Disadvantage and life-satisfaction

(pages 35-40 of the printed journal)

By Ian M. McDonald ⁱ

Background

In recent years, measures of life satisfaction or happiness based on self-reports have been of great interest to economists and other social scientists. These measures are a natural fit with economists' traditional focus on utility because they can be argued to be a measure of this concept.

Herein the prevalence and intensity of disadvantage are examined using measures of life satisfaction to identify disadvantaged groups and assess their degree of disadvantage. This approach follows earlier researchers, especially Blanchflower and Oswald (2004) and Layard (2005). Inter alia I also consider some policy challenges of improving the outcomes for disadvantaged groups.

Groups which experience low life satisfaction have a prima facie case to be considered worthy of support and help from the rest of society. Such assistance is consistent with the focus in welfare economics on the distribution of utility. Note that the case for using income as the measure of the neediness of a group presumes that income is a measure of utility. Life satisfaction surveys avoid this presumption and so allow us to get closer to the prime focus, that is utility itself.

The nature of utility as an analytical concept has been debated by economists since the latter half of the nineteenth century. One debate is over cardinal versus ordinal measures of utility. Another is the possibility of making interpersonal comparisons of utility. Herein, measures of life satisfaction are used to make interpersonal comparisons. The measure of life satisfaction adopted herein is based on responses to a question in the 2009 Household, Income and Labour Dynamics in Australia (HILDA) survey which asked: 'All things considered, how satisfied are you with your life … pick a number between 0 and 10 to indicate how satisfied you are'.

The disadvantaged groups on which I will focus in this paper are those experiencing low incomes, unemployment, disabilities, mental illness, discrimination, or, in the case of children, disadvantaged environments. The average life-satisfaction scores of people in these groups are calculated and compared.

Factors influencing life satisfaction

Comparing life-satisfaction scores for groups contrasts with the alternative approach of identifying the effect of one characteristic, such as the state of long-term (LT) unemployment, while controlling for all other factors, such as level of education achieved. The approach adopted has the advantage of being more closely related to particular groups of people. Welfare economics focuses on people; and human empathy is directed at people, rather than particular characteristics. Individuals possess many characteristics – for example, the LT unemployed probably have below-average levels of education. By considering the average life-satisfaction score of the LT unemployed, one is thinking of the LT unemployed as individuals with the average characteristics of those in LT unemployment.

Low income and unemployment

Now consider the life-satisfaction scores for the first of our disadvantaged groups, low-income recipients. The black dashed line in Figure 1 shows the average life satisfaction for each income decile. For example, the left-most point shows for the bottom decile an average life-satisfaction score of 7.7 on the 11-point scale.

Figure 1: Life satisfaction, Australia, 2009

For the population as a whole, Figure 1 shows a positive relation between life satisfaction and income. The difference between the average life satisfactions of individuals in the lowest and highest income deciles is 0.38 points.

Hopefully, overall government redistribution has made the life-satisfaction gradient less steep. (There may be exceptions due to welfare dependency and poverty traps.) Consider the pattern of net taxes minus benefits received for income quintiles shown in Figure 2 for Australia for 2003–04. This measure includes tax paid, income transfers received and government services received. Figure 2 demonstrates considerable redistribution in Australia through government intervention, especially from the top income quintile to the lower three quintiles. The positive relation between life satisfaction and income decile would presumably be much steeper were it not for this substantial redistribution.

Now consider the unemployed. Life-satisfaction studies show that unemployment is associated with a loss of life satisfaction. The life satisfaction of the short-term (ST) unemployed – defined as those unemployed for less than one year – in Australia in 2009 is shown as the unbroken black line in Figure 1.

For the ST unemployed, Figure 1 shows the life satisfaction for only the lowest seven income deciles. Not surprisingly, in the top three deciles there were too few people classified as ST unemployed to give meaningful average scores for life satisfaction.

