Mortality of firms in France from 2000 to 2017 : a longitudinal and cross-sectional demographic analysis[Mortalité des entreprises en France entre 2000 et 2017 : une analyse démographique longitudinale et transversale]

Business failure and survival is a subject of interest to researchers from a variety of backgrounds. Demographers have shown little interest in business demography and, more specifically, business survival and failure, although demographic analysis methods are increasingly being used in this area. Due to the lack of longitudinal data over a long period of time, researchers have very often studied the survival of firms at five years and rarely at ten years after creation. Previous research has also used smaller samples than might be desirable. Therefore, it is necessary to analyse the survival of firms over a longer period, in this case 17 years, and over several different cohorts, in this case 18 cohorts. Moreover, the results of previous studies on the factors of business failure do not take into account the period effect and the specific conditions prevailing in the year of creation, hence the need for an analysis of the factors of business mortality in terms of age, period and cohort. This study aims to analyse the survival of firms 17 years after they were created and to analyse the age, period and cohort effects on firm mortality in France. We use individual open data from the Bank of France. The study is based on a sample of 391,474 companies created between 2000 and 2017. Longitudinal and cross-sectional demographic analysis allowed us to establish lifetables by regions, by cohort and by period and to derive the statistics of the firm lifespans, including mortality rates and survival probabilities. Age-period-cohort modelling from aggregate data, classified by cohort and by age, allowed us to identify age, period and cohort effects on the mortality rates of the firms. Our results indicate a survival rate of 17 % seventeen years after creation for metropolitan France. The survival rates differ by region. They are negatively correlated with the hierarchical rank position of the regions. Finally, the results of the age-period-cohort analysis indicate, on one hand, a decline in the longevity of firms and, on the other hand, lower survival probabilities for the cohorts created during the 2008 financial crisis and those that went through it at younger ages. Conclusion With information on the cohorts and on the period of observation, the Age-Period-Cohort model would predict firm mortality rates by age, and vice versa. JEL codes: R11, G33, J11