This version was revised in March 2006.
Keywords: Distributional Economics; Health;
Introduction and Summary
This paper brings together some recent research about the relationship between health status and income inequality. It focuses upon a set of propositions which challenge the conventional wisdom. They are:
1. That in a rich country poverty – low material standard of living – probably does not directly impact on health, but does indirectly through stress which income differences generate.
2. The increase in household inequality in the period of the late 1980s and early 1990s was more due to changes in tax, benefit, and government spending policies than it was due to market liberalisation. However, the market liberalisation increased stress on New Zealanders.
3. There is some evidence that income inequality may be increasing, due to factors such as globalisation and technological change.
4. The most common poor New Zealand household is a couple with children who are of Pakeha ethnicity, who own their home (usually with a mortgage), and who depend upon wages for their main income. There are other groups who have higher incidence of poverty, but because they are smaller they do not involve as many people. This means that effective poverty eradication involves working on a broad front rather than targeting minority groups.
5. Illness does not correlate well with income, unless age is controlled for. The sick in New Zealand are the elderly, although the paper goes on to argue that policies aiming to reduce poor health in the long term need to target those with low incomes and low in the socioeconomic status hierarchy.
Does Poverty Affect Health?
Much of the international research on the social determinants of poor health in rich countries focusses on the effect of socioeconomic status (SES) – a composite of such things as occupation, education and housing conditions.
However there is not a lot of SES data readily available in New Zealand, so that often household disposable income adjusted for household composition (‘equivalised’) is sometimes used as a proxy, but also because it can be directly influenced by policy instruments. The two variables need not correlate well, for household income varies both in the short term and over the life cycle, while SES is largely for an adult’s life. Life time income would be a better indicator of SES than current annual income, but we do not have much data on that either.
What the international research shows is that there is an ‘SES gradient’, in which those with low SES experience higher morbidity from respiratory and cardiovascular diseases, ulcers, rheumatoid disorders, psychiatric diseases, dementia, a number of cancers, and so on. Often those at the bottom of the SES scale have life expectancies five to ten years shorter than those at the top of the scale (Wilkinson 2000).
Part, but not all, of the differences can be explained by differences in access to health care and by ‘unhealthy life styles’ such as the consumption of tobacco and excess alcohol, lack of exercise and obesity (all of which might be addressed by conventional health care and health promotion policies). Even so, there remains a significant residual which cannot be so readily explained.
The search for alternative explanations has paid much attention to socio-environmental explanations, perhaps intermediated through economic and psychological channels. However, in rich countries there is not a lot of evidence that income has a direct effect on health, even though – as we shall see – the poor tend to be sicker. Unlike those in much poorer countries, the affluent nation’s poor would appear to have sufficient to purchase healthy food, adequate clothing, reasonable housing and so on. (Sapolsky 2004, 2005)
For instance, today’s sickness benefit for a married couple is about 5 percent more in value, after adjusting for consumer price rises, than it was fifty years ago (and the beneficiary is likely to be entitled to more supplements). Yet there are foodbanks today, while there were none in 1956. (We do not know about the homeless, since they were not well monitored then.) The simplest explanation is that real wages (i.e. adjusted for price changes) have risen about 55 percent more in the period. In principle we need to make a host of other adjustments such as for households now earning two wagees but paying higher rates of taxation, but the basic conclusion is that the beneficiaries are relatively poorer and find it more difficult to cope.
This is not to argue that had they more income, they could not improve their consumption for a healthy life style, or that they spend their income wisely. (The same is largely true for those with greater affluence). At issue is that by itself it would appear that income is not a determinant of health, although it is a correlate. Once the minimal resources are available to sustain a basic level of health, absolute levels of income seem unimportant in determining health.
