By constructing the world panel dataset for the pollution emission embedded in international trade of 132 countries for the period between 1988 and 2008, we investigate whether the balance of embodied emission in trade (BEET) is consistent with the implication of pollution haven hypothesis (PHH). By using two differently constructed datasets, we are able to distinguish between the composition (i.e., changes in industry structure of international trade) effect and the technique (i.e., improvement in emission abatement) effect. We find that the composition effect is neither related with the income level nor the democracy level of countries. However, the empirical evidence, with consideration for the technique effect, provides a partial support that income level is negatively related with the BEET. Therefore, the BEET in fact has worsened for developing countries, but not by the composition effect assumed in the PHH.