Hit or miss? A quick look at the fees-free policy

Note: This article was published originally in the now-discontinued NZME journal Education Central on 17 June 2019

When the government announced in mid-May that $197 million had been reallocated from underspends on the fees-free tertiary education policy (to fund changes arising from the review of vocational education), it wasn’t the first time.  Eleven months earlier, the government announced that it had transferred an underspend on fees-free to fund an increase in funding rates.  My estimate of the four-year cost of that first transfer is around $116 million.

116 plus 197.  An over-costing of more than $300 million?  How come?

The reason was that the government expected a 15% increase in enrolments in response to the fee reduction.  That was simply unrealistic.  The elasticity of demand for tertiary education is very low – a complicated way of saying that fee changes don’t affect demand much.  One reason for that in the New Zealand context is that virtually all domestic tertiary students qualify for interest-free student loans when they enter tertiary education – the cash cost to a student of enrolling in tertiary education is close to zero.

The relationship between fee-levels and tertiary participation

The same occurs in all developed countries – fees don’t necessarily deter participation.  OECD data shows that fee levels don’t line up with participation levels; the majority of zero fee countries in the OECD have a lower rate of participation in tertiary education than New Zealand.

Figure 1 shows a weak relationship between fee-levels and participation rates in the core age-group for that level of study; other factors are much more important than fees in encouraging participation.  Of the 25 countries whose data is shown in Figure 1, all but one of those whose participation rate is below 30% have low or no fees.  Nine of the 25 have zero fees; of those nine, three have relatively high participation rates, three have low participation and three are around 30%.

The New Zealand evidence on the relationship between fee levels and participation also suggests that fees play only a small part in the decision to enrol.  A 2009 Ministry of Education report traced the relationship between fee levels in bachelors degrees and bachelors participation rates between 1994 and 2007.  As fees rose, so did the participation rate.

The problem of deadweight

Following the announcement of the second reprioritisation of fees-free funding, the Minister of Finance hinted that the government may give fresh consideration to whether to complete the full implementation of the policy.  Students’ association leaders expressed disappointment and stated that many people had reported that they had entered study only because of the fees-free policy. 

That may be so.  But the fact that the government has had to reprioritise so much fees-free funding suggests that the number of those attracted by the fees-free policy was very small.  Which means taxpayers have had to pay fees for 50,000 people in order to enable a very, very small number of people to enrol.  That the very great majority of people who benefitted from fees-free received a bonus for doing exactly what they always intended to do is the epitome of deadweight.  Further, students – especially those at universities – are disproportionately from the middle classes and above.  And the data suggests that they will earn higher than average incomes over their lifetimes. That amounts to a very regressive policy, a policy that favours the rich at the expense of the poor.

So what effect did fees-free have on 2018 enrolments?

On 12 June, the Ministry of Education released its latest enrolment forecasts.

These enrolment forecasts come from a set of statistical models that apply demographic and labour market variables – Statistics New Zealand’s population forecasts and Treasury’s unemployment forecasts – to the enrolment data.  They have very high predictive power – well over 90%.  The models produce the expected number of equivalent full-time student (EFTS) units, plus an error term that represents the lower and upper bounds of the 95% confidence interval.

Let’s look at the final full-year enrolments data for 2018 against those forecasts for all enrolments at levels 3 and above, funded through the main funding stream for teaching and learning – the Student Achievement Component (for short, SAC3+).  Of course, the fees-free policy extends beyond SAC3+.  For instance, it also includes industry training.  But SAC3+ is where most of the fees-free money has been spent. 

There are two published forecasts against which we can compare the 2018 actuals – the forecast done for the 2017 Budget Economic and Fiscal Update (BEFU 2017) and the forecast made for BEFU 2018.  BEFU forecasts are made early in the year, so they can feed into the May budget.  This means that BEFU 2017 was made before the election that brought in the fees-free policy.  The BEFU 2018 forecast is too early to have taken account of early enrolment data for 2018, and so is done without the influence of fees-free policy.  This enables us to estimate the effects of the fees-free policy on enrolments reasonably precisely.

