Deborah Swenson opened the discussion by questioning the appropriateness of treating knowledge (technical and managerial know-how) spillover as the only channel through which location proximity impacts on the performance of given firms. She argued that competitive pressure and the demonstration effect emanating from firms in proximity could have a significant positive spillover effect. Swenson also noted the importance of including time dummies in the estimation equation to capture potential business-cycle effects.
Chalongphob Sussangkarn agreed with Swenson that the model specification had failed to fully capture the impact of location proximity. He noted that the paper had missed a large part of the proximity effect by ignoring the spillover effects operating through the labor market and intermediate-goods channels. Lu Ming argued that the median total factor productivity of firms in a given locality, which is used as the key explanatory variable in the model, does not fully capture knowledge spillover. She alluded to the possibility of constructing a better explanatory variable by combining productivity of individual firms with a direct measure of proximity, if the data on the latter variable are available.
Suresh Narayanan noted that the strong results reported in the paper on the knowledge spillover effect need to be qualified for possible omitted variable bias. Some key variables relevant for explaining inter-firm variation in total factor productivity (TFP) such as the firm size, the nature of firm ownership (foreign versus local ownership), vintage factor (age), and export orientations are missing in the list of explanatory variables. If these variables were to be appropriately included, the locational knowledge spillover effect uncovered in this exercise could become much weaker or even disappear. Somkiat Tangkitvanich wanted the author to fully interpret the puzzling finding that firms that are relatively more productive in a given year tend to exit in the following year.