Note: Where available, the PDF/Word icon below is provided to view the complete and fully formatted document
Economics Legislation Committee

PHILLIPS, Associate Professor Ben, Australian National University

CHAIR: The committee will now resume. Welcome and thank you very much for appearing before the committee today. I invite you to make a brief opening statement.

Prof. Phillips : Thank you for the opportunity to come and speak today. The ANU Centre for Social Research and Methods has undertaken a detailed analysis of the proposed personal income tax cuts in the 2018 federal budget. We've provided a detailed submission outlining the distributional analysis of the tax cuts for Australian households where we focus on the impact of the tax cuts on household income distribution.

The analysis was undertaken by me and a number of colleagues of mine, including Professor Matthew Gray, who heads the Centre for Social Research and Methods, and fellow researcher Richard Webster. We've used the ANU model of the Australian tax and social security system called PolicyMod. The model is based on ABS survey data for 2015-16. The data is updated to the current financial year, for forward estimates and the years beyond that using detailed projections and benchmarking.

We attempt to mirror the assumptions of the budget as closely as possible. Broadly speaking, it comes under the guise of a microsimulation model. It's fairly similar to the models that you've heard already in the terms of the way it operates. It's fairly similar to the NATSEM STINMOD model. I previously was the manager of the STINMOD model at NATSEM. It's fairly similar to the Treasury's CAPITA version, which is another microsimulation model based on similar methodologies and similar data.

The model itself we benchmark to quite detailed ATO taxation statistics for 2015-16. The income distribution that we get largely comes from the ATO taxation statistics, so our results, probably not surprisingly, do mirror those quite closely of the government and also the PBO numbers that we've seen this morning.

The modelling takes account of the proposed personal income tax cuts. These include the proposed changes to tax thresholds in 2018-19, 2022-23, 2024-25; the introduction of the low- and middle-income tax offset for 2018-19; and the increase in the low-income tax offset, the pre-existing LITO, from 2022-23.

Our analysis is undertaken in a number of separate ways. We've also done a number of follow-up pieces—there was also a follow-up piece in the conversation. It was based on analysis at the individual level, which is similar to the analysis that the Grattan Institute has done and also to the analysis that Deloitte has done. However, the submission for which we've provided the analysis is at the household level. There are a number of ways of doing this. It doesn't really alter the results all that greatly. There are perhaps some academic arguments as to why one would use one over the other, but it doesn't matter for the purposes of today that much.

Our first analysis is a bracket creep analysis. This is where we consider the tax system of each year from the year 2000, and we model this up to 2017. So everything is in 2017 dollars—that's prices and wages. We take the system from, say, 2000 and we inflate all of the tax brackets to 2017. We do similar analysis for the out years—that is to, say, 2027. We bring all of the projected wages back to 2017. Let's say the threshold in 2000 for the top tax threshold was $70,000. We'd bring that forward, using wages growth to, say, $120,000 or $130,000. Likewise, I think the projection of the top threshold in 2024 is $200,000. We'd bring that back a little bit, into today's dollars effectively.

We also do a more standard distributional analysis that's for each year from 2017-18 to 2027-28. That's more aligned with the analysis that you will already have seen from various organisations that you've heard from this morning and from some that you'll hear from after me as well. This is the standard analysis where we use prices and wages for each year as projected by the budget. We try to stick to the assumptions of the budget as closely as we can. There's quite detailed information in the budget papers, so we usually think we can get quite close to the budget estimates. We consider average tax rates for each year and the impact on disposable income in comparison with the current trajectory of policy.

Our major conclusions were as follows: average tax rates peaked around the years 2002 to 2004. They dropped significantly due to very significant personal income tax cuts under the Costello treasurership, where tax rates were actually at a bottom between 2008 and 2010—about three percentage points lower than they were. Tax rates have slowly increased due to bracket creep between 2010 and 2017. Under the current policy, we expect that they will reach their historic peak by around 2021, and approach an even higher level by 2027. Even with these 'tax cuts' we are still seeing an increase in average tax rates across the board. That's lower-income, middle-income and, to some extent, higher-income earners. We're still seeing increases in taxes with these 'tax cuts'. That's due to the impact of bracket creep.

