Investment in Tertiary Education: Main Determinants and Implications for Policy
 |
Abstract |
|---|
 Top Abstract 1 Introduction2 Cross-country differences in...3 Structural and policy...4 Supply side: the...5 Demand side: the...6 Financing the individual...7 Explaining aggregate...8 Policies to enhance...9 Summary and conclusionsReferences |
|---|
Many OECD countries are aiming to reform their tertiary education (TE) systems. This work explores the determinants of the investment in TE, with a focus on institutional setting of TE systems and private incentives to undertake years of schooling beyond upper-secondary degree level. For this purpose the article first develops estimates of three main drivers of graduation patterns, namely institutional arrangements of TE supply, availability of funding for TE students and private returns to tertiary studies. Second, the article empirically assesses how these three factors affect graduation ratios. Based on this analysis, the article then discusses policy-levers of TE investment and explores possible routes of reform for TE systems in OECD countries. The main findings are as follows: graduation ratios increase with private returns to TE as well with the autonomy and accountability of the supply of education. Lack or insufficient financial help to tertiary students negatively affects graduation ratios. There is a number of policy-levers to stimulate investment in TE. They include policies affecting labour market premia, the degree of flexibility of TE provision and the availability of funding for students. (JEL codes: I21, I22, I28, J24)
Key Words: Investment in tertiary education • returns to education • supply of tertiary education
|
1 Introduction |
|---|
|
TopAbstract1 Introduction2 Cross-country differences in...3 Structural and policy...4 Supply side: the...5 Demand side: the...6 Financing the individual...7 Explaining aggregate...8 Policies to enhance...9 Summary and conclusionsReferences |
|---|
Tertiary Education (TE) is a key asset in knowledge-based economies: tertiary educated workers stimulate economy-wide productivity and growth, and are crucial for innovation and the use of new technologies (Aghion and Cohen 2004; Vandenbussche, Aghion and Meghir 2006). The role of higher education has often provided the case for massive public funding and regulation of this sector in many OECD countries. Yet, rising dissatisfaction with the performance of TE outcomes in a number of OECD countries has increasingly questioned the scope and the forms of public intervention. Calls for reform have been motivated by low quality and the excessive duration of studies, the substantial drop-out and the loose matching between qualifications and labour market-specific needs (Jacobs and van der Ploeg 2006). In this context, OECD countries face two main related challenges: how to make the most of public expenditure in TE and how to increase the overall resources invested in TE without further hinging on the public sector.
Against this background, this article aims at informing TE reform on several aspects. First, it documents TE outcomes, including the labour market rewards accruing to graduates. Second, it explores how policies and institutions affect private incentives to invest in tertiary human capital, the ability of individuals to finance this investment and the characteristics of university systems. Third, it provides some illustrative discussion of the possible routes of actions to reform TE system. In particular, the article provides guidelines on the governance of TE institutions and argues for flanking policies to the possible increase of private participation in the sector, with the objective of preserving or enhancing equality of access to higher education.
In assessing how policies can affect accumulation of tertiary human capital, the article draws on the extensive economic literature on the determinants of investment in TE. Traditionally, this literature has focused on demand-side determinants of investment (e.g. Becker 1967; Freeman 1986; Heckman, Lochner and Todd 2005) and, more recently, on the role of the supply structure (e.g. Rostchild and White 1995; Epple, Romano and Sieg 2006). Along these lines, we develop a number of indicators measuring the main demand-side investment determinants, namely the private rates of return to TE and the availability of individual financing. The role of TE supply side is assessed through specific indicators built to capture selected features of the institutional set-up of TE sector, such as the degree of autonomy, flexibility and accountability of universities. The article then tests for an empirical relationship between investment in TE, as measured by graduation rates, and its main demand-side and supply-side determinants. In this context, various issues are explored, as for instance, the relationship between short-term incentives to undertake higher education and the long-run feedbacks from the labour markets.
Main findings of the article are as follows: countries with incentive-based TE systems (i.e. characterized by higher educational input and output flexibility and higher accountability) display higher graduation ratios than countries with centralized and administrative-based systems. Private incentives to invest, measured by internal rates of return (IRR) reflecting net labour market premia, net replacement income and costs of education, are also positively related to the accumulation of tertiary human capital. High tuition fees do not systematically lead to lower accumulation of human capital, when comprehensive and consistent funding systems are put in place to defray schooling and living costs for students and when side-effects from greater reliance on household sector lead to efficiency gains in TE systems (e.g. through lower study duration and strengthened competition in the TE sector). From these results, the main policy conclusions of this work are that OECD countries with low levels of investment in TE can increase graduation patterns by: (i) increasing returns to education, through specific policy-levers; (ii) making individuals aware of both the cost and future returns of their investment; (iii) further development of individual funding system together with increased private participation in the TE sector and (iv) allowing for more autonomy and enhancing accountability in the TE sector. Reforming TE systems along these lines implies costs and trade-offs with other policy objectives, which vary from country to country and with the mix of policy options retained. While this article does not address the latter issues explicitly, it often argues that consistent and simultaneous policy measures are needed to achieve efficient and equitable TE systems.
The article is organized as follows. Section 1 describes some stylized facts on TE outcomes in OECD countries. Section 2 presents the analytical framework and discusses some pieces of the literature on determinants of graduation ratios. Section 3 discusses the key features of supply of TE in OECD countries and describes a summary indicator. Section 4 presents estimates of internal returns to TE. Section 5 discusses the affordability of TE in presence of financial market imperfections and presents an indicator measuring the availability of funding for tertiary studies. Section 6 empirically assesses the impact of the demand and supply indicators estimated on graduation ratios. Section 7 builds on these empirical findings to draw policy recommendations. Section 8 concludes.
|
2 Cross-country differences in TE outcomes |
|---|
|
TopAbstract1 Introduction2 Cross-country differences in...3 Structural and policy...4 Supply side: the...5 Demand side: the...6 Financing the individual...7 Explaining aggregate...8 Policies to enhance...9 Summary and conclusionsReferences |
|---|
We define investment in TE as the number of new graduates (ISCED-5/6)1 and expressed as a share of the cohorts of age 20–29.2 This measure is harmonized across countries in that graduates are recorded by their highest degree achieved. Thus, it makes it possible to look at the determinants of TE investment in countries with different structure of TE studies (e.g length and type of programmes). However, it is purely quantitative and neither accounts for the quality dimension of the investment, nor for its composition (i.e. TE fields). We choose to focus on the flow of the investment in TE rather than the stock of human capital, since factors other than current policy and TE settings could have a bearing on the latter. These factors are beyond the scope of this study, which is identifying the possible policy-levers of the investment.
During the last two decades, graduation ratios have strongly progressed in the OECD area, particularly so at the turn of the century. The increase in TE investment has been impressive for women, with female graduates almost doubling between 1994 and 2004 (Figure 1). This pattern reflects a likely catching-up with men in terms of the underlying stock of graduates. However, the companion paper Oliveira Martins et al. (2007) shows that the composition of investment across genders is still pretty uneven, with women graduating relatively more in Education, Health & Welfare and Humanities & Arts, while men's degrees being more concentrated in Science and Engineering. The accumulation of tertiary human capital has also been unequal among OECD countries, as shown in Figure 2. Despite a general tendency to increasing TE investment, differences in graduation ratios level remain substantial among OECD countries: New Zealand, for instance, records almost eight times as many tertiary graduates as Turkey and four times as many as Greece. Another interesting feature of graduation patterns over the 90s and early 2000s is that several small OECD countries have recorded stronger increases than big OECD countries with historically high levels of human capital. This is the case of Korea, New Zealand or Ireland, where investment in TE has been higher than in the United States or Canada.
