Just to give away the answer at the outset – I think that there is an important lesson to be learned, because the basic business model of a research university is similar to the business model of a traditional newspaper, in that it relies up the aggregation of two goods, one basically public the other basically private, along with the sale of the two as a forced “bundle.” What kills the business model is when customers figure out how to get one without the other. Thus the shopping mall killed the traditional department store by disaggregating the sale of different categories of goods. The internet is killing the traditional newspaper by disaggregating the different “sections” of the paper. The question for universities is whether they can preserve the “bundle” that they have traditionally been selling, or whether that two will inevitably be disaggregated.

So let me now take a stab at explaining what all that means. Some “goods” that people can produce generate extremely diffuse benefits, and so it is hard to get others to pay for them. A weather forecast is a good example. It takes a lot of time and energy to set up the equipment needed to generate accurate weather forecasts, but once you’ve made the investment, it can be very difficult to get people to pay for the information, simply because it’s so easy for others to reproduce it. Even if you could get a small group of subscribers willing to pay you for your accurate forecasts, they can just tell other people, who can tell other people, and so on. Thus a weather forecast, despite having significant economic value, will tend to be underproduced by private market actors. It is, as economists like to say, non-excludable, or difficult to exercise property rights over.

One way to encourage production of such goods is the well-known mechanism of taxation and government provision. So we tax the entire population (on the assumption that everyone benefits from weather forecasts) and we give some money to Environment Canada, which sets up weather stations across the country, and generates periodic weather forecasts (that everyone else is free to reproduce as they like).

There are certain advantages to this arrangement, but also certain well-known problems (not least of which is that many people just take the benefits they are receiving for granted, and so are perpetually grousing about the taxes required to produce them). The major disadvantage is that centralized state control is substituted for the decentralized production mechanism of the market, which can produce a variety of negative effects (lack of consumer choice, monopolistic behaviour, bureaucratic inefficiency, etc.)

There are, however, other models for the provision of such benefits – one in particular that is underappreciated. The basic idea is that you take a good that is non-excludable (i.e. relatively non-excludable, or very costly to exclude people from), you bundle it up with another good that is excludable and you sell them as a take-it-or-leave it package, at a price that is sufficient to cover the cost of production of both goods. (There’s a biggish economics literature on this, e.g. see here.) This of course must involve some degree of collusion among competitors, or barriers to entry, since you are in effect selling the excludable good at an above-market price. Thus one is in principle vulnerable to a new competitor coming along, who offers to sell the excludable good without tacking on the accompanying non-excludable good (and therefore is able to sell at a much lower price).

To give a concrete example, consider the traditional newspaper. The role that newspapers play in holding politicians accountable, through muck-raking, investigative journalism, and simply reporting on their behaviour, is essentially a public good, in that it generates extremely diffuse benefits. It is also very expensive. Consumers are, in general, not willing to pay for it, since they can easily free ride – either get the news second-hand, or else not even get the news, but enjoy the benefits that come from living in a relatively corruption-free state, under the careful watch of journalists. Even with advertising, there is no way to make a news section of a newspaper pay for itself.

So how have newspapers been able to afford to report the news? Traditionally they did it by bundled the news up with a bunch of other content, which was very cheap to produce, and which many people read. This is why newspapers contain an automotive section, a sports section, a “homes” section, and so on. They then sold the entire thing to advertisers and consumers, as a “package deal.” Crucial to this was an information asymmetry – newspapers gave advertisers a circulation number, but advertisers had no idea how many people were actually reading each section of the paper, and how many people were looking at their ads. We’ve all seen people pick up a paper, pull out the sports section and throw the rest away. Advertisers were generally aware of this phenomenon, but they had no hard numbers about how many people did it.

In summary: in a traditional newspaper, there was huge cross-subsidization of news by other sections, made possible by an information asymmetry that concealed the extent of cross-subsidization from those who were paying the bills.

One way of thinking of the traditional arrangement was in terms of blackmail: if you want your sports section, or your classifieds section, you’re going to have to pay for our newsroom as well. “Bundling” is a nicer way of putting it. The important point is that this was a beneficial arrangement overall, because it allowed for the production of a public good without state involvement. And since the goal of the press is in part to keep tabs on the state, and prevent abuse of authority, it is extremely important to be able to produce this good in the private sphere.

