Social Policy for the 21st Century: Justice and Responsibility Proceedings of the 1999 National Social Policy Conference, 21-23 July, 1999. Social Policy Research Centre Reports and Proceedings, 1999.
Keywords: Distributional Economics; Social Policy;
In a recent article, the London Economist describes the “bad point” of New Zealand’s economic reforms which began in the mid 1980s as “a big increase in inequality.”1 In fact the New Zealand economy has generally had a poor growth performance, higher unemployment, and a worrying current account deficit ever since the reforms (although price levels have been more stable). Table 1 is a comparison between the overall economic performance of the Australian, New Zealand and OECD economies since 1985. There is no doubt the New Zealand economy has done worse. 2 Why this has happened, and why the New Zealand economic performance has been inferior to the Australian one, belongs elsewhere. For this paper, The Economist’s observation emphasizes just how widespread is the view that New Zealand has a more unequal income distribution, as a result of the policy changes of the last one and a half decades. But The Economist comment gives no sense of the magnitude of the increased inequality, nor its causes, which are the focus of this paper.
|Table 1 – ECONOMIC PERFORMANCE: 1985-1998|
|Inflation Private Consumption Deflator (% p.a.)|
|Inflation GDP Deflator (% p.a.)|
|Unemployment (% of Labour Force)|
|Employment Growth (% p.a.)|
|GDP Volume Growth (% p.a.)|
|Labour Productivity Growth (% p.a.)|
|Terms of Trade Change (% p.a.)|
|Export Volume Growth (% p.a.)|
|Import Volume Growth (% p.a.)|
|Current Account Deficit (% GDP)|
OECD Economic Outlook, December 1998. The New Zealand figures do not always correspond to the official figures, but are used here for consistency. The OECD consists of 28 economies.
The 1998 data is estimated.
* G7 for unemployment.
The standard way in New Zealand to trace household income changes is to use household income reported in the Household Economic Survey, adjusted first to a disposable (after tax and benefit) income basis, and then adjusted for household composition using a household equivalence scale. There are difficulties with the resulting measures, but as far as is known the problems are not sufficiently strong to invalidate the results to be presented here. 3
A number of research teams have used this approach over the years. 4 There are two key studies. First, Mary Mowbray has provided estimates for the 12 available years between 1981/2 and 1995/6. 5 Second, and more recently, Statistics New Zealand (SNZ) has provided estimates for the four years 1981/2, 1985/6, 1990/1, and 1995/6. For a number of reasons the SNZ data might be expected to be more authoritative than the Mowbray (and other) studies. However it is not as comprehensive, so both are used here. In any case, all the studies tell broadly the same story.
The equivalent disposable income of a household might be thought of as a sophisticated per capita measure of the household spending power, in which household economies of scales and differences between adults and children are allowed for.6 The Mowbray data is summarised in Figure 1 and Table 2. They show the mean income falling from $29200 (for a household of two in 1991 dollars) in 1982 to $25710 in 1994, and then recovering slightly to $29420 following the upswing to 1996, a gain of .06 percent p.a. over the 14 years. This pathetic figure is further evidence of the poor performance of the New Zealand economy over the period. In summary, over a 14 year period of reforms, the spending power of households has been stagnant.
The median income falls from $22402 in 1982 to $19680 in 1996 (an average fall of .92 percent p.a. over the period). The divergence between the mean and the median is an example of the increasing inequality of incomes. It means that the increase in inequality has not been simply a share shift from the poor to the rich, but a shift from the middle incomes to high incomes also. As a result those in the bottom 80 percent of the income distribution have experienced a fall in their real incomes over the period.
The Distribution of Income
|TABLE 2: HOUSEHOLD EQUIVALENT DISPOSABLE INCOME|
|INCOME SHARES (percent)||MARCH SURVEY YEARS|
|DECILE AVERAGE ($1999 thousands)||MARCH SURVEY YEARS|
|BELOW RCSS POVERTY LINE|
SOURCES: Mowbray (1993) plus update, except gini coefficients are from SNZ (1999) .