The proposed tax cuts will see average tax rates increase significantly beyond their levels of 20 per cent by 2026, but they do remain significant cuts with that in mind. Compared to current average tax rates, projected rates increase for all income groups, although a little more for the low- and middle-income categories. From that perspective, we would see this as being a slight reduction in progressivity of the tax system moving out to those out years of 2027. Over the first tranche of cuts, middle-income households enjoy the larger cuts relative to the current policy by 2021-22. By 2024-25, which I think is the end of the major parts of the tax changes, the larger gains are for higher-income groups. When I say larger gains, that still means that many people are paying a higher rate of tax right across the income distribution; it's just that the work that's being done against bracket creep is a little stronger for high-income groups than it is for middle-income and low-income groups—although keeping in mind that low-income groups are not paying much tax, if any at all.

With lower wage growth, average tax rates will also be lower due to less bracket creep. Lower wage growth poses a risk to revenue, with significantly lower revenue in the order of $39 billion per annum. That's only on the revenue side. You will also find that if there are lower wages there will also be lower expenditure in the budget. We have not done an analysis of what the expenditure impacts will be. For example, the age pension will be less. It might be that Public Service wages will be less. There are other areas that complicate this issue. But certainly $39 billion is a large amount of money if wages were only to grow by 2.5 percentage points. You get less income growth, which leads to less tax revenue, and you get less bracket creep going on. Bracket creep is a very significant contributor to the budget bottom line. So, there is that significant risk.

A recent updated analysis using individuals as the unit of analysis was in The Conversation. That was published, I believe, on 31 May. This came to fairly similar conclusions. In regard to tax shares, we found that the share paid by the top 10 per cent of earners drops from 58 per cent to 54.3 per cent by 2027. This is indicative of a somewhat less-progressive tax system moving forward. The upshot is that the coalition's policy only partly overcomes bracket creep. I wouldn't necessarily view the tax cuts as being genuine savings to households, in that there is the expectation that bracket creep will be returned at some point in the future. Largely what this budget tries to do is return some of that, but I would argue that it returns more to high-income families and high-income individuals than it does to middle-income families.

The proposed policy does slightly more to overcome bracket creep for higher-income individuals, but it also locks in a higher tax share for those on lower and middle incomes and a lower share of the tax burden for higher earners. On that basis, the proposals will lead to a slightly less-progressive income tax regime than the one we currently have, but it will still be a long way short of a flat tax. Pretty much everyone is set to pay more tax into the future. We looked at this in terms of each five percentage point of the tax distribution, and I think we found that for all of those, except for the top five per cent, they'd be paying a higher average rate of tax. The top five per cent would basically be unchanged. That's when you compare to their current rate of tax in 2017-18. Certainly, when you compare it to the current trajectory to 2027-28, you will see that there are significant tax cuts. It's a complicated little story. Compared to today, rates are increasing. Compared to where they're headed anyway under the current tax trajectory, there are certainly significant tax cuts. Thank you very much.

CHAIR: Thank you very much, Professor Phillips. As stated in your submission, you find that the:

… overall fiscal impact is very similar to that claimed in the Federal Budget of around $140 billion in tax cuts over the decade from 2018-19.

I think you also mentioned that in your opening statement. Is that broadly correct?

Prof. Phillips : That's right.

CHAIR: I don't have too many questions for you. Your opening statement was very, very comprehensive. Were you previously at NATSEM?

Prof. Phillips : Yes, I was.

CHAIR: Were you responsible for this modelling?

Prof. Phillips : I was not responsible. I've never heard of this analysis. My apologies.