In recent years it has been proposed that the intermediating channel is stress. There is a substantial biomedical literature which has established that individuals are more at risk from stress-sensitive diseases – which include many of those listed above that were associated with the SES gradient – if they
(i) feel they have minimal control over stressors;
(ii) feel that they have no predictive information and intensity of the stress;
(iii) have fewer outlets for the frustration caused by the stressor;
(iv) interpret the stressor as evidence of circumstances worsening; and
(v) lack social support for the duress caused by the stressors. (Sapolsky 2004, 2005)
While there has been less research to test the hypothesis, it is plausible that the poor are subject to greater disease-inducing stress than those further up the income scale. This is not to argue that others do not suffer stress: but that they have better opportunities to cope with it (in terms of the previous paragraph) than the poor.
An intriguing insight is that while the objective state of being poor appears to adversely affect health, the subjective state of feeling poor seems even more important. (Adler 2003) In which case, the notion of relative poverty levels becomes very relevant. Many of the stresses – such as those of inferiority feelings because of low spending power – do not go away just because of a general rise in income. Thus the downside of low income in a rich country involves different mechanisms to that in a poor country. The notion of an absolute poverty line is no longer relevant, while relative poverty is.
It is possible that the observed connection between income inequality and poor health is a stress caused-effect. (Kawachi & Kennedy 2002) The most probable channel is that the lower in the social pecking order, the more likely that it causes stress.
However the stress hypothesis is not confined to a channel to sickness only through household disposable income. Other economic phenomenon cause health damaging stress. The best studied is unemployment where it is clear that psychological and financial stress puts at risk the unemployed and her or his partner. More generally, early life experiences, social exclusion, work and social support as well as unemployment have all been identified as sources of health damaging stress. (Wilkinson & Marmot 1998)
Changes in the New Zealand Income Distribution: 1980s to mid 1990s.
New Zealand household income inequality increased sharply in the late 1980s and early 1990s. Table 1 shows that the shares of the bottom four quintiles of (equivalised) household disposable income fell between 1987 and 1997, with the top quintile gaining the 3.3 percentage points the remaining 80 percent lost.
Table 1: Shares of Household Equivalised Disposable Income by Household Quintile
| Year (Sep or Dec) |
Bottom Quintile |
4th Q | Middle | 2nd Q | Top Quintile |
Income ($2004) |
| 1987 | 10.3 | 13.7 | 18.3 | 24.3 | 33.3 | 23,500 |
| 1988 | 10.2 | 13.5 | 17.7 | 24.0 | 34.7 | 23,300 |
| 1989 | 10.0 | 13.2 | 17.6 | 23.6 | 35.6 | 23,300 |
| 1990 | 9.9 | 12.6 | 17.7 | 24.1 | 35.7 | 22,700 |
| 1991 | 9.7 | 12.3 | 17.3 | 24.2 | 36.5 | 21,500 |
| 1992 | 9.5 | 12.2 | 16.0 | 24.2 | 37.3 | 21,600 |
| 1993 | 9.5 | 12.4 | 17.1 | 24.3 | 36.7 | 21,000 |
| 1994 | 9.8 | 12.3 | 16.7 | 24.0 | 37.1 | 22,000 |
| 1995 | 9.9 | 12.6 | 17.3 | 23.9 | 36.3 | 22,400 |
| 1996 | 9.3 | 12.1 | 16.9 | 24.4 | 37.3 | 22,600 |
| 1997 | 9.6 | 12.1 | 17.3 | 23.9 | 36.6 | 24,300 |
| … | … | … | … | … | … | … |
| 2000 | 9.4 | 12.0 | 17.2 | 24.3 | 37.2 | 24,600 |
| … | … | … | … | … | … | … |
| 2003 | 9.1 | 11.9 | 17.5 | 24.7 | 36.8 | 25,800 |
Source: Perry 2005.
On some comparisons the increase in inequality was the greatest in the Western World over the period. (Easton 1996)
Unfortunately in 1997, in a cost saving measure, the New Zealand government replaced the annual household survey from which the income shares were derived by a triennial one, so it is more difficult to discern trends afterwards. However, Table 1 indicates the likelihood that the bottom two deciles continued to lose disposable income share, the middle one made some gains but did not recover back to its 1987 level, while the upper two deciles seems to have had a share increase. If so, inequality continued to increase after the mid 1990s, although the following section argues that it may be due to different causes.