* 1997 = .033.
A common means of comparing changes in incomes over a period is to use the gini coefficient. A rise in the coefficient usually means there has been an increase in inequality. 7 The SNZ gini coefficient estimates are shown in Figure 2 and Table 3. There appears to be three phases. Between 1981/2 and 1987/8 (or possibly 1986/7) the gini coefficients are broadly constant indicative that the degree of inequality remained broadly constant too. The coefficients then rise sharply to 1990/1, from about .28 to .31. SNZ notes the increase is statistically significant. Afterwards, the inequality seems to rise more slowly up to .33 in a six year period. The standard deviation of the household distribution might be thought of increasing by just over a fifth to 1991, and by a half to 1997, compared to the what it was in the mid 1980s. 8
The picture from the gini coefficients is reinforced by Table 3 and Figure 3, which shows the household deciles. The top decile of households increases its share from 20.1 percent of total income in 1982 to 26.8 percent p.a. in 1996. The second decile holds its share (15.1 to 15.4 percent) and the remainder experience a decreasing share. Given that the average real disposable (equivalent) income hardly changed, we find that those in the top decile experienced a 34.8 percent in their spending power between 1982 and 1996 (equivalent to a 2.2 percent p.a.), the second to top decile experienced a 2.5 percent rise, and the remaining 80 percent had a 10.1 percent fall over the 14 year period. The bottom thirty percent have experienced an average fall of 12.6 percent, for the proportional reductions tend to be largest as one moves down the income distribution.
|TABLE 3: HOUSEHOLD EQUIVALENT DISPOSABLE INCOME|
|INCOME SHARES (%)||MARCH SURVEY YEARS|
|DECILE AVERAGE ($1996 thousands)|
SOURCE: SNZ (1999)
Why the rise in inequality? First, observe the most rapid increase occurs in the period when the government is cutting the top income tax rate – it was 66 percent in the year to March 1986 and was 33 percent by 1990. This lifted the relative income of those in the top decile, who were the main beneficiaries of the tax cuts. This is sufficient to explain most of their income increase. In order to fund the reduction in income taxation on those at the top, the government cut social security benefits and other government spending (sometimes by the imposition of user charges), withdraw tax concessions, and allow income tax rates to rise on lower incomes via fiscal creep. 9 Since there was little income growth, the net effect of the fiscal changes was to switch income from the poor and those on middle incomes to the rich. 10
Thus far we have explained the increasing inequality by the deliberate actions of government taxation and spending decisions. Did the market economy, especially market liberalisation, also add to inequality? When I last did a comprehensive review I had only data up to 1993. I concluded that there was no evidence of an impact of market liberalisation on overall income inequality. I argued that the measures were so widespread they impacted on everyone, and so no part of the market income distribution especially benefited or suffered. Instead there was considerable turbulence within the distribution. 11
The addition of more recent data allows some reassessment. Inequality appears to have continued to increase after 1991, despite there being no major changes in the fiscal stance. While income tax cuts tended to favour middle incomes and families and benefit eligibility continued to be tightened, the changes were minor in comparison to the earlier changes. Because there are only a few observations it may be that the increase in the observed inequality is explicable in terms of statistical noise or the business cyclical, and there is really stability of income inequality. On the other hand, the trend increase is sufficiently perceptible to raise the possibility that, in contrast to before 1988, there is a systematic market mechanism which is increasing inequality. If there is it can be traced in the ratio of the median to the mean. Up to 1988 the ratio hovered in the .78 to .80 range. As we would expect, following the pro-rich fiscal measures, it fell to about .7 in 1992. But it appears to be continuing to fall, and was at .67 in 1996. Those who believe that more liberalised markets generate inequality have the current evidence on their side.
