Note that the estimates should ideally internalize the fact that the rates of decline in
tariff (and hence the terminal year for achieving the zero tariff point) for different products
are different. There are also some products for which the zero tariff point will never be
reached as they are outside the ambit of the FTA. Pace of reduction of tariffs of a particular
industry will depend on the tract in which it is included. Though we have discussed this in
section 2.2 we have also added the year for which the tariff rate will actually become zero.
Finally it should be pointed out that the methodology that we have used has several
limitations. First is regarding data availability. For running the regressions we need
continuous trade and tariff data at six digit HS from 1988 to 2009. HS Combined, in which
the TRAINS dataset is reported, has a total of 5050 products. Table 3 column 2 lists the
number of products that could be used in this analysis – the rest had to be abandoned due to
non availability of data (especially continuous tariff data over time). Since the analysis could
not be conducted on many products it has imparted a bias to our results and the results are not
comparable to Bhattacharyya and Mandal (2009) which dealt exclusively with the aggregate
data. Secondly Ordinary Least Square method (adjusted for autocorrelation) was used and
hence the estimates are biased. The method is however justified in this case as we do not have a long enough time series to implement the standard methods of time series analysis such as determining their level of stationarity and their long run relationship. To recognize this problem we have reported the correlation between tariffs and imports (and tariff of the ASEAN countries and India’s export) wherever possible.