Why did the changes in the late 1980s and early 1990s occur? Aside form the complication of the business cycle, three groups of reasons have been considered:
1. The economic reforms involving sweeping liberalisation of the economy (increased use of the market mechanism) from 1984 to 1993.
2. Changes in tax and benefit (and other welfare policies).
3. Fundamental changes in the structure of the economy arising from technical change and globalisation (the external impact of those forces which are integrating the world economy).
Obviously all the explanations had some effect, but a careful analysis concluded the dominant impact came from policies which cut income taxes on the rich which were paid for by higher income taxes on the poor, cuts in social security benefits and cuts in other government services (Easton 1996, 1999).
This is not to say that the other effects were unimportant. Unemployment peaked at 11.1 percent in March 1992, although the percentage at a point in time can be misleading. In the 57 months between October 1988 and June 1993, 754,312 sought work by enrolling on the New Zealand Employment register. (Department of Labour 1994) To give some idea of this magnitude, those who registered as unemployed in the period represent about 47 percent of the the average size of the labour force over the period, or around 10 percent of the labour force each year when the average rate of unemployment at any point in time was 8.7 percent. This overestimates the likelihood of being unemployed because of inflows (from school leaving, returning to work, and immigration), and underestimates it insofar as not all unemployed registered.
Whatever is the true figure, it would appear that a high proportion of the labour force experienced unemployment in the period. Those who enrol more than once, are counted but once in the above total. In fact over 45 percent were enrolled at least twice, and 2.1 percent more than 5 times. This suggests that at least 21 percent of the labour force experienced repeated unemployment in the 4¾ year period, and 1 percent experienced it on five or more occasions. The average number of enrolments was 1.8 times for non-Maori, and 2.2 times for Maori. The average cumulative duration on the register was 59 weeks – the Maori averaged 69 weeks. The register does not pick up all the labour market churning, but it suggest it was substantial in the period.
However, because the stress was substantial it was more evenly spread than at first one might have expected. Similarly, while the liberalisation resulted in people both suffering and benefiting, there is a tendency to focus on the big losses. For instance, substantial reductions of protection on the clothing industry resulted in job losses, but the price of the now-imported clothing fell for the population as a whole. Did those small but widespread gains offset the large losses to a small set of workers? Or – and perhaps this is the more relevant – in the case of each individual did the myriads of small benefits from the ending of protection in many industries offset the one (or more) big employment losses?
If they did not, was there a particular social group of New Zealanders who were privileged by receiving only the benefits but not the downside. When we look at the available data – inadequate though it may be – we can conclude that those on high incomes benefited not from the change in their market incomes but from the marked reductions in income tax they paid. It turns out that many of the rich also suffered – from a loss of wealth, perhaps following the sharemarket crash, perhaps because they too lost benefits from the protection their investments had enjoyed. Equally, there were those on modest and low incomes who would have been better off from the changes, once the terrible transition of high unemployment had worked its way through, except they found themselves paying higher income taxes, or receiving less benefits from the state.
This account keeps to the conventional distributional approach, which largely ignores any economic growth and looks only at distributional shares. But in the six years from 1987 to 1993 average income per head fell. (See the final column of Table 1.) The distributional impact was such that real income fell for the bottom four household income quintiles, but remained (roughly) constant for the top quintile, because the tax cuts offset the fall in their market income. Falling incomes are a further source of stress (although again there is considerable churning).
Changes in the New Zealand Income Distribution: After the mid 1990s.
We need not report the details of the distributional policy changes which occurred after 1996, merely noting that there were both progressive and regressive changes, but none were of the redistributive magnitude of the earlier period. Because of the data gaps it is also harder to assess whether there have been major redistributional changes since 1996.
However there seems to have been a tendency for increasing inequality, but this time the changes involve more the bottom two quintiles losing and the top two (or three) quintiles gaining. This suggests a different redistributive mechanism from that of the previous decade. (See also Dixon (1998) for more clues.)
There is a vigorous debate about the extent to which the forces of globalisation and technological change are modifying the income distribution of rich countries. It can be only sketched here, beginning with a simplified account of how globalisation may impact unfavourably on the rich country’s income distribution.
