What does “decomposing income inequality” mean?

I was recently asked to explain the money-shot result from my PhD in a more accessible way – and I thought I’d share that here, just to get some feedback regarding whether the words I’m using actually make sense when they are put together:

I initially undertook this research due to my frustration when looking for information about tax and income inequality while working at Infometrics as an economist.  The 1980s and 1990s had seen dramatic changes taxes and payments to beneficiaries, and also to income inequality, so I thought it would be a valuable contribution to tease out how much these changes in tax and benefit settings had influenced income inequality in New Zealand.

To do this I worked with the models available at the New Zealand Treasury and Victoria’s Chair of Public Finance to calculate the payments that would have been received by someone in the early 2010s if taxes and benefits (including superannuation) were the same as they were in the late 1980s and early 1990s (adjusting these figures up for wage growth over that time). 

In other words, if New Zealand had not change the tax and benefit system and had instead increased payments at the same rate as wages since the late 1980s what would income inequality have looked like?

Analysing this information, and considering what would have happened in reverse – if current settings were applied in the late 1980s and early 1990s – I found that nearly 46% of the increase in income inequality appeared to be due to these changes in taxes and benefits.

However, this ignores the way that the change in taxes may have changed people’s behaviour – the new tax settings may have seen them work more, invest more in education, migrate to another country, change when they move out of home, or any of a myriad of other changes. 

As a result, I corrected the figure for the change in the way people worked due to policy.  This reduced my figure, suggesting that just over 39% of the increase in income inequality was due to the tax and benefit changes.

These things are hard as it involves trying to find a balance between accuracy and clarity – when there can be a trade-off.

“Transfers” is a more accurate term than benefits, given that a number of the payments are not seen as benefits (eg national superannuation), and the inequality number is disposable income inequality – the number after taxes and transfers. But I’ve also been told that these two terms get people stuck and make it hard for people to make sense of what is going on when trying to understand the result.

Also how do we discuss the shortcomings without overplaying them? There is a lot missing (how have some of the other characteristics of the population changed due to policy – how are these related to some other results in my thesis?), but given that the core of the result is itself of interest. This point estimate of 39% needs to clearly have uncertainty around it – but that uncertainty doesn’t imply its irrelevant.

Anyway, I’m not particularly good at research of communication – so trying to do both is a learning experience. Would be interested to here your thoughts about whether I’ve made appropriate choices about this trade-off, or in questions you may have about the research!