<>Draft for discussion; this version completed 24 May 2103.
Keywords: Political Economy & History; Social Policy; Statistics;
This note disinters a result I found in the 1970s and subsequently buried in Income Distribution in New Zealand. Years later it may have some relevance to some puzzles I am working on about the changing nature of family policy.
The burial was in a table which is reproduced in the appendix. It compares the income of taxpayers who claimed the dependants’ exemption with surveyed earnings, a measure of average wages and salaries.  The vast majority of dependants for this purpose were children (under 18).  For simplicity we call these taxpayers ‘fathers’, although they include some women who were heads of households with children. Throughout this paper ‘family’ means a household with pre-adult children and the adults who look after them (although variations such as three generation families also existed) .
The final column shows that the ratio of fathers’ income to wages decreases over time from almost a 70 percent premium in the early 1950s to below a 40 percent premium in the early 1970s. The pattern is persistent. The correlation coefficient of its trend-line is a high -0.95. 
Why the Declining Premium?
Practically the decline means that the standard of living of one income families with children was not rising as fast as average wages (there being few two income families before the 1970s). One can think of a number of reasons why this might happen but the most likely – the most dominant – is that lower income men were increasingly becoming fathers.
This does not mean that average one-income families were necessarily poorer. Adjustments have to be made for the incidence of income tax, and for the number and age of children (and also the cost of housing). Even so there must have been an increasing strain on family finances.
I leave to speculation why this eroding of the fathers’ income premium occurred, although one recalls the traditional practice was that men did not marry until they could support a family, although there were some households with dependents headed by taxpaying women. Instead we seem to have had a post-war ‘democratisation’ of marriage and child bearing.  By democratisation I mean a shift from decisions from being based on economic power to one where individuals thought they had a right to have children (increasingly) irrespective of their income. The focus of this note is whether this change tells us anything about what happened to social policy in the 1970s and 1980s.
The tax exemption for dependants was eliminated from 1973/74 following the recommendation of the 1972 Royal Commission on Social Security which wanted the savings to be used to (partially) fund a higher Family Benefit – it was doubled in April 1973. So we do not know what happened to the premium after that date. 
However there is no evidence in the existing series that the premium would have soon flattened out, although presumably it must have at some later time given that it is unlikely that fathers’ earnings would fall (on average) below that of other men. It seems safe to assume that the premium continued to fall, at least through to the 1980s, although perhaps not as fast.  Despite the caveats in a paragraph a couple of sections ago, I am going to assume there was a fall in family incomes relative to those of other households (after adjusting for household composition) – or, at least, that family income stress increased.
The Significance of a Lower Premium for Fathers’ Incomes
At this stage I want to briefly set down what may be the significance of a lower premium for fathers’ incomes as a means of stimulating a discussion on what was happening to social structure, the economy and income support in the period.
The first possibility is that women in low income families would be driven to work. The reasons for the rise in working mothers, which became important in the 1970s, are complex and not entirely economic; for instance, some woman wanted to maintain a career and have children too; others went out to work to reduce social isolation. Even so, the falling fathers’ income premium would have added to any other pressures.
A second possibility is that as some fathers found it harder to provide adequate income support the alternative of solo motherhood became less unattractive (particularly given the introduction of the Domestic Purposes Benefit).  Again there were certainly other social factors, such as changing attitudes – although there is more opinion and anecdote about them than rigorous analysis.
Having set out these possibilities I mention a curious incident which bothers me. Why was so little done to offset the apparently falling fathers’ incomes, especially given the central notion in the New Zealand welfare state has been that wages are responsible for funding children. (I have written repeatedly on this, and can provide material on request.) I accept that the policy makers would not have read my Income Distribution in New Zealand but the statistics apparently reflect a social reality of pressures on family income which seems hardly observed by policy makers and had little effect on their policies.
I can think of complicated explanations. One I have tended to favour is that the notion of the father’s wages supporting his children was so central to thinking that it was not to be challenged even by evidence. In any case, the main beneficiaries from challenging that policy – mothers and children – were poorly organised politically. However, another possibility is that there is something wrong with the fathers’ income premium series and in fact there is no dramatic fall. Or perhaps family size was falling at about the same effective rate. In either case the income pressure on families would be less than the data suggests.
Unfortunately we just do not have the data base to rigorously investigate family incomes in the first part of the post-war era. There may be opportunities in the Population Census data base, but a very useful compilation of long run demographic data does not shed any light on these issues.  (See also Appendix II for a limited exploration of age effects based on the available data from the 1966, 1971 and 1976 Population Census. It suggests that age- composition effects may not be great.)
Why this is currently important to me is that I am working on changes in the social security system in the 1970s and 1980s. I can show was that there was an unquestionably major change in the use of the system after the 1972 Royal Commission reported. Their analysis was based on a continuation of the stable social and economic trends of the previous three decades, but instead there were dramatic changes in the uptake of benefits. What were the drivers of those changes?
