Pension reforms and the lengthening of working life: An analysis of life-cycle exploitation in Italy

Pension reforms and the lengthening of working life: An analysis of life-cycle exploitation in Italy Description of the data and file structure This dataset contains the empirical material used to estimate the evolution of working life, retirement duration, necessary labour, pension-related transfers, and life-cycle exploitation in Italy. It accompanies a study of how pension reforms and distributive change have affected the division of lifetime labour between necessary labour and surplus labour. The dataset combines macroeconomic, demographic, labour-market, and pension information for Italy over the period 1960–2025, with some variables extended through cohort-based projections where required by the model. The file dataset_Italy.xlsx reports annual macroeconomic and distributive variables, including working hours, employment, net income, the Monetary Expression of Labour Time (MELT), average wages, average pension income, and the labour-time equivalents of workers’ and pensioners’ reproduction costs. The file AP.xlsx contains cohort-level estimates of age of labour-market entry, effective retirement age, duration of working life, expected retirement duration, expected year of death. Two auxiliary datasets provide the demographic and labour-market foundations for these calculations. The file employment_by_age_1960_2025.csv contains single-age employment-rate estimates for Italy from 1960 to 2025. Observed ISTAT age-class employment rates are used where available, while earlier missing values are reconstructed through a constrained backcasting procedure in logit space as described in the paper. The file istat_italy_life_tables_1960_2025.csv contains age-specific survival probabilities and life-table information for Italy, used to compute expected survival and retirement duration. Values before 1974 are reconstructed through a backcasting procedure in logit space as described in the paper. Together, these files allow the reconstruction of the paper’s main variables: the duration of working life $A$, the duration of retirement $P$, annual and life-cycle necessary labour for active workers $\alpha$, annual and life-cycle pension-related necessary labour for retired workers $\beta$, the mass of unpaid labour $\Pi$, and the rate of exploitation over the life cycle $s$. The dataset is intended to make the empirical procedure transparent and reproducible, including the construction of variables that are not directly available in official statistical sources, such as average age of labour-market entry, cohort-specific working-life duration, and labour-time equivalents of wages and pensions. The data draw on multiple official and scholarly sources, including ISTAT national accounts and historical statistical publications, OECD data on effective retirement age and employment, Penn World Table data on hours worked, Eurostat life-expectancy data, INPS pension statistics, and reconstructed estimates based on Fullin and Reyneri’s survey-based evidence on age of first labour-market entry. Details on the theoretical framework, variable definitions, and computation procedures are provided in the accompanying manuscript and appendices. Files and variables File: dataset_Italy.xlsx Description: The file dataset_Italy.xlsx reports annual macroeconomic and distributive variables, including working hours, employment, net income, the Monetary Expression of Labour Time (MELT), average wages, average pension income, and the labour-time equivalents of workers’ and pensioners’ reproduction costs. Variables Variable Description H Annual hours worked per worker N Total employed persons Y Net income (Euro) MELT Y / (N x H) W Average annual nominal wage CAT Average annual nominal pensione treatment TP Pension treatments per person b Average annual nominal pension income alpha W/MELT beta b/MELT alpha_adj alpha / H beta_adj beta / H File: employment_by_age_1960_2025.csv Description: The file employment_by_age_1960_2025.csv contains single-age employment-rate estimates for Italy from 1960 to 2025. Observed ISTAT age-class employment rates are used where available, while earlier missing values are reconstructed through a constrained backcasting procedure in logit space as described in the paper. Variables year age employment_rate File: istat_italy_life_tables_1960_2025.csv Description: The file istat_italy_life_tables_1960_2025.csv contains age-specific survival probabilities and life-table information for Italy, used to compute expected survival and retirement duration. Values before 1974 are reconstructed through a backcasting procedure in logit space as described in the paper. Variables Year Age Survival probability method of extrapolation for missing data annual_mortality_improvement_used for extrapolation source_file from ISTAT (where available) Survivors Deceased Death probability (thousands) Years lived Life expectancy File: AP.xlsx Description: The file AP.xlsx contains cohort-level estimates of age of labour-market entry, effective retirement age, duration of working life, expected retirement duration, expected year of death. Variables year_of_entry: year of entry in the labour market (1) age_of_entry : age of entry in the labour market (2) year_of_exit: year of exit from the labour market (3) Effective retirement age (4) A + U: (4) - (2) P: expected years lived in retirement Expected_year_of_death U: time spent out of employment between (1) and (3) A: time spent in employment between (1) and (3) File: exploitation_1960_1980_clean.ipynb Description: Jupyter notebook for producing the empirical figures and tables of the paper on life-cycle exploitation in Italy. It loads the annual macroeconomic and pension datasets, computes MELT, life-cycle averages, surplus labour, the rate of exploitation, and the decompositions of their changes. It also generates the main paper figures on retirement age, working life, life expectancy at retirement, annual/life-cycle variables, surplus labour, and Shapley decomposition results. File: employment_over_lifetime_clean.ipynb Description: Clean pipeline for reconstructing age-specific employment rates and computing expected years in employment by cohort. It downloads and processes ISTAT life tables, reconstructs single-age employment-rate profiles from ISTAT age-class data, backcasts missing employment rates to 1960, produces diagnostic figures for the appendix, and combines employment rates with survival probabilities to calculate cohort-level expected years in employment used by the main exploitation notebook. Access information Data was derived from the following sources: Information on the average effective retirement age comes from the OECD (2019; 2021; 2023; 2025a) and OECD (2025b). Data on life expectancy at retirement age come from Eurostat (2024b). There are no publicly available datasets on the average age of first entry in the job market, while this information is indispensable to compute $A$ and $P$. Therefore, we compute an estimate based upon ISTAT survey data as presented by Fullin and Reyneri (2015) to apply the procedure detailed in Appendix 4. Data on net domestic product were obtained from ISTAT, while data on total hours worked in the economy come from the Penn World Tables. The construction of α(t) and β(t) requires the availability of measures of the average annual pension and wage. Data on average wages and data on net social contributions are drawn from various issues of the Annuario Statistico Italiano published by ISTAT from 1961 to 2024. Employment data are obtained from OECD (2025c) Average pension data come from national statistical sources: ISTAT (2025) and INPS (2017; 2018; 2019; 2020; 2021; 2022; 2023; 2024).