A table provided in the recent SNZ Incomes supports this conclusion. 12 If households are ranked my market income, there has been an increase in the share of market income of the top two deciles, and a corresponding fall in the share of the bottom seven deciles. Market income deciles do not simply relate to equivalent household income deciles, and the relationship changes over time. Nevertheless, the data is suggestive that, at the very least, rising unemployment has undermined the income of those in the lowest deciles. Unemployment was markedly higher in the mid 1990s compared to the mid 1980s – probably more than double when labour force participation rate changes are allowed for. It is also possible that there has been a widening in the dispersion of pay rates (especially at the very top of the distribution), and that changes in rates of return on investments are impacting on the income distribution, but neither is yet evident in the available data.
What Has Happened to Poverty?
Given that the real disposable (equivalent) incomes of the bottom three deciles have been falling, it might be thought unquestioned that poverty has been rising in New Zealand.
However there are some who contest this common sense. The most notable recent contribution is from Roger Kerr, the executive director of the Business Roundtable, which consists of the chief executives of the large corporations which have been both major advocates of the reforms and their largest (relative) beneficiaries. 13 In a recent paper he argued:
What does the [SNZ] study tell us about poverty, as opposed to changes in the distribution of incomes? Between 1982 and 1996, according to Statistics New Zealand, there was no increase in the proportion of individuals or households with an income of less than 50 or 60 percent of median disposable income. On the basis of these poverty benchmarks, between 6 percent and 12 percent of households were in poverty over the period.
These measures of poverty are similar to measures used by Stephens, Waldegrave and Frater.14 Using a benchmark of 60 percent of median disposable income, the latter found that the percentage of households in poverty fell from 13.7 percent to 10.8 percent between 1983/84 and 1992/93. At a benchmark of 50 percent, poverty was stable at 4.3 percent of households. The period examined did not reflect fully the economic recovery that started during 1991. Nevertheless, the study suggests that fewer households were in poverty than reported by Statistics New Zealand. Neither study suggests a rise in reported poverty since the reforms began.
The Statistics New Zealand study contradicts three claims that have frequently been made. First, the claim that the rich are getting richer while the poor are getting poorer is simply not true. People on high incomes have increased their share of total disposable income while middle income earners suffered a significant loss of income share. However, there was no significant change in the share of income of low?income households. (original’s italics)
The point here is not that Kerr is seriously misrepresenting the statistics (especial Statistics New Zealand, who do not even imply that 50 percent of the median is a poverty line). Rather he is using a poverty line indexed to the median, which as we have seen, has been falling relative to the mean. Consider the situation where the government takes income from those in the middle of the income distribution and gives it to the rich, without affecting the total income. The effect will be to depress the median, and hence the poverty line. Thus the numbers in poverty will fall according to the Kerr poverty line, even though they have had no change to their incomes, and inequality has increased.
There is a case for indexing the poverty line for changes in mean incomes in the long run. 15 However since average incomes have hardly changed, in practice a constant price poverty line is satisfactory for New Zealand analysis over this period.
Figure 5 and Table 2, report poverty levels if the standard (constant price) poverty line is used, based on the assessment of the 1972 Royal Commission on Social Security. It shows poverty rising slightly up to 1990, and then increasing dramatically in the early 1990s following the benefit cuts and the severe economic downturn. As the economy went into a cyclical upswing, poverty levels fell (probably as a result of increased labour market engagement), but still remaining above the level of the 1980s. The level depends on the precise poverty line, but the pattern remains broadly the same. In summary using the RCSS poverty line, poverty numbers inched up from 11.8 percent to 12.6 percent in the 1980s, rose dramatically to 19.6 percent in 1993, and by 1996 were 14.5 percent. If a constant price poverty line is used, poverty has definitely increased (and that would be also true were it were indexed to mean real incomes).
Kerr contradicts himself a little later, writing:
Relative poverty can only be reduced by raising the income of those who are judged to be in poverty at a faster rate than that of other groups. A doubling in everyone’s income, for example, would have no effect on the reported level of poverty according to a relative poverty standard.
