1parameter,symbol,units,value,range_low,range_high,source,year_geography,moment_or_target,method,uncertainty,verified_url,notes2discount factor (annual),beta,per year,0.99,0.985,0.995,Computed targets: ex-post real 3-month T-bill 0.57% (1984-2025); 10y TIPS real yield 0.97% (2003-2025); latest TIPS 2.66% (2026-10-26),US 1984-2025,"real risk-free rate 0.6-1.0% p.a. (rows real_tbill_rate_expost_mean_1984_2025, tips_10y_real_yield_mean_2003_2025 in data/processed/calibration_targets.csv)",calibrated: beta ~ 1/(1+r) in a no-growth representative-agent steady state (0.990-0.994); in the Aiyagari block beta must be set below 1/(1+r) so that r clears the asset market -- modeling team solves for it,"moderate: r ranges 0.6% (bills) to 2.7% (current TIPS); with growth g and CRRA gamma, beta=(1+g)^gamma/(1+r)",https://fred.stlouisfed.org/series/TB3MS ; https://fred.stlouisfed.org/series/DFII10 ; https://fred.stlouisfed.org/series/CPIAUCSL,"fetched 2026-10-27; value is a model-side choice consistent with the measured moment, not a measured number"3relative risk aversion,gamma,dimensionless,2,1,10,"Mehra & Prescott (1985) JME 15(2):145-161, doi:10.1016/0304-3932(85)90061-3 (text read from author-hosted PDF: 'With alpha less than ten ...'; CRRA utility restricted to 0<alpha<inf); Kaplan, Moll & Violante (2018) AER 108(3):697-743, doi:10.1257/aer.20160042 (DOI verified via Crossref; parameter value in KMV not re-verified here)",literature,equity premium / precautionary saving,literature standard; value 2 is a conventional choice inside the Mehra-Prescott admissible range,high: 1-10 admissible; equity premium requires implausibly high gamma in the frictionless model (the puzzle),https://doi.org/10.1016/0304-3932(85)90061-3 ; https://www.academicwebpages.com/preview/mehra/pdf/The%20Equity%20Premium%20A%20Puzzle.pdf ; https://doi.org/10.1257/aer.20160042,"fetched 2026-10-27; gamma=2 is a choice, not an estimate"4Frisch elasticity of labour supply (intensive margin),eta_F,dimensionless,0.5,0.3,0.75,"Chetty, Guren, Manoli & Weber (2011) AER P&P 101(3):471-475, doi:10.1257/aer.101.3.471; NBER WP 16729 abstract: 'Frisch elasticities of 0.5 on the intensive and 0.25 on the extensive margin'; Hicksian 0.3 intensive / 0.25 extensive","US/OECD micro evidence, 2011 review",labour-supply response to wage changes,literature (meta-analysis),low-moderate for the intensive margin; aggregate Frisch incl. extensive margin ~0.75,https://www.nber.org/papers/w16729 ; https://doi.org/10.1257/aer.101.3.471,fetched 2026-10-275Frisch elasticity (extensive margin),eta_F_ext,dimensionless,0.25,0.2,0.3,"Chetty et al. (2011), same abstract",US,participation response,literature,moderate,https://www.nber.org/papers/w16729,fetched 2026-10-276capital share (gross),alpha,ratio,0.354,0.33,0.38,"Fernald quarterly_tfp.xlsx, annual sheet column 'alpha' (mean 1984-2025 = 0.354; 1948-2025 = 0.338); cross-check 1 - rescaled BLS labour share level (1984-2026) = 0.382",US business sector 1984-2025,"rows capital_share_fernald_alpha_mean_1984_2025, capital_share_1_minus_labor_level_1984_2026 in data/processed/calibration_targets.csv",computed from downloaded data (scripts/02),low-moderate: depends on treatment of proprietors' income and housing,https://www.frbsf.org/wp-content/uploads/quarterly_tfp.xlsx,fetched 2026-10-277labour share (level),1-alpha,ratio,0.618,0.546,0.64,"FRED:PRS85006173 (BLS nonfarm business index, 2017=100) rescaled by PWT labsh(2017)=0.5913 (FRED:LABSHPUSA156NRUG); BEA compensation share of GDI (FRED:A4002E1A156NBEA) mean 1984-2025 = 54.6%, 2025 = 50.9%",US 1984-2026,"rows labor_share_level_mean_1984_2026, compensation_share_gdi_mean_1984_2025 in data/processed/calibration_targets.csv",computed; RESCALING ASSUMPTION documented in script: BLS index level anchored to PWT 2017 economy-wide labour-compensation share (concepts differ),moderate: level depends on the anchor; the index itself shows a ~10 pt fall since 2000 (104.5 mean 1984-2026 -> 93.4 in 2026Q2),https://fred.stlouisfed.org/series/PRS85006173 ; https://fred.stlouisfed.org/series/LABSHPUSA156NRUG ; https://fred.stlouisfed.org/series/A4002E1A156NBEA,fetched 2026-10-278depreciation rate (private fixed assets),delta,per year,0.055,0.05,0.1,"BEA Fixed Assets via FRED: M1TTOTL1ES000 (current-cost depreciation) / K1TTOTL1ES000(t-1) (current-cost net stock); mean 1984-2024 = 0.0550, 2024 = 0.0543",US 1984-2024,row depreciation_rate_private_fixed_assets_mean_1984_2024 in data/processed/calibration_targets.csv,computed from downloaded data,"moderate: includes residential structures (low delta); nonresidential equipment+software alone is higher (~0.10), compute hardware much higher (see GPU row)",https://fred.stlouisfed.org/series/M1TTOTL1ES000 ; https://fred.stlouisfed.org/series/K1TTOTL1ES000,fetched 2026-10-279TFP persistence (annual),rho_z,per year,0.876,0.83,0.93,"Fernald utilization-adjusted TFP (annual dtfp_util), cumulated log level, linear detrend, OLS AR(1): rho=0.876 (1984-2025), 0.929 (1948-2025); raw dtfp: 0.830 / 0.903; HP(100): 0.62 / 0.47; HP(6.25): ~0",US 1948-2025 / 1984-2025,rows tfp_util_adjusted_ar1_rho_* in data/processed/calibration_targets.csv,computed (scripts/02); detrending method matters -- linear detrend reported as headline because annual RBC calibrations are usually stated for a trend-stationary process,high: 0.0 (HP 6.25) to 0.93 (linear trend); quarterly HP(1600) rho_q=0.70 (annualised 0.24),https://www.frbsf.org/wp-content/uploads/quarterly_tfp.xlsx,fetched 2026-10-27; Fernald (2014) FRBSF WP 2012-19 must be cited10TFP innovation s.d. (annual),sigma_eps,log units per year,0.0126,0.0074,0.0145,"Fernald annual dtfp_util: AR(1) residual sd after linear detrend 0.0126 (1984-2025), 0.0135 (1948-2025); first-difference sd 0.0127 / 0.0145; HP(6.25) residual sd 0.0074",US 1948-2025 / 1984-2025,rows tfp_util_adjusted_ar1_sd_* and tfp_util_adjusted_growth_sd_* in data/processed/calibration_targets.csv,computed (scripts/02),moderate,https://www.frbsf.org/wp-content/uploads/quarterly_tfp.xlsx,fetched 2026-10-2711"TFP process, literature benchmark (Cooley-Prescott 1995)","rho_z, sigma_eps",per quarter,NOT OBTAINED,,,"Cooley, T.F. & Prescott, E.C. (1995) 'Economic Growth and Business Cycles', ch.1 in Cooley (ed.) Frontiers of Business Cycle Research, Princeton UP, pp.1-38 (bibliographic details confirmed via IDEAS/RePEc; chapter text not accessible online)","US, pre-1995",quarterly Solow-residual AR(1),"literature; values commonly quoted as rho=0.95, sigma=0.007 quarterly but NOT verified from the chapter text on 2026-10-27 -- use the Fernald rows above",n/a,https://ideas.repec.org/a/eee/reveco/v6y1997i3p335-337.html,fetched 2026-10-27; weakly sourced -- do not cite numbers from this row12TFP growth (trend),g_z,log change per year,0.0095,0.009,0.013,"Fernald annual dtfp_util mean: 0.0095 (1984-2025), 0.0126 (1948-2025)",US,rows tfp_util_adjusted_growth_mean_* in data/processed/calibration_targets.csv,computed,low,https://www.frbsf.org/wp-content/uploads/quarterly_tfp.xlsx,fetched 2026-10-2713real GDP growth (trend),g,log change per year,0.0272,0.02,0.031,"FRED:GDPC1 annual average, log change: 0.0272 (1984-2025), 0.0307 (1948-2025); BEA quarterly SAAR mean 2.78% (1984-2025)",US,rows real_gdp_growth_mean_* in data/processed/calibration_targets.csv,computed,low,https://fred.stlouisfed.org/series/GDPC1 ; https://fred.stlouisfed.org/series/A191RL1Q225SBEA,fetched 2026-10-2714investment/GDP,I/Y,ratio,0.158,0.15,0.19,"FRED:GPDIC1 / FRED:GDPC1, mean 1984-2025 = 0.158; 2015-2025 = 0.182",US,rows investment_gdp_ratio_mean_* in data/processed/calibration_targets.csv,computed (ratio of chained-dollar aggregates; approximate),low-moderate,https://fred.stlouisfed.org/series/GPDIC1,fetched 2026-10-2715consumption/GDP,C/Y,ratio,0.66,0.64,0.69,"FRED:PCECC96 / FRED:GDPC1, mean 1984-2025 = 0.660; 2015-2025 = 0.680",US,rows consumption_gdp_ratio_mean_* in data/processed/calibration_targets.csv,computed,low,https://fred.stlouisfed.org/series/PCECC96,fetched 2026-10-2716CPI inflation (mean),pi,percent per year,2.79,,,"FRED:CPIAUCSL, annual-average log change 1984-2025",US,row cpi_inflation_mean_1984_2025,computed,low,https://fred.stlouisfed.org/series/CPIAUCSL,fetched 2026-10-2717idiosyncratic earnings persistence (annual),rho_y,per year,0.9136,0.91,1.0,"Floden & Linde (2001) RED 4(2):406-437, doi:10.1006/redy.2000.0121 -- Table 4 of the 1998 IIES WP 654 (PDF in data/raw/papers): US rho = 0.9136 (s.e. 0.0090); Storesletten, Telmer & Yaron (2004) JME 51(3):609-633, doi:10.1016/j.jmoneco.2003.06.005 -- NBER WP 7995 abstract: persistent component autocorrelation 0.98 to unity",US PSID (FL: 1980s-90s),AR(1) persistence of log wage/earnings residual,"literature (GMM estimates, read from Table 4 of the WP PDF)","moderate: 0.91 (FL) vs 0.98-1.0 (STY); FL's process is for hourly wage productivity, STY for earnings",http://su.diva-portal.org/smash/get/diva2:328071/FULLTEXT01 ; https://doi.org/10.1006/redy.2000.0121 ; https://www.nber.org/papers/w7995 ; https://doi.org/10.1016/j.jmoneco.2003.06.005,fetched 2026-10-2718variance of persistent earnings innovation,sigma_eta^2,"log units squared, per year",0.0426,0.03,0.05,Floden & Linde (1998/2001) Table 4: US sigma^2_epsilon = 0.0426 (s.e. 0.0048),US PSID,innovation variance of AR(1) component,literature (table read from WP PDF),moderate: implies sd 0.206; variable definitions taken from the table header -- check eq.(1) of the paper before use,http://su.diva-portal.org/smash/get/diva2:328071/FULLTEXT01,fetched 2026-10-2719variance of transitory (i.i.d.) earnings component,sigma_xi^2,log units squared,0.0421,0.02,0.06,Floden & Linde (1998/2001) Table 4: US sigma^2_xi = 0.0421 (s.e. 0.0039),US PSID,transitory variance,literature (table read from WP PDF),moderate (may include measurement error),http://su.diva-portal.org/smash/get/diva2:328071/FULLTEXT01,fetched 2026-10-2720"earnings persistence, STY (2004)",rho_y,per year,NOT OBTAINED (point),0.98,1.0,"Storesletten, Telmer & Yaron (2004) 'Consumption and risk sharing over the life cycle', JME 51(3):609-633, doi:10.1016/j.jmoneco.2003.06.005; NBER WP 7995 abstract: 'autocorrelation coefficient between 0.98 and unity'; 40% of lifetime uncertainty realised during working years",US PSID 1969-1992,persistence of persistent shock,literature (abstract only; full text 404 at the NYU mirror),n/a,https://www.nber.org/papers/w7995 ; https://doi.org/10.1016/j.jmoneco.2003.06.005,fetched 2026-10-27; point estimates of variances not read21earnings process reference (handbook),-,-,NOT OBTAINED,,,"Krueger, Mitman & Perri (2016) 'Macroeconomics and Household Heterogeneity', Handbook of Macroeconomics vol.2, doi:10.1016/bs.hesmac.2016.04.003 (DOI verified via Crossref)",US,discretised earnings process used in HA macro,literature; their calibrated values not read (Elsevier full text not fetched),n/a,https://doi.org/10.1016/bs.hesmac.2016.04.003,fetched 2026-10-2722"earnings dynamics, non-Gaussian reference",-,-,NOT OBTAINED,,,"Guvenen, Karahan, Ozkan & Song (2021) Econometrica 89(5), doi:10.3982/ECTA14603 (DOI verified via Crossref)",US SSA admin data,skewness/kurtosis of earnings changes,literature; not used for the baseline AR(1),n/a,https://doi.org/10.3982/ECTA14603,fetched 2026-10-2723borrowing limit,a_min,multiple of average annual earnings,0,-1,0,"Aiyagari (1994) QJE 109(3):659-684, doi:10.2307/2118417 (DOI verified); zero vs ad-hoc/natural borrowing limits are model conventions",model convention,bottom-50% DFA net-worth share 2.3% (2026Q2) can discipline it,"illustrative: baseline a_min=0, sensitivity to -1x average earnings",high (illustrative),https://doi.org/10.2307/2118417,fetched 2026-10-27; limit values not re-verified from the paper's text24top 1% wealth share,S_top1,percent of household net worth,32.5,22.8,33,"Fed Distributional Financial Accounts, dfa-networth-shares.csv, 2026:Q2 (TopPt1 15.0 + RemainingTop1 17.5); 1989:Q3 = 22.8",US 2026Q2,row wealth_share_top1_dfa_latest in data/processed/calibration_targets.csv,computed from downloaded DFA,low (DFA