In the following schematic figure the economy is characterised by quadrants, by dividing it into a tradeable and non-tradeable sector (the sector exposed to overseas competition to that which is not exposed) and a skilled and unskilled labour.
| Tradeable Sector | Non-tradeable Sector | |
| Skilled | ||
| Unskilled | Hollowed out? |
It is the tradeable sector/unskilled labour force quadrant (in the South-West of the block) which appears to be ‘hollowed out’. The most evident example is demise of the textile, clothing and footwear industry. There will still be unskilled jobs in the tradeable sector (e.g. janitors for high tech food processing) and there will also be unskilled in the non-tradeable sector (such as in rest-homes). However one would expect that major job losses in the hollowed-out quadrant would reduce unskilled wage rates elsewhere. Unskilled wage rates are low, and wages are an important component of low income households (see below). Such changes may have contributed to the increased inequality that appears to have ben occurring in the late 1990s. The effect of the hollowing out from globalisation is therefore increased income inequality unless there is an upgrading of skills in the labour force.
Broadly the same story applies for technological change except in this case the (relative) reduction in unskilled jobs is the result of technical change favouring skilled jobs (and it is unnecessary to make the tradeable versus non-tradeable sector distinction, since the technological change can occur in either sector or both of them). Again, unless there is sufficient upskilling of the labour force, there will be downward pressures on unskilled wage rates and increased household inequality.
Who Are The Poor Households in New Zealand?
I now report some research which Suzie Ballantyne and I did using the Household Economic Survey (HES) for the three year period covering 1994/5-1996/7 when the HES included questions on the respondents’ recent utilisation of health services together with a subjective assessment of each’s health status, as well as socioeconomic variables such as income and expenditure and personal characteristics. (Easton 2002, Easton & Ballantyne 2002)
Our work involved detailed econometric analysis, with considerable more care taken in the application of the household equivalence scale, which adjusts for household composition, for housing (since outlays are different for those who own their houses and those who rent), and for the choice of poverty line. The following tables are based upon the Michelini equivalence scale, income econometrically adjusted for housing circumstances, and a poverty line based on the benefit datum line of the Royal Commission on Social Security, although using other assumptions will not markedly change the broad conclusions.
Table 2 shows the proportion in poverty and the proportion of the poor by household type. While the highest incidence of poverty is among households consisting of a single adult with children – as is frequently commented upon – they make up only a sixth of the poor. In contrast households with two adults and children have less than half of the poverty incidence of the single parent households, but make up more than twice its numbers. This illustrates the danger of focussing on solo parents, and ignoring the vast majority of the poor, who are children and their parents, most of whom live in two adult homes. Households with children contain over four fifths of the poor.
Table 2: Poverty by Household Type
| Proportion in Poverty | Proportion of the Poor | |
| Adult not in Labour Force | ||
| Adult in Labour Force | 4.6 | 1.6 |
| 2 Adults, neither in Labour Force | 2.6 | 0.4 |
| 2 Adults, 1+ in Labour Force | 6.0 | 3.5 |
| 1 Adult with 1+ Children | 4.0 | 4.4 |
| 2 Adults with 1 Child | 38.7 | 16.3 |
| 2 Adults with 2 Children | 13.9 | 7.6 |
| 2 Adults with 3+ Children | 21.5 | 18.0 |
| All 2 Adult with Children Households |
18.4 | 42.3 |
| 3 Adults without Children | 7.2 | 8.0 |
| 3 Adults with Children | 22.8 | 23.5 |
| In Households with Children | 21.9 | 82.1 |
| In Households without Children | 5.5 | 17.9 |
| ALL | 14.2 | 100 |
Source: Easton and Ballantyne (2002)
Table 3 looks at the poor by ethnicity and shows the same pattern of differences between proportion in poverty and proportion of the poor. Non-Pakeha poverty is higher than Pakeha poverty. Yet almost three fifths of the poor are Pakeha (because there are a lot more of them).
