Re-Engineering the ISAE Manufacturing Survey

The Joint harmonized Manufacturing survey for Italy, carried out by the Institute of Studies and Economic Analysis (ISAE, formerly ISCO), began on a quarterly basis in 1959, becoming monthly in 1962. The survey was broadly modified in several occasions: in 1986 it was re-designed in order to provide data also at the regional level, adopting a new stratified random sample, the strata represented by the sector, region and size of the firm. In 1998, the sample was upgraded further, using an optimal allocation of the reporting units to the sample strata (Cochran, 1977). These changes satisfied the demand for more detailed and, at the same time, better harmonized data. However, the processing of the results was still based on a very detailed industry grid based on the old NACE1970 classification, re-codified to obtain harmonized data for the Main Industrial Groups and total manufacturing. Size weights were used in the processing of the results, but there were still some differences in the elaboration of the data at the national and regional level, resulting in a not fully-fledged comparability between local and national data. In 2003 ISAE started a re-thinking of the manufacturing survey processing phase. It has reached two main relevant goals: i. The underlying industrial structure for the aggregation of survey results is now based on the NACERev1.1 classification, at the 3-digit level, adapted to take into consideration the structure of Italian economy. ii. The weighting scheme is now based on a coherent system of size weights, based on a four-stage method in which, firstly, the balance Ba,j for question a, firm j, is aggregated in each strata, using the j-firm employees as weights; in the following stages, the result for each strata is progressively aggregated to calculate the Industry total, using value added weights, provided by an external source (i.e., the National Institute for Statistics, ISTAT). The main consequence is that now results at the regional and dimensional level are fully comparable to the ones for the entire industry. Historical data up to 1991 have been recalculated accordingly to the new aggregation scheme.