In 2011 a comprehensive reform of the Norwegian public pension system was implemented. A key feature of the new pension scheme is the decoupling of the decision to retire from the labour force and the decision to claim old age pension benefits. Agents eligible for early pension take-up can claim pension benefits in the age range 62 - 75, regardless of whether they retire from the labour force or not. When a potential claimant delays pension take-up by, say, a year, she forgoes pension benefits this year - but annual pension benefits for the rest of her life are increased. The decision to delay claiming can therefore be thought of as buying an annuity.
Actuarial neutrality of a pension scheme requires that the expected value of future benefits is the same regardless of the timing of the pension take-up. In Norway, pensions are actuarially adjusted based on average longevity measures specific to each birth cohort. Since individual expected longevities may differ from the average longevity of a birth cohort, there is a potential scope for adverse selection. Individuals with lower than average expected longevity may increase their expected lifetime income by claiming as early as possible, while individuals with higher than average expected longevity may increase their expected lifetime income by delaying the pension take-up.
When the pension scheme was implemented in 2011, all four combinations of retirement and claiming turned out to be rather common. Brinch et al. (2013) find some positive correlation between retirement and the claiming of pension benefits, but the relationship is far from perfect. In particular, the authors find that claiming is strongly associated with predictors of expected longevity.
Individuals who retire without claiming pension benefits must rely on previous savings or some income stream to finance current consumption. Being in a couple may facilitate this option if agents can rely on the partners' income while delaying the pension take-up. This thesis investigates whether empirical evidence suggests that couples coordinate on claiming pension benefits.
This thesis relates to two strands of the economic literature: the demand for annuities and economic models of household behaviour. There is no consensus among economists on the determinants of annuity demand, nor on the extent to which the household cooperates and pool their resources. The theoretical parts of the thesis investigate both issues thoroughly.
The starting point for the empirical investigation of couples' take-up decisions are all Norwegian citizens who were born in 1949, members of a couple and eligible for early pension take-up in 2011. For these individuals I have data on annual incomes (2010 and 2011), wealth (2010), partner's income (2010) and pension take-up. I also have data on the relative money's worth (RMW) of delaying pension take-up from age 62 to 67. In their study of adverse selection in the Norwegian pension scheme, Brinch et al. (2013) estimate a mortality model and simulate the life span of each individual in the 1949-cohort. The authors use the expected longevity to calculate the RMW of delaying pension take-up for each individual who was eligible for early pension take-up in 2011. The RMW is defined as the lifetime expected benefits conditional on claiming at age 67 in terms of expected benefits conditional on claiming at age 62.
All regressions are done in Stata. The first empirical strategy is to investigate the decision to retire without claiming pension benefits. First, I estimate a linear probability model where the dependent variable is a dummy variable taking the value 1 if the individual retired without claiming pension benefits. I use own income and wealth, partner's income and the RMW of delaying pension take-up as regressors. The only significant regressor is a dummy variable indicating whether the individual is in the upper quartile of the income distribution in the sample. Importantly, wealth, partner's income and the RMW of delaying pension take-up are not significant. In the second step, I exclude individuals who did not retire from the sample, and use the same regressors to estimate the probability of not claiming pension benefits. Three regressors are significant: a dummy variable indicating whether the individual is in the upper quartile of the income distribution, a dummy variable indicating whether the individual is in the upper quartile of the wealth distribution and the RMW of delaying pension take-up. Partner's income is not significant. It is tempting to interpret the estimated result as the probability of not claiming benefits given that an individual has retired. Since the decision to retire and the decision to claim pension benefits may be simultaneous decisions, the estimated result should, however, be interpreted with caution.
The second empirical strategy is to investigate couples where both members were born in 1949 and eligible for early pension take-up in 2011. First, I estimate a linear probability model on the sample of couples where only one member claimed benefits, using the difference in RMW as the sole regressor. The difference in RMW is not significant. Thereafter I estimate the probability of wife claiming while the husband did not claim, using the RMW of delaying the take-up of both the wife and the husband as distinct regressors. None of the regressors are significant, and we would not reject the null hypothesis that all coefficients are zero. Similarly, I estimate the probability of husband claiming while the wife did not claim, using the RMW of delaying the take-up of both the wife and the husband as distinct regressors. In this regression the RMW of delaying the husband's take-up is significant.
At last, I investigate whether empirical evidence suggests that individuals in the sample understand the incentives in the pension scheme. I estimate the probability of the wife claiming benefits, disregarding the husband's decision. I use the RMW of delaying the take-up of both the wife and the husband as distinct regressors, and find that only the RMW of delaying the wife's take-up is significant. I run a similar regression estimating the probability of the husband claiming pension benefits, regardless of what his wife did, and find that both regressors were significant. The findings suggest that claiming is strongly associated with the RMW of delaying pension take-up, and are thus in accordance with the findings of Brinch et al. (2013).
None of the empirical findings suggest that Norwegian couples coordinate on claiming pension benefits. Empirical evidence suggests, however, that the individuals understand and respond to the incentives in the pension scheme. Lack of coordination should therefore not be interpreted as lack of understanding of the incentive structure..