One quirk of the HELP repayment system is that, on reaching each repayment threshold, the debtor pays a % of their entire income. England and New Zealand followed Australia in creating income contingent student loans. But their repayment systems are based on a % of their income above the threshold – they have marginal rate systems. The Australian income tax system also uses marginal rates.

The Universities Accord final report HELP repayment recommendations include ‘moving to arrangements based on marginal income’.

As explained below, this Accord change would reduce ‘effective marginal tax rates’ – the loss of disposable income on each dollar above the threshold. Under the current system, EMTRs can exceed 100% – so that earning an extra dollar reduces rather than increases a HELP debtor’s annual disposable income. The Accord final report calls this an ‘unfair situation’.

The Accord final report does not specify a marginal rate – an issue I discuss in another post. England and New Zealand have marginal loan repayment rates of 9% and 12% respectively above their repayment thresholds.

The problem with high effective marginal tax rates

For 2023-24 the first income threshold for repaying HELP debt is $51,550. Someone whose HELP repayment income – the same as their taxable income in most cases – is exactly $51,549 repays $0. But someone who earns $51,550 pays 1% of that amount, $515.50.

Due to this system the effective marginal tax rate, the money lost from each extra dollar above the threshold, is high on incomes just above each threshold, of which there are currently 18. This is most extreme at the first threshold, with the chart below showing marginal tax rates on some incomes well above 100%.

I use ‘effective marginal tax rate’ here because it is standard policy terminology, but HELP repayments differ from other types of EMTR, such as the combined losses of additional tax and reduced means-tested benefits as income increases. These represent money lost without future benefit. For people who will eventually repay all their HELP debt, however, repayments in the 1% band – $500 to $600 a year – will increase their future disposable income, as they will clear their debt slightly earlier than otherwise.

Costs of % of all income repayment systems

Despite early repayment’s future benefits, some HELP debtors would prefer the money now. Of this group some engineer their income to avoid or minimise a HELP repayment. The Accord final report published a chart, copied in below, showing income ‘bunching’ below threshold repayment changes, especially at the first 1% threshold. HELP debtors with incomes close to a threshold can manipulate their repayment liability by reducing hours worked or claiming tax deductions they would not otherwise have made.

These analyses are better done comparatively with non-HELP debtors (in case earnings around the thresholds correspond with incomes that are common for other reasons). However income bunching under the threshold repayment cliffs has also been observed previously, so there is little reason to doubt that the Accord final report chart identifies a real practice.

Income manipulation saves the HELP debtor money, at least temporarily. But it has costs to others, including reduced labour supply and lower income tax revenue.

Some people dodge repayment but most do not

While I believe that the current system causes some manipulation of income, its scale needs to be put in context. In 2018-19 and 2019-20 the first thresholds were reduced. Starting with $55,783 in 2017-18, with a 4% of all income rate, the first threshold went to $51,597 in 2018-19 with a 2% of all income rate, and then $45,811 with a 1% rate in 2019-20. The overall policy goal was to reduce HELP’s costs to government. The new lower % of income rates were intended to reduce the incentive to manipulate income. The cost of crossing the first threshold fell from about $4,200 to $460.

ATO tax return statistics show that lowering the thresholds resulted in big increases in the number of people making repayments in the $55,000 and under income brackets, as seen in the chart below.

This is not surprising. Only some people can vary their income to hit precise target amounts, doing so requires greater knowledge of the thresholds and rates than most people are likely to have, and some of them would recognise that, annoying as a 100%+ EMTR is, it will not cost them anything in the long run.

On the Accord final report chart the absolute numbers of people manipulating their income look low, but I’d want to know more about the data source. If it is the 2019-20 ATO 2% sample file the numbers could be multiplied by about 50 to show their true scale (the report just says ‘internal administrative data’).

But even HELP repayment dodgers are 50 times the number reported that is well below the additional people who are repaying. Between 2017-18 and 2019-20 the number of people with taxable incomes of $55,000 or below who made a HELP repayment went from 23,698 to 228,475.

Benefits of % of all income systems

The current % of all income system undoubtedly has disadvantages, but I have not in the past called for it to be changed. My reason was the potential effect of marginal repayment systems on HELP’s total costs. Repayment delays if the marginal rate reduces repayment revenue would increase the interest subsidy cost of HELP and elevate the risk that debts will not be repaid in full.

New Zealand’s marginal rate system results in only moderate bad debt, but their repayment rate of 12% starts with a repayment threshold of only NZ$22,828, equivalent to about A$21,300. A threshold this low is not politically viable in Australia.

My support for the % of all income system was in the context of an earlier system where the first threshold was higher than now – e.g. the $55,783 in 2017-18 – and the first repayment rate was 4% of all income. Although the EMTRs just above the threshold were huge, anyone whose income consistently reached the threshold was likely to repay in full.

But say, for example, we had the English repayment system of 9% of income above the threshold. Someone earning $100 above the threshold would repay only $9. People who consistently earned just above the threshold would never reduce their balance, let alone repay in full. Indexation would increase how much they owed.

Persistent income in the current first threshold of 1% of all income, however, would also lead to increasing rather than decreasing HELP balances in many cases. Indexation of 2.5% (around what we can expect as the long-term average) on a $25,000 debt = $625, slightly more than the amount repaid.

Does a failed reform open the way for a marginal repayment system?

In my view the repayment threshold levels and rate reforms introduced in 2019-20 were bungled, for reasons explained in this post. They failed in the policy objective of significantly reducing the system’s costs to government because they reduced annual repayments for many debtors with incomes above the old thresholds. The additional repayment revenue from lower-income debtors was probably not worth the policy and political grief.

But would going to a marginal rate be worth its own case of policy and political grief?