Harris Value research: Market value down, appraised value up (2026) Published September 4, 2026 (Central). Source snapshot August 30, 2026. Report: https://harrisvalue.com/research/2026-market-value-down-appraised-value-up SOURCE HCAD's public 2026 Real_acct_owner.zip: https://download.hcad.org/data/CAMA/2026/Real_acct_owner.zip HTTP Last-Modified: Sun, 30 Aug 2026 21:23:46 GMT Archive bytes: 211812293 SHA-256: 80397483d1e427d75761f62c263ba5905b8342a92d72321ff323a9609ff0c5e9 Members: real_acct.txt and real_neighborhood_code.txt. The URL can change in place; the hash identifies the analyzed snapshot. REPRODUCTION Download the standalone standard-library Python analysis script: https://harrisvalue.com/research/2026-market-value-down-appraised-value-up-analysis.py Use Python 3.11 or later and the archive with the hash above. Choose an output directory: python 2026-market-value-down-appraised-value-up-analysis.py Real_acct_owner.zip --root ./research-output --source-last-modified "Sun, 30 Aug 2026 21:23:46 GMT" The command writes JSON and CSV aggregates under research-output/frontend. Generation time changes on reproduction; the cohort calculations do not. The script never accesses an application database. The public JSON contains only aggregate statistics, area labels, and source metadata. No property identifiers or owner information are published in these research downloads. COHORT AND EXCLUSIONS Read Latin-1 tab-delimited records with exact column counts. Of 1,628,241 source rows, 1 malformed row was omitted before class/year classification. 470,500 rows were outside the 2026 A1 single-family residential cohort. Within 1,157,740 well-formed target records, 14,240 had a missing, invalid, or nonpositive current/prior market/appraised value. There were no duplicate target accounts, invalid target account numbers, or appraised-above-market exclusions. The final matched cohort contains 1,143,500 unique accounts. DEFINITIONS Market down: tot_mkt_val < prior_tot_mkt_val. Appraised up: tot_appr_val > prior_tot_appr_val. Affected: both of these conditions on the same record. Percentage change: 100 * (current - prior) / prior. Headline count: 66,758, or 5.8380% of all 1,143,500 usable records. Market-declining count: 555,039; 12.0276% also had appraised value rise. Sensitivity: (prior market - market) * 100 >= prior market, and (appraised - prior appraised) * 100 >= prior appraised. 50,193 records satisfy both 1% thresholds, applied before rounding. Affected-group medians: market change -3.4129%; appraised change +8.4731%. These are separate medians of individual percentage changes. LOCAL RESULTS Use site_addr_3 as situs ZIP. Accept exactly five digits or explicit ZIP+4; 190 usable records have no valid ZIP. Use Neighborhood_Code joined to the official cd/dscr fields in real_neighborhood_code.txt for HCAD area names. Publish groups with at least 100 usable records: 137 ZIPs and 2,917 areas. 8 ZIP groups (172 usable records) and 2,863 areas (119,146 usable records) are below the threshold. All these records remain in countywide totals. ZIP and area tables are alternate groupings, not additive populations. All local-table percentages use the usable-record count for that group. Area names describe HCAD groupings, not municipal or civic boundaries. LIMITATIONS This is a descriptive appraisal-data snapshot, not a tax-bill, protest-outcome, or causal analysis. The usable source records are labeled Noticed (1,143,484) or Value Unchanged, Notice Not Required (16); these labels do not establish a final certified roll. Prior columns/current values may reflect different stages of appraisal or review. Ownership, improvements, exemptions, corrections, and protests can affect individual records. No property-by-property homestead eligibility verification or attribution of cause was performed. The example on the report page is hypothetical and assumes the residence- homestead cap applies in both years with no qualifying new improvements. CORRECTIONS Analysis and publication: Harris Value, which also operates a property- comparison tool. Send reproducible questions to hello@harrisvalue.com.