University Challenges
The Trilemma of Higher Education Policy in
Advanced Industrial States
Ben Ansell
Harvard University
This paper analyzes the politics of higher education policy in advanced industrial states, examining both partisan preferences over spending and three different institutional forms of higher education provision: the Anglo-American, Continental, and Scandinavian models. The paper applies the well-known ‘trilemma’ framework developed by Iversen and Wren (1998) to the area of higher education policy. I demonstrate that the logic of the trilemma can be extended to a three-way trade-off between the extent of coverage, the degree of subsidization, and the overall cost of higher education. The Anglo-American model leads to a mass, partially private, and publicly inexpensive system. The Continental model leads to an elite, fully public, and inexpensive system. Finally, the Scandinavian model leads to a mass, fully public, but highly expensive system. This set of choices facing both governments and voters is modeled formally and the derived trade-offs tested on a panel dataset of OECD states from 1990 to 2000. Three case studies, of the United Kingdom, Sweden, and Germany are then analyzed in greater detail to show the dynamic partisan trade-offs facing states as they expand higher education. The paper concludes by offering some suggestions about the long-term impact of these decisions on the likely success of newly-elected Continental right wing parties in achieving their education and labor market goals.
Presented at the Midwest Political Science Association Annual Conference
April 2006, Chicago
Please do not cite without author’s permission ()
In contemporary Europe, the rhetoric of both the Left and the Right appears to have converged on a policy of creating a flexible, high-skilled workforce so as to compete with newly industrialized states in a globalizing world. However, this strategy can only work in combination with a mass higher education system that provides these flexible high-end skills. Scandinavian and Anglo-American states have moved firmly towards this model over the past two decades, albeit adopting different strategies to this end. Continental states, however, remain wedded to an elite model of higher education, which has led to an over-whelming reliance on specific, trade-sensitive jobs at the intermediate level. Moreover, the current set of conservative governments in these states appears unwilling to facilitate this expansion since this hurts their core constituency of the upper middle class. Conversely, in the Anglo-American and Scandinavian systems, expansion to a mass system was made possible. However, even in these systems, the partisan profile of government determined the strategy that was chosen. Ironically, traditionally neoliberal England facilitated expansion through Labour’s fiscally progressive strategy of tuition fees, whereas traditionally social democratic Sweden facilitated expansion through the Moderate coalition’s fiscally regressive decision to expand public funding.
This paper develops the argument that higher education policy in the OECD is driven by a set of partisan choices within a ‘trilemma’ between the extent of coverage, the degree of subsidization, and the overall public cost of higher education. Consequently, states can achieve a maximum of two out of the following three points: a mass higher education system, a fully publicly subsidized system, and a system that keeps public spending on higher education less than around 1.5% of GDP. For a fixed budget constraint, moving from an elite to a mass higher education system requires a large reduction in the degree of public subsidization, requiring a large injection of private financing. For a fixed level of subsidization, expanding from an elite to a mass system means breaking through the budget constraint. Thus some countries, like England, will choose mass, partially private, inexpensive higher education systems. Other, like the Scandinavian states, have mass, fully public, but expensive higher education systems. Finally, Continental countries, have inexpensive, publicly funded, but elite higher education systems.
Partisan politics drive any transition between these three types of system. In many states, governments have been unable to prevent a huge increase in enrolment since the 1980s, typically because they have funding formulae that prevent quotas. However, even during this transition from an elite to a mass system, political choices emerge over how to fund such a change. A transition from a Continental to an Anglo-American system is more likely to emerge under left-wing governments since tuition fees, with accompanying grants for low-income families, constitute a more progressive use of public funds than fully subsidized higher education. Changes from a Continental to a Scandinavian system are more likely under center-right governments, who can channel this increased funding to the upper middle class. Even if, as in the Continental states, governments can prevent the expansion of the higher education system (for example, through quotas) partisan politics will still matter. Left-wing governments in Continental countries will look to expand access to the working class and to channel money more generally away from higher education to primary and secondary education. Right-wing governments in Continental states will instead look to limit further enrolment, will reject fees, and shift funding from lower levels of education towards higher education. Thus, it appears that an electoral ‘turn to the right’ in Continental Europe will lead to neither tuition fees nor expansion of the university system.
