Learning-by-Exporting: Micro-dynamic Evidence from Taiwan

Do firms become more efficient after becoming exporters? Widespread empirical evidence indicates that exporting manufacturers achieve higher productivity levels than non-exporting manufacturers. Two explanations, self-selection and learning-by-exporting, are proposed in the literature, with the self-selection theory arguing that firms with higher productivity levels self-select into the highly competitive export market, whilst the learning-by-exporting theory argues that firms become more efficient after becoming exporters. Although the self-selection effect has the support of many empirical studies, the learning effect still has little or no empirical support, and is inconsistent with many of the micro-survey studies. We illustrate numerically that since it is heavily reliant upon an examination of productivity differential between exporters and non-exporters, the commonly-adopted approach may suffer from an underestimation of the learning effect in the export market. Such underestimation is also apparent when the underlying economy becomes more open to world markets. As a complement, we provide an alternative empirical strategy based upon active learning theory, and apply this to Taiwanese manufacturing census data for the years 1986, 1991 and 1996. We find that learning effects are empirically supported for both the export and non-export markets. Furthermore, we find that the learning effect in the non-export market is much stronger than in the export market.