Household Demand Analysis under Selectivity and Endogeneity: Evidence from the Czech Consumers

Abstract

This dissertation examines consumer demand analysis, with a particular focus on the econometric challenge of selectivity. The first essay reviews methods for addressing selectivity linked to censored data in the single-equation framework, including the to- bit, two-part, double-hurdle, and sample selection models, the latter implemented with three estimators (FIML, Heckman two-step, and Cosslett's semi-parametric). An ap- plication to the Czech Family Budget Survey data demonstrates that even low levels of zero consumption lead to bias and that elasticity estimates vary substantially across methods, with the Heckman and Cosslett's procedures performing best. The second essay extends the discussion to demand systems, reviewing, among others, the Rotter- dam demand system, the Almost Ideal Demand System (AIDS), and the Exact A!ne Stone Index (EASI), and shows how selectivity corrections can be incorporated into demand systems and illustrates these approaches using the AIDS model. The third essay provides the first Czech evidence on household demand for bottled non-alcoholic beverages. Using a quota-based representative survey from 2022, a QUAIDS model is estimated for five categories. Censored data are treated with the two-step estimator of Shonkwiler and Yen, and missing prices are imputed by...

Tato dizerta!ní práce se zab"vá anal"zou spot#ebitelské poptávky se zam$#ením na problém selektivity. První esej shrnuje metody #e%ení selektivity u cenzorovan"ch dat v rámci jednorovnicového p#ístupu, a to v!etn$ tobitu, two-part, double-hurdle a sample selection modelu, u kterého uvádí t#i zp&soby odhadu: FIML, Heckman&v dvoustup'ov" a Cosslett&v semiparametrick". Jejich aplikace na !eská data ze Statis- tiky rodinn"ch ú!t& ukazuje, (e i nízk" podíl nulov"ch pozorování vede k vych"len"m v"sledk&m a (e odhady elasticit se mezi jednotliv"mi metodami v"razn$ li%í. Nej- lep%í v"sledky poskytují Heckman&v a Cosslett&v postup. Druhá esej roz%i#uje diskuzi na poptávkové systémy, p#i!em( se v$nuje mimo jiné Rotterdamskému modelu, mo- delu Almost Ideal Demand System (AIDS) a Exact A)ne Stone Index (EASI). Na p#í- kladu AIDS modelu ukazuje, jak lze do poptávkov"ch systém& za!lenit korekci selek- tivity. T#etí esej p#iná%í první anal"zu poptávky po balen"ch nealkoholick"ch nápo- jích na !esk"ch datech. Na základ$ reprezentativního %et#ení dle kvótního v"b$ru z roku 2022 odhaduje QUAIDS pro p$t kategorií. Cenzorovaná data jsou o%et#ena pomocí metody Shonkwilera a Yena a chyb$jící ceny jsou nahrazeny dv$ma alterna- tivními zp&soby. Ukazuje se, (e zvolená metoda v"znamn$ ovliv'uje v"sledné odhady. V%echny...