To develop this argument, this paper proceeds in three steps. I begin by examining the preferences of left and right-wing parties over funding for higher education. We shall see that right-wing parties typically prefer to fund higher education vis-à-vis primary education. This assertion is empirically tested on a set of twenty OECD countries in the 1990s. I then move to a more aggregated analysis of the differences between three types of higher education system: the Anglo-American model of partially privatized, mass higher education; the Scandinavian model of fully public, mass higher education; and the Continental model of fully public, elite higher education. I develop a formal analysis of these trade-offs at both the aggregate level and in terms of the preferences between these three systems of different individuals. I then conduct an empirical analysis of how far this trilemma is manifested in the empirical trade-offs between coverage, subsidization, and overall public cost. The paper then moves to a case analysis of the politics of higher education reform in England, Sweden, and Germany. The paper concludes by arguing that the recent ‘turn to the right’ in Continental Europe may have little success in reforming higher education.
Section Two: Partisan Preferences and the Composition of Education Spending
Though education is one of the rare policies that is lauded by politicians of all partisan stripes, in fact there are clear partisan patterns both to promises over increased spending and to actual aggregate investment in education (Ansell, 2006b; Boix, 1998). Since universal public education implies the granting of a uniform good funded through a progressive tax system, the poor end up receiving absolutely more than they put in, with the rich paying out more than they receive in return. Hence, education at its core looks very similar to a standard Meltzer-Richard public good model, where redistribution is more likely when the median income voter is poorer vis-à-vis the mean income (Meltzer and Richard, 1984). Of course, there are a variety of other important economic forces (let alone socio-cultural factors) at work in determining the impact of education on individual utility, for example, education may alter the future income distribution, or it may produce large externalities. Still, the basic redistributive logic of aggregate education spending is powerful, and is apparent in a broad variety of studies linking regime type or partisanship to spending (see, for example, Lake and Baum, 2001; Brown and Hunter, 2004, Lindert, 2004, Ansell, 2006a).
However, the logic of higher education spending reverses that of aggregate education spending. Unlike primary and most secondary schooling, university education has traditionally been limited to less than half of the population and access has been correlated with parental wealth. For example, even by 2000, fewer than twenty percent of students with parents in blue-collar professions went on to university in the UK, as compared to seventy percent of the children of professionals, managers, and associate professionals (DfES, 2003). Given this distortion of coverage, we find that where left-wing parties are the advocates of overall education spending, it is right-wing parties, representing voters of substantially higher income, who prefer to increase higher education spending. Thus, where education spending can be targeted, we see a reversal of partisan preferences to the general case of universal provision.
The logic behind this assertion is simple. If a right-wing government comes to power and they face a fixed budget but no constraints on the composition of that budget, we would expect their preference to be to target education funding towards their own constituents, and since right-wing voters are more likely recipients of higher education than left-wing voters, they will bias spending towards higher education. In fact, public funding of higher education is generally a highly regressive form of public spending, the clear beneficiaries of which are the children of the wealthy; even though there may be more general externalities that recommend at least some public funding of higher education. Thus, for a given budget we might expect right-wing parties to take resources away from the primary education sector and transfer them to the higher education sector and vice versa for left-wing parties. In fact, right-wing parties spend generally less on aggregate education than do left-wing parties. Thus, budgets are not, in practice fixed, and the change in the relative funding of higher and primary education may come from freezing the former and cutting the latter. Similarly, left wing parties typically increase overall spending on education but will choose to target increased funds at primary and secondary education rather than higher education (as was the case during Tony Blair’s first two terms – see Section Five).
The case of secondary education is somewhat more complex than those of tertiary and primary education. The latter is clearly universal in all OECD countries, whereas in many states tertiary education is provided to a minority of the population. However, while all students in the OECD will obtain at least some secondary education, almost all states limit compulsory education to the age of sixteen, two to three years before tertiary education begins. Thus, while all students receive secondary education between the ages of eleven and sixteen, a significantly smaller proportion will continue on in upper secondary education between sixteen and eighteen / nineteen. This proportion varies wildly across even the wealthiest countries in the OECD – with the UK losing around one third of students to the workforce or unemployment at age sixteen and Finland retaining all but three percent. This inconsistency means that secondary spending is not directly comparable across OECD countries and thus neither is the degree to which secondary spending is a universal or targeted form of spending. As we shall see, this weaker theoretical prediction about secondary spending will meet with fairly weak findings in empirical tests on the ratio of tertiary to secondary spending, at least in terms of the full sample. However, our expectations about the ratio of tertiary to all other forms of education, and in particular, tertiary to primary, should be that right-wing parties will target funding towards tertiary where they are able to do so.
Table One examines whether the partisanship affects the composition of education spending across an array of twenty OECD states during the 1990s. Model A uses tertiary spending over primary spending as its dependent variable, Model B uses tertiary spending over primary plus secondary spending, and Model C uses tertiary spending over secondary spending. To measure the effects of partisanship I use the cabinet center of gravity index constructed by Thomas Cusack and Lutz Engelhardt from Budge et al’s (2002) ‘Comparative Manifestoes’ dataset. This variable takes Budge et al’s codings for the ideology of each political party in each OECD state, as measured by the composition of their manifesto statements. Cusack and Engelhardt then adjust this measure to reflect the partisan composition of cabinets in each country-year, weighting each member of a coalition appropriately. This provides us with an index ranging from minus one hundred (the most left-wing cabinet possible) to positive one hundred (the most right-wing cabinet possible) with an average partisan range across countries of around fifty points. I also control for a variety of economic and demographic characteristics that might impact the composition of education spending: overall spending on public education; the proportion of the population under fifteen; GDP logged; the square of logged GDP; logged population; all other non-education government consumption; and a linear time trend. I also include a lagged dependent variable to capture temporal dependence in the data.
To estimate the impact of changes in cabinet partisanship on education composition, I conduct a set of OLS cross-sectional time-series regressions, using country-fixed effects. These fixed effects mean that we are solely examining the within-country effects of partisanship (and, indeed, the other independent variables) on the composition of education spending not on the difference between, say the Social Democrats in Sweden and Labour in the UK. Unfortunately, data availability for education composition is fairly poor, with in total 106 observations, or around five per state. This is reflected in the relatively small coefficient on the lagged dependent variable, which ranges between .35 and .47 and in the general statistical insignificance of most of the control variables in the analysis.
Nonetheless we do see a significant effect for cabinet partisanship, even in these rather unpromising data conditions. Model A shows the strongest and most statistically convincing effect of cabinet partisanship. In this analysis, a one point shift in partisanship is associated with a 0.007 increase in the ratio of tertiary to primary education spending. Given that the within-country standard deviation of this variable is 0.98, this implies that a fifty point shift from left to right on the cabinet center of gravity scale (the average range across countries) would lead to an increase of 0.35 in the tertiary / primary ratio, or just over a third of one standard deviation. The potential impact of this shift can be interpreted visually in Figure One, an added variable plot of the fixed effects analysis.[1] A fairly tight slope of moderate incline indicates a robust and substantively significant relationship between cabinet partisanship and the tertiary / primary ratio. The mean ratio in the dataset is 2.05 – thus a shift from left to right of fifty points would mean a seventeen percent increase in the ratio. If we include long-term effects, these predicted first differences would double, leading to an increase in the tertiary-primary ratio of two-thirds of a standard deviation.