Reference period for the Community Innovation Survey: two or three years?

This paper analyses the effect of the length of the reference period on the results provided by a Community Innovation Survey (CIS). Since the first CIS survey (CIS1) to the last one (CIS4) a three year reference period has been used so as to identify innovative firms. Only for the CIS light survey (carried out between CIS3 and CIS4) a small number of countries, including Luxembourg, have collected innovation data based on a two year reference period. The inclusion of our CIS light national results in that analysis gives insights on the impact of a two year reference period on the CIS results. To do so, we compare some CIS light results to some CIS3 and CIS4 results. In doing so, it has to be pointed out that comparison between results from different survey has to be made with care. Indeed, some modifications in the concepts, methodology and definitions used may have a significant impact on the results provided and as a consequence lead to difficulties in the comparison process. More economic environment, which changes over time, is also likely to significantly impact firms' innovation activities. In order to circumvent these difficulties and to have additional insight, additional data on innovation activities launched in 2005 or planned to be launched in 2006 were collected through our CIS4 survey (the field of the survey took place in the beginning of 2006). Based on these data we build a second two year reference period (2005-2006) in line with the methodology and definitions used in three year reference period (2002-2004). As an introduction to this paper, we present advantages and disadvantages of respectively a three year reference period and a two year reference period. We also put our attention on a first insight regarding the persistence of innovation. To our opinion, this persistence and its impacts are key points in the resolution of two issues related to the length of the reference period that next CIS surveys will have to face: (1) should we alternate full CIS survey based on a three year reference period and light one based on a two year reference period, (2) should we rather opt for a specific reference period for all the next surveys (full or light) and in that case which one? 1 Paper to be presented at the 32nd CEIES seminar on ?Innovation indicators ? more than technology?, Aarhus, Denmark, 5 and 6 February 2007. 2 A first part is dedicated to present the methodological aspects. More precisely, we present, over the last CIS data collections, some changes in the methods used for collecting and producing the datasets and variation in firm's environment. Indeed these aspects are likely to affect the results and as a consequence have to take into account in order to make reliable comparisons over the CIS results. A second part aims to the examination of the impact of the length of the reference period on the propensity to innovate by size and sector (manufacturing industry / service). A third part is dedicated to the profile of the innovative and non innovative firm through different lengths of the reference period: with a shorter reference period are the innovative firms, in comparison to the non innovative ones, more often involved in international markets, do they more often belong to a group, do they more often belong to specific sectors? In a last part we compare innovative firms' behaviour between CIS3, CIS light and CIS4. In order to do so, we put our attention on the following themes related to innovation process: sources of information, innovation activities and effects of innovation.