The internet changed all this, by disaggregating the various “sections” of the paper. Apart from the fact that competitors could provide the content of just one section online (e.g. Craigslist), putting the newspaper online also made it possible for advertisers to see exactly how many people were reading what (and how many people were viewing their ads). This meant that newspapers could no longer sell ads based on the circulation figure for the paper as a whole, they had to sell placement in particular sections, beside particular stories, with full transparency about how many people were actually viewing them.

Turn now to universities. There are important similarities between university research and investigative journalism. It produces extremely diffuse benefits, which are relatively non-excludable. Much of it is very difficult to monetize, and even if you could, it is not clear that this would be the most efficient arrangement. (Patents impose huge efficiency losses in subsequent markets, and so it is great to have a model of intellectual production where fundamental discoveries can be put out into the public realm. Apart from that, there are the diffuse benefits produced by basic science, social sciences and humanities, none of which are patentable at all).

So research is pretty much a non-excludable good. How then do we finance its production? Partially through granting agencies, many of them funded through taxation. But we also secure its production by bundling it up with an excludable good, namely, teaching. Unlike research, it is easy to charge the primary beneficiaries of teaching – we keep a list of who is in the class, we take attendance, and we kick out people who are not registered. The university then cross-subsidizes from teaching to research (mainly by paying faculty to do both).

Again, one way of thinking about this arrangement is in terms of blackmail (understood in the nicest sort of way). Basically, the supereducated get together and say, “if you want us to teach, then you’ll have to pay for our research as well.” The important point again is that this is a beneficial arrangement overall, because it allows for the production of a public good without state involvement.

What stops others from “unbundling” the two, offering the excludable good without the non-excludable one? It seems to be two things. First, the prestige of a degree is linked to the status of the faculty, and the status of the faculty is determined by their research. So teaching-only universities have less prestige. When you unbundle the non-excludable good from the excludable one, you get quality degradation in the excludable one (often not real quality degradation, but merely perceived quality degradation – but perceived quality degradation can be just as important, particularly insofar as it affects the signalling value of the degree). Second, you do also get real quality degradation. There’s this thing called the “research frontier” — which is the set of unsolved problem in a field that people are currently working on. With every decade that passes, the amount that you have to know, in order to get to the research frontier, increases dramatically. Professors who are not involved in research are typically able to educate students to the point where they are within 10 years of the research frontier. But there is something of a “last mile” phenomenon, where taking that extra step is the most difficult, because you need to be still reading the journals, and actively involved in research to be able to keep up. Typically only research faculty (not all, some!) are able to bring students right up to the frontier.

In any case, university professors have a tendency to forget which side their bread is buttered on, and so imagine that because their research is so fabulous, that society is willing to actually pay them to do research. As a result, they strive to cut down their teaching as much as possible – something that is individually rational, but not in their (our) collective interest. There has been a lot of commentary about how many courses are now being taught by graduate students and sessionals (although it should be noted that it’s not nearly as bad in Canada as it is in United States). This is driven by a desire on the part of universities to save money, although it’s important to note that, as course instructors, sessional instructors are not much cheaper than faculty. (At my university, in my field, the hourly rate that we pay graduate students is higher than that of many junior faculty.) The difference is that you have to pay faculty to do research, whereas when you hire someone to teach, all you have to pay them to do is teach. So the problem is not that contract faculty are being paid much less for their teaching, it is really that they are not being paid to do research.

I think the basic business model (involving cross-subsidization from teaching to research) still has a lot of life in it, mainly because of the way that the prestige system works. For example, the growing importance of national and international rankings in the choices that students make about where to go to university has the indirect effect of locking them into subsidizing research, because these rankings are primarily driven by faculty research. And I have serious doubts about capacity of the internet to change very much – although in principle it could be a powerful force driving unbundling. I suspect that if the business model does die, it will be through the shortsighted choices made by faculty, who will over time negotiate their way out of more and more teaching, and thereby deprive themselves of the one major bit of leverage they have, when it comes to persuading “society” to pay them to do research.