Studies seeking to measure inequality across households typically use a methodology which produces results that we argue are seriously misleading if they are intended to be used in the formulation of tax or transfer policy.
Households are clearly very heterogeneous in terms of size, age composition, labour force status, presence or absence of individuals with care needs, and so on. The standard methodology deals with this heterogeneity essentially by homogenising it. The variable used as a measure of household wellbeing is typically household money income or consumption. In order to deal with varying needs of different household types, and the idea that there are ‘household economies of scale’ – the cost of meeting these needs per household member falls as their number increases – a formula is used to deflate household income to try to achieve comparability across households.
For example, one typical ‘equivalization index’ is constructed by counting the first adult as one, the second as half an adult, and the first child and any other children as 0.3 of an adult, so that a household with two adults and two kids has an equivalence index of 2.1. A single individual household has an index of 1. The household’s income is then divided by the index number and the result becomes the measure of wellbeing upon which inequality measurements are based. This implies for example that the couple with two pre-school children will be judged just as well off as a single individual if their household income is 2.1 times as high as hers. There is a large literature on the limitations of this methodology, but our critique is more fundamental and wide-ranging.
Two fundamental weaknesses of equivalization
The first and most fundamental mistake is to take household income or consumption as the basic measure of wellbeing when there is so much heterogeneity in second earner labour supply at a given primary income. This assumption effectively sets the value of household production to zero, which is a serious mistake when there is high degree of substitutability between market and household produced goods and services, for example childcare. The approach also assumes that the allocation of household consumption and leisure is such that household members are equally well off. There is no attempt to model within-household decisions concerning individual time allocation and consumption, despite the fact that there is a large economics literature on this. The approach ignores the need for a conceptual framework based on an empirically-relevant model of household behaviour, such as the one we provide in our paper.
A second serious mistake is to suppress the significance of the household life cycle. In the above example of an equivalence index, a single mother with two children is regarded as just as well off as a retired couple if their equivalized incomes are about the same.
We argue that households should be disaggregated and separately identified according to observable characteristics that impact significantly on wellbeing, including most importantly:
- The household’s lifecycle phase: younger couples who have not yet had a child; couples with pre-school children, couples with older children; couples still of working age but without children at home; retirees.
- Single-person households; single-parent households; households with special care needs
This classification appears to us to capture the main determinants of needs/preferences and constraints that are relevant for policy analysis. These are the main household groups across which the trade-offs inherent in tax/transfer policy have to be made.
Empirical analysis of the Australian tax and transfer system
Drawing on Australian data we illustrate the way in which the equivalization methodology provides support for policies that shift the tax burden towards lower and middle income families and widen the net-of-tax gender pay gap. We select a sample of two parent families of working age and with two children under 15 years. Under these criteria all records have the same equivalization index of 2.1.
We first show that equivalized household income is a seriously unreliable welfare measure because it implicitly sets the value of household production to zero across families that exhibit a high degree of heterogeneity in second earner labour supply.
Given that the distribution of participation rates tends to be roughly constant across primary income (the income of the highest earning household member), the primary earner’s income would provide a far more reliable measure of household wellbeing because it is more closely correlated with the determinants of household living standards in a realistically formulated model. This is broadly equivalent to assuming that at a given primary income, a second earner’s contribution to the household, the sum of her market income and value of her household production, are on average about equal across households. This seems to us to be a far better approximation than that which says that the value of the latter is zero. Moreover, data on primary income is readily available.
The data show that the profile of primary earnings (as opposed to joint income), up to around the 80th percentile of the wage distribution, is approximately linear and relatively flat. It is in part due to this that an alternative ranking of household wellbeing on the basis of joint income results in that of low to middle wage two-earner households being grossly overstated relative to that of higher wage households with a zero-to-low second earner income. The former households are wrongly characterized as being much better off than the latter. This mis-ranking biases policy toward means-testing family benefits on the basis of joint income and away from fairer and more efficient policies, in particular more progressive taxes on individual incomes.
We demonstrate this by providing an in-depth analysis of the Australian tax/transfer system, one which takes full account of key elements that have been introduced in recent decades. The first and most notable feature of these is the way in which changes in the structure of effective marginal tax rates, set not only by the personal income tax and the low income tax offset (LITO), but also by joint income-targeted family payments under Family Tax Benefit Part A, have resulted in a major shift in the tax burden away from top income earners towards those on lower and middle incomes. This has had an especially strong negative impact on second earners, primarily women. These changes have been enacted at a time when inequality in pre-tax incomes, particularly at the top of the distribution, has increased considerably. Instead of designing a tax policy to offset the adverse impact of these on fairness and efficiency, the tax policy reforms have exacerbated them.
The results of our empirical analysis support the argument that Australia does indeed need a tax reform, but one that works in the direction of reversing these effects rather than intensifying them.
This is a summary of our working paper, ‘Inequality Measurement and Tax/Transfer Policy’, IZA Discussion Paper No. 13326.
Recent Comments