As recorded in Figure 1, the difference between the life satisfaction of the ST unemployed and the population in the same income decile averaged 0.21 points over the seven income deciles for which a meaningful comparison can be made. This difference is labelled the ST unemployment 'toll'. The ST unemployment toll represents about half the difference between the life-satisfaction measures of the lowest and highest income deciles.

Figure 2: Net benefits received, Australia

Now consider the LT unemployed, that is, those unemployed for more than one year. Note that Figure 1 shows an observation only for the lowest income decile for the LT unemployed. Only in this decile were there sufficient numbers of LT unemployed to provide a meaningful score for life satisfaction.

LT unemployment imposes a substantial toll, equal to 0.54 points, which is over twice the ST toll and greater than the difference between the life satisfaction of the lowest and highest income deciles. LT unemployment is shown to be an unattractive state, a phenomenon which casts doubt on the idea that the LT unemployed are, in a meaningful sense, choosing to be LT unemployed.

But even if LT unemployment imposes large losses of life satisfaction, other evidence suggests that the LT unemployed are often employable. Consider the relation between the aggregate and LT unemployment rates shown in Figure 3 (updated from McDonald, 1993). The data in Figure 3 is monthly data for the period February 1978 to March 2011, with observations for the GFC recession shown in black. The Figure shows that decreases in the aggregate rate of unemployment have been associated with decreases in the LT unemployment rate. This is especially evident for the period following the major recessions in the early 1980s and the early 1990s. For example, the most north-easterly observation of 11.2 per cent for the aggregate unemployment rate and 3.7 per cent for the LT unemployment rate for December 1992 was followed by a contemporaneous decline in aggregate and LT unemployment over the next 16 years, to August 2008. It appears that the expansion in economic activity absorbed the LT unemployed into employment. By August 2008 the LT unemployment rate had fallen to 0.6 per cent.

Figure 3: The relation between the rate of long term unemployment and the aggregate rate of employment, Australia, February 1978 to March 2011 2003–2004

Currently, LT unemployment is in the upper reaches of the data shown in black. The employability of the LT unemployed, as suggested by the record since 1978, implies that it would be reasonable to expect LT unemployment to decrease if aggregate unemployment decreased. Indeed, were unemployment reduced to 2.5 per cent, then the historical relation suggests that LT unemployment would become negligible.

The relation between the rate of long term unemployment and the aggregate rate of employment, Australia, February 1978 to March 2011 2003–2004 The relation between the rate of long term unemployment and the aggregate rate of employment, Australia, February 1978 to March 2011 2003–2004 Figure 3 demonstrates that increases in aggregate unemployment cause increases in LT unemployment with a lag. Consider the two big loops that show the recessions in the early 1980s and early 1990s. As aggregate unemployment rose, LT unemployment initially increased slowly, but eventually, as the unemployed became LT unemployed, the LT unemployment rate increased more rapidly. The same phenomenon is also evident in the current GFC-induced recession. Aggregate unemployment increased from 4 per cent in February 2008 to 5.8 per cent in May 2009 and LT unemployment followed with a lag.

But can aggregate unemployment be reduced to 2.5 per cent? History suggests it can. Aggregate unemployment averaged less than 2.5 per cent in the 1950s and 1960s. Since then there have been extensive microeconomic reforms and a substantial decline in trade union power. These changes should, according to economic analysis, make low unemployment easier to achieve. Indeed this belief was part of the case for making the microeconomic reforms. This does raise the question of why, since the 1960s, we are unable to achieve the low unemployment rates of that decade (McDonald, 2007).

Even those economists who, in spite of this puzzle, call the current rate of aggregate unemployment 'full employment' recognise that policies that prevent increases in unemployment are beneficial for a range of disadvantaged people. It has become clear that financial crises are the major threat to macro-economic stability. Thus financial-market reform indirectly assists the disadvantaged who bear the brunt of increases in unemployment.

