From 1951 to 1981 the labour force participation rates of men in Britain aged 60-64 fell from 87.5% to 74.6%, and of men aged 65-69 from 47.2% to 17.0%. A number of competing explanations for this reduction cite health, the economic status of older workers, and `structural' conditions in the labour market. None of these monocausal explanations has proven convincing, however, and there have been surprisingly few attempts to test the competing hypotheses empirically. In Discussion Paper No. 284, Research Fellow Paul Johnson conducts such an empirical analysis, using cross-section data on age of workforce in 34 industrial sectors from the decennial censuses of 1951 to 1981.
The censuses contain no explicit data on retirement. Johnson constructs an an `inferred retirement rate' for each sector by comparing the size of the workforce aged 60-64 at the time of each census with that aged 65-69. The early retirement rate is constructed by comparing the size of the workforce aged 55-59 with that aged 60-64. Johnson then regresses the inter-industry variation in these inferred retirement rates on inter-industry variation in the standardized mortality rate, the average weekly earnings of male adult manual workers in each sector and the proportion in each sector covered by an occupational pension scheme. The rate of unemployment in each industry is a structural indicator of general labour market conditions, while trade union density may indicate how far entry to and exit from the workforce is governed by agreements between employers and unions.
The regressions show that health and economic factors have little influence on retirement at age 65, which is dominated by the structural factor of high trade union density and, to a lesser extent, by high unemployment. This suggests that changes in labour force participation at age 65 are the outcome of external structural influences which stem from the desire of unions and employers to manage exit from the labour force. In contrast, retirement before age 65 is dominated by the economic variables of high average weekly earnings and membership of occupational pension schemes, suggesting that early retirement was largely voluntary. In 1981 there is also a strong positive relationship between high standardized mortality rates and early retirement; the labour shake-out of the late 1970s and early 1980s may have encouraged employers increasingly to use health status as a reason for redundancy.
By showing that retirement and early retirement behaviour are influenced by different factors, and that the importance of these factors has changed over time, this research demonstrates why earlier work that focused on monocausal explanations could not generate robust results. The paper also has a clear policy implication, Johnson notes. An increase in the normal age of retirement may become necessary as pensioner dependency ratios rise in the first three decades of the next century. The increasing infl uence of trade union density on the cross-industry pattern of retirement suggests that this would require the agreement of both trade unions and employers' groups. Yet there is little incentive for employers to raise retirement ages when, in order to manage their internal labour markets, they tend to pay older workers a remuneration greater than their marginal product; and there is little incentive for unions to bargain for an extended working life for members who are increasingly opting for retirement before age 65

The Labour Force Participation of Older Men in Britain, 1951-81
Paul Johnson

Discussion Paper No. 284, December 1988 (HR)