A comparison of micro and macro expenditure measures across countries using differing survey methods Kevin Milligan ~ Vancouver School of Economics ~ University of British Columbia

A comparison of micro and macro expenditure measures across countries using differing survey methods

with Garry Barrett and Peter Levell
Google Scholar entry.

In Christopher Carroll, Thomas F. Crossley, and John Sabelhaus (eds.) Improving the Measurement of Consumer Expenditures.
Amazon.com page for book, to be published December, 2014: link.
Conference website, December, 2011: link.


Household-level consumption lies at the center of research into many important economic questions. The measurement of microeconomic phenomena such as household poverty requires the observation of consumption choices made by households to provide useful information on economic hardship. At the macroeconomic level, the understanding of responses to booms and busts is enhanced by observing household consumption responses. However, there are ongoing concerns about the reliability of expenditure surveys in many countries.

In this paper, we aim to contribute to these discussions by providing an international comparison of the performance of household expenditure survey data across four 'Anglosphere' countries: Australia, Canada, the United Kingdom, and the United States. Our international comparison is a useful way to gather some evidence on the potential sources of problems with expenditure surveys, as differences in experience and methodology provide sources of variation that may give insights into the importance of factors influencing the performance of expenditure surveys.

Our strategy is to compare household expenditure survey data to expenditure measured in the national accounts of each country. While this 'coverage' approach is frequently adopted in country-specific studies of expenditure behaviour, the novelty of our contribution is to produce comparable results across four countries. Coverage rates are highest in Canada and the UK; for Canada, and Australia coverage remained fairly stable over the past three decades. In contrast, in the UK and the US coverage rates have sharply declined over the past three decades. Next, survey response rates and top income shares were considered in tandem with coverage rates. From a series of graphical comparisons and regression models it is found that the fall in response rates and top income shares over time is quite predictive of changes in coverage rates within countries. The last component of the analysis examined coverage rates for specific components of expenditure. Individual expenditure items considered were food at home, alcohol purchased in stores, new and used motor vehicles, and furniture appliance and household equipment. Most evident is the high and stable coverage of regularly purchased items, along with the more volatile coverage of irregular and larger expenditure items.

Final Draft, March, 2012: PDF.

Return to Research Page.
kevin.milligan@ubc.ca ~ Home Page ~ Vancouver School of Economics ~ UBC