Quantifying the psychological costs of unemployment: the role of permanent income

Abstract

Unemployment causes significant losses in the quality of life. In addition to reducing individual income, it also creates nonpecuniary and psychological costs. We quantify these nonpecuniary losses by using the life satisfaction approach. In contrast to previous studies, we apply Friedman's (1957) permanent income hypothesis by distinguishing between temporary and permanent effects of income changes. This allows us to account for intertemporal spillovers of income compensations. Our results show that the nonpecuniary costs of unemployment are only half as large compared to a standard estimation without this distinction. Nevertheless, the nonpecuniary costs of unemployment calculated with this modified quantification method are still about two times higher than its pecuniary costs. This confirms the high value of work for life satisfaction.

Notes

¹ The relationship between subjective well-being and utility is explored, inter alia, by Frey and Stutzer (2002) and Clark et al. (2008).

² We will use the terms life satisfaction, well-being and happiness interchangeably.

³ See Clark and Oswald (1994), Gerlach and Stephan (1996), Korpi (1997), Clark et al. (2001), Frey and Stutzer (2000, 2002), Clark (2003, 2006), Di Tella et al. (2003) and Frijters et al. (2006).

⁴ There is strong empirical evidence for the PIH. For example, DeJuan and Seater (1999, 2006) show that permanent income has a highly significant influence on individual consumption decisions. For comprehensive surveys of the literature on empirical tests of the PIH, see Deaton (1992), Browning and Lusardi (1996), Browning and Crossley (2001) and Meghir (2004).

⁵ We follow van Praag et al. (2003) in defining permanent income by the average income over all years a person is in the panel. Intuitively, and abstracting from impatience and interest effects, if the individual knows his past and future income streams and wants to smooth consumption, he will consume his average lifetime income in each period. We use logarithmic income to account for the nonlinear influence of income on individual happiness.

⁶ According to Equation 1, the explanatory variables on both sides of Equation 4 should also include LTUE_it if unemployment lasts longer than 1 year. Our empirical estimates show, however, that LTUE does not have a significant effect on well-being. To ease the exposition, we thus do not consider it in this section anymore.

⁷ For the time horizon h, one could assume that a person anticipates the monetary compensation in case of unemployment, so that the increase in permanent income is effective for well-being over the entire lifetime (h equals life expectancy). Alternatively, one could also assume that people realize the increase in permanent income only from the point of time onwards at which they become unemployed and receive the compensation. In this case, the time horizon h comprises the remaining lifetime after entering unemployment. The consumption-relevant permanent income rises as given by (8) because the individual will spread the compensation only over future periods.

⁸ Equations 8 and 9 yield exact results, rather than approximations, if the transitory income component is constant over time.

⁹ Clark et al. (2001) show that unemployment can cause lower life satisfaction even after one has returned to employment. In the Appendix, we show how this ‘scarring’ effect can be incorporated into our monetarization methodology.

¹⁰ The data used in this publication were made available by the GSOEP Study at the German Institute for Economic Research Deutsches Institut für Wirtschaftsforschung (DIW), Berlin.

¹¹ Dropping individuals who are observed for less than a certain minimum number of years is necessary to make the calculation of permanent income meaningful. Our results are robust with respect to other cut-offs, e.g. 8 or 12 years. Panel attrition is fairly low. For a detailed analysis of the attrition rate and its causes, see Kroh and Spieß (2006).

¹² Figure 2 does not distinguish between the life satisfaction of employed men and women because both are almost identical during the time period examined.

¹³ The reference categories are ‘full-time employment’ and family status ‘single’. When interpreting the coefficients, we focus on the results obtained from the extended model.

¹⁴ We controlled for repeated unemployment spells in a separate regression, but did not find significant results either.

¹⁵ We also ran a separate estimation including a child dummy and an interaction term to account for the possibility that unemployed suffer less if they have children at home. Whereas the child dummy was positive for all individuals as in the estimation above, the interaction term was negative but insignificant. If there is a difference at all, it seems that becoming unemployed causes a stronger drop in the life satisfaction of people with children than of those without.

¹⁶ This standard method has been used by Winkelmann and Winkelmann (1995, 1998) as well as by Blanchflower and Oswald (2004).

¹⁷ Since we model income in a logarithmic form, the marginal impact of absolute income is decreasing. This means that our method yields a compensation that lies in between the so-called ‘Willingness To Accept’ (WTA) and the ‘Willingness To Pay’ (WTP). The WTA tells us how much additional income a person would need to receive to keep his/her life satisfaction at the same level as before becoming unemployed. Formally, this is calculated by exp(− β ₃/β ₁) − l in the truncated model and by exp(− β ₃/(β ₁ + β ₂))− 1 in the extended model. The WTA in the truncated model is thus 344% (412% for men, 270% for women) and 97% in the extended model (128% for men, 74% for women). The WTP refers to the income share a person is willing to give up to avoid becoming unemployed. The WTP is calculated by l − exp(β ₃/β ₁) in the truncated model and by 1 − exp(β ₃/(β ₁ + β ₂)) in the extended model. In the truncated model, the WTP is 78% (81% for men, 73% for women). In the extended model, the WTP is 49% (56% for men, 43% for women).

¹⁸ This refers to Germany, where the other 60% are typically replaced by the unemployment insurance, welfare benefits etc. so that an individual who loses his job experiences a typical pecuniary loss of 40% of his previous income (see Winkelmann and Winkelmann, 1998).