Study findings conducted at the macro level using aggregate industry data or at the micro level using firm level data, have not shown convincing evidence that investment in automation will improve productivity and provide high enough return on investment. Problems of mismeasurements, mismanagement, time lags, and redistribution of profit may account for part of the reasons for the so-called 'productivity paradox'. Complexity, interactions, mixtures of strategic and operational issues makes it extremely difficult to isolate the effect of automation investment even at the firm or enterprise level. This empirical study is based on a survey conducted by the Statistics Division, Ministry of Economic Affairs in Taiwan on the status of automation in the manufacturing sector at the factory level that allows a better resolution of productivity. Two variations of the Cobb-Douglas are specified to model the production function of the 18 industries in the manufacturing sector. The results show that automation investment provides high enough return on investment (ROI) and return on productivity (ROP). Cross industry analysis indicates that for those industries with higher output capital ratios, the ROI and ROP will also be higher. Investment in automation may provide synergistic effect, if it will improve the output-capital ratio of the automation capital that in turn will lead to increasing ROI and ROP for both the automation and non-automation capital.