Note: Where available, the PDF/Word icon below is provided to view the complete and fully formatted document
Standing Committee on Tax and Revenue
17/07/2017
Taxpayer engagement with the tax system

BRADON, Mr Edward, Senior Advisor, The Behavioural Insights Team (Australia)

GALLAGHER, Dr Rory, Managing Director, Asia-Pacific, The Behavioural Insights Team (Australia)

CHAIR: Welcome. We thank you for your attendance to assist the committee with this inquiry. Although the committee does not require you to give evidence under oath, these hearings are formal proceedings of the parliament and warrant the same respect as proceedings of the House of Representatives. As such, the giving of false or misleading evidence is a serious matter and may be regarded as a contempt of parliament. It is unlawful for anyone to threaten or disadvantage a witness on account of evidence given to a committee, and such action may be treated by the House of Representatives as a contempt. The evidence given today will be recorded by Hansard and attracts parliamentary privilege. We will obviously proceed to questions, but I invite one or both of you to give an opening statement or remarks.

Dr Gallagher : Thank you very much for the invitation to speak to the committee today in its inquiry into taxpayer engagement on behalf of the Behavioural Insights Team, or the BIT. The BIT started its life inside 10 Downing Street as the world's first government institution dedicated to the application of behavioural sciences to improve the design and delivery of government policy. In 2014, we spun out of government and are now a social purpose company co-owned by the UK government, its employees and Nesta, which is a social innovation charity. That has enabled us to partner with more governments and organisations around the world. We now have six offices in London, Manchester, New York, Singapore, Wellington and here in Sydney.

In 2012, we began a partnership with the NSW Department of Premier and Cabinet, and have been working with them since then to establish their own behavioural insights unit and to run a series of trials, including in the payment of fines. We now have a team of 12 here in Sydney and have successfully partnered with a number of Australian and state government agencies, local councils and non-government organisations to build their own capacity in the use of behavioural insights and evaluation methodologies.

At the core of our approach is a more realistic model of human behaviour and a more empirical- and evidence-based approach to government. I know you have already heard from our colleagues at BETA about the use of these tools in government more broadly, so I won't go into further detail about that approach for now. In relation to tax and revenue specifically, we have demonstrated that the application of behavioural insights can help provide a more nuanced understanding of taxpayer behaviours and enhanced regional approaches focused on deterrents, enforcement and voluntary compliance, often at low cost. We have proof that it is feasible to run robust trials, including randomised control trials, in tax authorities to find out what actually works in practice.

Many traditional approaches to tax compliance emphasise deterrence, where the goal is to increase the costs and reduce the benefits of evading tax. This has largely proved effective. Indeed, it is important to note that the vast majority of people in companies do pay their tax without need for significant sanctions or interventions. However, Behavioural Insights challenges models that assume people will always rationally weigh up the costs, benefits and risks of whether to pay tax on not. In contrast, Behavioural Insights emphasise that people are heavily influenced by the context and environment in which that decision is made.

Behavioural Insights Team's work on tax compliance started shortly after we were established in 2010. It originally centred on recovering tax debt, but more recently we have trialled interventions that help people avoid being late with their tax. We have also shown that behavioural insights can influence corporations and organisations, not just individuals, and that it can be adapted for new and emerging channels such as SMS or text messages. Finally, we have now been developing a more nuanced understanding of what influences tax compliance and how this might vary across groups and individuals—for example, how different messages may work for first-time versus repeated debtors or for different income groups. We have now run tax trials in a range of countries and places, from local councils in the UK to Costa Rica, Guatemala and here in New South Wales.

As you have already heard from our colleagues at BETA about some of our key trial results, we won't go into detail in this opening statement, but we have demonstrated that people can be encouraged to pay tax and fines by making the process easy, by highlighting key messages, by prompting honesty at key moments and by emphasising what most people are already doing—the social norm that most people do pay their taxes on time. We believe that behavioural insights can be used to sharpen both traditional enforcement and compliance tools. Indeed, our EAST framework for understanding how to influence behaviour by making it easy, attractive, social and timely applies equally well to tax and revenue as it does to any of the other policy areas that we work in.

