Using Scanner Technology to Collect Expenditure Data*

Abstract

In terms of collecting panel expenditure data, there are trade-offs between the demands imposed on respondents and the detail and coverage of data collected. Comprehensive spending data tend to be cross-sectional whilst panel studies include only limited, aggregated expenditure questions. Recently, economists have begun to use detailed, bar-code-level spending data from household panels collected by market research companies. However, there has not been a detailed assessment of the strengths and weaknesses of this collection method or its implications for the recorded data. This paper seeks to fill this gap by providing an in-depth examination of data from one company, Taylor Nelson Sofres (TNS), recording grocery purchases over five years. We assess how far the ongoing demands of participation lead to 'fatigue' in respondents' recording of their spending and the implications for household attrition, and we provide a detailed comparison of the expenditure data collected by TNS and the wellestablished Expenditure and Food Survey (EFS). Broadly, we suggest that problems of fatigue and attrition may not be particularly severe, though there are differences in expenditures that cannot be attributed to demographic or time effects and may be suggestive of survey mode effects.

Goods versus characteristics: dimension reduction and revealed preference

“…Imputing missing unit prices at a much finer regional level, i.e., postcode, is not always possible in the data. 16 Note that these figures are based on observed data only.…”

“…The relative weaknesses come primarily as a consequence of attrition and non-response. See Griffith and O'Connell (2009) and Leicester and Oldfield (2009) for further descriptions of the Kantar data.…”

Estimating households' willingness to pay

“…However, 37% of households spend more than 1% of their budget on organic products, and there are a small number of households (7%) that spend over 5% of their budget on organic products. These numbers illustrate the tremendous heterogeneity in demand for organic products, and that organic is an important 13 See Leicester and Old…eld (2009) Table 4. In principle, we could estimate a nonparametric hedonic function h g for each product category g: However, the characteristics vector z is high dimensional.…”

Design of optimal corrective taxes in the alcohol market

“…In particular, we show that for all ethanol purchases and on-trade only ethanol purchases, households that buy relatively large quantities of ethanol get a relatively high share of the their ethanol from spirits and a relatively low share from beer and from cider. For a more detailed description of the data, see Griffith and O'Connell (2009) and Leicester and Oldfield (2009);Griffith et al (2013) contains more information on the alcohol segment of the data. We use a sample of 10,289 households, which we observe repeatedly throughout calendar year 2011.…”