Table 3: Poverty by Ethnicity
| Proportion in Poverty | Proportion of the Poor | |
| Pakeha | 10.7 | 58.5 |
| Maori | 23.7 | 19.9 |
| Pasifica | 33.4 | 11.8 |
| Asian | 26.3 | 9.8 |
| ALL | 14.2 | 100 |
Source: Easton and Ballantyne (2002)
Similarly, Table 4 cutting the data by housing tenure, shows that renters have the highest incidence of poverty, but that over half of the poor live in their own homes.
Table 4: Poverty by Housing Tenure
| Proportion in Poverty | Proportion of the Poor | |
| Rent | 27.3 | 47.4 |
| Own with Mortgage | 13.5 | 37.5 |
| Own without Mortgage | 6.0 | 15.0 |
| Rent Free | 1.2 | 0.1 |
| ALL | 14.2 | 100 |
Source: Easton and Ballantyne (2002)
It is messier, but it can also be shown that the poor households are more likely to receive market income (wages) then depend on benefits. (Nevertheless beneficiaries with children are more likely to be in poverty.)
In summary, the most common poor New Zealand household is a couple with children who are of Pakeha ethnicity, who own their home with a mortgage and depend upon wage income. There are other groups who have higher incidence of poverty, but because they are smaller they do not involve as many people.
There is a tendency for poverty rhetoric to emphasise single parent households, ethnic minorities, renters, and beneficiaries. The reality is while a household with any of these features is more likely to be poor, the poor are more likely to be mirror images of the characteristics. The policy implication is that poverty eradication involves working on a broad front rather than targeting minority groups. If there is a typical poor household, it is one which contains children.
The Location of the Sick in the Income Distribution
The particular advantage of the years the study used was that respondents were asked to rate their health status on a scale ‘excellent ’, ‘good’ ‘fair’ or ‘poor’ health. Very few categorised themselves at the bottom, so we combined the two into a single category. Some 8.6 percent of the population were in it. Note the health rating is self reported and subjective.
Again there is an apparent paradox in the results, for the sickest are more likely to be in the middle quintile than the bottom quintile of the overall household distribution. (Table 5)
Table 5: Percent of Total Who Rate Themselves As ‘Fair’ Or ‘Poor’ Health
| Bottom Quintile |
4th Q | Middle | 2nd Q | Top Quintile |
ALL | |
| ALL | 21.7 | 25.3 | 25.5 | 16.3 | 11.2 | 100 |
Source: Easton and Ballantyne (2002)
However an examination of the characteristics by age and gender shows that the poorest in each category have the highest incidence of poverty. (Table 6) The old are the sickest – over half of those in the fair or poor categories are aged over 65 – but they are not as poor as the young. As far as poor health is concerned, age is a more important determinant than income.
Even so, within age groups, those in the lowest income quintiles tend to be the sickest. The gradients vary (and are erratic, presumably because of the low numbers involved) and they seem to be stronger for females than males, which itself constitutes a puzzle – perhaps the females are under more stress.
What the gradients suggest for those under 65, is that health levels in those in the middle three household quintiles are much the same. But poor health is markedly more common – typically about double – for those in bottom quintile households.
Those in the top income households tend to have fewer who are in inferior health. Does their income enable them to purchase better health care? Are they suffering less illness inducing stress? It cannot simply be that those adults with poor health tend to sink into lower income groups because their earning power suffers, since children in such households who are less likely to compromise the family income, are also in better health.
Table 6: Incidence (%) Who Rate Themselves As ‘Fair’ Or ‘Poor’ Health
| Bottom Quintile |
4th Q | Middle | 2nd Q | Top Quintile |
ALL | |
| Under 15 Female | 7.4 | 3.6 | 2.9 | 3.1 | 1.2 | 4.5 |
| Under 15 Male | 5.8 | 5.8 | 3.4 | 3.0 | 3.1 | 4.7 |
| 15-64 Female | 19.7 | 11.0 | 8.7 | 7.2 | 3.9 | 7.9 |
| 15-64 Male | 10.2 | 8.3 | 7.4 | 4.2 | 3.8 | 6.2 |
| Over 65 Female | 42.5 | 29.4 | 27.5 | 22.3 | 21.2 | 26.5 |
| Over 65 Male | 40.2 | 29.8 | 25.2 | 21.3 | 19.4 | 25.5 |
| ALL | 9.3 | 11.0 | 11.0 | 7.0 | 4.8 | 8.6 |
Source: Easton and Ballantyne (2002)
Conclusion
This chapter has focused upon a set of propositions which may be summarised as :
1. That in a rich country poverty – low material standard of living – probably does not directly impact on health, but does indirectly through stress which income differences generate.
