A chapter of Globalisation and Welfare State
Keywords: Distributional Economics;
Note that this Chapter currently lacks the figures and tables which it discusses
In the 1980s, the objective of reducing inequality was given an increasingly low priority in policy implementation. Why this happened is in some ways a puzzle, because nominally the party in power was at first Labour who had once had a strong commitment to social justice and reducing inequality. Yet as we shall see, they steadily abandoned that commitment, although they did not go the distance of the succeeding National government.
The most evident step, was in 1987, when Labour’s election manifesto demoted full employment from being a top priority. There are no accounts of why it abandoned this key element of social policy for almost half a century. Perhaps it could be explained by a loss of faith in the ability of government’s ability to promote full employment, but the changes it made to the redistribution of income mechanisms – taxation and benefits – are more difficult to explain. There is a curious element to the story though. It was widely believed that market liberalisation would lead to an increase in economic inequality. Leaving aside the impact of unemployment, that is not particularly true, aside from those just mentioned redistributional mechanisms. This will astonish many readers, so I need to summarize the evidence of a much longer paper. First we need some sort of formal model of the determination of the overall income distribution.
The central features of the reforms, relevant to this chapter are shown in boxes in Figure 1, as are the other exogenous (for these purposes) influences. They reforms are separated into market liberalization, macroeconomic policy, and fiscal policy. Two further elements which is also taken as exogenous in the world economy, and the factor endowment.
The world economy, and the outcomes of the market liberalization, and the macroeconomic policy come together to determine the domestic factor prices, and the factor distribution. Meanwhile the outcomes of the macroeconomic policies, the factor endowment, and the factor prices themselves, plus social factors (such as the willingness of women, the elderly, and others to work) determine the utilization of the factors, a marker of which is unemployment. The interaction of the factor prices and the factor endowment generates the personal market distribution. Adjustment for taxation and benefits gives the personal disposable market income, which combine into household disposable incomes, also influenced by social factors such as demographic change and household composition. Full income or the social wage, where effects such as government spending, prices, quality of services, and hours worked are taken into account, is explored in the next chapter, while the focus of this chapter is on the 1980s and after, it is necessary to start earlier in order to identify long term trends. For various reasons – including the beginning of key data series and the report of the Royal Commission on Social Security (1972) – as far as possible the data begins in 1971/2.
The World Income Distribution
Key to understanding globalisation and the economic distribution is that a small trading economy cannot isolate itself in the long term, or perhaps even the medium term, from the rest of the world. The prices at which goods are traded reflect factor prices, and world interest rates affect domestic capital transactions. One might also argue that competitive market pressures would also affect the ability of a country to choose its own tax regime – certainly the attitudes and conventional wisdom of the world financial community seems to argue this.
Tracking the story of the world income distribution is not easy, although there is data on the key elements of wages and interest rates, and some recent data on international comparisons of the household distribution. To understand what was happening in New Zealand we begin by considering these.
The largest element in the factor income distribution, and on some measures the greatest contribution to distributional inequality, is labour earnings, assessed in part by wage rates. Table 1 shows that real wages rose rapidly in the OECD between 1968 and 1973, including in New Zealand. But after 1973 real wage growth was much slower. In the G7 countries they rose less than 1 percent a year. The New Zealand experience is even more miserly, with the real wage income falling in the post 1979 period. Comparing the real wage changes to this crude measure of productivity change (measured by real GDP per person employed), we observe that after 1973 workers in the G7 had real wage increases less than average labour productivity increases by about ½ percent annually over the entire 22 years. The gap was greater in the 1980s. The New Zealand real wage income to real GDP per worker outcome is not too different, although the relative gap was larger in the 1980s. Thus while the New Zealand economy performed more poorly than the G7, so its real wage path was lower, the broad picture is the same: in most of the rich economies there has been downward pressure on workers real wage income since the mid 1970s. We cannot simply attribute the declining real wage path we observe during the 1980s to internal changes in New Zealand and to the reforms.
The story of the real wage path is reinforced by that of real interest rates, for they were higher in the 1980s than in the 1970s. Table 2 records real rates for the G7 countries, and New Zealand. Up to the end of the 1970s, world and New Zealand real interest rates, whether short or long term, were below the volume GDP growth rate. They were low, and even negative. Then the G7 rates rose dramatically, on average by more than 5 percentage points between 1974/1979 and 1980/1990, while the New Zealand domestic rates were over 7 percentage points higher than they had been in the previous decade.