CHAIR: All right. I'm trying to find the person responsible for the modelling of Bill Shorten's proposed abolishment of the tax free threshold and creation of three rates of tax—$50,000, 10 per cent; $100,000, 20 per cent; and anything above $100,000, 30 per cent—but I can't seem to find the person that did the modelling. They've gone to ground. I thought maybe it was you.

Prof. Phillips : Good luck.

CHAIR: Never mind.

Senator KETTER: To lay some groundwork here: the effects of this tax package change over time. Quintiles 3 and 4 stand to gain the most over stages 1 and 2, and then, by the end of the package, it's quintile 5 that's getting the most benefit out of the package—is that correct?

Prof. Phillips : That's correct, yes. That's considering it from a household perspective. The bottom half of the income distribution pays virtually no tax, the middle income pays some tax and the top end of the income distribution pays a lot of tax. We do find that the benefits, certainly in the medium term, say by 2021, are there a little bit more so that the bracket creep is being returned—I'd probably prefer to say that rather than to say the benefits of the tax cuts—more for middle-income earners than for high-income earners. Low-income earners are not really paying that much tax, so there's not that much to return. By the end of tranche 3, the benefit is more for the higher-income families and higher-income individuals than for middle-income.

Senator KETTER: You also do a specific analysis in terms of the effect of lower-than-expected wage growth?

Prof. Phillips : That's correct.

Senator KETTER: You used a figure of 2½ per cent. Where does that come from?

Prof. Phillips : I think the wage projection that the government is using at this point is 3½ per cent from the year 2020 onwards. That's probably generally considered to be optimistic, although I understand their reasons for doing it. For the out years, they tend to use projections based on the previous record of average weekly earnings, which is a lot higher than what it currently is. The 2½ per cent is just a different projection that's based on, effectively, what the current growth of average weekly earnings is. It's actually a little bit lower than 2.5 per cent at the moment; it's currently running at about two per cent. It's just a different way of thinking about where wages growth could go—based on a more realistic assumption, I would expect; based on today's wage growth.

Senator KETTER: Even 2½ per cent might be considered to be slightly optimistic, unfortunately?

Prof. Phillips : I don't think so. Not necessarily, no.

Senator KETTER: That's your prediction?

Prof. Phillips : I don't have a prediction for it. It's just a scenario, really. It's one possibility of many potential possibilities. I understand why Treasury and the government go with 3.5 per cent, because it's based on what the overall long-term trends are. By 2025, that's crystal balling that's beyond my capability, I'm afraid. By that point in time, 3.5 is as good a guess as 2.5, to be quite frank about it. Certainly, at this point in time, it looks a little optimistic, but that's not to say that I don't think we could possibly get there. It's just standard budget convention, what they're doing. I don't really have a problem with it. Also, it's nice to have some scenarios around that.

Senator KETTER: Sure. Can I get you to take us through some of your estimates as to what might happen, because you do identify that this reduces revenue by $39 billion. Can you elaborate on that for us?

Prof. Phillips : When you look at, say, 2.5 per cent wage growth compared to 3.5—obviously, that's a difference of one per cent per year. When you put that out to, say, 10 years into the future, it's roughly an 11 or 12 per cent reduction in wages overall. Already you've got 10 to 12 per cent less tax revenue from that. Then you've also got bracket creep. On top of that, you're also missing out on bracket creep, so you're looking at a considerable reduction in terms of the overall tax take. That's where we get the estimate of around $39 billion per annum.

Senator KETTER: In terms of the fact that we're not being given the year-on-year analysis by Treasury, and given that you're a modelling expert, I want your view as to what Treasury's saying, that year-on-year analysis provides for unreliable figures. You're quite proficient in this area—you've got a lot of experience—is there any excuse for Treasury withholding that year-on-year analysis?

