Slovenia affordability analysis — snapshot retrieved 2026-10-09

Run `python analyze.py` with Python 3, numpy and scipy installed. Archived inputs
are self-contained. The script writes annual-metrics.csv and results.json under
dist/data. Open dist/index.html for the archived report, charts and methods.

inputs/reviewed_series.json contains reviewed linked wages, CPI and HPI.
Wages: SURS Yearbook 2005 table 13.1; tables 0772610S and 0701023S.
Link old/new wage levels at overlapping 2023; use new-source growth thereafter.
HPI: SURS 0419005S; 2007 mean of quarterly 0419001S. Base 2015=100.
CPI: World Bank FP.CPI.TOTL, IMF IFS source.
inputs/mortgage.csv: ECB MIR.M.SI.B.A2C.A.R.A.2250.EUR.N.
income_all.json: Eurostat ilc_di03?geo=SI&sex=T&age=TOTAL.
Select statinfo=MED_EI, unit=EUR; shift survey year to income year minus one.
burden.json: Eurostat ilc_lvho07c?geo=SI&unit=PC.
rents.json: SURS PxWeb 0400600S group041, measurement50, all months.
rents_meta.json records original source metadata.

Index formula: (cost_t / income_t) / (cost_2015 / income_2015) * 100.
Mortgage index: price/wage index * annuity(rate_t,term)/annuity(rate_2015,term).
Constant LTV cancels. Monthly annuity has principal+interest; actual EU-SILC
overburden includes interest but excludes principal. These are distinct metrics.

Exploratory regression: 100*Delta ln(HPI/wage) = alpha + beta*Delta rate_pp + error.
18 annual changes, 2008–2025. OLS, Bartlett Newey-West lags0/1/2, n/(n-k)
covariance correction, approximate Student-t(16) intervals. Not causal.
No forecasts, composite scores, missing-year imputation or microdata claims.