These higher real interest rates impacted throughout the world. The high real interest rates of the 1980s may offer a prosaic, but more relevant, account of the pressure for privatization of government assets. A government which is borrowing at negative real interest rates can afford the luxury of not very profitable commercial assets. When interest rates increase substantially, the attractiveness of such investments diminishes.
There is no entirely satisfactory wage index. Figure 2 shows the track of average hours earnings relative to the Consumers Price Index. The average hourly wage rate is based on an industrial survey – basically it is the wage bill divided by the number of hours worked. As an index measure it suffers from a variable composition in the work force. In particular in a period of rising unemployment and stagnant pay rates, the index could show a rise because lower paid workers are more likely to become unemployed. real wages rose between March 1978 and March 1982. Under the freeze they experienced a decline, but of only 11 percent to March 1985, bottoming earlier. Since the nadir they have recovered by a total of 7 percent to March 1993.
Work by Sylvia Dixon on wage dispersion reinforces the general picture of the critical trend beginning before 1984. (1) Her data suggests that wage dispersion narrowed until the late 1970s, but increased steadily thereafter. There is no discernable change in the increase in the rate of wage dispersion after 1984. There seems to be two processes operating here. One is the ~bounce back’ effect from wage compression of the early 1970s, which appears to be completed by the early 1980s. The other process is the secular increase in dispersion in the late 1970s and thereafter. this may be due not only to rising unemployment (which began in the late 1970s), but also to the impact of globalisation with its greater supply from low wage, low skilled, labour markets.
The pattern of the share of labour income in added value (Net National Income, NNY) (2) shows that the factor share of labour in NNY was relatively low in the early 1970s (near 70 percent), probably because the terms of trade were high and farming profitable. After they collapsed following the end of the commodity boom of the early 1970s, the factor share steadily rose, reaching 77.3 percent in 1990/1. As we might expect from the wage path, the factor share also fell rapidly during the freeze to 69.3 percent in 1984/5. Since then it has recovered a little, but the factor share was broadly flat averaging 73.2 percent between 1985/6 and 1991/2. So the dramatic change in the factor distribution occurred before the reforms. Subsequently there has been some stability in the factor share. In contrast I found there was a tendency for the factor share of labour to rise up to the 1970s. (3)
The information on aggregate returns of wealth is skimpy. However it is clear that the yield gap between return on capital and the interest rate narrowed dramatically in the early 1980s. (4) Bonds and fixed interest debt became more attractive, and despite a slightly higher return, investment in productive capacity become less attractive. Equity investors looked for higher return activities, of which the property and share market boom of 1986 and 1987 was the most spectacular example, even though it was not underpinned by any genuine change in the underlying rate of return.
In summary it is not possible to attribute significant changes in the factor distribution to the market liberalisation. The shift in the real interest rates occurred in the late 1970s, in events external to New Zealand, and had already impacted on the economy by the early 1980s. The wage and price freeze reduced the labour share which enabled some rise in the return on capital. There would have been market pressures for this rise anyway, and it may well be that the reforms disabled the alternative pressures for a recovery in the real wage.
Regrettably, there is little information on the distribution of personal wealth and human capital.
We can say more about the utilization of human capital – unemployment. Table 5 shows the available Population Census figures back to 1896 (Interpretation should allow for definitional and social change over time), while the March unemployment rate since the 1970s is shown in Figure 3. (5) There appears to have been a sharp rise in unemployment in the 1980s, to levels for males comparable to those of the late stage of the depths of the inter-war depression. The female ones are even higher, but this reflects social change as more women choose a career in the labour force.
From one perspective, a higher unemployment rate might be interpreted as an increase in the inequality of the allocation of paid work in the economy. Systematically relating the allocation of employment by income is more complicated. The only useful consistent series are from the population census, but they are not available before 1986. (A further complication is that there is good reason to believe that disguised unemployment, that is people who would like a job but are too discouraged to seek one and are thus classified as not-in-the-labour force, is affected by socioeconomic variables, which locate people in the income distribution. Moreover the unemployment refers to the preceding week whereas the income refers to the preceding year.)