Prof. Phillips : From one perspective, I can understand why they might wish to do that, in that forecasting into the future, even thinking about wages growth into the future, is very difficult. There are other factors that also make it difficult to predict. If you can put out a figure of $143 billion—obviously, that comes from the individual years. I personally wouldn't have any major issues with putting out individual years with the usual caveat, that there's obviously a margin of error around these sorts of forecasts. I would be happy to put my numbers out there. I see that the PBO has put numbers out. My numbers line up almost exactly the same, with a little bit of difference, in that theirs are, I think, based on cash accounting and mine are based on accrual accounting. I get $140 billion; they get $143 billion. I don't really see a big problem with doing individual years, with the caveat that of course forecasting into the future that far out is always very difficult.

An important part is that the analysis that we're doing is distributional analysis, so I'm not so concerned about whether it's exactly $140 billion or $130 billion; it's the overall shape of the distributional impact. Does this impact high income, middle income, low income? I don't think that will change dramatically as a result of whether it's 2½ per cent or 3½ per cent. Those distributional impacts are quite robust in terms of where we will be in 10 years time.

CHAIR: I have a follow-up question to that one. You did say, Professor Phillips—and this is on page 12 of your submission—that:

… caution be taken in interpreting the distributional analysis presented in this submission since it is unlikely, in reality, that our tax system would stay unchanged with no adjustment for bracket creep for the next 10 years.

Is that correct?

Prof. Phillips : That's correct. One concern I do have with the distributional modelling that we're doing, and that anyone would be doing, is that you've got to pick a counterfactual. The counterfactual we're looking at is, where will the current system be by 2027? It's unusual to go out that far. Normally we would just go over the forward estimates, and I'm comfortable with that. Going out to 2027—it's a long way out and it may well be that the tax system will change anyway. We don't know who will be in government in future years. Different governments and different leaders will have different perspectives on where the tax system should go, and there will be different economic circumstances that will take place. It's very difficult to use that as a counterfactual. From that perspective, I do think that you need to take a fair bit of caution doing analysis that far out as with the counterfactual being the current system.

Senator McALLISTER: I think you said that you use ATO data for your model. That data is very comprehensive, is it not? It speaks to all of the participants in the tax system?

Prof. Phillips : Yes.

Senator McALLISTER: It's able to be disaggregated into fine-grained assessments about where people are in the income distribution?

Prof. Phillips : It can. It has its limitations, but the model we use combines that data, which is effectively two per cent of all tax filers, which is about 250,000 individuals, with the ABS survey data, which is much richer, in that it provides not only income data but also a lot more information around households. That's how we're able to do a household analysis. The analysis of just using the tax data is based on individuals, so it only has information on one particular person; it says nothing about their spouse or their kids.

Senator McALLISTER: But you're able to link those two datasets?

Prof. Phillips : Yes.

Senator McALLISTER: The ABS data, I imagine, also has information about occupational categories that you could link to.

Prof. Phillips : It's got a lot of very detailed information—a lot more information than we can get through the ATO.

Senator McALLISTER: During estimates, Senator Cormann tabled some material. He was uncertain as to its origin, but I think he indicated that it was the basis for the material on the front page of The Australian. It discussed the impact on certain occupations. The assumptions are listed at the bottom. They chose 2018-19 wages from a sampling of enterprise agreements. Is that a reliable way to establish wage levels, given that you have these alternative datasets?

Prof. Phillips : Yes, I am aware of that article. As an example, I think there was a forklift driver who was earning, on average, $165,000 per year.

Senator McALLISTER: Correct—starting at $145,000 in the base year and projected to earn $165,000 by 2024-25.

Prof. Phillips : I would not view that as realistic. I have also looked at ATO taxation statistics. Anybody can download this information from the web. I think the average for a forklift driver was around $150,000. Their category may have also included some heavy-machinery drivers beyond forklift drivers, but you were looking at $50,000 or $60,000 per year as the more typical income. I thought it was an unusual selection of incomes.