Disability

Consider now disability support pensions (DSPs). There is a concern that too many receive the DSP. In 2009 there were over 750,000 DSP recipients, some 7 per cent of the labour force. This rate is greater than the unemployment rate.

Figure 1 shows that DSP recipients suffer from low life satisfaction. The DSP toll is 0.71 points. Notice that these individuals are concentrated in the lower five income deciles. The DSP toll exceeds the LT unemployment toll. Thus these people are on average suffering greater disadvantage.

Welfare dependency has perhaps trapped some people into DSP recipiency. The measure of this would be whether – if they shifted to sustainable employment – they would achieve greater life-satisfaction.

Figure 4: Disability rate and aggregate unemployment, Australia, 1972–2009

Cai and Gregory (2004) show that inflows into the DSP are sensitive to the aggregate rate of unemployment, being lower when aggregate unemployment is falling. Cai, Vu and Wilkins (2008) show that there are negligible outflows from DSP recipiency to sustained employment. In consequence, in comparing Figure 4 with Figure 3, the relation between the DSP rate and the aggregate unemployment rate is a marked contrast to the relation between the LT unemployment rate and the aggregate unemployment rate. The DSP rate does not seem to respond to changes in aggregate unemployment – it simply trends upward. However, as Figure 5 shows, the DSP rate may be approaching a plateau.

Figure 5: Disability rates as percentage of labour force, Australia, 1972–2009

A recent survey (OECD, 2010), of the large number of DSP programs offered by member countries of the OECD found that measures to induce shifts from DSP to employment were, in general, not very successful. A few measures such as unpaid work trials and temporary earnings-supplements show some success. The OECD concludes that more work should be done to try to develop successful programs. Life-satisfaction data suggests that DSP recipients are deserving of support. It also hints that some DSP recipients at least would be better off if they could find work.

Mental illness

Figure 1 shows that mental illness is associated with low life satisfaction. The mental illness toll is 1.39 points, a shortfall that is fairly similar across all income deciles. This very large difference in life satisfaction strongly supports the case for allocating more funds to treating mental illness.

Discrimination and racism

Discrimination is deplored on equity grounds because it violates the right to equal treatment. However, discrimination can also be criticised for lowering life satisfaction and thus should be treated in the same way as the other types of disadvantage considered herein. Indeed, it may be that people support the 'right' of equal treatment because they think it leads to greater life satisfaction. In this vein, McDonald and Mitchell (2010) explore the application of the utility approach to the work of the Victorian Equal Opportunity and Human Rights Commission.

The HILDA survey asks people whether they feel they have been discriminated against in their current job. This is only one potential source of discrimination. As reported in volume 6 of the HILDA reports (Wilkins et al., 2011), the average life satisfaction of people who claim to have been discriminated against in their current job is 0.5 points less than the average life satisfaction of all respondents to the HILDA survey. However, this is not very reliable evidence since some who claim discrimination probably have not suffered the experience, and others who have suffered discrimination probably do not report it. More reliable evidence comes from a US study by Blanchflower and Oswald (2004) who find that being 'black' (African-American) in the US causes a substantial life-satisfaction toll.

The calculations made by Blanchflower and Oswald (2004) are based on multivariate regression analysis and so the differential between the life-satisfaction levels of blacks and whites captures the 'pure' effect of skin colour. For two hypothetical individuals in the US who have the same income, marital status, education level etc., and only differ by the colour of their skin, Blanchflower and Oswald find that the person with black skin is substantially worse off than if he had white skin. They calculate a dollar value on this loss of well-being of $US30,000 per year. This compares with a gain of $US100,000 per year from a lasting marriage. Undoubtedly this comparison, by controlling for all other variables, strikingly demonstrates the 'pure' effect of racism. However, the 'people approach' of this article suggests that it may not be an estimate of the cost of discrimination as measured against the yardstick of human empathy. If we feel empathy for people, then how they cope overall may be the true object of our concern. For blacks, their shortfalls in other areas such as education and marital stability suggest that the cost of racism is higher than the 'pure' effect.