It is possible to show that the unemployment began to penetrate into higher income strata in the late 1980s.(6) Moreover the churning observable in Labour Department employment register is indicative that it was substantial in recent years, and probably affected much of the bottom half of the employed income distribution. (7)
In summary there has been a rise in unemployment during the reforms, but there is some evidence that it was more “equally shared” in the late 1980s.
Personal Income Distribution
There is not a lot of personal income distribution material available. However Table 7 is derived from the various population censuses and shows the distribution of income by deciles for the adult (over 15 years old) population. It tells us, for instance, that in 1991 the top ten per cent of adults reported receiving 30.3 per cent of the total reported income. In simple terms, their income averaged three times the average adult income. Over the same period the bottom thirty per cent reported receiving 7.3 percent.
The definition of income is critical. Until 1981 the census income question was in terms of the individual’s market income. Since then the question asked about total income, with benefit income is added to market income. (Income tax is not deducted.) Miraculously both questions were asked in 1981. The overlap gives some possibility of comparison over the whole of the post-war era.
Excepting the 1945 to 1951 period, where as the table footnote observes data definitions make comparisons difficult, the general post-war picture is one of mildly reducing inequality, with variability about the trend. The top decile share falling from 38.5 per cent in 1951 to 34.9 per cent in 1981, followed on the wider income definition, from 31.3 per cent in 1981 to 30.3 per cent in 1991. Conversely the share of the bottom half of adults rises from 4.0 per cent in 1951 to 7.7 per cent in 1981, and then from 15.5 per cent in 1981 to 18.8 per cent in 1991 (although some of this increase may be due to the grossing up of social security benefits to include taxes in 1986).
It is possible that the comparisons are still subject to a measurement artefact because of the grossing up of social security benefits from 1987. The apparent improvement for low income deciles in 1991 may be due to this. One must be cautious. The data suggests there may well have been in the 1980s a slight increase in inequality in market incomes, probably as a result of rising unemployment. This was offset in part by social security benefits (of which unemployment may have been crucial) giving a slight decrease in inequality of total incomes, through to the 1990/91 year.
This and other more fragmentary data series over the 1980s perhaps tell the following story. (8) From them it is difficult to argue that there was a major change in the personal (market) income distribution in the period from the market liberalisation. But there was probably a shift towards greater inequality in the personal income distribution. 13, a conclusion supported by data presented in section 7. This can be attributed to three effects. First, in the early 1980s there were factor price shifts which followed world trends. Second, there was rising unemployment. Third, and partly offsetting the first two, there was the rising earnings of women, who are positioned lower in the income distribution.
None of these effects may be directly attributed to the market reforms themselves. This might seem surprising. However, suppose that the reforms affected the utilization of the factor endowment across all groups, as much those with high as well as low incomes. The total impact might be collective impoverishment, but no overall change to the personal income distribution. Because we see easily the hardship of the reforms on, say, workers who were laid off, we might underestimate the distributional effects of those who kept their jobs, and were promoted. Thus there may have been considerable impact on individuals, but the impact on the aggregate distribution seems to have been broadly neutral. In the end, the reforms may have been “democratic”, with all classes suffering equally. (9)
Benefit, Taxes, and the Reforms
Traditionally, the social security benefit entitlement has been on the basis of need, so the level was the same for all beneficiaries in similar domestic circumstances. Exceptions include accident compensation from its inception in 1974, and national superannuation which replaced the age benefit in 1976. However in the 1990 benefit reform, the traditional parity between domestic purposes, invalids, sickness, and unemployment benefits was broken. The recent relativities for a married couple are shown in Table 8. The rate they would have been if the pre-1991 regime had been retained is shown as the Royal Commission Benefit Datum Line (BDL).
Figure 5 shows the sickness benefit since 1972 measured in constant 1992/3 prices. The recommendations of the 1972 Royal Commission on Social Security led to the social security benefit increasing in those prices from around $236 a week for a married couple in 1971/2 to around $281 in 1974/5. This level was maintained through to 1990/1 with a perhaps a slight upward drift, although there was a weakening of the level in the late 1980s as the government began to address the budget deficit. In April 1991 the sickness, and most other benefits were cut by varying amounts to levels comparable or lower than those rejected by the 1972 Royal Commission, the invalid benefit excepted.(10)
If the real level of the benefits hardly rose in the 1980s, the total real net spending on benefits increased. Partly this was the result of demographic pressures: the rise of the numbers of the elderly and solo parent families. But the rise in unemployment – which also affected numbers of sickness and domestic purposes benefits, and the elderly who involuntary retired – also contributed to upward pressure on government spending.
We can use the underlying tax index series on which the RDII is based. (11) The index measures the burden of taxation on various points in the employee distribution. The effects of the introduction of GST and the abolition of sales tax, was included, assuming that all disposable income was consumed. Figure 6 shows the tax rate for three points in the wage and salary distribution, which change over time according to the method used in the RDII. In 1981 the wage rates reflected the average of the top, middle, and bottom quintile, although they are unlikely to do so today. Not surprisingly, given the fiscal pressures, the effective tax rate has risen for all three earnings levels. However it has risen much more for the bottom quintile than the top one, with the middle quintile experience between the two. In particular the bottom quintile’s effective tax rate rose from 18.0 percent in 1981 to 28.1 percent in 1993, or 10 percentage points, while the top quintile rose from 39.1 percent in 1981 to 40.8 percent in 1993 or by just 1.7 percentage points. (The middle quintile rose from 26.6 percentage points to 35.4 percent in 1993, 8.8 percentage points.) This is not unexpected given the flattening of the income tax scale.
What is unexpected is that the sharpest narrowing in the tax rates occurred in October 1982, before the reforms although the cuts in the top income tax rate in October 1986 and 1988 are also evident in the graph. The reason we cannot see any dramatic impact from the latter two reforms is that the largest beneficiaries from the later tax cuts were only those at the very top. In 1984/85, the point at which tax payers started paying 45 percent on their incomes was $25,000, equivalent to $40,500 in 1990/1 if indexed by changes in hourly earnings. The point at which the 66 percent rate cut in was $38,000 in 1984/5 (equivalent to around $62,000 in 1991/2). I estimate from the 1991 census that only 4½ percent of all wage and salary earners had incomes exceeding that figure. Thus the big gainers were not even a quarter of the top quintile, while some at the bottom end of the top quintile may be paying more relatively more tax. The upper quintile figure is thus an average, which obscures how well the very rich have done. This is a persistent problem in distributional analysis in New Zealand. The top quintile or even the top decile is too broad, to trace the experience of the very rich. On the other hand we have little systematic information on the top, say, 1 percent of the distribution.
In summary the period up to 1990, the government faced severe fiscal stress. As a result the burden of taxation rose and the real value of benefits was not markedly increased. However the burden of taxation was shifted towards low income earners, who experienced much larger tax rises than those further up the scale, while that on the richest fell. Thus the taxation system became less progressive. At the same time, in terms of the measure used here, beneficiary real incomes were not undercut, or perhaps only a little in 1989 and 1990. The difficulty of drawing conclusions is such a lot happening in terms of detailed changes, that it is difficult to trace each effect. Alternately we can look at their overall impact on incomes.
For the record, the top one percent of those over 15 reported incomes for tax purposes in excess of $81,000 in 1989/90. Their share of total reported (assessable income) was 8.5 percent. Using a pareto coefficient for the upper tail of 3, we calculate the following as the lower points of the upper percentile of personal incomes from other data sources as follows:
1990/1 using assessable income for taxation purposes: $81,000;
1990/1 using Population Census reported income: $81,000;
1990/1 using Household Expenditure and Income Survey: $83,000.
Household Disposable Income
As explained in Chapter 7, the standard means of comparison of household standards of living involves “equivalent disposable income”. Disposable income is market income adjusted for benefits and direct taxes (to give “total income”). Incomes are further adjusted for household composition, to take account of the economies of scale of living in a larger households, and the fact that children generally do not require the same income as adults to attain the same standard of living.
Mary Mowbray provides data on equivalent household incomes between 1981/2 and 1990/1, which she has updated to 1992/3. (12) Figure 7 shows the average income of each household decile as a proportion of the mean income for the year. The following are discernable patterns in order of how obvious they are.
– The bottom decile shows some fluctuation which may be around a flat trend. This probably reflects the sampling variability arising from the small number of household which report substantial losses.
– The change at in the top decile is more systematic, with a near flat trend to 1987/8, and then the share rising from an average share of 20.4 percent (i.e about double the average) up to 1987/8 to an average share of 24.5 percent in the four years from 1988/9. This represents an increase in the share by one fifth.