Senator McALLISTER: On a separate question, your modelling work goes to households and to individuals. It doesn't, as far as I can tell, go to distinctions between women and men in the economy. Do you have any remarks in addition to those you provided in your written remarks about gender impacts of the tax package?

Prof. Phillips : I think my comments would be very similar to the previous speaker—that is, Jinjing Li from NATSEM. I have not done any analysis of the gender impact of the budget. You certainly can do that analysis, although there are some complications. In terms of the tax cuts, you can certainly see the gendered impact. Like Jinjing, I would expect the 'tax cuts' would go more towards high-income males, but, at the same time, they're also the ones paying a higher amount of tax. I'd need to do some more analysis to be more certain of that.

Senator McALLISTER: Your modelling has a capacity to consider other streams of income in two families, including the transfer system, essentially. It is the case that, if you were seeking to augment or support household budgets, tax relief is not the only mechanism by which relief could be applied?

Prof. Phillips : That's correct—yes.

Senator McALLISTER: Whilst you are logically correct that those who pay the most tax will benefit the most from tax cuts, it's not the case that tax cuts are the only way to deliver relief to household budgets?

Prof. Phillips : That's correct.

Senator McALLISTER: If you wanted a fairer approach to Australian women, you could target income support through other means?

Prof. Phillips : You certainly could—yes.

Senator McALLISTER: And they would include the family tax benefit arrangements, childcare support or other measures that impact directly on working women?

Prof. Phillips : There's no doubt that if you, for whatever reason, wanted to change the income distribution of Australia and tilt it more towards the low-income spectrum, tax cuts are probably not the way to go. You would have to go through the transfer system—that's correct.

Senator McALLISTER: Chair, that's fine.

Senator HANSON-YOUNG: I'm interested in the conclusion of your submission, where you talk about the risk of actually having lower revenue overall because of a lower projected jobs growth rate.

Prof. Phillips : I think it's wages growth.

Senator HANSON-YOUNG: Yes, wages growth. Can we tease that out a little bit, because one of the big concerns that many people have about this huge amount of tax cuts in the first place is there is already a lack of revenue. If we have low wages growth compounding that, what types of figures are we talking about in terms of a lack of money to pay for public services?

Prof. Phillips : I would be very surprised, as I said before—with that counterfactual—if we went out to 2027 there wouldn't be some form of tax cuts regardless of who was in charge. I don't necessarily view them as tax cuts in a sense. They're really just returning bracket creep.

Senator HANSON-YOUNG: Even at the higher end?

Prof. Phillips : The analysis that we did did show that there was more support for the higher end compared to the middle element of the income distribution. This is a problem that the budget has had for many years in that we've been projecting that wages growth would increase to 3½ per cent and it hasn't done so at this point. That's part of the reason why you continue to get large deficits year in and year out. We see four years into the future we'll get a tremendous amount of revenue coming in but it never actually appears. That's partly because wages growth has not actually materialised as we would hope. I've got no idea whether it will materialise in the coming years or not.

Senator HANSON-YOUNG: But isn't it a bit of a fallacy to suggest that we should be giving tax cuts to compensate for low wages growth, as opposed to finding ways to actually lift wages, such as a boost in the minimum wage?

Prof. Phillips : These are not things that I've looked at in terms of my submission. I can't comment at this point on other ways that we could potentially boost wages. It's still worth remembering that even with the 'tax cuts'—and I keep saying that in inverted commas—we're still looking at higher average personal income tax rates into the future whether you're on a low, middle or high income. It's really only the top income group that perhaps isn't looking at higher tax rates. They're looking at about the same tax rates. There is a bit of confusion I think in the debate around what you might call nominal tax savings as opposed to real tax savings. I expect that there would need to be tax cuts in the future. You would naturally expect that would go to middle- and high-income earners because they're the ones who do pay tax.

Senator HANSON-YOUNG: Which does very little then for tackling inequality.