In summary, the use of behavioural insights in tax compliance is becoming increasingly accepted and mainstream. Indeed, the UK's own tax department, the HMRC, and the ATO now have dedicated BI units. Both of those departments have pioneered the use of BI in this field. The Behavioural Insights Team is delighted to provide input and evidence to this inquiry and look forward to the questions and discussion this morning.

CHAIR: Thank you. As you said, we spoke to some of your colleagues earlier. I would like to go into some specifics. You talked about it being easy, attractive, timely et cetera. From the observations that you have made, given the experts that you are in this field, what is something that is being done—either here, in the UK or elsewhere—that is the best, most effective thing that you have seen in getting more compliance in the area we are talking about?

Dr Gallagher : Many of our biggest wins have simply been in simplifying the process and communications to customers. Many of our earliest wins and biggest results and effect sizes were from simplifying multipage letters and very dense and legal language into more-simple language that people can understand and respond to. We managed to then build on that by drawing on different parts of behavioural evidence. I mentioned the use of social norms. One of our earliest wins in this space was to highlight that most people do pay their tax on time. Nine out of 10 people do pay their tax on time. It was proved to be more effective than our traditional, more enforcement-focused methods. As we have repeated and replicated these trials over time, we have begun to be able to trial different variations of that message. We now know that for most people, for example, the message that nine out of 10 people pay their tax on time is more effective if you can localise it. Rather than saying nine out of 10 people in Australia pay their tax on time, it would say nine out of 10 people in Sydney—or even Western Sydney—pay their tax on time. Maybe to pre-empt a question later on, of course those statistics need to be correct. You need to make sure you do them and tailor them to the context. But the localisation and personalisation of the message makes it even more effective. For most people, then highlighting that they are part of a small minority who have not paid can again give you some additional impact. We have seen across the piece that, if you can simplify and personalise messages, that can increase compliance significantly, and reminding people what most other people are doing is certainly one of the other most effective tools that we have seen.

CHAIR: In looking at this whole inquiry we have divided it into two bits. There are people who are in legal professions providing legal services and goods but who may be involved in the cash economy, if you like, and not necessarily behaving in a way that is declaring all their income to pay tax. Then separately, but as part of this inquiry as well, we have people who are engaged in illegal activity. As a behavioural insight person, would you have anything to offer on what we could do in that space? Obviously making it simpler might not work with them.

Dr Gallagher : Sure. We have been involved in a variety of work that looks at things like corruption as well as priming and prompting honesty. In terms of the black economy, we have provided some advice and support in some trials on trying to encourage different segments of the economy to comply. For example, we did some work around encouraging plumbers and electricians to comply through focused disclosure periods and focused, targeted campaigns, as well as doctors and dentists. Again, we think it is important that you have these targeted campaigns throughout the income bracket, where you do have information and data that suggests that there is unreported income. Indeed, in those areas, again, having a time-limited window and focus, and highlighting, for example, that you do hold third-party information which leads you to suspect that they are not fully declaring can be very important. I am sure data analytics will increase in future, where you will be able to target that and be able to pick up where people are outside the norm relatively more easily. In terms of the determined, illegal activity—more deliberate—we have done less work in that space. I am not sure whether Ed wants to talk a little bit about that, but there certainly are behavioural insights you can use in that area.

Mr Bradon : An approach we have frequently used for more deliberate noncompliance, illegal activities, is to think about not just the individual who is themselves engaged in that activity but also those connected to them and around them, particularly where you have whistleblowing or the opportunity to inform authorities to aid enforcement activities. Looking at those processes in detail, is it extremely easy to do that? Do you need lots of extra information? Do you feel like you are putting yourself at risk if you are reporting this activity? Often when you look at those processes you find it is a very cumbersome and difficult path to go through. There are people who could report it but don't, often for quite prosaic reasons. So it is looking at that full network and seeing what the opportunities are there, especially if you've got individuals who are determined not to comply. It is also the network where those opportunities exist.

CHAIR: What do you think we could do in that space to make it easier or more comfortable for them to be a whistleblower, if you like?

Mr Bradon : I guess there are a couple of thoughts there, one in terms of proactive whistleblowing. We find that small friction costs, even tiny friction costs, as much as a click on a website, can have a meaningful effect on behaviour—so reviewing step by step what it feels like and looks like to be a whistleblower, if it easy at every stage. And even with small frictions you could remove from that, we would expect to see changes. Another approach which we thought of in a different context is trying to change their default. This is a common behavioural insights idea. At the moment the default would be: if you don't report, that's it and you never ask the question. If you have interactions with individuals, perhaps when they are filling in their tax returns, there may be opportunities to proactively ask a question, 'Have you seen behaviours of this type?' which may undermine the tendency of 'don't ask, don't tell' in terms of some of these behaviours.

Dr Gallagher : That approach is based on some good behavioural evidence about people's propensity to lie by omission via commission. So people find it easier not to tell the truth if you don't ask them directly. They can forget to report, for example. We have run trials in that space to highlight, for example, continuing to claim a single person discount—that continues for many years unless you contact the government. People find it easier, if those circumstances change, not to report, whereas if you encourage them or have a default where they are forced to reaffirm that they are still living on their own or that they still have that amount of income every year, then they are much more likely to tell the truth.

Ms BUTLER: One of the ongoing issues we have here in Australia is the interaction between the Child Support Agency and the tax office. Your child support obligations are assessed by reference to your tax return. A lot of payers just do not file a tax return. The tax office does not chase them because they are not payers who are likely to have a debt to the revenue even though they are likely to have a debt to an individual child support payee. Do you think that if we were able to find some convenient way of asking people in the child support system whether they have filed their tax return that might encourage compliance in a fairly cheap way? Is there any analogous to that in the UK?

Dr Gallagher : It is not a system that we have been directly involved with, but I would think that a simple and timely moment to ask that question would bring people into the system. As Ed has highlighted, there will be a trade-off: the more you are asking people to input and fill, the more likely it is that people will get out of it. So, yes, I think that would be the case but it would need to be a simple question for people to answer at a timely moment.

Mr LEESER: Could you talk about how your randomised control trials might work in the tax space? Give us a few examples of where you have used it and how that works.

Dr Gallagher : I refer to the trials I mentioned earlier around social norms. Effectively, how that would work is that we would analyse the evidence, look at what we think might be the most effective letters based on that evidence and randomly select a group of taxpayers to receive different types of letters. For example, people would receive what is called a control letter. That would be business as usual. They would receive the letter that is usually sent out that year. You then might have two variations. One of them would be a behavioural letter, a simplified letter that highlighted the key actions up-front. People would be randomly selected into that group. And then there would be a third group where there was a social norm at the top—for argument's sake, 'nine out of 10 people in the UK pay their tax on time'. Through that process of randomisation you will then have a spread of people across all groups, who equalise themselves out. So you would know that if people respond more to the treatment letters, that is down to the content of that letter rather than differences within the populations themselves. Indeed, when we put out our first paper in this space, in 2012, there were some concerns about whether the public would accept that: is it okay to be 'experimenting' on the population? Actually, we found that the public were supportive. If you view this more as continuous improvement then of course the tax office would send out millions of letters, so why wouldn't we try to find out the most effective way to do that? That was explained to the public. We were transparent about what we were doing and we received a lot of support.

Mr LEESER: Four or five years ago the government started issuing a receipt with people's tax returns in Australia to show how much of your tax was going to which agencies and so on. I suspect that part of that was about the government trying to start a discussion about government spending more broadly. One of the things it shows is the large slab of the budget that goes on welfare spending—in particular, the age pension. To your knowledge, has that receipt had any effect on taxpayer compliance more broadly? What do you view that receipt as doing in a behavioural sense?

Dr Gallagher : I am not aware of any evidence that has robustly trialled that. It has been rolled out. But because there was no randomised controlled trial there has not been the ability to look at what that does for compliance over time.

Mr LEESER: Does anyone else around the world do something like that?

Mr Bradon : We have done some trials not trying to present the full list of information to taxpayers but instead the social norm message about nine out of 10 taxpayers. We have found that reminding recipients of the letter that their taxes pay for schools, hospitals and other services can be effective in improving compliance. That has been the case in particular for taxpayers with large debts—30,000 pounds or more. We hypothesise that that is because those numbers are large enough that you can plausibly see the link between that money and specific services. In that case, we found that including information about what the tax is being spent on is effective.