|
|
|
3 Structural and policy determinants of investment in TE: a short literature review |
|---|
|
TopAbstract1 Introduction2 Cross-country differences in...3 Structural and policy...4 Supply side: the...5 Demand side: the...6 Financing the individual...7 Explaining aggregate...8 Policies to enhance...9 Summary and conclusionsReferences |
|---|
The economic literature has put forward several demand-side determinants of investment in TE (see Becker 1967; Freeman 1986 for a seminal review and Heckman, Lochner and Todd 2005, for a survey on recent developments). These include: (i) the standard model where investment depends on the expected returns from an additional year of schooling net of direct and opportunity costs of schooling; (ii) liquidity constraints and financial market failures that prevent individuals from financing their tertiary studies through borrowing; (iii) any cyclical, structural and demographic effects on expected future earnings, not contemplated in the standard model (Card and Lemieux 2000; Heckman, Lochner and Todd 2005); (iv) the disutility of school versus work (Card 2001; Heckman, Lochner and Todd 2005); (v) the quality of TE investment, as a function of peers’ ability and resources specifically directed to enhance quality (Hoxby 2005; Epple, Romano and Sieg 2006); (vi) gender-specific social and behavioural determinants of the investment in TE, including the rise in divorce rates, women's greater responsibility for children, girls’ earlier maturity and higher level of non-cognitive skills (Goldin, Katz and Kuziemko 2006). Some of these determinants can be estimated for OECD countries, as for instance the returns to schooling and liquidity constraints, or at least controlled for (e.g. demographic effects and structural trends); however, due to the lack of data on other demand drivers (preferences, abilities, behavioural determinants), we have to neglect these latter aspects in the analysis.
Concerning the supply determinants of TE, the literature is more recent and the empirical evidence less well-established. There are at least two aspects of this literature that deserve mention. First, there is some debate on whether the Industrial Organization approach can be applied to the supply of TE and in particular, whether objective functions of TE institutions are identifiable (Winston 1999). Second, TE supply is rather heterogeneous among OECD countries, with English-speaking countries closer to a market system while continental European countries being typically administratively based supply systems. In the latter, governments set tuition fees almost irrespective of the production costs and of the quality of students enrolled.3 This is opposite to the experience of TE system in the United States (Hoxby 2005; Epple, Romano and Sieg 2006), where students match universities along quality and quality is a function of both initial students’ level and the resources invested by the universities. Two issues then challenge the use of a standard TE supply-demand framework of the schooling decision. First, as argued above, this would not be relevant for the majority of OECD countries. Second, the data required for the estimation (measures of ability, TE investment by family income level, production costs at TE institution level, etc.) are lacking. Therefore, the approach adopted here consists of explaining investment decisions by its main drivers, without imposing a structural relationship. The following determinants of investment in TE are considered: (i) the institutional set-up of TE systems; (ii) the expected private returns from engaging in TE studies and (iii) individual financing opportunities that are made available to students. We present estimates of these three determinants first and assess their impact on graduation ratios in turn.
|
4 Supply side: the institutional set-up of TE |
|---|
|
TopAbstract1 Introduction2 Cross-country differences in...3 Structural and policy...4 Supply side: the...5 Demand side: the...6 Financing the individual...7 Explaining aggregate...8 Policies to enhance...9 Summary and conclusionsReferences |
|---|
There is a consensus that the performance of TE institutions critically depends on three main aspects: (i) freedom in managing resources and setting objectives; (ii) performance-based allocation of resources; (iii) and incentive-compatible public funding rules (OECD 2003; Kis 2005; Teixeira et al. 2004). Along these lines, we use a summary indicator of supply of tertiary developed by Oliveira Martins et al. (2007). This indicator is based on a survey to OECD member countries and covers three main sub-categories (Figure 3): (i) input flexibility; (ii) output flexibility and (iii) accountability of institutions.
|
Input flexibility measures the extent to which TE institutions are free to allocate their resources and to shape their "production" function. Input flexibility puts together criteria for students’ selection, autonomy to decide on sources and structure of funding (e.g. level of tuition fees) and staff policy (e.g. hiring/firing rules, wage setting, etc.). Output flexibility reflects the capacity of TE institutions to diversify their products and provide educational services as to better accommodate demand, such as the possibility to decide on course content, structure (short-term, part-time, distant learning studies). Possible restrictions to access universities are captured by the degree of regional mobility of students and the existence of numerus clausus for the number of diplomas attributed each year. Accountability summarizes features of TE evaluation and funding. Accountable systems provide incentives to excellence, by allocating resources on a performance basis and by sanctioning unsatisfying outcomes. Accountability is gauged through evaluation rules (independent agency, stakeholders) and funding rules (grand-fathering, input or output based).