2. The increase in household inequality in the period of the late 1980s and early 1990s was more due to changes in tax, benefit, and government spending policies than it was due to market liberalisation. However, the market liberalisation increased stress on New Zealanders.
3. There is some evidence that income inequality may be increasing, due to factors such as globalisation and technological change.
4. The most common poor New Zealand household is a couple with children who are of Pakeha ethnicity, who own their home (usually with a mortgage), and who depend upon wages for their main income. There are other groups who have higher incidence of poverty, but because they are smaller they do not involve as many people. This means that effective poverty eradication involves working on a broad front rather than targeting minority groups.
5. Illness does not correlate well with income, unless age is controlled for. The sick in New Zealand are the elderly, although the paper goes on to argue that policies aiming to reduce poor health in the long term need to target those with low incomes and low in the socioeconomic status hierarchy.
The policy conclusions might be that there is an socioeconomic (SES) gradient on health status, services whose purpose is to provide health care to individuals need to take more into consideration the age of the local population.
However where services are aimed at the entire population need to take more into consideration the age of the local population, aiming to raise health status in the long term, the international research evidence suggests SES effects appear to be important.
How to target the SES effectively is another matter, especially as it is not the same as income (or even lifetime income, although that would give a better correlation). It seems likely that reducing income inequality may help – especially perceptions of income inequality, and or its importance. (If so the largest single cause of inequality is having children.)
But it may be also possible to address some of the stress channels directly. Certainly it makes sense that for any policy change, that there needs to be an accompanying audit of what stress it will generate and to what extent the further stressed will be able to deal with it.
References
Adler, N.E. (2003). ‘Looking beyond the borders of the health sector: The socioeconomic determinants of health’ in E.R. Rubin & S.L. Schappert (ed) Meeting Health Needs in the 21st Century (Association of Academic Health Centres Washington, DC) p.14-27.
Department of Labour (1994) An Analysis of the Dynamics of the Registered Unemployed: Exit Probabilities and Repeat Spells (Department of Labour, Wellington).
Dixon, S. (1998) ‘Growth in the Dispersion of Earnings: 1984-1997′, Labour Market Bulletin , p.71-107.
Easton, B.H. (1996) ‘Income Distribution’, in B. Silverstone, A. Bollard, & R. Lattimore (ed) A Study of Economic Reform: The Case of New Zealand (North Holland) p.101-138.
Easton, B.H. (1999) ‘What has Happened in New Zealand to Income Distribution and Poverty Levels,’ in S. Shaver & P. Saunders (ed) Social Policy for the 21st Century: Justice and Responsibility (SPRC, Sydney) Vol 2, p.55-66.
Easton, B.H. (2002) Validation and the Health and Household Economy Project , Paper to the Wellington Health Economists Group,
Easton, B.H. & S. Ballantyne (2002) The Economic and Health Status of Households , (Wellington School of Medicine)
Kawachi, I & B.P. Kennedy (2002) The Health of Nations: Why Inequality Is Harmful to Your Health (New Press).
Perry, B. (2005) Social Report Indicators for Low Incomes and Inequality: Update from the 2004 Household Economic Survey (Ministry of Social Development, Wellington)
Saplosky, R. (2004) Why Zebras Don’t Get Ulcers : A Guide to Stress, Stress Related Diseases and Coping (Harry Holt & Company)
Saplosky, R. (2005) ‘Sick of Poverty’, Scientific American , December 2005, p.72-79.
Wilkinson, R. (2005) The Impact of Inequality : How to Make Sick Societies Healthier (Routledge)
Wilkinson, R & M. Marmot (1998) The Social Determinants of Health: The Solid Facts (WHO)