methodology),https://www.federalreserve.gov/releases/z1/dataviz/dfa/distribute/table/,fetched 2026-10-2725top 10% wealth share,S_top10,percent,68.9,,,DFA 2026:Q2 (top1 + Next9 36.4),US 2026Q2,row wealth_share_top10_dfa_latest,computed,low,https://www.federalreserve.gov/releases/z1/dataviz/dfa/distribute/table/,fetched 2026-10-2726bottom 50% wealth share,S_bot50,percent,2.3,2.3,3.5,DFA 2026:Q2; 1989:Q3 = 3.5,US 2026Q2,row wealth_share_bottom50_dfa_latest,computed,low,https://www.federalreserve.gov/releases/z1/dataviz/dfa/distribute/table/,fetched 2026-10-2727top 0.1% wealth share,S_top0.1,percent,15.0,,,DFA 2026:Q2,US 2026Q2,row wealth_share_top0p1_dfa_latest,computed,low,https://www.federalreserve.gov/releases/z1/dataviz/dfa/distribute/table/,fetched 2026-10-2728wealth Gini,G_w,Gini,0.751,0.75,0.87,Lower bound computed from DFA 5-group shares (2026:Q2) = 0.751 (within-group equality assumed); SCF-based and WID Gini NOT OBTAINED (WID page renders client-side; SCF bulletin publishes no Gini),US 2026Q2,row wealth_gini_dfa_grouped_lower_bound_latest in data/processed/calibration_targets.csv,computed lower bound; the true Gini is higher (literature SCF-based values ~0.85 are NOT verified here),high,https://www.federalreserve.gov/releases/z1/dataviz/dfa/distribute/table/ ; https://wid.world/country/usa/,fetched 2026-10-27; weakly sourced upper range29SCF 2022 median / mean family net worth,-,2022 USD,192900 / 1063700,,,"Federal Reserve Bulletin, 'Changes in U.S. Family Finances from 2019 to 2022' (Oct 2023): 'real median net worth surged 37 percent to $192,900'; 'real mean net worth increased 23 percent to $1,063,700'",US 2022,mean/median wealth ratio = 5.51,"literature (official survey report, read 2026-10-27)",low,https://www.federalreserve.gov/publications/october-2023-changes-in-us-family-finances-from-2019-to-2022.htm,fetched 2026-10-2730real risk-free rate,r_f,percent per year,0.57,0.56,0.97,"FRED TB3MS - CPI inflation, 1984-2025 mean 0.57% (sd 2.31); 1948-2025 0.56%; 10y TIPS 2003-2025 mean 0.97%, 2015-2025 0.63%, latest 2.66% (2026-10-26); real fed funds 1984-2025 0.83%",US,"rows real_tbill_rate_expost_mean_1984_2025, tips_10y_real_yield_mean_2003_2025",computed,moderate (ex-post vs ex-ante),https://fred.stlouisfed.org/series/TB3MS ; https://fred.stlouisfed.org/series/DFII10,fetched 2026-10-2731"equity premium (arithmetic, over T-bills)",E[r_e - r_f],percent per year,9.17,6.18,9.7,"Computed from Shiller ie_data.xls (S&P composite total return) and FRED TB3MS: 9.17% (1934-2025), 8.94% (1948-2025), 9.66% (1984-2025); Mehra & Prescott (1985) Table 1, 1889-1978: equity 6.98%, riskless 0.80%, premium 6.18% (read from the paper PDF text)",US 1934-2025 / 1889-1978,rows equity_premium_over_tbill_* in data/processed/calibration_targets.csv,computed (arithmetic means of annual nominal excess returns; geometric premia are ~1.5-2 pts lower),"moderate: sd of annual excess return 17.7%, so s.e. of the 92-year mean ~1.8 pts",https://shillerdata.com/ ; https://doi.org/10.1016/0304-3932(85)90061-3,fetched 2026-10-2732equity return volatility,sigma_e,percent per year,17.9,15.7,18.9,"Shiller data: sd of annual real total return 17.9% (1872-2025), 18.9% (1926-2025), 15.7% (1984-2025); Mehra & Prescott (1985) 1889-1978: 16.54%",US,rows equity_real_return_sd_* in data/processed/calibration_targets.csv,computed,low,https://shillerdata.com/ ; https://doi.org/10.1016/0304-3932(85)90061-3,fetched 2026-10-2733mean real equity return,E[r_e],percent per year,8.64,6.98,9.9,Shiller data 1872-2025: 8.64% (arithmetic); 1926-2025: 9.05%; 1984-2025: 9.94%; Mehra-Prescott 1889-1978: 6.98%,US,rows equity_real_return_mean_*,computed,moderate,https://shillerdata.com/,fetched 2026-10-2734Sharpe ratio (equity),SR,ratio,0.52,0.5,0.61,"computed: 0.52 (1934-2025), 0.53 (1948-2025), 0.61 (1984-2025)",US,rows sharpe_ratio_equity_*,computed,moderate,https://shillerdata.com/,fetched 2026-10-2735investment-grade corporate OAS (mean),s_IG,percent,0.88,0.8,1.33,"FRED:BAMLC0A0CM, mean over the only