An Analysis of Alternative Financing Systems for Ireland
Two main studies exist that undertake exante^{11} empirical analyses of higher education financing alternatives for Ireland; these are Flannery and O’Donoghue (2011) and Chapman and Doris (2016). In conducting ex ante analyses, the key challenge is to obtain valid predictions of graduate earnings for many years into the future. Moreover, these lifecycle earnings predictions must be obtained not just for the typical—or median— graduate but for graduates throughout the earnings distribution; this is important because graduate hardship and inability to repay are the result of low graduate earnings, so such earnings must be modelled carefully. Flannery and O’Donoghue (2011) and Chapman and Doris (2016) use different datasets and different econometric approaches to modelling graduate earnings. In addition, they differ in the scheme parameters that they model. It is therefore of interest to compare the result they obtain. In this section we first give an overview of the results contained in these two papers. We then discuss some supplementary results regarding ICL schemes obtained using the Chapman and Doris (2016) methodology that may be of additional interest in the context of the Irish debate on higher education funding.
Flannery and O’Donoghue’s (2011) paper compares a GT with an ICL. Here, lifecycle graduate earnings profiles are obtained from an Irish dynamic microsimulation model (The Lifecycle Income Analysis Model [LIAM]) based on data from the Living in Ireland Survey, which was the Irish component of the European Community Household Panel (ECHP) that ran from 1994 to 2001.^{12} The alternative GT and ICL schemes are then applied to these earnings profiles.
The GT is modelled as either a 1% or a 2% increase in PayRelated Social Insurance (PRSI) contribution rates, with real rates of interest of either 0% or 2%. It is assumed that 20% of graduates emigrate and, because the GT is collected only from earnings in Ireland, no repayments are made by these graduates. Assuming that the portion of government spending on higher education that is repayable by the GT is €10,000, the government subsidy can be calculated. Results for the version of the tax with a 2% real interest rate show that a small government subsidy of 4% would be required if a 2% surcharge on PRSI rates were imposed, whereas a 51% subsidy would be required if a lower surcharge of 1% were applied. The results indicate the importance of emigration to the yield from a GT, with smaller subsidies of 0% and 40% required for the 2% and 1% surcharges respectively in the absence of graduate emigration.
The ICL scheme modelled in Flannery and O’Donoghue (2011) assumes a loan of €10,000 repaid at a rate of 10% on marginal earnings over a threshold of €35,000 and 15% on earnings over €42,000. Two interest rate regimes are again modelled—one with a zero real rate and another with a 2% real rate. Again, it is assumed that 20% of graduates emigrate but since the ICL results in graduates owing a debt, it is not assumed that emigrants repay nothing; instead, these emigrating graduates repay 40% of their debt. Finally, any outstanding debt is written off at retirement. Applying this ICL scheme to the estimated lifecycle earnings profiles, the average subsidy is found to be 26% if a 2% per annum real interest rate is charged, and 40% if not. The analysis of the repayment patterns of the graduates does not include an explicit analysis of
RBs. However, it is found that those in higher lifecycle earnings deciles repay more of their loans than those in lower deciles. Moreover, the present value of repayments rises strongly with the earnings decile.
Table 10.3 summarises the Flannery and O’Donoghue (2011) results. It is notable that female graduates repay a much lower proportion of the €10,000 that is repayable, whether under the GT or the ICL. In addition, under the ICL, fewer women than men repay their loans in full, and those women who do repay in full take longer to do so than men who repay in full. These findings are all, of course, the result of lower earnings
Table 10.3 Revenue and repayment analysis of graduate tax and income contingent loan system for Ireland
Graduate tax system 

Graduate tax revenue as % of total repayable (€10,000) with 2% real interest rate and 20% graduate emigration 

Females Males Total averag 
Yield of 1% graduate tax 42.7 55.2 e 49.0 
Yield of 2% graduate tax 81.6 109.0 95.6 

Income contingent loan system 

Repayment patterns for graduates with two different interest rates and simulated graduate emigration with some repayment (debt of €10,000) 

% of borrowers who repay in full 
Average repayment period in years 
Average NPV of repayments (€) 
Average subsidy as a % of loan 

0% real interest rate 

Females 
66 
16.2 
5328 
46.7 
Males 
82 
14.2 
6482 
35.2 
Total average 
75 
15.1 
5907 
40.1 
2% real interest rate 

