Comparison of non-survey techniques for constructing regional input-output tables

On the grounds of the long discussion in the literature for the construction of regional input-output (I/O) tables, the present study endeavors to identify the best performing method among the most applied location quotient based non-survey techniques. The analysis uses actual data-in particular, the EU input-output tables for the years 2010 and 2014-and by employing three different statistics, it compares the tables created by the quotients FLQ, AFLQ, SLQ, CILQ and RLQ. Comparison is made in both the technical coefficient tables and the Leontief ones. The results indicate that the AFLQ and FLQ provide better results for values of delta from 0.1 to 0.3, whereas for values of delta from 0.4 to 0.9, the results of the two quotients are not satisfactory. At the same time, using another statistic, the SLQ-CILQ technique seemed to yield satisfactory results. Similar results are also presented for the Leontief tables. Finally, using a linear regression model, the imports do not appear to be related to the value of delta in the FLQ.