When Ignorance Isn’t Bliss

My oped today is on the role of information in tax reform. Full text over the fold.


Make Taxpayers Literate, Australian Financial Review, 21 April 2009

Would the rich pay more tax if we replaced the current income tax system with a flat tax? What about if we abolished the income tax and raised the GST?

If you answered ‘no’ to both these questions, then you’re in line with what virtually all tax experts – left and right – believe is true in developed countries. But when researcher Joel Slemrod put these questions to a sample of the US population, 41 percent said ‘yes’. They thought that under a flat income or consumption tax, the rich would pay more tax. Assuming the experts are correct, it looks like 4 out of 10 Americans fundamentally misunderstand their tax system.

What is startling about this is that most tax analysis generally begins from the premise that people have a perfect understanding of the system. For example, a major policy focus in Australia over recent years has been on ensuring that the tax and benefit system does not overly penalise the move from welfare into work. Implicitly, most of those who argue for lowering ‘effective marginal tax rates’ assume that people are well-informed about the system, and can calculate how getting a job or changing their hours will affect their tax bill and welfare cheque.

In many contexts, the economic assumption of ‘perfect information’ is a reasonable approximation of the real world. But when it comes to complex systems like taxes and family benefits, it becomes increasingly tenuous to claim that the typical person knows the system.

Lately, economists working in the field of ‘behavioural public finance’ have begun to implement a series of randomised experiments to test the extent to which salient information matters. One of the leaders of this field is Raj Chetty, a lanky whiz kid who at the ripe age of 29 is a full professor at Harvard University.

With co-author Emmanuel Saez, Chetty ran an experiment in which tax agents were randomly required to inform low-income workers about the parameters of the Earned Income Tax Credit (EITC), an in-work benefit for poor families. Tax agents who complied with the program induced a substantial increase in earnings among their clients. Indeed, Chetty and Saez estimate that a small amount of additional information (a two-minute tutorial and a follow-up letter) led to a big increase in earnings. ‘Teaching the tax code’ had the same impact as a one-third increase in the generosity of the credit.

The results of this experiment fly in the face of conventional economic wisdom. Under the US EITC, poor families can be eligible for thousands of dollars annually. For economists, observing taxpayers who are ignorant about the basic parameters of the system is like seeing pedestrians stepping over a pile of $20 notes strewn on the footpath.

In another exercise, Chetty and his co-authors Kory Kroft and Adam Looney showed that it is not only information about taxes that matters, but also whether that information is ‘in your face’. For this experiment, they exploited the fact that product prices in the US do not include sales taxes. Working with a grocery store, they posted tax-inclusive prices on a series of randomly selected products, and watched to see how it affected consumer behaviour. When surveyed, consumers typically knew that tax would be applied at the checkout – yet posting a tax-inclusive price on the shelf still reduced demand by 8 percent. (Subsequent Harvard research backs up the finding that sales taxes are especially salient: incentives to buy a hybrid car are seven times more effective if delivered through sales tax waivers than via income tax credits.)

Although I know of no Australian surveys that have comprehensively measured ‘tax literacy’, there are reasons to think that Chetty’s findings might apply here. The expansion of family tax benefits and child care benefits to the middle class – and the sheer frequency with which these policies change – make it pretty improbable that everyone is up to date with the latest policy. And Australians are world-leaders in our use of tax agents. When three-quarters of individual taxpayers require professional help to lodge their return, it’s surely a hint that the system has grown too complex.

Speaking to the National Press Club last November, Treasury Secretary Ken Henry told the tale of his conversation with ‘Jim from Jericho’, and emphasised the wisdom of ‘bar room conversations of practical people’. But as well as listening to what voters already know, tax reformers should focus on issues of information and salience. Indeed, the way a policy is designed and understood might be as important as its price tag.

Andrew Leigh is an economist in the Research School of Social Sciences at the Australian National University.

About these ads
This entry was posted in Behavioural Economics, Tax. Bookmark the permalink.

6 Responses to When Ignorance Isn’t Bliss

  1. Kevin Cox says:

    There is no reason that a tax system given the information about my circumstances cannot automatically determine and show my options and suggest an appropriate action.

    Rather than trying to reform the tax system first why not get the current system to operate with little effort to the individual. The tax office probably has the makings of such a system for its internal use and the electronic tax return system goes a ways towards this objective.

    Let me have a way of setting up my personal profile. When something changes then the profile is changed.

    If the tax system changes then I do not have to worry about the rules because the tax system will tell me the effect on the changes on me.

    So rather than the tax system requiring me to understand its rules the tax system asks me to provide information and it works out the tax.

    We can set up systems to automatically supply our transaction data.

    So, when I get paid the amount of my pay automatically is sent to the tax office. When I pay an allowable tax deduction the amount is automatically sent to the tax office.

    My personal information about me is not kept with the tax office but is kept under my control. When the tax office wants to calculate my tax then it only gets the data needed at the time from me automatically does its calculations and throws away the personal information. My personal information is kept by me and there are ways it can be automatically updated when my circumstances change (I have a child, get divorced, get married etc.)

    If we have such a system it then becomes much easier to simplify the tax rules and to “experiment” with variations.

    So my suggestion to the tax office is to make the existing system work more efficiently for the end user so that the introduction of new rules will be simpler and easier.

  2. Paul Abbott says:

    You ask:

    “Would the rich pay more tax if we replaced the current income tax system with a flat tax? What about if we abolished the income tax and raised the GST?”

    The answer is actually yes — because the rich (the really rich, not just the well off) can structure their affairs, through companies or trusts, so as to avoid paying tax now.

    “What is startling about this is that most tax analysis generally begins from the premise that people have a perfect understanding of the system.”

    This is clearly a silly premise!

    “Implicitly, most of those who argue for lowering ‘effective marginal tax rates’ assume that people are well-informed about the system, and can calculate how getting a job or changing their hours will affect their tax bill and welfare cheque.”

    Surely an online tax/welfare/salary calculator would be easy to make?

    “In many contexts, the economic assumption of ‘perfect information’ is a reasonable approximation of the real world. But when it comes to complex systems like taxes and family benefits, it becomes increasingly tenuous to claim that the typical person knows the system.”

    “And Australians are world-leaders in our use of tax agents. When three-quarters of individual taxpayers require professional help to lodge their return, it’s surely a hint that the system has grown too complex.”

    Strongly agree. However, I think this point is well addressed by Kevin Cox’s excellent suggestion.

    Finally, you do not mention salary packaging, which distorts the current tax system. In particular, academics, doctors, and hospital staff can arrange their affairs to greatly reduce their tax burden. I see this as unfair in that most PAYG workers do not have the same access to this tax reduction measure.

  3. derrida derider says:

    Paul, experts already take into account the ability of “the rich” to evade tax when comparing different tax systems – in fact it has been the focus of a lot of empiric research for that reason. “The rich” would still pay much less tax under a flat tax (direct or indirect) than the present system.

    Apart from the gaping hole of superannuation for over 60s (only available since 2006), there is far less benefit to salary packaging than most people think. If you travel lots of kilometres you can get a tax benefit through packaging a car. But the major users of salary packaging these days are workers in charities (which are exempt from Fringe Benefit Tax), and charities are hardly infested with high income employees.

  4. Sinclair Davidson says:

    Would the rich pay more tax if we replaced the current income tax system with a flat tax?

    Without specific details of the proposed flat tax (and then an understanding of the subsequent dynamic response) I would have thought the answer is ‘It depends’.

  5. Marcin says:

    I think you’re ignoring how knowledge about the effect of the tax and welfare system are generated and spread.

    When a person makes a move e.g. into employment out of benefit receipt, even if they did not anticipate the loss of benefits, if they don’t actively try to avoid it, they will probably end up revealing what has happened, and the state will move appropriately to adjust benefits. Thus, the person knows what happens, and complains (or crows) to their society (or “social network” in the current jargon). Thus, information about the most likely welfare-affecting events becomes disseminated.

  6. Pingback: Links: April 2009 « Consider the Evidence

Comments are closed.