– The third change is there is a definite weakening in the other deciles (excluding the second upper quintile) after 1988.
Thus there was after 1987/8 an increase in inequality of equivalent disposable income.
Perhaps this is not surprising given the substantial cuts in top tax income rates which occurred in October 1988, while tax concessions were withdrawn on all incomes. What is surprising is that we cannot observe any major effects from the October 1986 cuts. Probably, while the cuts were as deep at the top (from 66 to 48 percent, in comparison to the 48 to 33 percent in 1988), they were not as broad. The October 1982 cuts do not appear to have affected the household distribution markedly either. If there had been data for the years immediately after the earlier tax changes an effect might have been observable, but two years later the cuts had been offset by the fiscal drag.
However the tax cuts were not the only cause of the increased inequality. It can be shown there was a relative decrease in the incomes of those in the second to bottom quintile after 1988, which does not seem to be due to either tax nor benefit changes. The most likely explanation is that it reflects some deterioration in the labour market.
A third major effect on the income distribution was the benefit tax cuts of 1991, discussed below. Note that the effect mentioned in the preceding paragraph occurs before these cuts, as illustrated in Figure 9.
Figure 8 shows the Mowbray estimates for households which have at least one Maori or one Pacific Islander in them, or of “other ethnicity” (mainly Pakeha/European). The figures bounce around a bit, more so than those in Figure 7, which suggests there is a serious sampling error for the small ethnic populations.
Insofar as there is a trend it would seem that those living in Pacific Island households (i.e. one adult in the household was a Pacific Islander) experienced a falling position in the distribution in the 1980s (and on average were about 86 percent of the national average). Meanwhile the Maori trend seems to be slightly rising to 1987/8 and lower after 1988/9, although the peak in 1988/9 is extraordinarily close to the national average. The Maori mean over the period is 88 percent of the national average.
Figure 8 also shows the position of households with children. The rise in 1988/9 probably reflects the improvement in family support in that year. The average for the period for is 90 percent, consistent with the picture that there are many households with children in poverty.
The 1990 and 1991 Welfare Reforms
In addition to the 1988 income tax cuts, the most important redistributive package was that of the 1990 Economic and Social Initiative and the closely related 1991 Financial Statement , described as the “redesign of the welfare state”. (13) For this chapter’s purposes they involved a major reduction in most social security benefit levels and entitlements, although there were other changes which affected government spending, and introduced – or increased – user charges, while the Employments Contract Act of 1991 was also seen as an integral part of the package.
The nominal reason for the measures were the actual and projected budget deficits, although the official statements involved the usual exaggerations to justify the harshness of the cuts. The cuts were on the poor because, apparently, as the Prime Minister promised, although not pursued with much diligence thereafter (with the exception of tertiary education fees):
As I said earlier, in moving to reduce the fiscal deficit, we have rejected the simple option of increasing taxes. However, top income earners are on a low rate compared with most overseas countries. That being so, we will be looking to this group to pay for more of the social services they currently receive free or with heavy subsidy from the state. (14)
There was perhaps an unarticulated argument which arose from the a belief that a behavioral response from the unemployment benefit level was discouraging active job search. Perhaps too it was thought the benefit was propping up the lowest wages, which were discouraging job creation by firms. Undoubtedly the government saw its benefit cuts and Employment Contracts Act as a part of the same package. Unfortunately there is not a lot of official documentation as to any underlying economic principles, yet it would be surprising if some advisers did not have some such economic account underlying their thoughts.
Whatever the intention, the sickness benefit cuts were deep (Figure 5), and the unemployment benefit cuts even deeper. Up to March 1991, both benefits were close to the Royal Commission’s BDL. After, they were 7 and 18 percent lower (Table 8). In addition the stand down period for the unemployment benefit was increased to six months, and other benefit eligibility rules tightened.
While there is not the ongoing data series to evaluate the cuts, we can use an ad hoc procedure to update them. It involves the assumption that the change to market incomes and other government policy impacts did not affect the underlying distribution between 1987/8 and 1991/2 – a conservative assumption given the evidence of the chapter this far. Given these assumptions we can update the 1987/8 “Government Income and Outlay by Market Decile” table of the Fiscal Impact on the Income Distribution report to 1991/2 ,(15) using actual events and Treasury forecasts (which may differ from outturns).