Mr LEESER: How have you communicated that information in that instance?

Dr Gallagher : Again, that was through a randomised controlled trial. People received that information in their standard tax letter. That makes a really good point in terms of segmentation. As we mentioned earlier, telling them that nine out of 10 people pay their tax on time and you are part of a small minority works for the majority of taxpayers but not for the very wealthy or those with the highest tax debts. But the message that your tax goes to pay for schools, roads and hospitals was much more effective for that group on those sorts of salaries than for the group with smaller tax debts and low incomes.

Mr Bradon : That was probably the most general version of that statement we could make. In a lot of other contexts, we found it was more effective to be specific and use dollar values. In the trial that we did, we did not have that level of data available to be 100 per cent confident that what we were saying would be accurate. So we used that general message. But I think my hypothesis would be that if you have that more detailed data available it could be personalised and thereby made even more effective.

Mr LEESER: You mentioned before that perhaps nine out of 10 people in Sydney, or Western Sydney, comply. Would you not do it by postcode? There may be some postcodes where there is horrendously poor compliance? Or is it too difficult to get the data on that basis?

Mr Bradon : Generally speaking, the more localised the area the better. But it is important for it to be an area that you yourself identify with and understand. A local town might be a slightly larger geographic area than their specific postcode. If they have a stronger identification with their town than they do with their postcode, that can mean the larger area is a better choice for that reference group. It depends on the data. If we have data almost down to the street level, that can potentially be very effective.

Ms FLINT: I think you mentioned that this started in the UK in 2010.

Dr Gallagher : That's correct.

Ms FLINT: Which agencies and which areas have had the most success in the UK? Have you been able to identify significant savings to government? Have you seen a reduction in the size of some areas of enforcement or agencies themselves? And what is the quantum of savings that you have achieved?

Dr Gallagher : In the UK, HMRC is a good example of how this approach has now become embedded. They have their own dedicated unit which is bigger than the size of the original team inside No. 10. I think they have 20 staff now. That is because they have seen good return on investment. Indeed, we have seen hundreds of millions of pounds brought forward. Many of these interventions are very low cost; it is the effect of changing a letter. That is not absolutely without cost, but it is pretty cheap. Or it might be using other types of channels such as text messages; again, that is at very low cost. The return on investment has been very high, particularly in the tax space. We have seen it grow throughout government in the UK. Most departments now have a dedicated unit. We have seen some big wins in things like the employment space. We ran lots of trials with the Department of Work and Pensions to help people find work more quickly by changing the nature of the conversations from being overly focused on compliance—around what someone had done in the previous fortnight—to being focused on what they were going to do in the next fortnight. There is good behavioural evidence that that encourages people and makes it much more likely that they will follow through on their good intentions. We asked them to be very specific about what they will do and when—'On Tuesday morning, after walking the dog, I will look for three retail jobs.'. We have seen a lot of gains in that space. More recently, we have done lots of work with the health department in trying to reduce obesity. In the education space, it has been used to encourage kids to turn up to school to achieve more in their maths and literacy cases. So it has pretty much become embedded and mainstream across the UK government. And it is fair to say that that process is now underway in Australia. The ATO has its own unit. We have been working very closely with the New South Wales government for the last four or five years. They have a 10-person team now. We apply it across different policy areas ranging from the payment of fines through to things like tackling domestic violence.

Ms FLINT: And what was the process of that rollout across government? Was it an organic thing because such good results were achieved in one agency that they then shared those findings and processes? How did it unfold?

Dr Gallagher : It was a little bit of both in both the UK and in New South Wales. We identified areas where we thought we could have an impact and measure results relatively quickly. We started in areas like tax and fines, in both the UK and in New South Wales. That is an area where you can get results relatively quickly and improve the return on investment. We went through a process of trying to identify across government where these approaches could be used and where we thought they would have the most impact. But when departments approached us we were also open to working with them. So the model is both proactive and reactive in both those places.

