Measurement Errors in Trade Statistics

Trade data obtained from the UN Comtrade and IMF DOTS yields very different estimates of the trade effects of the Euro. For example, the OLS estimator yields coefficients of the euro with opposite signs for the two data sources, when a sample covering all countries is applied. Trade statisticians have done lots of work to maintain trade data and to improve the quality of the data. I hope to provide some feedback to the data from the users' side. Actually, there are some obvious inconsistencies in trade statistics for a given country pair in a given year. Taken export value from Angola to Australia reported by Austrilia in 2006 as an example, the value in Comtrade is 0.01 while in DOTS is 52760.47, which is obviously wrong given values in other years (as can be seen in the following table). There are other similar cases which might affect empirical estimations using trade data heavily. You can find a summary of inconsistency in trade statistics in the two data sources here. I believe some efforts to reasonably correct such inconsistency is helpful for empirical analysis. You can find more details regarding measurement errors in trade statistics in my forthcoming paper.