Since we also know the individual components of the table (such as spending on the unemployment benefit by decile), the individual items can be adjusted according to the Treasury estimate of the savings. Thus the impact on actual household spending power can be calculated. The result is shown in Table 13.
The top quintile suffered little, not surprisingly since the only significant reductions were the loss of family benefit at $6 p.w. per child, and a slightly higher charge for some medical treatment. (Tertiary fee hikes come later.) On the other hand those in the lowest quintile, who are very dependent upon social security, suffered heavily from the benefit reductions of level and eligibility.
Changes in Poverty Numbers
Just as we use a mean or coefficient of variation to characterize some feature of an income distribution, we can also use a poverty line, as discussed in chapter 7. For our purposes a poverty level provides a useful insight into the overall impact of the 1990/1991 packages on those with low incomes. The poverty line used here is that set by the 1972 Royal Commission on Social Security, known as the BDL (Benefit Datum Level).If it had been increased in line with long term real income growth, it would have been 12 percent higher.
Because of the Mowbray data it is not necessary to use a single poverty level, not provide it in a particular year. Figure 9 provides estimates of the proportions of people (not households) below seven different poverty lines, the middle one being the Royal Commission BDL for each of the available years between 1981/2 and 1992/3.
All the poverty lines show a show a major increase in poverty in the early 1980s, with some deterioration after 1988. In proportional terms the Royal Commission BDL records one of the lowest increases in poverty in the range. The numbers rose from about 430,000 in 1994 to 593,000 in 1991/2, an increase of just over 35 percent.
This is a very big increase, although one supported by anecdote and surveys of increasing hardship. It probably cannot all be attributed just to the benefit cuts. Real per capita GDP fell about 4.0 percent over the two years, while unemployment rose from an average rate of 7.1 percent in 1989/90 to 10.6 percent in 1991/2. At the very least, one could say that the poor were not protected from the economic deterioration, but probably it would be fairer to say the cuts worsened their circumstances when the economy was deteriorating.
On the systematic evidence available, supported by anecdote, there can be no doubt that the welfare reforms of 1990 and 1991 increased economic inequality and generated hardship among the poorest in the community.
New Zealand in the OECD Distribution of Household Disposable Income
In recent years the Luxembourg Income Study (LIS) has assessed changes of household income. New Zealand is not a member of the study, but the LIS researchers have used available New Zealand data, to locate New Zealand as best as possible. Fortunately a New Zealand-Australian comparison exists, whereby we can cautiously locate New Zealand in a wider picture. The income unit used has market income supplemented by social security entitlements, with taxation deducted to give equivalent disposable income (see section 8 for details).
Using the LIS data base, Atkinson, et al examined trends in income inequality in 13 OECD countries and concluded “for the nations we can study over the [1980s], inequality rose in a majority of nations studied and particularly in the United States, the United Kingdom, and Australia”. (16) Thus there was a general tendency for disposable household income inequality to rise in the 1980s, but there were considerable differences between countries. Given our earlier identification of downward pressures on real wages and upward shifts in interest rates we should not be surprised, but reductions in tax progressivity also played a part. No country can completely isolate its tax system and levels from international trends. One of the pressures on New Zealand was the international lowering of top tax rates. They used gini coefficients to group countries by inequality of household incomes in the 1980s (Table 3). It was not possible to include New Zealand fully. However Peter Saunders (with Helen Scott) compared New Zealand and Australia by adjusting the latter’s data parallel to the published New Zealand data. (17) He found that the equivalent disposable distribution in Australia and New Zealand were becoming less equal in the 1980s, but that the New Zealand distribution is less unequal than the Australian one. On the basis of the estimated gini coefficients, New Zealand would still be in an intermediate position between the “Northern Europeans” and the “Southern Europeans plus Commonwealth”, as Atkinson, et al had also judged.
If it difficult to place New Zealand in an international hierarchy, it is even more hazardous to assess the extent it has changed its placing in the hierarchy. One of the best attempts was by the Joseph Rountree Foundation. (18) Their New Zealand data only covers the 1980s and shows increasing in the latter part of the period, consistent with the above findings).
The Rountree study found that while the UK income distribution was growing more unequal in the 1980s, during the late 1980s, the New Zealand inequality was growing faster. The Report concludes “income inequality has been growing more rapidly in the UK than in any other country [from a total of 20] except New Zealand.” (19) almost certainly this conclusion would apply for New Zealand in the early 1990s.