Ms FLINT: You said you started in Downing Street. So it was driven from the top down? This initiative was led by the Prime Minister?

Dr Gallagher : It was in the UK, yes. David Cameron and his political advisers helped set up the unit. In New South Wales, it was a less political appointment. Certainly, the Premier at the time supported that. We were brought in by the secretary of the New South Wales government to start it over here as well.

CHAIR: You have mentioned things we could do to make the tax system simpler to adhere to. We set company tax rates, personal tax rates et cetera. How can we make people adhere to those rules? Do you look at the types of things governments should do to start with that would make people adhere to it? Forget about making it easier; would it be right to assume that if tax rates were lower, more people would adhere to it? Is that a true general statement? If we had bigger gaps as people go up the scale, do you think people would not adhere to tax rules because they are going to a much higher level that they do not like? For personal income tax, there are trusts and other entities you can form. Do you look at that?

Dr Gallagher : Yes, we do. We have not worked with the tax agency here or in the UK to do a systematic review of the tax system. But there is fairly good evidence that where you make a system easier to comply with and simpler, that is the most effective method you have for increasing compliance—rather than trying to communicate in as simple as possible—

CHAIR: I suppose what I am saying is that it can be easier and simpler to comply at a 75 per cent tax rate than at a 35 per cent tax rate. Do you know what I am saying?

Mr Bradon : This is not a specific issue that we have studied, but the general evidence is that people do obviously pay attention to financial incentives and a very large difference can make big differences to behaviour. Especially at the margin, the changes you get from the design of the system are much larger. One example we have worked on in the UK is around pensions contributions. There are generous incentives to save towards your pension in the UK. But there is some evidence suggesting that the amount you have to be offered in terms of tax incentives is very small relative to the amount of extra savings you get; it might be as little as six pence per extra pound of extra saving at the extreme end. The policy change of switching to a more Australian style system of defaulting most individuals into workplace pensions has had a huge effect on the proportion who are enrolled. If you were to try and achieve that same difference of behaviour through purely financial incentives, it would be very expensive. Our approach is not to say financial incentives do not matter. But, especially in terms of return on investment, the design of the system—opt in, opt out and whether it is easy—matters a lot as well.

Ms BUTLER: Do you find that explaining to people how much things cost encourages them to make a contribution to them? For example, if you tell people that it costs $50 million to build one kilometre of highway or rail link, do people appreciate the value for money they are getting by making a contribution to the public purse?

Dr Gallagher : There is a fairly good evidence on things like operational transparency. When people recognise not just the money but the effort that goes into something, they are then more supportive of those measures. Indeed, we have run some trials here in New South Wales and in the UK to highlight that, for example, if you highlight the cost of a missed hospital appointment, people are more likely to turn up. Actually trying to make it very real, highlighting the cost to the system, will make it much more likely that people turn up and comply with the requests.

Ms BUTLER: You mentioned, Chair—or it might have been you, Dr Gallagher—that we have in the Australian Taxation Office a sort of nudge unit. A Second Commissioner, Andrew Mills, is responsible for behavioural economics and behavioural insights in the Australian Taxation Office. Are you working with the Australian Taxation Office on how they might develop their own behavioural insights work?

Dr Gallagher : No, not at the moment.

Ms BUTLER: Have you had any chance to have a look at what they are doing?

Dr Gallagher : I haven't. Quite often our teams will talk at international conferences and share intel at those sorts of forums. I haven't recently. When I first came over, I presented at the ATO and had some conversations. In recent months and years, I haven't had any detailed conversations.

Ms BUTLER: You guys, the Brits, kind of invented nudge units—right? That's what you were saying just before.

Dr Gallagher : Yes. We were the first unit that was set up, and indeed one of the reasons that I came over and the team established here was to try and help build units and help spread this capability across different contexts but also to learn for ourselves what works in different contexts. Part of the reason, for example, we have set up in Singapore is to understand how different interventions may play out differently in those different cultural contexts. Indeed, we ran one trial there that highlighted, for example, that, when you are requesting tax payments and people are late, actually using a pink letter is very effective because, in that sort of cultural context, that signals late payment. That would probably be totally ineffective or counterproductive over here if you suddenly received a pink letter through the letterbox.