downloadable window 2023-10-30..2026-10-26; max 1.33 (2023-11-01); latest 0.90",US 2023-2026,rows ig_oas_* in data/processed/calibration_targets.csv,computed (3-year window only; FRED restricts ICE BofA history since April 2026),high as a long-run mean (window is a low-spread period),https://fred.stlouisfed.org/series/BAMLC0A0CM,fetched 2026-10-2736high-yield OAS (mean),s_HY,percent,3.08,2.6,4.61,"FRED:BAMLH0A0HYM2, 2023-10-30..2026-10-26 mean 3.08; max 4.61 (2025-04-07); latest 3.30",US 2023-2026,rows hy_oas_* in data/processed/calibration_targets.csv,computed (3-year window),high as a long-run mean,https://fred.stlouisfed.org/series/BAMLH0A0HYM2,fetched 2026-10-2737"credit-spread stress peaks (2008, 2020)","s_HY^peak, s_BBB^peak",basis points,NOT OBTAINED from data,800,2000,"Federal Reserve, Financial Stability Report Nov 2020, accessible version (text description of Figure A / 1-6): HY spread 'jumps to its peak of almost 2000 basis points in late 2008'; March 2020 'from 400 to roughly 800'; triple-B 'peak of almost 680 in 2008', March 2020 'from 150 to roughly 400'",US 2008 / 2020,stress-scenario spread levels,"literature (chart descriptions, not series values); FRED history unavailable (ICE licence)","high (chart-read, +/-100 bp)",https://www.federalreserve.gov/publications/2020-november-financial-stability-report-accessible-version.htm,fetched 2026-10-27; weakly sourced38bank lending standards (SLOOS) stress peaks,-,net percent of banks tightening,83.6 (2008Q4) / 71.2 (2020Q3),,,FRED:DRTSCILM; mean 1990-2025 = 6.6,US,rows sloos_ci_tightening_* in data/processed/calibration_targets.csv,computed,low,https://fred.stlouisfed.org/series/DRTSCILM,fetched 2026-10-2739margin requirement (initial),m,fraction of market value,0.5,0.5,1.0,"Regulation T, 12 CFR 220.12(a): required margin for a margin equity security is '50 percent of the current market value of the security' or the percentage set by the regulatory authority where the trade occurs, whichever is greater; short sales 150% (220.12(c)(1))","US, current CFR",collateral constraint on levered equity positions,regulation (text read on LII mirror; eCFR redirected to an unblock page),none (statutory floor); exchanges/brokers may require more,https://www.law.cornell.edu/cfr/text/12/220.12 ; https://www.ecfr.gov/current/title-12/chapter-II/subchapter-A/part-220/section-220.12,fetched 2026-10-2740repo haircut (normal / stress),h,fraction of collateral value,0 -> ~0.5,0,0.5,"Gorton & Metrick (2012) JFE 104(3):425-451, doi:10.1016/j.jfineco.2011.03.016; NBER WP 15223 text: 'This index rises from zero in early 2007 to nearly 50 percent at the peak of the crisis in late 2008' (repo-haircut index on structured-product collateral; Treasuries excluded)",US 2007-2008,collateral haircut path in stress,literature (WP PDF text read),moderate: index is an average over nine structured asset classes; Treasury/agency haircuts stayed ~0-2%,https://www.nber.org/system/files/working_papers/w15223/w15223.pdf ; https://doi.org/10.1016/j.jfineco.2011.03.016,fetched 2026-10-2741bank equity/assets (book),e_B,ratio,0.0946,0.08,0.11,"FRED H.8: RALACBW027SBOG / TLAACBW027SBOG = 0.0946 (2026-10-14); mean 2010-2025 = 0.106; 1984-2025 = 0.080; FDIC QBP 2Q2026 reports ROA 1.37%, net income $90.1bn, 4,238 institutions but no equity ratio on the summary page",US commercial banks,rows bank_equity_to_assets_* in data/processed/calibration_targets.csv; leverage = 10.6x,"computed (H.8 residual = assets - liabilities, a book-equity proxy)",moderate (book vs regulatory capital),https://fred.stlouisfed.org/series/RALACBW027SBOG ; https://fred.stlouisfed.org/series/TLAACBW027SBOG ; https://www.fdic.gov/quarterly-banking-profile/quarterly-banking-profile-2q-2026,fetched 2026-10-2742bank credit growth (nominal),g_B,log change per year,0.06,,,"FRED:TOTBKCR annual averages 1984-2025: mean 0.060, sd 0.027",US,rows bank_credit_nominal_growth_*,computed,low,https://fred.stlouisfed.org/series/TOTBKCR,fetched 2026-10-2743task-based production framework,-,-,structural