Over the last two years high CPI-driven HELP debt indexation – 3.9% in 2022, 7.1% in 2023, probably in the vicinity of 5% this year – has been a major issue. It has brought to public attention HELP debt issues that had been waiting for their trigger.

The last time HELP debt indexation exceeded 5% was in 2001, as the the new GST flowed through into prices and an indexation rate of 5.3%. At that time 1.1 million people had HECS (as it then was) debts of about $7.2 billion. A Factiva search shows that this unusually high indexation was newsworthy at the time. But with much lower HELP balances per person than now the average debt increase was only $350. The indexation issue then went quiet for two decades, except when the government wanted to charge more than CPI.

The reason for HELP indexation’s long low media profile was that between 2002 and 2021 it averaged 2.4%. It was below 2% between 2016 and 2021. This extended period of low inflation left HELP indexation as a latent issue, as increases in HELP debtor numbers and average debts gave it far more potential to cause significant personal cost and political trouble than it had in 2001.

In 2022 just over 3 million people owed $74.4 billion – nearly three times as many debtors as in 2001, and more than seven times as much debt. 7.1% indexation in 2023, on average about $1,700 per debtor but much more for the over 300,000 people with debts exceeding $50,000, made it a big issue in mainstream and social media. If Accord final report enrolment and attainment recommendations turn into reality then HELP debt and debtor numbers will increase into the distant future.

Why indexation should stay – but with anomalies corrected

Holding HELP debt on its books is a cost to government. In most years it borrows on the bond markets and re-lends to students at a lower rate, CPI. Outstanding HELP debt is equivalent to 8% of all Australian government debt. HELP debtors and the government consequently see CPI indexation in different ways; the debtors as a cost and the government as a subsidy. That’s why we have seen various failed government attempts over the years to impose a ‘real’ rate of interest on HELP debt, and in recent times debtors calling (with so far a similar lack of success) for indexation to be abolished or reduced.

With finite tax dollars for infinite demands on government significant increases in HELP subsidies should not be a high priority – certainly below Accord proposals to assist more students to finish their degrees.

CPI indexation is usually a sensible compromise between competing policy and political considerations. It partly offsets the cost to government of student loans while usually keeping interest low for student borrowers. But CPI can produce significant anomalies. While a precise interest subsidy amount is hard to calculate, as outstanding government bonds have a range of interest rates, the chart below uses the 10-year bond rates for the year in question as a rough guide. In 2022 and 2023 especially this methodology generates a substantial interest surplus, due to the rare situation of CPI being well above the bond rate.

By accident, HELP debtors have more than covered the government’s cost of borrowing – contrary to the policy decisions of all the Cabinets and parliaments that have previously rejected this idea. While this situation should self-correct in the next couple of years it has changed the politics of HELP debt and needs a solution.

Lower of CPI or some other indicator

To reduce the chance of further spikes in HELP indexation various lower-of CPI or some other indicator formulas have been suggested. I summarised these in a blog post last year: lower of CPI or the bond rate, the RBA cash rate, or some measure of wage growth.

The Accord final report recommended the last of these options, in particular the wage price index. The WPI measures changes to hourly labour costs. The final report says that ‘such a cap will ensure that the indexation of HELP debts no longer outstrips the growth in wages and the servicing capacity of debtors does not go backwards overall.’

I’m not convinced by this logic. To ‘service’ the loan, the most important aspect of which is annual repayments, there would be more logic in indexing the repayment thresholds by some wage indicator than total debt. Indeed, the thresholds were indexed to average weekly earnings rather than the CPI until 2019 (a Grattan-era recommendation of my team to stop thresholds rising too much that backfired).

For HELP debtors WPI will typically increase by less than their wages, because WPI excludes higher wages due to reasons such as changing occupations and working more hours. Due to these factors early career graduate earnings grow by much more than general wage increase indicators.

Conceptual issues aside, WPI won’t reliably remedy the problem we are trying to fix. While Australia is not in a 1970s style wage-price spiral, inflation inevitably influences wage demands. So to some extent the WPI will track CPI, contrary to the goal of moderating CPI indexation during periods of high inflation. In the 2010s, as seen in the chart below, CPI and WPI typically moved in the same direction. Growth in real wages in the 2000s pushed the WPI above CPI.

In the last 25 years the WPI has been lower than CPI indexation only four times, including 2022 and 2023. Only very pessimistic assumptions about future real wage growth would make it a better indexation method for HELP debtors than CPI. I’d bet on CPI as the lower long-run indexation rate. There is a strong policy commitment to keeping CPI in the 2-3% range (not coincidentally, the long-run indexation average since the RBA was given inflation targets in the 1990s is in this range). By contrast, real wage growth is usually seen as a desirable policy goal.

Either CPI or a fixed maximum indexation rate

A stronger political and policy fix than any of the proposed lower-of alternative indicators to CPI is a maximum indexation rate. This sends a simple and clear message: if you borrow under HELP you will not pay more than X% a year. No worrying about outlier years in the CPI or any of the other possible indicators.

In the example below I suggest 4%. This is not a complete political fix, as 3.9% indexation in 2022 started indexation as a political issue. But 4% is a rate designed to make this policy low or zero cost to government in most years, which would work strongly in its favour if it goes to the Cabinet expenditure review committee. I’m assuming here that bond rates stay below their historical averages – not my area of expertise but the RBA notes long-term trends towards lower rates. Compared to CPI, however, the 4% maximum would have left the government worse off in only two of the last 25 years.

Conclusion

We can’t work out this year’s HELP indexation rate until the ABS releases the March quarter CPI in late April. Based on December quarter data I think this year’s indexation could be in the 5% to 5.5% range. So a 4% maximum indexation rate would cost the government money this year compared to the normal 1 June indexation going ahead, the third exception to the general pattern.

There is probably not, in any case, enough time left to legislate and implement a changed indexation rate for this year. But announcing a policy change would help the government manage the politics of indexation. It remains to be seen if the 2024 indexation rate will be the second or third highest this century, but either way another round of negative media stories will need political management.

====================================================================

A technical point about how CPI is used

As noted in a blog past last year, the higher education funding legislation has several different ways of applying CPI indexation. For HELP debt indexation, for reasons I am unclear on, it uses two years of CPI data. It adds up the CPI index numbers for the previous four quarters and divides it by the same quarters in the prior year. Although the long-run averages are very similar to simple year-to-year CPI increases, the formula slows the flow of CPI changes through to indexation. As a result, initial high inflation does not immediately fully flow through to indexation. For example, 2022 indexation would have been 5.1% rather than the actual 3.9% using a simple March quarter 2021 to March quarter 2022 increase. However the formula means that indexation can still be elevated after year-on-year CPI returns to its normal range. Perhaps there are some compound interest benefits to slowing the initial increase. But a simpler calculation should be considered as part of overall HELP reform. If the government chooses the WPI it will also have to consider the relevant time period to use.