In summary, the rich world generally, but not uniformly, experienced increasing household disposable income inequality during the 1980s (with data not yet available for the 1990s). New Zealand seems to have been a middling country in terms of ranking inequalities, but form the mid 1980s it appears to have suffered the greatest increase in inequality among the 20 OECD countries for which data is available.
The Increase in Inequality
The market liberalisation was among a number of things which caused a dramatic change in the distribution of wealth, income, and spending power in the 1980s and 1990s. While individuals and groups were affected by the change, which can be traced at the level of anecdote and case study, these microeconomic changes need no accumulate to a macroeconomic change of the aggregate distributions. However there was such aggregate change, most evident in some of the available measures of the household disposable income distribution.
This chapter’s task has been not only to describe these changes but to try to identify the causal processes which generated them. In the introduction we identified six exogenous groups of influences. We can no assess each’s importance, in terms of the evidence available. The order of presentation is from least important to most important. Some processes not mentioned in the main text because there was no useful evidence a mentioned here as hypotheses.
There is no evidence of an impact from the changes in the factor endowment over the period. It does not seem necessary to use it to explain the changes in the economic distributions. One possible process, not explored here, is the possibility that exogenous technical change affected the value of types of human capital. In particular, an increasing demand for skills was not matched by an increasing supply, which generated unemployment among the unskilled. This hypothesis will have to be investigated when the data becomes available.
The role of social factors such as women’s work preferences, household composition, and demographic change remain shadowy. Undoubtedly the former affected the gender distribution, and probably reduced aggregate inequality – although perhaps not by much. There is no published work yet on the impact of changing demography on the economic distribution, although smaller households mean that the average real incomes on a household equivalence basis are rising more slowly than they do on a personal basis, reflecting the reducing household size and the resulting loss of economies of scale.
There is a popular claim that the rise in solo parent families (or domestic purposes beneficiaries) has contributed to rising policies, but there is no systematic evidence to support what at the moment amounts to a popular prejudice or, at best, an untested hypothesis. It does seem likely that the changing age distribution – at the moment towards the elderly and young people, may also affect the distribution. but the change is slow, and the magnitudes not large enough to generate the spectacular changes we saw in the late 1980s.
With a few, probably minor, exceptions it has not been possible to demonstrate that the market liberalization affected the aggregate distributions. While undoubtedly they affected the experiences of individuals, it appears that they impacted on all income strata roughly equally. This perhaps was a consequence of the effects of the old regulatory regime of detailed market intervention being so widespread that almost all social classes were beneficiaries from, and contributors to, the cross-subsidization.
The exceptions were that it seems possible that the change in relative prices tended to favour the rich, who also benefitted more from the improved quality and choice that liberalization generated. It is also possible that some sectors benefitted more than others. Probably part of the spectacular growth of the finance sector in a time of economic stagnation, was due to the liberalization of monetary markets.
The one liberalization change which we have not been able to trace, because it is too recent and the data is not yet available is the impact of the 1991 Employment Contracts Act. It is possible that the new industrial relations regime contributed to a widening of dispersion in labour earnings, although it could be argued the enactment facilitated a change which was already underway from other influences. We need longer data runs, and the record of at least one entire business cycle to assess this.
Undoubtedly changes in the world economy affected domestic factor prices, which would have increased to increasing inequality, as occurred elsewhere. However the external relevant changes occurred in the late 1970s and early 1980s, but at best do not seem to have impacted markedly until the late 1980s when other changes were much more important.
The effects of the four groups of exogenous influences thus far identified have all been negligible, small, or slow. The last two influences were far more significant.
Insofar as the macroeconomic policies of the reform contributed to a rise in unemployment and thus reduced utilization of labour, we would expect this to impact on the income distribution. However the study observed that there was a sense in which unemployment became more dispersed through the income distribution. Nevertheless there is clear evidence of a weakening of incomes of those just above the social security benefit levels, in the period after 1988. Typically these are households dependent upon wages, in the most marginal parts of the labour market. More vulnerable to unemployment and to relative wage reductions, the evidence is that they suffered relative to other groups in the community. It is especially significant that the changes occurred before the 1991 changes.