Prof. Phillips : Our analysis would show, as I think I said in my opening statement, in The Conversation piece that the tax share of the top 10 per cent of earners drops from 58 per cent to 54.3 per cent. A lot of that has to do with the continuation of bracket creep. Some of it has to do with the policy change. But, yes, you're right: if we let the current system go as it is at the moment—if these tax cuts weren't to pass—you would find that that share of high-income earners would drop because of bracket creep. And even with these tax cuts you will still see that that tax share will be reducing at an even greater rate. So, yes, the progressivity of the system is reduced as a result of (a) bracket creep and (b) the tax cuts that have been proposed.

Senator HANSON-YOUNG: In terms of what can be done for low-income earners and tackling inequality, which all sides of politics acknowledge and want to do something about, you're suggesting that these tax cuts as they've been presented will not really tackle inequality?

Prof. Phillips : It will not tackle income inequality, no. It will not improve income inequality. If anything, it will make it ever so slightly worse. I would caution that it is ever so slightly; it's not a big change.

Senator HANSON-YOUNG: What about the argument that giving a tax cut to high-income earners will mean they get to pay off their house faster or have an overseas holiday and it's not necessarily money that's going to be going into the economy?

Prof. Phillips : Most of the literature would suggest that middle-income earners and low-income earners are more likely to spend a dollar that you give them or a tax cut that you provide to them. So, in terms of the economic impact, it tends to be middle- and low-income earners that tend to give a bigger bang for your buck.

Senator HANSON-YOUNG: Lifting wages growth would obviously help with that too.

Prof. Phillips : Of course, yes.

Senator HANSON-YOUNG: Thank you.

Senator STORER: Professor Phillips, in your conclusion you note that it's unusual for budgets to legislate tax changes that start beyond the forward estimates. You said that where this happens it should be expected that tax cuts would be applied to temper bracket creep. If that were to occur, what would that look like?

Prof. Phillips : The point I was making is that, certainly historically, when we look at tax packages, the tax changes start over the forward estimates. That is usually in one or two years time. It is unusual to have large-scale tax changes that are six, seven or eight years into the future. Sorry, what was the next part of your question?

Senator STORER: You said that a natural consequence of a progressive tax system is that such tax cuts tend to be higher income households.

Prof. Phillips : That's largely because you tend to find that high-income earners and middle-income earners are that ones who pay a large share of tax. If you are returning bracket creep, if that is what the goal is, then you would certainly expect that the nominal dollars will basically flow to middle- and, in particular, high- income earners.

Senator STORER: You say the targeting of the tax cuts more strongly favours higher income households over lower income households, where the cuts are the most modest. But you note average tax rates are still projected to increase over the next decade.

Prof. Phillips : That's right. The strongest force that we have going on over the next 10 years or so, if wages growth is 3½ per cent, is going to be bracket creep. The proposed tax cuts do unwind that to some extent, but not to the full extent. So we are still seeing increases in taxes; it is just that it has been tilted towards the high-income households. The proposed tax cuts—the increase of the threshold from $180,000 to $200,000, the increase from $120,000 which effectively becomes $200,000—benefits high-income earners more than low-income earners. The work of dragging back bracket creep mostly benefits high-income earners and not so much the middle-income earners. It certainly doesn't benefit low-income earners. But they are not paying much tax, so you would need to go down the path of the Social Security system to improve their disposable income.

Senator STORER: You note that budgets are always subject to future risk, one of which is wages growth. If we hit wage growth of 2.5 per cent rather than the 3.5 per cent in the budget projections, it would lower income tax rates and projected revenue by $39 billion per year by 2027-28. If that was seen to be happening in the forward estimates, do you think it would be prudent to undertake the stage 2 and stage 3 changes?

Prof. Phillips : I don't really have a view on whether it is prudent to undertake the stage 2 and stage 3 tax cuts.

CHAIR: Professor Phillips, thank you very much for joining the committee today.