Backus et al. (Am Econ Rev 84:84-103, 1994) introduced a new theoretical interpretation of the short-run relationship between the terms of trade and the trade balance: movements in net exports and the tendency for the trade balance to be negatively correlated with current and future movements in the terms of trade, but positively correlated with past movements. There have been several papers aimed at testing the S-curve hypothesis in the related literature. The most common empirical approach used in those papers is the cross-correlation analysis. This paper highlights several deficiencies of the cross-correlation analysis when being used directly to test for the validity of the mentioned hypothesis. The purpose of this paper is to offer an alternative approach so as to combine the causality theory in econometrics with the testing procedure of the S-curve hypothesis via the cross-correlations. The paper's empirical results reveal that disaggregated bilateral trade data between the United States and China almost never support the S-curve hypothesis, contrarily to a previous paper employing the same data, but utilizing the conventional cross-correlation analysis. Why this difference may have appeared is also discussed.