Nevertheless the largest impact was from the change in tax and benefit policies. The 1988 income tax changes, which cut top rates, without comparable changes in bottom rates, but with the elimination of a range of tax concessions, gave a substantial boost to the incomes of the highest 10 or 15 percent of households to the detriment of the rest. Unlike the 1983 and 1986 cuts however, fiscal creep did not pull those benefits back. In any case the income tax system at the top end is getting so flat that fiscal creep would have been small even had the inflation rate been higher.
A second change, which reduced inequality albeit in a small way, was the family support package of 1989. However its real value depreciated with inflation, and so the boost to families becomes less noticeable shortly after.
The third change, of a magnitude and permanency (thus far) comparable to the 1988 income tax reform, was the 1991 benefit cuts, although we observe some weakening of real benefit levels from 1989, presumably a fiscal consequence of the funding of the substantial income taxes cuts.Without question the benefit cuts reduced the incomes of the poorest, increased household income inequality, and added to the social hardship.
It may be that the 1991 benefit cuts should be seen as a consequence of relieving the fiscal stress generated by the 1988 income tax cuts (which had been temporarily covered by assets sales). However the political explanation was a “redesign of the welfare state”, although it is still difficult to see past the rhetoric to a coherent social vision, other than a notion of a higher degree of income inequality than in the past. Perhaps an account could be constructed based on the widening of the earnings dispersion from the higher levels of unemployment, the changing world factor price relatives, and perhaps the changing domestic factor endowment. It has yet to be done, and in any case it may find difficulty explaining the high marginal tax rates that were left on beneficiaries and others on low incomes.
In summary, the big impact on the distribution of income were the tax cuts on upper incomes in the late 1980s (under the Labour Government), and the benefit cuts in the early 1990s (under the National government). The impacts were substantial, and they substantially increased income inequality.
Next Chapter Ch 12: The Social Wage
1. Dixon (1995).
2. Not published here. See Easton ibid, Figure 3.
3. Easton, Income Distribution in New Zealand (1983).
4. Easton ibid (1997), Figure 4.
5. The data is an amalgam of census data and the Braae-Gallacher series before 1986 and the HLFS thereafter.
6. Easton, ibid, (1997), p.111-2.
7. See Chapter 8.
8. For the other series see Easton ibid (1997), p.112-115
9. Frustratingly, it is not possible to trace, in the available data sets, the impact of the speculative financial and property boom of 1986 and 1987, and the bust thereafter. This is partly because they are not precise enough to capture short term shifts. Many people assess the impact on their finances from the height of the boom to the bottom of the bust, whereas the data is better assessing the longer term of three or four years. Moreover much of the wealth increases (and decreases) were fictitious, and would not have been reported in the income data bases used here.
10. Bolger et al (1990:57).
11. We do not use here the “Real Disposable Incomes Index” (RDII) despite its wide use – or, rather, misuse. The index assumes that all parts of the labour earnings distribution experience the same increase, so there are no changes in the pre-tax earnings dispersion, and it ignores unemployment, despite this impacting on different parts of the earnings distribution to different degrees, with low income earners more likely to experience unemployment.
12. Mowbray (1993). The adjustment for household composition is based on the Jensen Household Equivalence Scale (1988), but note the scale may over estimate the strength of economies of scale, and thus under estimate the effective incomes of small households relative to large ones (Easton 1995). Insofar as the household equivalence scales overestimate economies of scale, children will be worse off than the level reported here suggests.
13. Bolger et al, (1990); Richardson (1991).
14. Bolger, ibid, p.10-11.
15. (1990: table 33).
16. Atkinson et al (1993:20).
17. Saunders (1994).
18. op cit (1995:65-66).
19. ibid (1995:66).
APPENDIX: MAORI INCOMES
We can report here on trends in Maori total income relativities in the 1980s. (Total income is market income plus benefit income before tax.) Unfortunately the only reliable data source, the population census, uses different definitions in each census. Table 9 reports the average income of Maori over the age of 15, by gender, relative to the income of the total equivalent population. Not unexpectedly, the Maori relativities are much lower than the non-Maori, although Maori women are closer to the female average, than Maori men are to the male average.
Comparing the three definitions involving Maori descent or ancestry we observe that Maori male income relativities were falling over the period, while Maori female relativities were rising. It has not been possible to assess whether this is an age effect, although that seems unlikely. It is likely this is an unemployment effect, but it is not possible to check this with a reasonably consistent definition of maoriness.