reference,,,"Acemoglu & Restrepo (2018) AER 108(6):1488-1542, doi:10.1257/aer.20160696 (abstract verified on AEA site); Acemoglu & Restrepo (2019) JEP 33(2):3-30, doi:10.1257/jep.33.2.3 (DOI verified)",theory,automation lowers labour share and may lower wages; new tasks offset,literature,n/a,https://www.aeaweb.org/articles?id=10.1257/aer.20160696 ; https://doi.org/10.1257/jep.33.2.3,fetched 2026-10-2744task displacement share of US wage-structure change,-,share,0.50-0.70,0.5,0.7,"Acemoglu & Restrepo (2022) Econometrica 90(5):1973-2016, doi:10.3982/ECTA19815 (abstract on Econometric Society site): 50-70% of changes in the US wage structure over four decades due to relative wage declines of groups specialised in routine tasks in rapidly automating industries",US 1980-2016,validation target for displacement channel,literature,moderate,https://doi.org/10.3982/ECTA19815,fetched 2026-10-2745elasticity of substitution between tasks / between capital and labour,sigma,dimensionless,0.7,0.5,1.25,"Oberfield & Raval (2021) Econometrica 89(2), doi:10.3982/ECTA12807; NBER WP 20452 abstract: aggregate capital-labour elasticity for US manufacturing 'stable since 1970 at about 0.7'; Karabarbounis & Neiman (2014) QJE 129(1), doi:10.1093/qje/qjt032 argue for an elasticity above 1 (their point estimate not read from the abstract)",US manufacturing / global,substitution vs complementarity in the task aggregator,literature; NB: AR's task-level elasticity is a distinct object -- treat 0.7 as a capital-labour benchmark and run sensitivity 0.5-1.25,high,https://www.nber.org/papers/w20452 ; https://doi.org/10.3982/ECTA12807 ; https://doi.org/10.1093/qje/qjt032,fetched 2026-10-27; KN upper value weakly sourced46ML hardware price-performance doubling time,T_2(FLOP/s per $),years,2.46,2.07,2.95,"Hobbhahn & Besiroglu (2022) 'Trends in GPU price-performance', Epoch AI, 27 Jun 2022: FLOP/s per dollar doubles every 2.46 y [95% CI 2.24-2.72] (470 GPUs 2006-2021); ML GPUs 2.07 y [1.54-3.13]; top GPUs 2.95 y; FP16 2.30 y; 0.122 OOM/yr",global 2006-2021,compute rental price decline: price per FLOP halves every ~2.5 y => -24.5%/yr (=1-2^(-1/2.46)),literature (page read),moderate; post-2021 trend may differ,https://epoch.ai/blog/trends-in-gpu-price-performance,fetched 2026-10-2747LLM inference price decline (constant capability),g_p^inf,factor per year,10,9,900,"Cottier, Snodin, Owen & Adamczewski (2025) 'LLM inference prices have fallen rapidly but unequally across tasks', Epoch AI Data Insight, 12 Mar 2025: declines from 9x to 900x per year depending on milestone; 40x/yr to reach GPT-4 level on GPQA Diamond; example MMLU-86 threshold: $37.50/M tokens (GPT-4-0314, Mar 2023) -> $0.18 (Gemini 2.0 Flash, Feb 2025). Appenzeller (a16z, 12 Nov 2024) 'LLMflation': ~10x per year; GPT-3-level (MMLU 42) $60 (Nov 2021) -> $0.06 per M tokens (2024)",global 2021-2025,delegation-fee price path,literature (pages read),high: rate depends on capability threshold and may not persist (authors' caveat),https://epoch.ai/data-insights/llm-inference-price-trends ; https://a16z.com/llmflation-llm-inference-cost/,fetched 2026-10-2748frontier model list price (input / output),p_tok^frontier,USD per million tokens,10 / 50,4,50,"Anthropic pricing page (platform.claude.com), read 2026-10-27: Claude Fable 5.1 $10 / $50; Claude Opus 5.5 $4 / $20; Claude Sonnet 5.5 $2 / $10. OpenAI developer docs pricing, read 2026-10-27: gpt-6-astra $10.00 / $50.00 (flagship); gpt-5.5 $5 / $30",US list prices 2026-10-27,per-token delegation fee (frontier),provider pricing pages (fetched today; prices change frequently),low on the day; high over the paper's horizon,https://platform.claude.com/docs/en/about-claude/pricing ; https://developers.openai.com/api/docs/pricing,fetched 2026-10-2749small model list price (input / output),p_tok^small,USD per million tokens,0.10 / 0.50,0.05,0.5,"Anthropic: Claude Haiku 5.5 $0.10 / $0.50 (prompts <= 100k tokens), read 2026-10-27; OpenAI: gpt-5-nano $0.05 / $0.40, gpt-6-luna $0.10 / $0.50, read 