Uneven growth: detailed empirical appendix
This appendix details the empirical results presented in the paper. The appendix is divided in to 3 main sections: (1) Point-in-time, (2) Lifetime and (3) Lifecycle statistics.
Point-in-time
Sample
Our sample includes individuals aged 20 years and above with non-negative market and post-government income. The following table provides a summary of the sample composition by sex and age for each year.
Table 1: Point-in-time sample composition by sex and age
Summary statistics
All income and tax variables are expressed in terms of 2020 Australian dollars. The consumer price index is used to adjust nominal values. The following table provides summary statistics by year and gender. The following table presents summary income and tax statistics from 1991-2020. To examine a specific income component, type its name in the search bar under the “Income” column.
Table 2: Summary income and tax statistics
Mean income growth
This section examines trends in mean income and tax over time. There are three tabs in this section. The first presents trends in levels, the second presents trends in year-on-year growth and the third presents in terms of cumulative growth.
Growth incidence curves
We now turn towards examining growth across the distribution. For this purpose, we take market income as the base income concept. A good starting point to examining the distribution of growth is plotting growth incidence curves (GIC).
The following growth incidence curves plot the growth rate in the respective mean income component over time within each quantile of market income. The first tab plots the cumulative GIC for the 30 years. The other tabs break this down into decades and plot the average (annualized) growth rate.
Growth by quintiles
Trend in means
Cumulative growth
Growth at the very top
Inequality metrics
We present additional measures of income inequality along with trends in the Gini coefficient.
Redistributive effect
We measure the redistributive effect of tax/transfers using the Reynolds-Smolensky index. Redistribution is largely determined by the average size and progressivity of the tax/transfer system. We measure progressivity using two indices - the Kakwani index and the Suits index. Average size is given by the total tax (transfer) divided by total net income. We transform transfer progressivity indices by taking their absolute values. This means that an increase in the index value for either transfers or taxes imply and increasing level of progressivity.
It is worth noting that, in the case of the Suits index, the value ranges from 0 to 1. If the index is close to this upper bound (as is the case with the transfer system) it indicates that the system is highly progressive.
Lifetime
We examine 10 cohorts, each over a span of 20 years from the year they turn 30 to the year they turn 50. Lifetime income is the sum of real income over those 20 years.
Sample
The cohort sample consists of individuals who consistently filed tax returns from 30-50 years, who had non-negative lifetime market and post-government income.
Summary statistics
For the sake of comprehensibility, represent dollar values of lifetime income in their annualised form (divided by 20 years).
Mean income growth
This section examines trends in mean income and tax over time. There are three tabs in this section. The first presents trends in levels, the second presents trends in growth between cohorts and the third presents in terms of cumulative growth.
Growth incidence curves
We now turn towards examining growth across the lifetime income distribution. For this purpose, we take lifetime market income as the base income concept.
The following growth incidence curves plot the cumulative growth rate in the respective mean lifetime income component over the 10 cohorts across the lifetime market income distribution.
We break down this cumulative growth across the distribution into growth from one cohort to another.
Growth by quintiles
Trend in means
Growth at the very top
Inequality metrics
We present additional measures of lifetime income inequality along with trends in the Gini coefficient.
Redistributive effect
Using the same indices as we do for the point-in-time approach, we measure the redistributive effect of tax and transfers over lifetime.
Lifecycle
Income growth by age groups
Lifecycle profiles for our lifetime cohorts
Robustness checks
We conduct robustness checks of some of our main point-in-time results with 2 alternative samples (B and C) to our main sample (A).
Samples and their composition
The following list details the samples and their respective assumptions.
A (Main sample)
Use variables that are consistently reported in all years.
Negative income tax coded as a transfer.
Restrict to non-negative market and net income.
B
Use variables that are consistently reported in all years.
Exclude capital gains from capital income
Negative income tax coded as a transfer.
Restrict to non-negative market and net income.
C
Use variables that are consistently reported in all years.
Restrict to non-negative market, net income, labour, capital and tax.
Sample frequencies and percent excluded
Gender and age composition
Overall, all 3 samples have the same balance of gender and similar age compositions.
Growth incidence curves
We check whether the cumulative growth incidence curves that we obtain from the alternative samples are similar to that from our main sample.
The shape of the GIC curves are similar across samples. The magnitude varies slightly.
Inequality
We also check if trends in distributional indices are different across samples.
Further statistics using payment summary data
We extend our ALife sample for 2002-2020 with a sample of “non-lodgers” (those who did not lodge a tax return) from ATO’s client register. The sample of non-lodgers contain 2 income variables - labour income and public transfers. (Work-in-progress)