Females 
57 
16.0 
6652 
33.5 
Males 
77 
15.4 
8167 
18.3 
Total average 
67 
15.6 
7413 
25.9 
Notes: The average repayment period for the ICL system includes only those that had paid their loan in full. The NPV of repayments are repayments discounted to the year of graduation of each graduate Source: Adapted from Flannery and O'Donoghue (2011) by female graduates compared to males; the LIAMsimulated earnings streams indicate that the present value of lifecycle earnings for women with tertiary education are about twothirds those of men educated to that level.
Chapman and Doris (2016) compare a mortgagestyle GGBL with an ICL. In both cases the loan amount is €16,000, which would represent, for a fouryear degree, a moderate increase in fees from the current level of €3000 per annum. The lifecycle earnings profiles in this case are provided by unconditional quantile regression analysis of 2006 National Employment Survey data. The resulting earnings profiles indicate that female graduate earnings are significantly lower than male earnings, particularly at the top of the earnings distribution; this echoes the pattern seen in the profiles estimated using the microsimulation model of Flannery and O’Donoghue (2011).
The GGBL that is modelled in Chapman and Doris (2016) is based on repayment over ten years, with a real interest rate of 2% applied from the date of graduation, and with repayments beginning two years after graduation. The analysis shows that although the RBs are moderate for a graduate with median lifecycle earnings, for working graduates at the bottom of the earnings distribution RBs are very high, particularly in the two or three years after repayments begin—as high as 83% for males at the 10th percentile of lifecycle earnings. When account is also taken of the fact that some graduates are not in employment and so have no earnings, the proportion of graduates for whom repayments are problematically high rises further. Even five years after repayments begin—and so seven years after graduation—over a quarter of graduates have RBs in excess of 18% of gross annual earnings, a conservative threshold that has been used to indicate excessively high RBs (Chapman and Lounkaew 2015b). These high RBs lead the authors to reject GGBLs as a feasible alternative for higher education funding in Ireland.
In their analysis of ICLs, Chapman and Doris (2016) model four alternative schemes by varying two parameters, the repayment rates and the interest rate. Two repayment schedules are modelled, one entailing a flat rate of 8% on marginal income above an earnings threshold of €26,000 and the other entailing rates of 28% on total income once this threshold is reached, starting at 2% and rising in increments of 1% for every
€5000 of additional earnings over €26,000, up to 8% on earnings above €56,000. In addition, two alternative interest rate regimes are modelled, one entailing a zero real rate of interest, and the other with a 2% real rate of interest in periods when income rises above the €26,000 threshold, but a zero real rate otherwise. A 20% rate of graduate emigration is also assumed: half of these emigrate permanently and it is assumed that no repayments are ever made by these graduates; the remaining 10% emigrate temporarily but recommence repayments once they return to Ireland. This is arguably a more pessimistic emigration scenario than that presented by Flannery and O’Donoghue (2011), since repayments are not made by any graduates living abroad. On the other hand, assuming that half the emigration is transitory does allow for some emigrant repayments.^{13}
Various measures of graduate affordability are reported in the paper, as well as the government subsidy implied by nonrepayments under the four alternative schemes. The affordability issue matters because it illustrates the repayment burdens of debtors as a proportion of their aftertax earnings. The authors conclude that all four schemes show reasonable levels of affordability for graduates, with repayments representing up to 8.6% of net earnings for men and up to 6.3% for women. The government subsidy required under the four schemes ranges from 26% to 37%, depending on the particular scheme, with the schemes that include a positive interest rate found to have subsidies at the lower end of this range. The importance of emigration patterns to the size of the subsidy is also noted, with emigration adding 10 percentage points to the subsidy required. It is noteworthy that, despite differences in the methodology used to simulate graduate earnings profiles, the estimated subsidies in these two papers are very similar for the schemes that are most alike.
We now report some additional results based on the data and methodology used by Chapman and Doris (2016)—see Table 10.4. Here, we focus on varying the loan amount: as well as results for a loan of €16,000, we show results for a loan of €20,000, equivalent to €5000 per annum for a fouryear degree. This would entail a more substantial increase in higher education funding from the current fee level of €3000 per annum
Table 10.4 Repayment analysis for two alternative loan amounts: €16,000 and €20,000
25th percentile 
50th percentile 
75th percentile 