An absolute income or expenditure threshold, on the other hand, focuses debate on the explicitly identified commodities and expenditure patterns that are necessary to avoid hardship. It is likely to show substantially less poverty than that reported in the studies discussed earlier. An absolute standard recognises that an increase in real income reduces poverty. This is simply common sense.
So Kerr has now switched to favouring an absolute standard for a poverty line. I will not rehearse the case for and against this view, with which every social policy analyst is familiar. The point is on page 3 of his speech Kerr relies on an indexed relative poverty line (plus some misreading of the statistics) for his argument. By page 5, he is opposed to an indexed poverty line, favouring a constant price one.
So while there is considerable agreement that the income distribution has got more unequal, it would have been less than comprehensive to have implied that there is a unanimous view on the course of poverty. Nevertheless real incomes have fallen at the lower end of the income distribution. Poverty must have risen on any commonsensical definition.
1. 10 April, 1999.
2. B.H. Easton, The Commercialisation of New Zealand (Auckland University Press, 1997), p.142-7; B.H. Easton, In Stormy Seas: The Post-War New Zealand Economy (Otago University Press, 1997), p.257-8.
3. S. Carson & B.H. Easton, The Economic Status and Health Status Project, Paper to the 1999 conference of the New Zealand Statistical Association, July 1999.
4. M. Mowbray, Incomes Monitoring Report: 1981-1991 (Social Policy Agency, Wellington, 1993); V. Krishna, “Modest but Adequate: An Appraisal of Changing Household Income Circumstances in New Zealand”, Journal of Social Policy of New Zealand, Issue 4, July 1995 p.76-97; B.H. Easton, “Poverty in New Zealand: 1981-1993”, New Zealand Sociology, Vol. 10, No 2, November 1995, p.182-213; R. Stephens, C. Waldegrave, & P. Frater, “Measuring Poverty in New Zealand,” Social Policy Journal of New Zealand, Issue 5, December 1995, p.88-112; N. Podder & S. Chatterjee, Sharing the National Cake in Post Reform New Zealand: Income Inequality in terms of Income Sources, Paper presented to the New Zealand Association of Economists, 1998; Statistics New Zealand, Incomes: New Zealand Now, Wellington, 1999.
5. The 1995/6 year is for households who reported their previous year’s income between April 1995 and March 1996. This gives an average of the incomes for the year ended September 1995.
6. Indeed dividing by numbers of household inhabitants to give per capita income is using a crude household equivalence scale.
7. For the conclusion to be unambiguous the lorenz curves must not cross.
8. To convert gini coefficients into coefficients of variation see B.H. Easton, Income Distribution in New Zealand (NZIER Research Paper No 28, Wellington, 1983) p.33.
9. i.e. the effect of raising real taxes by not changing tax brackets for inflation.
10. An Australian audience is likely to ask what was the effect of GST. The research evidence suggests that GST did not in itself increase inequality. However, the resulting income tax cuts were skewed towards the rich, so the total package of GST plus income tax cuts increased inequality.
11. B.H. Easton, “Distribution”, in B. Silverstone, A. Bollard, & R. Lattimore (eds) A Study of Economic Reform: The Case of New Zealand, (North Holland, 1996).
12. Figure 4.9; page 60.
13. R. Kerr, Equalising Incomes or Reducing Poverty: Which Basis for Welfare Policies? (New Zealand Business Roundtable, 1999).
14. R. Stephens et al op. cit. (1995). See B.H. Easton, “Measuring Poverty: Some Problems” Social Policy Journal of New Zealand, 9, Nov 1997, p.171-180, for a commentary, and R. Stephens, C. Waldegrave, & P. Frater “Measuring Poverty: Some Rebuttals of Easton” Social Policy Journal of New Zealand, Issue 9, November 1997, p.181-185, for a reply.
15. B.H. Easton, “Poverty in New Zealand“: Five Years After, Paper for the Conference N.Z. Sociological Association, 1980; Easton (1995) op cit.