Oil prices and the cost of IT, 1996-2026: data files Servnet research study: https://www.servnetuk.com/research/oil-prices-it-costs-2026 Build: 2026-09-19T11:47:56Z. Produced by scripts/research/oil-it/analyse.py (standard-library Python, deterministic). This is original analysis of open official data. It contains no forecasts of oil, energy, freight or IT prices and no investment advice. Correlation is not causation. FILES oil-prices-it-costs-2026.csv One row per series and period (full history). Columns: series_id, series_title, publisher, date (YYYY-MM; quarterly and half-yearly rows are dated to the first month of the period), value (as published, or as derived by Servnet where the title says so), unit, licence, licence_url, attribution. oil-prices-it-costs-2026-episodes.csv Index responses after oil price episodes: % change at +3/+6/+12 months, and the excess over each index's own trailing trend. Derived statistics; each row also carries the source series' licence and attribution (source_licence, source_licence_url, source_attribution) and a derived_statistic_licence note. oil-prices-it-costs-2026-passthrough.csv Cumulative 12-month elasticity of each index to Brent, 95% and conservative intervals, p and Benjamini-Hochberg q values, verdict. Derived statistics, with the same source_* and derived_statistic_licence columns. oil-prices-it-costs-2026-now2026.csv Latest reading against the pre-conflict baseline (February 2026, or 2025 Q4 for quarterly series) and a year earlier. Contains third-party levels, so every row carries licence, licence_url and attribution, exactly as in the main CSV. LICENCE Servnet's derived statistics and compilation: CC BY 4.0. Each third-party series keeps its own licence, given in the licence and attribution columns. ECB values are unmodified and may not be modified. BMRS-derived values carry the Elexon attribution requirement. CC BY 4.0: https://creativecommons.org/licenses/by/4.0/ Credit Servnet's compilation as: "Servnet, Oil prices and the cost of IT, 1996-2026 (2026), servnetuk.com". When you re-use any third-party series or a statistic derived from it, keep its attribution line (the attribution column of the main and now2026 CSVs, or the source_attribution column of the episodes and passthrough CSVs). REQUIRED NOTICES Office for National Statistics, Department for Energy Security and Net Zero, HM Revenue & Customs: Contains public sector information licensed under the Open Government Licence v3.0. https://www.nationalarchives.gov.uk/doc/open-government-licence/version/3/ ONS System Average Price of gas: Source: National Gas Transmission. Elexon BMRS (ONS system price of electricity, and the Elexon-derived system sell price series): Contains BMRS data © Elexon Limited copyright and database right 2026. Licence: https://www.elexon.co.uk/bsc/data/balancing-mechanism-reporting-agent/copyright-licence-bmrs-data/ Anyone who re-uses or sub-licenses these values must keep this attribution statement and require any further sub-licensees to do the same (Elexon BMRS licence). World Bank: World Bank Commodity Price Data (The Pink Sheet), licensed under CC BY 4.0. Changes: none (values as published). Eurostat (prc_hicp_minr, sts_sepp_q, nrg_pc_205; accessed 19 Sep 2026): Source: Eurostat. Values as published (unmodified). Eurostat has no responsibility for the results or conclusions drawn from these data, or for any changes made to them by Servnet. Copyright notice: https://ec.europa.eu/eurostat/help/copyright-notice European Central Bank: Source: ECB statistics. The values are unmodified and may not be modified. Japan METI: 出典: 石油統計 (経済産業省) (https://www.meti.go.jp/statistics/tyo/sekiyuka/), processed by Servnet (monthly kl volumes combined from three issues of the confirmed time-series table; Middle East shares calculated by Servnet); the processed figures are not published by the Government of Japan in this form. Public Data License 1.0. Taiwan (Open Government Data License, version 1.0; without this notice the licence is treated as never granted): Energy Administration, Ministry of Economic Affairs (經濟部能源署), 2026, Monthly Data on Crude Oil Import Sources (原油進口來源月資料), data.gov.tw dataset 161212, version as retrieved 19 Sep 2026. The Open Data is made available to the public under the Open Government Data License, User can make use of it when complying to the condition and obligation of its terms. Open Government Data License: https://data.gov.tw/license United States: U.S. Bureau of Labor Statistics, U.S. Energy Information Administration (STEO Energy Security Data Supplement, 12 Aug 2026, EIA analysis based on Vortexa tanker tracking; Electric Power Monthly, 24 Aug 2026), Board of Governors of the Federal Reserve System (H.10), U.S. EPA ENERGY STAR and U.S. Geological Survey: U.S. federal government works in the public domain; please cite the agency as the source. NOT INCLUDED EIA spot prices (Europe Brent spot price FOB, WTI, US Gulf Coast jet fuel) are EIA-published with an upstream source of Refinitiv (LSEG); redistribution is pending EIA confirmation. Their observations are not in these files. The pass-through file includes derived statistics for jet fuel, and the now2026 file gives % changes only (levels blank). Proprietary sources (Drewry, Xeneta, TrendForce, Uptime Institute, JMIC ship counts) are cited on the study page as single dated facts only; no series from them is included. Bank of England exchange rates and IMF PortWatch were not used. COLUMN NOTES trailing_trend_pct_per_year True % a year: (exp(12g) - 1) x 100 for monthly series, (exp(4g) - 1) x 100 for quarterly, where g is the average log change per period over the 24 months (8 quarters) to the baseline. excess_*_pts Actual % change minus the change implied by continuing that trend. index_baseline_period Episodes file: the baseline period of each index. Monthly indices use the baseline month. Quarterly indices use the last quarter ending in or before the baseline month (for a December baseline, Q4 of that year); +3/+6/+12 months means +1/+2/+4 quarters from it. residual_rho1 Pearson correlation of consecutive residuals of the main-sample regression. pct_per_10pct_brent (1.1 ^ elasticity - 1) x 100. verdict robust: conservative 95% interval (widened by 1.25) excludes zero in the main sample and both robustness samples with the same sign, and BH q < 0.05. significant-not-robust: 95% interval excludes zero in the main sample, but the result fails at least one robustness condition. null-precise: interval includes zero and the conservative interval lies within -0.1 to +0.1. null-imprecise: interval includes zero and the conservative interval is wider than that. change_vs_baseline_pct_or_pts % change, or percentage-point change for share series (Japan/Taiwan crude shares, UK aluminium and polymer shares by weight). memory_sensitive Series thought to carry the 2025-26 rise in memory-chip prices (a flag, not an estimate). Data are as retrieved on 19 September 2026 and are subject to revision by the publishers.