Identification of the statutory retirement dates in the Sample of Integrated Labour Market Biographies (SIAB)

"We analyse how administrative data on the labour history of individuals can be used to identify financial incentives within the pension system, even though these data do not include information on pension-relevant periods. We apply the Sample of Integrated Labour Market Biographies (SIAB 7514). The data consist of a two percent sample of the population of the Integrated Employment Biographies from 1975 to 2014 and are provided by the German Federal Employment Agency. We present a method for identifying the pensionable periods for old age pensions. In addition to birth date and gender, we show how to identify the qualification periods to determine whether an individual is eligible for one of the old age pension types (standard old age pension, old age pension for women, old age pension for the unemployed or under a progressive retirement plan, old age pension for persons with a long insurance record). Eligibility for a pension type then determines the earliest statutory retirement dates (normal retirement age (NRA) and early retirement age (ERA)). The knowledge about eligibility for a pension type enables us to compare the actual labour market exit age with the NRA and ERA for each birth cohort from 1936 to 1948 and to calculate the proportions of employees for the different paths out of the labour market. First, we explain the information that is necessary to identify the statutory retirement dates. We cannot identify periods of illness, inability to work, maternity, parenting, caregiving on a non-commercial basis or voluntary insurance payments from the SIAB. To assess the accuracy of pensionable periods calculated using the SIAB, we therefore use a high-quality administrative biographical dataset (Biographical Data of Selected Insurance Agencies in Germany (BASiD 5109)) that combines information on individual employment biographies (including qualification periods) with retirement information from the statutory retirement insurance records. We use the BASiD to collect information on employment states and other relevant variables that are not available in the SIAB. Moreover, we show that we can reduce the errors in identifying the relevant eligibility criteria for old age pension types to a negligible amount when we restrict our sample to employees with a high labour market attachment and short gaps in their labour market histories. We argue that the employees in our reduced sample are the employees of interest for analysing the impact of the financial incentives of the pension system on the labour market behaviour of older employees. Only these employees have a real choice of whether to work another year or to retire. We conclude that we can reliably identify individual statutory retirement dates in conventional individual labour market history datasets that do not directly contain retirement information. The additional information we generate makes these data sets a valuable alternative for the analysis of the labour market behaviour of older employees." (Author's abstract, IAB-Doku) ((en))