Loan amount: €16,000 Males 

Number of years of repayments 
12 
13 
8 
Mean % net income 
4.2 
4.0 
5.6 
Mean monthly repayment 
127 
119 
180 
NPV of repayments 
13,126 
14,208 
15,077 
Females 

Number of years of repayments 
22 
12 
11 
Mean % net income 
3.0 
4.3 
4.5 
Mean monthly repayment 
76 
127 
137 
NPV of repayments 
13,388 
13,929 
14,782 
Loan amount: €20,000 Males 

Number of years of repayments 
14 
14 
10 
Mean % net income 
4.5 
4.5 
5.6 
Mean monthly repayment 
139 
141 
183 
NPV of repayments 
16,407 
17,759 
18,846 
Females 

Number of years of repayments 
25 
15 
13 
Mean % net income 
3.3 
4.3 
4.7 
Mean monthly repayment 
88 
130 
148 
NPV of repayments 
16,735 
17,411 
18,477 
Notes: The mean monthly repayment is calculated only over years in which the repayment is positive. The discount rate used for NPV calculations is 2% Source: Adapted from Chapman and Doris (2016), with additional results provided by Bruce Chapman and Aedfn Doris
and may be more attractive to policymakers in light of the funding challenges that will result from demographic changes in the coming years. For both loan amounts, the results refer to a scheme based on repayments of 8% on marginal income over an earnings threshold of €26,000, and with interest charged at a real rate of 2% per annum when earnings exceed the threshold.
The results reported in Table 10.4 show clearly that increasing the loan amount has very little effect on the measures of affordability reported. The percentage of net income accounted for by loan repayments varies through the graduate earnings distribution to a similar extent for both loan amounts. For men, the range for a loan of €16,000 is from 4.2% at the 25th percentile to 5.6% at the 75th percentile; for the €20,000 loan, the range is 4.5% to 5.6%. For women, the range is greater for both loan amounts, going from 3.0% of net income at the 25th percentile to 4.5% at the 75th percentile for the €16,000 loan, compared to 3.3% to 4.7% for the €20,000 loan. As was the case in Flannery and O’Donoghue (2011), the differences between men and women are entirely accounted for by the lower earnings of women, and in particular their flatter lifecycle earnings profiles. In all cases, the RBs are moderate.
The pattern in the figures for the average monthly repayments is similar to that for the RBs; average monthly repayments are generally lower for women than for men at corresponding points in their respective earnings distributions and higher the further up their lifecycle earnings distribution the individual lies.^{14} However, the absolute figures are remarkably similar and as expected with ICL, mean monthly repayments are hardly affected by having incurred a bigger debt.
The differences for alternative loan amounts arise only in the number of years of repayment: since the amounts being repaid monthly do not differ according to the loan amounts, the impact of the increased loan burden falls on the number of years over which repayments are made. Whereas the female with median lifecycle earnings repays her €16,000 loan in 12 years, the €20,000 loan takes 15 years to repay. The difference for males with median earnings is an increase of just one year, from 13 to 14 years; at other points in the distribution, the increase in years of repayment is greater, at two years. For both loan amounts, the years of repayment are quite similar to those found in other countries. A final unsurprising point is that the net present value of the loan repayments is higher for the higher loan.
The analysis of the €20,000 loan also shows that the government subsidy implied by the ICL increases very little compared to the €16,000 loan; allowing for nonparticipation and for the emigration of 20% of graduates, the subsidy increases from 27% for the €16,000 loan to 28% for the €20,000 loan.
The careful simulation of graduate earnings profiles is central to the reliable assessment of any higher education funding scheme that is based on graduate earnings, and such simulations necessarily entail many assumptions. The simulations used in Flannery and O’Donoghue (2011) and Chapman and Doris (2016) are quite different in their methodologies, and yet their conclusions in respect of ICLs—which is where the two papers overlap—are very similar. For ICLs based on positive real interest rates and repayments calculated on marginal income over some threshold, the estimated government subsidies are very close—26% versus 27%. It is noteworthy that both analyses find that the extent of nonrepayment arising from graduate emigration is very important in driving the size of the government subsidy, with results in both exercises indicating that the subsidy rises by about 10 percentage points when emigration of 20% of graduates is allowed for.