Some of these things are culturally specific, and part of our reach now across the world is to try and understand what works in different contexts. Fascinatingly, we are seeing pretty similar effects whether that is in the UK, Australia, Costa Rica or Guatemala. Some of these effects are highlighting what most people do, highlighting that, if, for example, you don't comply and you've already been given a chance, that is a deliberate choice, can be very effective; highlighting, for example, that you are monitoring someone: 'We have information on you and we will be monitoring what you do,' seems to be very effective. Many of these things seem to operate despite vast differences in cultural context and the way that those tax systems are set up, which is something that we are really interested in and want to find out more about.

Ms BUTLER: That is fascinating, because I would have thought that our convict history might have made us more anti-authoritarian and less willing to voluntarily comply with obligations here in Australia.

Dr Gallagher : One of the questions that I was often given was: will these interventions work in Australia in the same way that they did in the UK? That was really interesting for us to test. Many of these messages seem to be effective in both contexts but, if you ask people, they won't always necessarily understand why. We are not always very good at what predicts our own behaviour. If you ask someone—and we have done—'Do you think that receiving a letter saying, "Nine out of 10 people pay their tax on time," will influence your behaviour?' most people will say no. It is only through running the sorts of randomised control trials that we have talked about that we have actually been able to find the evidence for what did work. Definitely when I turned up and said, 'Hey, we've got these trials based on social norms; do you think it'll work?' Often there was a sense that it wouldn't, whereas we've seen evidence to the contrary.

Ms BUTLER: How about a letter that indicates that a large proportion of your personal income tax goes into paying what is described in government as welfare, which isn't necessarily disaggregated into the age pension? Do you think that would have an impact on people's propensity to pay their tax?

Dr Gallagher : The example that Ed drew on earlier, I think, is the pertinent one. It is much easier to engage when you think about the specific things that that might be paying for. It is easier to conceptualise schools, roads, hospitals, nurses, teachers than it is a sort of amorphous overall government expenditure. Trying to make those things real for people is probably a more effective mechanism.

Ms BUTLER: You talked about making things easy for people and attractive as well. We have had the opposite here, of course. There have been some pretty well-publicised failures of the ATO's website right at the time when people are trying to do their taxes. In fact, I think there was a particularly spectacularly badly timed website failure. There was a five-hour website outage on the same day as the tax commissioner was giving a flagship speech to the Press Club in Canberra. Presumably making it harder to comply would have the opposite effect to the effect that you are driving for?

Dr Gallagher : That is exactly right. In those sorts of questions, we know around the world that, when you have a deadline day, you get a huge spike in activity. One of the things we have looked at is whether you can bring some of those things forward, bring some of those payments forward—again, trying to think creatively about how you target different groups or even different mechanisms. One example that has been used overseas is tax lotteries. For example, if you pay your tax early, on time and online, you are entered into a lottery to win an amount of money or an experience. We have seen that in Australia with a local council where if people paid early, in full and on time, they were entered into a prize draw to win a local experience. That is actually very effective at driving online payments and helping to smooth some of those peaks you otherwise get on deadline day.

Ms BUTLER: Imagine if we did that and the tax office website was down. We would end up having to pay everyone compensation for the loss of the chance to win the prize—wouldn't we?

Dr Gallagher : The thing is you're bringing them forward, so that master would be three months in advance. If you manage to get such a big group forward that it crashes the website, I think that would be a spectacular success.

Ms BUTLER: They seem to manage to crash regardless of what's going on. It seems to be easy for it to crash.

Dr Gallagher : That broader point about how to smooth the peak when people are applying and trying to use different types of rewards and incentives is one that is definitely worth looking into.

CHAIR: Thank you for your attendance. You will be sent a copy of the transcript of your evidence and will have an opportunity to request corrections to any errors. If you have been asked to provide additional material or if there is any further information that you would like to provide, please forward this to the secretariat by Monday, 31 July. Additionally, the committee may provide you with further written questions on notice during the course of its inquiry and your responses will be taken as additional evidence. Your timely assistance with this would be greatly appreciated.