Skip to main content

Consumer Price Index — Statistical Methodology

A technical annex for a fictional national statistics office, from one declarative JSON source. It runs the in-house math engine through the estimators a real price-index methodology depends on, dressed as an official publication.

  • Index theory. The Jevons geometric mean (prod_{i=1}^{n} frac{p_{i,t}}{p_{i,0}})^{1/n}, the Laspeyres aggregate, and the superlative Tornqvist index as a share-weighted sum of log-relatives.
  • Sampling precision. Unbiased sample variance, the standard error of the mean, and a two-sided confidence interval at level 1 - alpha using the normal quantile z_{1 - alpha/2}.
  • Hedonic regression, in matrix form. The ordinary-least-squares estimator hat{beta} = (X^{T} X)^{-1} X^{T} y, with the design matrix laid out as a real 2-D matrix.
  • Closed-form identities. A nested radical and a continued fraction both converging to the golden ratio, and continuous-time discounting int_{0}^{infty} e^{-delta t} {}_{t}p_{x} dt.
  • Dressed as a publication. A drawn agency crest, a SPECIMEN watermark, a Han Xin verification mark, a vertical EAN-13 ISSN spine mounted via a rotated canvas group, and a results table.

Glyphs are positioned from the font's OpenType MATH table and written into the PDF as selectable, searchable text — never a flattened image.

Features used: Shortcodes, Tables, Columns, Canvas, Barcodes, Watermarks, Headers & Footers, Styles