In a previous post, I suggested that the treatment of housing in the official income distribution data massively underestimates inequality. This is because it fails to account for imputed rent (the non-monetised value of housing services enjoyed by owner-occupiers) and for capital gains. My calculations in that post were illustrative of why this extended income was important (renters became relatively worse off, homeowners better off), but those illustrative calculations were not analysis of real data. What I wanted to see was an income spectrum which included monetary income, imputed rent and capital gains, but also including an imputation for “social transfers in kind” – that is, the receipt of public services such as education, public health care, child care subsidies as well as a range of rebates and concessions. From there, the standard income inequality questions could be asked to properly analyse inequality in this extended income.
My Wish Granted
Since writing my original post, my attention has been drawn to the work of Yuvisthi Naidoo, from the UNSW Social Policy Research Centre. Her PhD and subsequent articles calculate many of the above changes (and more). She starts with standard household disposable income and adds values for net imputed rent and social transfers in kind to produce a measure of “full income”. She then converts household wealth into income flows in the form of imputed lifetime annuities (that is, an income equivalent to drawing down on capital to leave zero at the end of life). This is done for household financial assets to produce a measure of “potential consumption”, and then more controversially for owner-occupied housing to get “adjusted potential consumption”.
The incorporation of an income stream from wealth is important because it overcomes a fundamental problem in standard inequality data. Simply focusing on standard money income (and considering wealth separately or not at all) gives a false picture of which households have the most and least economic resources, and the gap between them. For instance, ABS data tells us that 24% of low-income households have moderate wealth, and 9.8% have high wealth, while 60% of households in the highest income bracket do not have high wealth, yet these households are ranked solely on income in most analyses.
The use of annuities to calculate income flows from wealth holdings is not the capital gains accounting I was envisaging: a simple capital gain = income equation. However, it incorporates capital gains into the base on which the annuities are calculated. For instance, in the annuities model, a $50,000 capital gain is not considered income of itself, but the annuity is calculated on the increased asset value and for a 20 year annuity would add $2,500 to the income stream. The annuities approach probably under-estimates the value of capital gains income to households, but it is a better overall accounting of capital wealth and income than simply capital gains – and is certainly better than the nothing in the usual income accounts.
Results
The table below shows Naidoo’s calculations for median household incomes at each step in the process. Surprisingly (to me) the overall result is that extended income is more evenly spread than disposable (cash) income. This is evident in the final column which shows the ratio of median income in the lowest income quintile to the median in the highest quintile decreases. When just standard money income is considered, those in the top income quintile average 4.2 times the income of those in the lowest quintile, but when extended income is considered (as “adjusted personal consumption”), this figure falls to 3.14 times.
Median Incomes $p.a | Income Quintiles | Ratio Q5/Q1 | ||||
Q1 | Q2 | Q3 | Q4 | Q5 | ||
Disposable Income | 18200 | 30021 | 40461 | 52923 | 76399 | 4.20 |
Full income | 33259 | 44644 | 54438 | 65677 | 89487 | 2.69 |
Potential Consumption | 33982 | 46538 | 57763 | 71747 | 103002 | 3.03 |
Adjusted PC | 36058 | 50693 | 62728 | 79006 | 113154 | 3.14 |
The reason for this more equitable distribution in extended income is the impact of social transfers in kind, the benefits of which flow disproportionately to lower-income households. The step from disposable income to full income (that is, the inclusion of net imputed rents and social transfer in kind) produces an 83% increase in income for the lowest income quintile, but only a 17% increase for the highest quintile. By contrast, the inclusion of wealth annuities only adds a further 8% to income in the lowest quintile, but 26% in the highest quintile – so (unsurprisingly) these capital incomes increase inequality.
Naidoo focuses on older-age households, and she goes on to investigate other measures, but for me the impact of government services in reducing inequality (and the quantification of that) is the first standout political point coming out of the analysis of extended income distribution.