2026-10-27",US list prices 2026-10-27,per-token delegation fee (small),provider pricing pages,low on the day,https://platform.claude.com/docs/en/about-claude/pricing ; https://developers.openai.com/api/docs/pricing,fetched 2026-10-27; frontier/small ratio ~100x50firm AI adoption rate,a_F,share of businesses,0.198,0.17,0.23,"US Census Bureau, Business Trends and Outlook Survey; Census story 'Large Firms With at Least 20 Employees Biggest AI Users' (26 May 2026): 19.8% of businesses used AI in the prior two weeks as of 3 May 2026 (range 17-20% Dec 2025-May 2026); expected use 20-23%; 37% for firms with 250+ employees; information sector 39.7%. Fed FEDS Note (Allen, 3 Apr 2026): ~18% of firms (firm-weighted) at end-2025; 78% of labour force at AI-using firms (Atlanta Fed SBU)","US, May 2026",adoption share in the delegation block,official survey (story read; data page JavaScript-only),moderate: question wording changed Nov 2025 (series break); firm-weighted vs employment-weighted differ hugely (18% vs 78%),https://census.gov/library/stories/2026/05/ai-use-businesses.html ; https://www.federalreserve.gov/econres/notes/feds-notes/monitoring-ai-adoption-in-the-u-s-economy-20260403.html ; https://www.census.gov/hfp/btos/data,fetched 2026-10-2751generative-AI use by adults / workers,a_H,share,0.4,0.23,0.41,"Bick, Blandin & Deming (2024) 'The Rapid Adoption of Generative AI', NBER WP 32966, doi:10.3386/w32966 (rev. Feb 2025; Management Science 2026): nearly 40% of US adults 18-64 use generative AI; 23% of employed used it for work in the prior week; 9% every workday; 1-5% of work hours assisted. FEDS Note (Apr 2026): ~41% of workers use GenAI at work (RPS, Nov 2025)",US 2024-2025,household delegation participation,literature (NBER abstract read),moderate,https://www.nber.org/papers/w32966 ; https://www.federalreserve.gov/econres/notes/feds-notes/monitoring-ai-adoption-in-the-u-s-economy-20260403.html,fetched 2026-10-2752hyperscaler capital expenditure 2025 (4 firms),I_AI,"USD billion, calendar 2025",376.1,357.5,380,SEC XBRL us-gaap:PaymentsToAcquirePropertyPlantAndEquipment (data/raw/sec_xbrl): Alphabet CY2025 91.447; Meta CY2025 69.691 (72.22 incl. finance-lease principal per 8-K ex.99.1); Amazon FY2025 131.819 (earnings-release cash-flow statement; 128.32 net of proceeds); Microsoft FY2025 (Jul-24..Jun-25) 64.551 and CY2025 derived from quarterly XBRL = 16.745 + 17.079 (FY2025 minus three 10-Q quarters) + 19.394 + 29.876 = 83.094. Sum CY2025 = 376.1 (with MSFT FY2025 instead: 357.5),US-listed hyperscalers 2025,AI/compute investment flow (~1.2% of US GDP),"computed from SEC filings (verified, except the one derived MSFT quarter)",low-moderate: definitions differ (finance leases; non-AI capex included),https://data.sec.gov/api/xbrl/companyconcept/CIK0001652044/us-gaap/PaymentsToAcquirePropertyPlantAndEquipment.json ; https://www.sec.gov/Archives/edgar/data/1326801/000162828026003832/meta-12312025xexhibit991.htm ; https://s2.q4cdn.com/299287126/files/doc_earnings/2025/q4/earnings-result/AMZN-Q4-2025-Earnings-Release.pdf ; https://data.sec.gov/api/xbrl/companyconcept/CIK0000789019/us-gaap/PaymentsToAcquirePropertyPlantAndEquipment.json,fetched 2026-10-2753data-centre electricity use,E_DC,TWh per year,415,415,1200,"IEA (2025) Energy and AI, executive summary: data centres ~1.5% of world electricity in 2024 = 415 TWh; ~945 TWh by 2030; Base Case ~1,200 TWh by 2035; United States 45% of global data-centre consumption in 2024 (=> ~187 TWh, own arithmetic)",global / US 2024,energy cost of compute,literature (page read),moderate (projections),https://www.iea.org/reports/energy-and-ai/executive-summary,fetched 2026-10-2754US industrial electricity price,p_E,cents per kWh,9.64,8.98,9.64,"EIA Electric Power Monthly, Table 5.6.A: US total industrial 9.64 (Aug 2026, preliminary) vs 8.98 (Aug 2025); all sectors 14.87 (Aug 2026)",US Aug 2026,energy input price,official statistics (HTML table saved in data/raw/eia),low,https://www.eia.gov/electricity/monthly/epm_table_grapher.php?t=epmt_5_6_a,fetched 