2025 & calibrated\\38Investment-grade / high-yield OAS & $s_{IG}$, $s_{HY}$ & 0.88\% / 3.08\% & FRED BAMLC0A0CM, BAMLH0A0HYM2, 2023--2026 & calibrated\tnote{b}\\39Initial margin & $m$ & 0.5 (0.5--1.0) & Regulation T, 12 CFR 220.12(a) & regulation\\40Repo haircut, normal / stress & $h$ & 0 / $\approx$0.5 & Gorton--Metrick (2012), 2007--2008 & literature\\41Bank equity / assets (book) & $e_B$ & 0.095 (0.08--0.11) & FRED H.8, 2026-10-14; mean 2010--2025: 0.106 & calibrated\\42\addlinespace43\multicolumn{5}{@{}l}{\textit{AI}}\\44Hardware price-performance doubling; implied compute-price decline & $T_2$; $g_Q$ & 2.46 y (2.07--2.95); 24.5\%/y & Epoch AI (Hobbhahn--Besiroglu 2022); $g_Q=1-2^{-1/T_2}$ & literature; derived\\45LLM inference price decline & $g_p^{\mathrm{inf}}$ & 10$\times$/y (9--900) & Epoch AI (2025); a16z (2024) & literature\\46Frontier / small token price (output) & $p_{\mathrm{tok}}$ & \$50 / \$0.50 per M & Anthropic, OpenAI list prices, 2026-10-27 & calibrated\tnote{c}\\47Firm AI adoption & $a_F$ & 0.198 (0.17--0.23) & Census BTOS, 2026-05-03 & calibrated\\48Adult generative-AI use & $a_H$ & 0.40 (0.23--0.41) & Bick--Blandin--Deming (2024) & literature\\49Hyperscaler capex 2025 (4 firms) & $I_{AI}$ & \$376.1 bn (357.5--380) & SEC XBRL, earnings releases & calibrated\\50Industrial electricity price & $p_E$ & 9.64 \textcent/kWh & EIA EPM Table 5.6.A, Aug.\ 2026 & calibrated\\51Compute depreciation & $\delta_Q$ & \textsc{n/o} (0.20--0.33) & derived from $g_Q$ and 3--5-year life & illustrative\\52Compute ownership, top-4 share & -- & \textsc{n/o} (0.5--0.9) & no public measure & illustrative\\53Supervision cost of delegation & $\kappa_s$ & \textsc{n/o} (0--0.3) & no public measure & illustrative\\54Inference intensity, Model B & $\bar q$ & 0.02 & scenario parameter & illustrative\\55Delegation fee, Model B & $\phi_0$, $\phi_1$ & 0.02, 0.002 & frontier/small ratio $\approx$100$\times$ & illustrative\\56\bottomrule57\end{tabular}58\begin{tablenotes}\scriptsize59\item Values are the baseline choices of the numerical sections; the parenthesis is the admissible range recorded in the ledger. \textsc{n/o} = not obtained. Full ledger: \texttt{docs/parameter\_ledger.csv} (56 rows, all sources read on 2026-10-27).60\item[a] Baseline rounded away from the ledger point estimate; see the text of Section~\ref{sec:calibration} for the deviation and its justification.61\item[b] Mean over the three-year window FRED distributes for ICE BofA indices (2023-10-30 to 2026-10-26); weak as a long-run target.62\item[c] Input prices \$10 / \$0.10 per million tokens; prices change frequently.63\end{tablenotes}64\end{threeparttable}65\end{table}66