I am certain that the higher ‘normal’ level of unemployment was one, albeit less certain as to why it happened. Aside from the longish peak in unemployment arising from turbulence of the industrial restructuring of the 1970s and 1980s, my best guess is that there was a mismatch between the increasing technological complexity and the skills the workers offered, which added three or more percentage points to the ‘normal’ level of unemployment after the mid-1970s. It is also easy to trace demographic and policy shifts which had some effect..
I am not sure what is happening to overall family incomes and the related social conditions from the mid-1970s to the mid-1980s (when the household survey becomes available). As I have said, there is no lack of opinions and anecdotes, but they do not give me the foundation for a rigorous analysis. It may be that the falling fathers’ premium before 1973 may be useful for providing an account of what happened after 1973, but it may not.
 Economic and Social Trust On New Zealand; www.eastonbh.ac.nz
 The book used this measure as the key indicator of individual market earnings.
 Wives and housewives were covered by another exemption.
 There is no evidence that the data breaks in the late 1950s (at the introduction of PAYE) make much difference; presumably because almost all taxpayers with dependents were already reporting taxable income.
 An issue was the extent marriage was chosen rather than forced on the couple by premarital pregnancy.
 A further complication is that equal pay and the rising proportion of women in the earning labour force changes the interpretation of surveyed earnings.
 Analysis of the Household Expenditure Survey (from 1981/82) does not suggest falling relative family incomes in the early 1980s. After that, other changes, including rising unemployment and the tax benefit switches, obscure the market income trends.
 As a discretionary benefit in 1968 and a statutory benefit in 1973.
Mean Taxpayers with Dependents Income Relative to Survey Earnings
Year Taxpayers’ Income Surveyed Earnings Column1
$ % p.a. $% p.a. Column 2
1949/50 1368 862 1.69
1950/51 1618 909 1.78
1951/52 1702 1034 1.65
1952/53 1824 1092 1.67
1953/54 1940 1159 1.67
1954/55 2080 1264 1.65
1955/56 2142 1336 1.60
1956/57 2244 1386 1.62
1957/58 2358 1447 1.63
1959/60 2200 1475 1.49
1960/61 2430 1525 1.59
1961/62 2498 1617 1.55
1962/63 no data
1963/64 2582 1738 1.49
1964/65 2736 1805 1.52
1965/66 3039 2023 1.51
1966/67 3154 2120 1.49
1967/68 3233 2193 1.47
1968/69 3293 2318 1.42
1969/70 3562 2489 1.43
1970/71 3929 2837 1.39
1971/72 4463 3297 1.36
1972/73 5062 3662 1.38
1. Black lines indicate data break
2. Half Yearly Survey Earnings calculated at annual rate.
Source: B.H. Easton (1983) Income Distribution in New Zealand, page 240, Table 14.12.
APPENDIX II: The Effects of Changing Ages of Fathers.
Diane Owenga pointed out that the changing age of fathers could affect the average of father’s incomes since male incomes vary with age. (By fathers the note is referring to all fathers with dependent children, and not just the age at the time of birth.)
The effect is complicated because it is not monotonic. In particular male incomes peak in the early 40s. Thus having children earlier and fewer children will increase the numbers of fathers in the lower age groups who have lower incomes, but it will also reduce the numbers in the higher age groups with lower incomes relative to those in the middle.
Investigating the effect is even more complicated because most of the required data does not exist. Here is what I did.
There are tabulations of ages of the male (and female) head of households from the 1966 census by number living in the households. I have assumed that all those, other than the first two are dependent children, which gives me a distribution for fathers by age for 1966, 1971 and 1976. (I did not do later because that is well out of range of the table in appendix I.
However there are only four age groups: the very small numbers (less than 1 percent) of fathers under 21, those who are 21-24, 25-44 and 45-64 (assuming those of 65 are not relevant for this calculation).
Their average ages are 41.8 years for 1966; 41.3 years for 1971 and 41.0 years for 1976. This is consistent with the expectation that fathers were getting younger in much of the early postwar era. I am reluctant to assume this in the early 1950s in the early stages of post-war family formation. However the projection that I am about to do overlooks this and assumes a linear trend. (It would predict an average age in 1951 of 43.)
I then applied the 1971 male average incomes to the estimated father’s age distribution. The resulting averages were $38,600 in 1966, $38,400 in 1971 and $38,400 in 1976. This is a lower relative change than for the age itself because while there are relatively more fathers who are young, there are also fewer older fathers.
I then projected the ages back to 1951 on the assumption of the linear projection. Applying 1971 rates the average income would be about $39,700 or only slightly more (less than 1 percent) than the 1971 rate.
I’ve made all sorts of heroic assumptions here, but at the end of the day I conclude that any age composition effect is unlikely to be great.