However the cost of racism is calculated, it can be inferred that laws and organisations that reduce discrimination can increase life satisfaction.

Childhood disadvantage

Children from disadvantaged environments can suffer from a whole range of disadvantages. Unemployment, disability and mental illness can all result from disadvantaged childhood backgrounds and will lead to low life satisfaction in adult life.

Some early intervention programs to tackle disadvantaged childhoods have been very successful (see for example, Cunha et al., 2005). Studies of programs from the 1960s and 1970s have followed the children through their subsequent adult lives and found huge benefits in monetary terms. These monetary benefits probably understate the true benefits. For example, the benefit from being able to lead a life that does not run a significant risk of being imprisoned is probably greater than the saving on the costs of running prisons and the additional wage income earned.

Policy implications

Table 1 brings together the tolls for the various disadvantaged groups. We see that mental illness imposes the biggest toll, followed by disability, LT unemployment, low income, and ST unemployment.

The variety of life-satisfaction tolls recorded in Table 1 raises the question: why does social policy fare better in some areas, such as income distribution, than others, especially aiding the mentally ill? To address this question, the economists' demand- and supply-side concepts are useful. The demand-side is the value of improving outcomes for the disadvantaged. Improving outcomes for the mentally ill would be particularly valuable because mental illness causes such a large loss of life satisfaction. The supply-side recognises our ability to make improvements. Improving mental health is difficult, ⁱⁱ and improving life for many people on disability support pensions through shifts to sustainable employment is not easy, as the OECD (2010) report shows.

Table 1: The life-satisfaction tolls for various groups of disadvantaged people

It would seem that income redistribution has been well-applied in Australia. The small toll from being in the lowest compared with the highest income decile suggests that this income redistribution is a great achievement. But helping the groups that incur high life-satisfaction tolls is more difficult. How can we get aggregate unemployment down to 2.5 per cent? How can the life satisfactions of the disabled and the mentally ill be increased? Indeed the fact that these groups suffer such large life-satisfaction tolls probably reflects the difficulty of improving their situations. Improving the productivity of supply-side programs in these areas requires research, experimentation and program evaluation.

References

Blanchflower, DG and Oswald, A 2004 'Wellbeing over time in Britain and the USA', Journal of Public Economics, 88, pp 1359-86.

Cai, L and Gregory, RG 2004 'Labour market conditions, applications and grants of disability support pension (DSP) in Australia', Australian Journal of Labour Economics, 7, 3, 375-394.

Cai, L, Vu, H and Wilkins, R 2008 'The extent and nature of exits from the disability support pension', Australian Bulletin of Labour, 34, 1, 1-27.

Cunha, F, Heckman, JJ, Lochmer, L and Masterov, DV 2005 'Interpreting the Evidence on Life Cycle Skill Formation', Working Paper No. 111331, National Bureau of Economic Research, Cambridge, Massachusetts.

Frey, BS and Stutzer, A 2002 'What can economists learn from happiness research?', Journal of Economic Literature, XL, June, 402-35.

Layard, R 2005 Happiness: Lessons from a new science, Penguin: Allen Lane, London.

McDonald, IM 1993 'Long-term unemployment and macroeconomic policy', Australian Economic Review, 2: 31-4

McDonald, IM 2007 'Where is full employment?', Dialogue, 26, 2, 81-92.

McDonald, IM and Mitchell, H 2010 'Equality, wellbeing and the work of the Victorian Equal Opportunity and Human Rights Commission', Insights, 8, November 2010.

OECD (2010) Sickness, disability and work: Breaking the barriers, OECD publishing, www.oecd.org/publishing.

Wilkins, R, Warren, D, Hahn, M and Houng, B 2011 Families, Incomes and Jobs, Volume 6: A Statistical Report on Waves 1 to 8 of the Household, Income and Labour Dynamics in Australia Survey, Melbourne Institute of Applied Economic and Social Research, University of Melbourne.