Figures 2 and 3 below set out the results of the comparison between actual SAC3+ enrolments and those two forecasts.

If the actual value falls between the lower and upper bounds of the forecast, that means that we can say that the model predicted that result.

As well as looking at all SAC3+ enrolments, it’s also interesting to look at enrolments at bachelors degree level because bachelors students make up more than 60% of SAC3+ (ignoring the postgraduates, who are excluded from fees-free).

What can we infer?

In 2018, there were 203,380 EFTS in SAC3+ – compared with a 2018 forecast of 200,630 and a 2017 forecast of 208,750.  That is, the 2018 actual was 1.4% up on the forecast made in early 2018 but 2.6% below the 2017 forecast. (Note that it is likely that the 2017 forecast was wrong because the actual unemployment rate in 2018 was lower than the Treasury forecast that is used in the Ministry of Education’s enrolment forecast model – meaning that the 2017 forecast overestimated the number of enrolments in 2018).

What is important, however, is that the 2018 actual comes within the upper and lower bounds of the 2018 forecast and below the lower bound of the 2017 forecast.    In simple terms, that means there is no evidence of a statistically significant enrolment increase as a result of the fees-free policy.  The actual 2018 outcome is 1.4% up on the forecast done in early 2018, but that is within the 95% confidence interval for the forecast. It is within the margin of error.

Looking at Figure 3 where the forecasts for bachelors degree enrolments are shown, we see that the actual 2018 enrolment is between the upper and lower bounds of the two forecasts, so that the result can be seen as within the margin of error for each of the forecasts.  Again, no evidence of a fees-free effect.  Even though the downward trend in bachelors enrolments stabilised in 2018, the change isn’t statistically significant.

The fact that the 2018 full-year actuals are all above the forecasts done in 2018 is, however, interesting.  It is possible that the take up of the fees-free benefit may have gained momentum as the year went on.  Given the breakneck speed of the implementation, perhaps 2019 may be the year to watch …

A caution

This is a very superficial, high-level analysis.  What the Ministry of Education analysts are likely doing right now is redoing the 2017 forecast with updated unemployment data and checking the 2018 actuals against that.  And they will be doing a cohort analysis – to narrow in on the cohort of students who were beneficiaries of the fees-free policy.  They will be examining participation rates to control for the effects of the change in the size of the age cohort.  Their analysis will be the definitive one.    

So what does it mean?

Has the fees-free policy led to an increase in enrolments in these forms of tertiary education? On the evidence of the data in this analysis, no. Has it “stopped the rot” – that is, stemmed the drop in enrolment numbers that has resulted from demographic change and labour market trends, from a falling tertiary age population and a stronger employment market?  We can’t be exactly sure.  More analysis and another year will reveal the answer.

The fees-free policy may not be costing the government as much as originally planned.  But it does cost plenty.  The participation effects appear to be only marginal.  That means that the main effect of the policy has been to increase the government’s share of the cost of an eligible student’s enrolment – from around 85% of full-cost to 100% – but for little return in terms of access or participation. 

And what is the outlook for 2019 and beyond?

The Ministry’s new forecasts show a small decline in enrolments in 2019 and 2020, then stabilisation.

There is no forecast increase in enrolments.  Would enrolments have plummeted in the absence of the fees-free policy?  This first analysis suggests the effects are marginal.


It’s easy now to be critical of the government’s fees-free policy.  There was already plenty of evidence in the public domain that the policy would have high deadweight: that financial factors are not a barrier to tertiary education whereas cultural factors (including parental education) are; and – as discussed above – that fees aren’t an impediment to access.

But look at the context.  It was developed in 2016 by a party in opposition; it was an attempt to present a policy with a responsible, conservative costing.  But, because it was an election policy, that costing had to be done without the expertise of, or advice from, public service advisors.  The participation assumptions that drove the costing were carried forward to the post-election mini-budget, resulting in a substantial over-costing. 

But now that we’ve seen how this policy is playing out, isn’t it time to rethink?  And to look at the real issues for access to tertiary education?

Part 2 of this article looks at that question.

%d bloggers like this: