Over the years, I have concluded that one of the reasons policy debate can be so stifling is we’re usually debating options within existing policy parameters: that is, “fixes” to existing policy. It’s pretty rare that anyone talks about “greenfield” policies in areas where no policy ever existed. This is kind of a shame, because it means people don’t really understand the process of trade-offs that go into original policy-making.
So, just for fun, I am going to spend this week talking about how to create a student aid program from scratch – what decisions have to be made, in what sequence, etc. Of course, not many countries really do this truly from scratch anymore: a few in Africa and Asia and that’s about it (Russia is probably the largest country without a functioning national student loan plan). But let’s put ourselves in the position of a country that is just starting out in the whole student aid business, with levels of technology that are not quite Canada 2019. And see where this leads us.
Ready? OK, let’s go.
We’ll start by assuming that if this country is new to the whole business of student assistance, it’s probably not all that wealthy. If it wants a student aid program, it’s probably because it wants to expand access to higher education and become a richer country. The problem in this situation is that higher education is always relatively more expensive in poor countries than in rich ones, mainly because of labour costs. Because professors are somewhat internationally mobile, they need to be paid on a scale high enough to keep them from decamping to another country where academics are better paid. This can lead to professors being paid quite highly, relative to the average national income, which in turn makes higher education difficult to support publicly, particularly in countries where the tax take is not a very high proportion of the economy (common in both Africa and Asia – and indeed was also the case in North America prior to WWII). In order to expand, universities need fees, but in order to pay fees, many (most?) students need some assistance. And that’s where this exercise comes in.
We can take it for granted in this situation that the government cannot afford to provide everyone with grants sufficient to cover both fees and living expenses. Even in places like Scandinavia, non-repayable aid doesn’t cover everything; there is always some mix of grants and loans. So, the default policy position is going to be loans plus grants with the mix to be determined by some combination of financial circumstances and politics. I will come back to the loan/grant balance in a couple of days, but for now let’s stick to the loans part.
The very first question to ask is “where is the loan money going to come from”? There are really only two choices here: it can come from government current accounts or it can come from banks; if the latter, government can either guarantee the loans (which is how student loans started in North America), or they can pay the bank a fee for every dollar loaned (which is how it works in China and in Canada from 1995-2001). There are no other choices, really. So which option should a government choose?
(Some governments have tried to get cute and tried to use public-but-non-governmental sources of funding instead. In the 1990s, Ghana, for instance, tried running its student aid system using the one of the state pension funds as a source of capital. Let’s just say that the rioting which followed the news of substantial fund losses was not entirely unforeseeable).
For a whole bunch of reasons, the loan guarantee option makes an enormous amount of sense to most governments, because banks have something most governments do not: a functioning system of loan recovery. This matters a lot: the ability to lend money depends crucially on how much money ends up getting repaid. A country with 90% loan repayment is going to be able to be a lot more generous with loans than a country with 10% repayment (at that point, you might as well just stick with grants). Plus, the cost of a guarantee lies far in the future – under a guarantee system, you can get the first four or five years of a program basically for free. The costs pile up later, sure, but if you’re a government which tends not to think about the long term this is neither here nor there.
So why don’t all governments go that route? You might think it’s because it’s cheaper in the long run, but this is usually not the case. Yes, it’s true that a government’s cost of capital is lower than a bank’s cost of capital, but the relevant question is actually whether government cost (capital + loan recovery) is lower than bank cost (capital + loan recovery), and you have to have a pretty deep level of financialization of the economy and/or a very efficient tax system to make that work. Most OECD countries can do it; most non-OECD countries can’t.
No, the more common reason is that banks turn governments down because they don’t believe in the government guarantee. This is particularly the case in parts of Africa or Australasia where the there is no indigenous banking sector and the financial sector is in the hands of transnational agents (South African, French, Australian, British or Nigerian as the case may be). Basically, if you are Standard Chartered or ANZ or a similar institution, do you actually believe that the government of Madagascar (or the Solomon Islands, or wherever) will live up to its guarantee commitment? If not, you’re not going to put your shareholders’ money at risk.
As you can have probably gathered from the foregoing, this “choice” may not always be a choice – circumstance may force a government into one option or the other. And the nature of this “choice” may in fact constrain the amount of credit a government can extend to students. But even were this not the case, governments do tend to have to pick and choose between students when it come to aid. How they decide to ration aid is therefore the subject of tomorrow’s blog: see you then.
Now that we’ve decided on a loan system, we have to start thinking about how we are going to recover any money that we lend.
For decades, student loans only worked one way: on a regular amortization basis. If you borrow, you repay on a regular monthly basis after graduation, just like one pays a regular amount each month on a mortgage. Even within this framework, there are a lot of policy elements one can adjust. The length of the repayment period can be fiddled with: in Asia, they prefer short periods like four years for loan repayment, we prefer longer periods like ten or fifteen or even longer in the United States. Basically, the only thing going on here is a trade-off between individual monthly payments and higher total interest payments. There is also the question of a grace-period and how long it takes for repayment to actually start. In most countries the period is somewhere between six and twelve months; in its re-election platform the Liberal government suggested two years, and in Germany it is five. Assuming the government is paying for interest during the grace period (not always the case), this is simply a matter of generosity on the part of government. However, assuming funds are limited, government “generosity” in one area may mean fewer loans overall.
Another big question is the nature of interest rates on loans. Do you charge them? If you charge them, at what point do they start accumulating? In most countries where interest accumulates, it does so from the moment the loan is taken out; in North America, interest is zero (i.e. negative real interest) while the student is in school. At what rate do you charge them? Equal to inflation like in Australia? Equal to the government cost of borrowing as in the Netherlands? Some higher level which is lower than market rates, but which still produces enough surplus to offset loan losses, as in Canada? These are all important decisions that need to be taken. Again, generally speaking, the more “generous” the government happens to be in any one area, the less able it is to be generous in others. It is, as always, about balance.
The other alterative is to do something called “income-contingent loans”. The exact definition of this is vague, but usually these are loans collected through the tax system, calculated as a percentage of total income (or marginal income above a given threshold), as in the UK, Australia and New Zealand. This is not the only way to run income-contingent systems; in fact, most systems have some kind of income threshold below which repayment is suspended. Some – like Canada and the US – have “income-sensitive” systems where low-to-middle-income borrowers are relieved from paying the full amount of what is owed. These programs tend not to use the tax system for collection.
(Basically: income-contingent loans don’t need to be collected by the tax system, and you can use the tax system to collect loans that aren’t income contingent. But people confuse the two issues all the time).
The “full” income-contingent system is sometimes seen as the most generous and flexible system and holding the level of student debt constant that’s probably true. Where income-contingent debt gets weird is in the way it effectively abolishes the possibility of loan default. On the one hand, this is good for students; on the other, it means government has a hard time getting a signal about how well repayment is going. This is a problem particularly if you have high debt levels and a generous forgiveness clause, like the UK does: this results in a system where something like half of all debt will never be repaid and something like 70% of all borrowers will receive some forgiveness. This is either a feature or a bug, depending on your point of view.
Whether you have an income-contingent or a mortgage-style loan, you have to make some decisions about collections. In the few countries with simple, effective tax systems where the unit of taxation is the individual (as opposed to the family, as it is in the US and France, for example), it makes a certain amount of administrative sense to use the tax system to recover loans, and the use of the tax system certainly makes repayment based on income easier. But it is not fool-proof, because not everyone pays taxes, some people leave the country, etc. In the UK, Australia and New Zealand, there are still 10-25% of borrowers who are outside the payment system; in income-continent countries with laxer tax systems (Ethiopia, say), it is 50% or higher.
But say the tax system isn’t an option – what then? Well, in many countries this question tilts the playing field towards having the banks do it because they are already expert in recovery – and this in turn tends to lead to a decision to bring the banks in at the front end as well through a loan guarantee (see yesterday’s blog). But while that might work in places like, say, Malaysia, in countries with no national ID systems, or phone directories, or fully-functioning social security or even cadastral systems – most of Africa and developing Asia, in other words – the banks don’t necessarily have an advantage because under normal conditions they restrict their lending to a more stable clientele. In theory, you could hire debt collectors, but they are working under the same restrictions.
What many governments now do – especially in Africa where mobile money and online payment though platforms like M-PESA has become a dominant means of money transfer – is just create their own repayment agencies outside the regular tax system. And they are pretty good at it: loan repayment in some African countries is now over 50% or so, which may not sound like much but is a massive improvement over the norm a decade ago. This is allowing a lot of new borrowing to occur.
So, to sum up:
- Income-contingent vs mortgage-style repayment is only one of many decisions to make in terms of loan repayment design.
- Having the administrative capacity to track students and secure repayment is maybe more important, since loan recovery rates condition the availability of future loans – whether that capacity comes through the tax system or somewhere else is secondary.
- Decisions on repayment subsidies (especially interest subsidies) can have significant financial implications.
To this point, we’ve looked at debt recovery in the absence of any discussion about the absolute level of debt itself. To do that properly, we need to start looking at the question of loans vs. grants, which is our subject for tomorrow.
Welcome back to this little series. On Monday and Tuesday we looked at loans – how to pay for them and how to design repayment systems. Today, I want to introduce grants into the mix (to be clear, I’m only talking about grants where need is the primary criterion – there’s a whole other set of policy considerations about merit-based aid, which I’m going to leave to one side during this discussion).
Theoretically, the grants vs. loans debate is one of the most important and most fraught in the design of a student aid system. It is not really a matter of “generosity” unless you assume away the problem of scarcity. Loans and grants actually address quite different problems and don’t always need to be integrated into the same program. Here’s why:
Though we speak of “financial barriers” to higher education, these actually come in two quite distinct varieties. The first is what you might call the “rate of return barrier”: that is, where a program is so expensive, many people do not want to take it because they believe the costs outweigh the benefits. The current poster-child for this kind of program is the new Columbia J-school degree in data journalism which is retailing at $106,000 US. Given that this mostly gives graduates an entrée into a profession which is widely viewed as ailing if not failing, it seems unlikely many people will want to attend. Lending people money in this situation really does no good – you need grants to reduce the price.
Now, to be clear, just because a program has an expensive list price is not a particularly good reason to subsidize students; if a program genuinely generates negative private returns, there is no real reason to publicly subsidize it. But now imagine that large numbers of people start to think that moderately-priced programs which actually have reasonably good rates of return are “too expensive”. What then?
As it turns out, students from lower-income backgrounds tend both to overestimate the costs of higher education programs and underestimate the long-term benefits, relative to students from higher-income ones. This is likely one reason (there are others) why low-income students have a systematically lower propensity to opt into higher education. It’s also a problem that can be solved with targeted grants.
The more common financial barrier in higher education, though, is the liquidity barrier. That is to say, people want to go to higher education, they think it’s a good deal, but they don’t have enough money both to pay fees (remember, in this scenario we’re assuming the existence of fees) and keep body and soul together. They don’t need a grant to be convinced to go: they need a loan to tide them over until the end of their studies.
(To be clear: you could achieve the same thing with grants – it’s just more expensive. How much more? As we discussed a bit yesterday, it depends on how efficient your loan repayment systems are. If you’re recouping 90 cents on the dollar after losses, interest subsidies and administration, then a dollar in grants costs ten times a dollar in loans. If you’re only recouping 50 cents on the dollar, then a dollar in grants costs twice a dollar in loans. This is one of the many, many reasons why repayment rates matter)
Now, it’s not quite as simple as “grants to reduce prices where necessary” and “loans to solve liquidity problems”. Almost, but not quite. The third issue is debt, which is actually fiendishly complicated from a policy perspective.
Many people will claim that too much debt is itself a third type of barrier. There is some evidence for this (I wrote about it here), but it seems like a relatively small phenomenon. Most of what people mean when they say some students are debt averse is actually “this course’s benefits are perceived to outweigh the costs”. Now you may want to deal with this problem through grants, as suggested above – but it’s not a separate problem.
Even if debt aversion isn’t (much of) a thing, debt management is. Borrowers who start their working careers with very high debt-to-disposable income ratios tend not to have very good outcomes (it’s the ratio rather than debt per se that’s the problem: medical students tend to graduate with a lot of debt but usually pay it off fine, whereas drop outs with low levels of debt often default). There are a variety of ways to handle this problem. One approach is just to make default impossible by adjusting the rules (i.e. impose income-contingency), another is to extend the length of the loan so as to make a given debt load more bearable. But another set of approaches involve grants in one of three forms:
- grants at the time of enrolment, such that the amount of loan given in any year is not excessive (which is what we tend to do in Canada);
- grants at the end of studies to reduce accumulated debt to a manageable level (in Germany, for instance)
- subsidies during the repayment period to make loads more manageable (in Canada, this could mean things like the Repayment Assistance Program, and in the UK it could be massive debt forgiveness as in their current ICR system).
In an idealized, rational public policy system, a government would have some sense of various students’ price sensitivities and liquidity constraints, give out loans and grants accordingly and then, maybe, provide some non-repayable aid debt management grants. In reality, no government in the world operates this way. Instead, rules for loans and grants either get made separately with little account for balance (for instance, in the United States, or Australia), or someone goes in with a hard-and-fast rule about loan:grant ratios (50:50 in Germany, 70:30 in Sweden, etc.) and then makes the overall provision of aid as generous as the overall budgetary picture allows. And all the while, student groups are arguing for more grant money because hey, who wouldn’t want to pay less when the alternative is paying more?
At the end of the day, the loan/grant mix is 100% about politics and macro-budgetary priorities. There’s very little that’s “rational” about it. When it comes to deliberate policy making, the rational aspects tend to be less on the overall mix, and more on the specifics of how both loans and grants are rationed. Which will be the subject of our fourth and final blog on this topic tomorrow.
If you’re joining late, we’re talking about the policy decisions that need to be made when creating a student aid system. Read up on student loan origination, student loan repayment parameters and the loans/grants balance.
So now we’ve got all the big pieces in place – where the money comes from, how much is going to be loan vs. grant, and how loans are going to be recuperated. Now we get to the really fiddly bits: how to ration the aid (warning: this is a stupidly long post)
To some, this might seem a horrible question: why ration? The reality is, no country has a genuinely universal loan program. Even in the few places where, in theory, 100% of undergraduates can access loans, this access is usually limited by degree (you can only get one, or at least one at any given level) or to a certain number of years. Elsewhere, a variety of limits exist. Here are few of the limitations on eligibility for loans and grants:
- Type of institution. This type of limit is more common on grants than loans, but there are systems of assistance (Chile’s for instance) where loans are more easily available for students at public institutions than at private ones.
- Field of study. It’s rare that loans systems are only available for certain fields of study, but there are certainly a number (Tanzania, for instance) where they are preferentially available to students in high demand fields, such as STEM.
- Merit. This is an obvious filter for grants, though it can affect loans. In Japan, having higher high school grades gives you access to a different loan system with lower interest rates (I’m not sure how important this is in a country where interest rates are routinely negative anyway, but still). And in some of the earliest proposals for loans in the US (pre-1964), many thought grades should have an effect on loan availability.
- Time/Degrees. Most systems have some kind of time limit for years of borrowing (in Canada, it is usually 340 weeks, or roughly twenty academic terms; in Europe it can be as short as ten academic terms), or only allow borrowing for one degree at the same level (no getting two different bachelor’s degrees), or in a few cases, not providing for anything above a bachelor’s degree (the UK, until quite recently).
- Study Intensity. Nobody makes loans available to part-time students on the same basis as full-time ones; many don’t allow aid for part-time study at all.
- Prior Criminal records. In the US, anyway.
- Prior credit events. Easily the most offensive part of the Canadian Student Financial Aid system is a mid-90s addition to the program to deny aid to independent students taking out loans for the first time if they have three credit events in the past three years on amounts over $1,000. Yes, really.
And all of this is before we start rationing based on financial criteria. Some places don’t do this part: in Scandinavia, as long as you are eligible for aid, you simply have access to a standard set amount. In most places, there is some kind of income test which usually involves determines whether i) someone is able to receive assistance and ii) how much assistance they can receive. The process for doing this can be completely different for grants and loans; in Australia, loans for tuition are universal but grants for living expenses are stringently means-tested.
Complicated? Wait, we’re not close to done yet. There’s the question of which types of income (and, occasionally, assets) to measure. In many places, it’s simply “parental family taxable income”. But how long should you include parental income in the calculation (i.e. when should a student be considered “independent”?) What if you want to adjust for family size? Or the number of children currently in PSE? What about parental assets? Student income (Canada has among the world’s most arcane rules here, although we are gradually simplifying them)? Student savings? Student assets (cars, for instance)? Scholarship income? Spousal income? You can really tie yourself up in knots here. And it gets worse if you decide you want your aid to match up with student costs. For instance, you can try to adjust aid amounts for things like where the student lives (at home/away from home; travel allowances; different living allowances depending on exact location), presence of children, etc.
Now, most countries don’t get to this level of granularity. The prevailing global norm is that people who qualify get access to a set amount of loans and that’s that. In North America – and Canada in particular, we are really big on trying to engineer micro-equity among students. Student X gets $8000 in aid, but student Y gets only $6000 in aid, because they have a slightly higher scholarship, study in a different program and has slightly higher-income parents. It’s frankly exhausting and costs a fair bit of money to administer.
Then – I know, this is getting crazy complicated – there is the question of who gets to decide the amount of the aid and, if there are multiple sources of aid, who gets to package it. In unitary states with no significant philanthropy to speak of, it’s usually a single government agency that decides how much money to give and then hands it to students. But this isn’t always the case. In some countries, governments actually prefer to let institutions work out what “need” is, judging them better placed to do so since they are closer to actual student conditions. But this working means telling each institution what their overall allocation budget is going to be in advance, which, depending on the formula for giving out grants to institutions, may mean that some institutions get a lot more than they need to meet need and others a lot less.
In North America, federal rules assess each student individually, but for a variety of reasons the actual assessment and packaging of aid gets done by someone else. In the US, it’s individual colleges and universities who perform this function, whereas in Canada, it’s provincial governments. On both sides of the border, the entity that packages aid usually gets to be the “last dollar”, which is a big advantage. Say the provincial formula says that student X, based on a full assessment of income and assets, should get $7000 in loans and $2000 in grants. Then someone – say a new scholarship foundation, or some local Rotary club – gives that student another $1000. Well, providing they know about the transaction, the province is perfectly within its rights to say “ah, student X’s income just rose $1,000. So that reduces need by $1,000. Guess we can reduce our aid by $1,000 then.” Being the “last dollar” means you save every time someone else gives a student money. Pay attention to this over the coming months, because this could become an issue when the federal government moves to implement its promised increase in Canada Student Grants.
Long story short, there are a whole load of levers one can use in order to ration aid. Jurisdictions with relatively sophisticated student aid systems can broadly predict how much money each policy adjustment will cost or save. So, if a minister says, “I want to get rid of required parental contributions”, the ministry staff can usually cost this out reasonably well. Or, if the Minister says, “I got us another $50 million this budget round”, staff can provide options about various ways one could use that money to either increase aid under existing rationing parameters, or tweak the rationing parameters, or both.
But remember, we’re designing a program from scratch and that creates a completely different set of problems. How, exactly, do you cost a program if you have no prior experience of running such a program? The answer is “it’s extremely tricky”. In Africa, governments often just set criteria and hope for the best. But often they a backstop measure, which is simply to impose a hard limit on spending each year. In Canada (and many other OECD countries), this is unthinkable because we inscribe our rationing rules in legislation and regulation, meaning that once the rules are set, governments are required to hand out however much money the rationing formula requires. If that amount turns out to be too much, governments can alter the criteria for the following year, but in the current year there isn’t a lot anyone can do. Developing countries, on the other hand, often use first-come, first-serve rules and at a certain point just say: “sorry, we’re out of money”. It’s very rough and harsh justice, but it works (n.b. because of the very odd rules which governed the Millennium Scholarship Foundation – requiring it to spend a more or less fixed amount of money in a fixed number of years, it broadly had to work this way, though it was integrated with provincial systems in a way that largely disguised that this is what it was doing)
So, after four days of this, you may well be saying to yourself: Good Lord this is complicated. And yes, it is. Absurdly so. Aside from pensions, student aid is the only policy field where you have to work out both how to distribute funds and how to collect them (just in reverse order), and it has a stronger redistributionist flavour than pensions.
If you’ve managed to read this far, you’re probably a serious policy nerd, and so you can probably see why this makes student assistance so much fun. The amount of detail over which one can and should geek out while mastering these files is immense. It’s wonk heaven, really, and I have always felt quite privileged to be able to work in it. I hope after the last four days you can see why.