On comparing cost of living of cities using expatriate price survey

ABSTRACT

Commercial cost of living indexes, which are nominal Laspeyres price indexes derived from expatriate price surveys, are increasingly used in policy debate. However, these indexes contain two systematic errors – (1) using nominal exchange rate, instead of the purchasing power parity equivalent, cause upward biases in high income cities and (2) using a fixed basket causes upward biases in cities with different consumption pattern – that make them unfit for the purpose of policy debate if the errors are serious enough. Using a real Fisher index corrects these systemic errors. We showed high congruence in pair-wise comparison between these two indexes, but substantial differences in ranking arising from the systematic errors. Commercial indexes are useful for designing expatriate compensation, but we need a real Fisher index to calculate and rank the cost of living among cities.

KEYWORDS:

• Cost-of-living index
• substitution effect
• Penn effect
• Laspeyres index
• Fisher index

Notes on contributor

Dr Tan Boon Seng is an Assistant Director at the Institute of Singapore Chartered Accountants. His research applies microeconomic theory to investigate managerial and policy problems, and have been published in international journals in the areas of public policy, business and accounting. Dr Tan has held research and academic positions in universities in Hong Kong, Singapore and the USA. He has also served in marketing and strategic planning roles in healthcare companies.

ORCID

Boon Seng Tan http://orcid.org/0000-0002-3856-0084

Notes

1. See http://help.worldwidecostofliving.com.

2. EIU explains that their price relatives uses the arithmetic means as the denominator, as oppose to the usual method of just using the base city price, to obtain the desirable base-invariant property. This explanation is incorrect because base invariance at the level of price relative is achieved with geometric means and not arithmetic means. Either way, base invariance is not achieved at the index number level unless all the weights are equal.

3. The same result can also be developed for the case of the host city with a lower cost of living where budget B2 is to the right of B3.

4. Calculating the real and nominal Fisher indexes requires the corresponding Laspeyres and Paasche indexes.

5. See http://go.worldbank.org/PQ5ZPPYSY0, the 2005 data is the latest available in 2013.

6. See http://www.imf.org/external/pubs/ft/weo/2013/02/weodata/weoselgr.aspx. The data is also available from World Bank's World Development Indicator, but the IMF source updates more quickly.

7. The database reported exchange rate is 1 Cambodia Riel per USD. The reported exchange rate in 2013 is around 4000 Riel per USD. The reported prices appeared to be USD prices.

8. http://data.worldbank.org/about/country-classifications/country-and-lending-groups.

9. We did not report results for the Penn and substitution effects in percentage terms because the theory is built on the absolute terms and not percentage terms.