However, equally important is the fact that there are different households in each quintile once we incorporate extended income. As I argued previously, the inclusion of imputed rents as income moves homeowners up the income spectrum while renters will be even more clustered among those with low incomes. But it is not just housing tenure. Naidoo’s research showed that nearly a half of all older people (65+) were in the lowest standard (money) income quintile, but the inclusion of imputed rent and social transfers in kind reduced that to 22.5% (because older Australians are disproportionately more likely to own homes and benefit from health services). By the time imputed wealth annuities were included in the analysis, only 17% of older people were in the lowest income quintile, while 26% were in the highest quintile (up from 7.1% when only disposable income was taken into account) (Naidoo, Appendix Tables C8-11) .
Implications
The data in Naidoo’s PhD is now dated (2010 HILDA income data), but the two key implications highlighted above are clear and remain relevant:
- expenditure on public services has a major impact on reducing inequality, and
- a more comprehensive income analysis changes who we see and understand as being on the lowest (extended) incomes, and potentially the most vulnerable.
The first point (and the quantification of that impact) is important because it not only makes a further case (beyond the direct health, education and other outcomes) for funding public services, but it may also add to our perception of tax-and-transfer policies. Progressive tax is usually understood as taxing high income earners at higher rates than those on lower incomes. This is a key instrument for limiting income inequality. However, the inclusion of social transfer in kind shows that spending this tax revenue on health, education and community supports is a further transfer from rich to poor.
That said, the second point above confounds the impact of progressive taxation. Since income taxation is largely based on money income, the amount of tax paid will depend partly on the type of income rather than the amount. Those with relatively higher cash incomes may pay much more income tax than those with significantly higher extended incomes but lower cash incomes. The money-based tax system does not follow households as they shift up and down the extended income spectrum and so some progressivity is lost (or mistargeted).
There are of course issue around taxing non-cash income as it would require a conversion of some wealth to cash to pay the tax. This itself may cause hardship – unless of course the wealth was maintained and simply taxed at realisation or end-of-life. Another argument for inheritance taxes – but I digress!
There is a further complication in the progressive tax story. A flat rate tax like the GST is generally regarded as regressive because it impacts disproportionately on lower-income households (who spend a great proportion of their income on GST-taxable consumption). While this is undoubtedly true, the fact that social transfers in kind disproportionately benefit those on low incomes, and that as a state tax the GST goes fairly directly to the provision of those services, provides something of an offset against its regressiveness – although this is probably only partial and not an argument for increasing or broadening the GST.
Finally, the movement of people up and down the extended income scale suggests that when we focus services and concessions on those in the lowest money (disposable) income brackets, we may in some cases be providing services and concessions to those who are relatively better-off, and missing out on people with fewer economic resources (just because all their resources are cash incomes). However, this is not as straightforward as the similar argument above about tax because in some cases it is the provision of concessions and services (i.e. social transfers in kind) which lifts households out of the lower income brackets. There may be a circularity in targeting based on extended incomes, but targeting services and concessions based only on money incomes is equally flawed.
The extended income analysis is not a policy panacea, but I think it does provide a better window on income distribution from which to do policy analysis.
Conclusion
My previous post asked for an analysis of inequality based on a broader understanding of income, one that combined income and wealth into one metric. Having now got data thanks to Yuvisthi Naidoo’s great work, I can see that my first estimate of increased inequality was wrong (because the impact of wealth inequality is offset by the progressive impact of social transfers in kind). But the analysis throws up a whole range of questions and challenges for progressive tax and transfer policies. The discussion above is only the beginning of an analysis, and it is fraught in an era when we must fight for even the basic principle of progressivity. In that sense, my take-home message from all this is: be careful what you wish for!
And yet …
I continue to believe that categories and statistics are socially/politically constructed, not neutral reflections of reality. Uncritically using definitions and statistics that are designed for other purposes or other theories limits our vision and the potential for change.