RDP 2007-10: Trade Costs and Some Puzzles in International Macroeconomics Appendix A: Notes on Data

The following variables were used in the analysis:

1. Final consumption expenditure (2000 US$, converted using exchange rates) from World Development Indicators (World Bank) (transformed into per capita terms using GDP and GDP per capita data). This includes household final and general government final consumption expenditure. Tables 5 and 6 each include a regression which includes the statistical discrepancy in the consumption measure.
2. GDP per capita (2000 US$, converted using exchange rates) from World Development Indicators (World Bank).
3. Investment (gross capital formation as a percentage of GDP) from World Development Indicators (World Bank).
4. Real effective exchange rate (value in 2000 indexed to be 100) from World Development Indicators (World Bank). Available from 1975. Because of this and the regressions including the lagged real exchange rate, the analysis for this is limited to 1976–2001.
5. Saving (gross domestic saving as a percentage of GDP) from World Development Indicators (World Bank).
6. Trade-cost data: US imports FAS, CIF and duties by SITC 5-digit code, country and year from 1974 to 2001 from <http://cid.econ.ucdavis.edu/data/sasstata/ usiss.html>. The data have been aggregated to the 3-digit level. For all of the analysis I have dropped problematic observations such as where trade costs or FAS costs are negative. FAS cost is measured by the custom value. Before 1989, the FAS (customs value) and CIF numbers are reported directly. After 1988, the CIF numbers are calculated as the sum of FAS and charges. Trade costs without duties are the percentage CIF is above FAS. Trade costs with duties include duties as well. The percentages are approximated using the natural log function. Note that similar trade-cost data are not available for US exports. FAS cost does not include loading costs while FOB cost does.
7. Trade-related data from Glick and Rose (2002). This included information about distance and geography used in some of the analysis. These data are available from Rose's website (<http://faculty.haas.berkeley.edu/arose/>). Glick and Rose's data are only available up to 1997.
8. Capital restrictions data from the IMF's Annual Report on Exchange Arrangements and Exchange Restrictions. Data for Switzerland are not available over the whole sample.
9. Penn World Table versions of consumption and output. Output is real GDP per capita (Layspeyres) in 2000 constant prices and consumption is derived from consumption share for this series. These measures are in PPP.

In addition, some analysis used demographic and real GDP data from the World Development Indicators (such as the last row of Table 2, where some observations were lost in calculating GDP growth as a proxy for productivity). Also, some analysis used bilateral (with respect to the US) real exchange rates based on data used by Imbs et al (2005), and bilateral real exchange rates with respect to the US, calculated from national sources.

The analysis is for Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Greece, Iceland, Ireland, Italy, Japan, the Netherlands, New Zealand, Norway, Portugal, Spain, Sweden, Switzerland and the United Kingdom (the United States is not part of our sample as it is not covered by United States import data). Note that because of data availability, the sample varies somewhat across regressions. Because the Glick and Rose data do not include data after 1997, the coverage of information on bilateral trade and free trade agreements is more limited in the more recent sample. Also, the Glick and Rose data are not comprehensive (for example, it has few observations for Belgium). However, it is unclear how this would bias the results and it seems likely that the Glick and Rose data should be relatively comprehensive given the available data sources.