Exchange Rate Policies and USA–China Trade Balance

We investigate the effects of the real exchange rate on the trade balance between the USA and China by using a median threshold in non-linear modelling, controlling for several factors. Non-linear autoregressive distributed lag (NARDL) models are estimated with zero threshold, as well as with median threshold, and this is the first time in the USA–China literature of balance of trade. Quarterly data are employed, and the examined sample covers the period 1995–2023. It appears that the threshold choice matters when it comes to policy implications. Specifically, NARDL with the conventional zero-threshold analysis could lead to misleading policy proposals, especially when there are strongly unequal probabilities in the USD appreciation and depreciation regime. In our case, the traditional analysis based on zero threshold leads to diametrically opposite policy suggestions of what the alternative methodology of median threshold would suggest. We find that a nominal USD appreciation suggests a perpetual improvement of the US trade balance, and such an improvement could be further magnified if the US prices increase more rapidly than China's prices (e.g., as in the period from 2020 onwards). Evidence also shows that the US balance of trade could be temporarily benefitted by medium or large USD depreciations in real terms, and such an improvement could be notably facilitated when the US–China prices’ differential is downward sloping (e.g., as in the 2006–2019 period).