2026-10-2755compute ownership concentration,HHI_C / top-4 share,share,NOT OBTAINED,0.5,0.9,no verified public measure of installed AI-compute ownership found; the four-firm capex row above is a flow proxy only,global,H3 ownership-concentration experiments,illustrative,illustrative: modeling team to set a range (e.g. top-4 share 0.5-0.9) and run sensitivity,,fetched 2026-10-27; ILLUSTRATIVE56supervision / oversight cost of delegation,kappa_s,share of delegated task value,NOT OBTAINED,0.0,0.3,no public measure; related evidence: Bick-Blandin-Deming report 1-5% of work hours assisted and time savings of 1.4% of hours (not a cost measure),-,H1 delegation threshold,illustrative,illustrative: sweep 0-30% of task value,https://www.nber.org/papers/w32966,fetched 2026-10-27; ILLUSTRATIVE57compute capital depreciation (hardware),delta_C,per year,NOT OBTAINED,0.2,0.33,no direct public estimate fetched; implied economic depreciation from the Epoch price-performance trend (price per FLOP -24.5%/yr) plus physical life of 3-5 years,-,compute capital law of motion,illustrative (derived range),illustrative,https://epoch.ai/blog/trends-in-gpu-price-performance,fetched 2026-10-27; ILLUSTRATIVE58risky return s.d. (Model A portfolio mapping),sigma_R,per year,0.16,0.10,0.20,"Shiller ie_data.xls, real equity s.d. 1984-2025 = 0.157",US 1984-2025,equity volatility,calibrated,+/-0.03 across windows,https://shillerdata.com,Model A baseline59predictable return component s.d.,sig_theta_R,per year,0.06,0.02,0.10,none (illustrative),-,-,illustrative,swept in robustness,,Model A; no accepted empirical counterpart60human forecast noise s.d. (returns),sig_H_R,per year,0.10,0.05,0.20,none (illustrative),-,-,illustrative,swept,,Model A61foundation-model forecast noise s.d. (returns),sig_A_R,per year,0.04,0.02,0.095,none (illustrative),-,-,illustrative,swept (Fig. threshold),,Model A; sig_A < sig_H means the model beats the median human62effective number of independent foundation models,M,count,1,1,8,Spiro 2026 (cross-vendor error correlation ~0.77) motivates small effective M,2026,-,illustrative,1..8 swept,,Model A/C63delegation disutility distribution,kappa,share of consumption per year,"lognormal(mu=-6, sigma=1)",,,none (illustrative),-,-,illustrative,swept,,Model A64supervision time per delegated decision,tau_s,share of time endowment,0.01,0,0.05,none (illustrative),-,-,illustrative,,,Model A65compute per delegated decision (inference intensity),qbar,compute units per household-year,0.002 (A) / 0.02 (B),0,0.08,none; scenario parameter,-,-,illustrative,swept (H6),,Model A/B66mean cost of self-managed investment mistakes,m_H,per unit of wealth per year,0.01,0,0.02,household-finance evidence on underperformance (not re-estimated),-,-,illustrative,,,Model B67dispersion of self-managed idiosyncratic return,sigma_rH,per year,0.04,0.01,0.08,return heterogeneity literature (not re-estimated),-,-,illustrative,,,Model B68dispersion of delegated idiosyncratic return,sigma_rA,per year,0.01,0,0.02,none (illustrative),-,-,illustrative,,,Model B69"delegation fee, fixed / proportional",phi0 / phi1,goods per year / per unit saved,0.02 / 0.002,,,order of magnitude of robo-advisory fees (25 bp) and flat subscriptions,2026,-,illustrative,,,Model B70"dividend mean, persistence, innovation s.d. (Model C)","dbar, rho_d, sigma_d",per period,"0.05, 0.5, 0.01",,,none (illustrative),-,-,illustrative,,,Model C71safe rate per period (Model C),r_f,per period,0.01,,,none (illustrative),-,-,illustrative,,,Model C72perceived payoff variance (Model C agents),sigma2,goods^2,0.01,,,none (illustrative; must satisfy dbar*cash0 > gamma*sigma2*hbar),-,-,illustrative,,,Model C73private / model forecast noise s.d. (Model C),"sig_H, sig_A",dividend units,0.02 / 0.005,,,none (illustrative),-,-,illustrative,,,Model C74"market-maker depth, tick, levels, inventory limit","hbar, varsigma, K, I_max","shares, relative, count, shares","0.5, 0.0005, 10, 10",,,none (illustrative),-,-,illustrative,,,Model C75