A composite indicator is then built by aggregating these three sub-categories using uniform weighting. This composite indicator classifies TE systems as ranging from administratively based (low input and output flexibility, low accountability) to incentive-based systems.
In general, continental European countries are found to have relatively rigid supply systems, while English-speaking countries are more often characterized by incentive-based systems (Figure 4). Many OECD countries are, however, not statistically different from the average, as the 95 percent confidence interval around the point estimate of the indicator would show.4 Exceptions to this are New Zealand, Australia, United Kingdom and Mexico (on the right side of the spectrum) and Greece, Germany, Turkey and France (on the left side of the spectrum).
|
This supply indicator comes with a number of caveats. First, the indicator should not be interpreted as a measure of outcomes, but rather of whether TE systems are endowed with the means to achieve performance and quality. Moreover, while this indicator gathers together many institutional aspects of TE supply, it has some limitations. For instance, it could be less informative for federal countries, such as the United States, Canada, Belgium and Germany, where the organization of TE can differ substantially across local states/regions. In addition, in countries where provision of educational services is market based, incentives to excellence are transmitted through mechanisms other than public funding/evaluation. These market dimensions of accountability are not captured in this institutional indicator.5
|
5 Demand side: the Internal Rate of Return to education and its drivers |
|---|
|
TopAbstract1 Introduction2 Cross-country differences in...3 Structural and policy...4 Supply side: the...5 Demand side: the...6 Financing the individual...7 Explaining aggregate...8 Policies to enhance...9 Summary and conclusionsReferences |
|---|
IRRs are a comprehensive measure of private incentives to undertake TE. While several methods exist to compute IRR (Psacharopoulos 1995), we follow here De la Fuente and Jimeno (2005). They developed an unified framework combining a standard discount method with the estimation of labour market premia on micro-data. In their model, individuals are supposed to choose the optimal level of schooling by maximizing the present value of the expected life-time income, net of costs associated to education. Individual wages are a function of the number of years of schooling and evolve over time at a constant rate given by productivity growth and accumulation of experience. Individuals are entitled to unemployment benefits when they are unemployed. At the end of their working life, they receive some retirement benefits according to statutory replacement rates. Individuals pay taxes on wages, unemployment benefits and retirement income. In this context, the profitability of pursuing education beyond the upper-secondary degree is measured by the ratio comparing marginal benefits from TE to marginal costs6 (Boarini and Strauss 2007, for a detailed presentation of the formula used to compute IRR). Marginal benefits comprise a net wage premium, a net employability premium and a net pension premium. The marginal costs are given by the opportunity costs and the direct costs of TE.7
The various components of the IRR are either estimated on individual-level data by multivariate regressions8 (labour market premia) or drawn from various OECD tax and benefit models. IRRs are computed for 21 OECD countries, between 1991 and 2005 (but with unbalanced time-coverage) and separately for men and women.
Wage and experience premia are estimated trough Mincerian wage equations, where the log of gross hourly wage is regressed on educational attainment (gender-specific), number of years of experience in the labour market, working in the public sector, working part-time, tenure, having an indefinite-term contract, size of the company, right qualification for the job, gender and marital status.9 Wage equations are estimated by country with repeated cross-sectional OLS. Gross wage premia are found to vary substantially across countries; in 2001, for instance, they ranged from 27 to 92 percent (Figure 5). In 2001, women's tertiary wage premia were higher than men's (positive interaction coefficient) in 9 of 21 countries, the difference being significant in Poland and Portugal. By contrast, male graduates appear to yield significantly higher wage returns than their female counterparts in Australia, Austria, Finland and Italy. The experience premium per year of accumulated labour market experience also shows large cross-country variation. It is the lowest in Germany (0.23 percent) and the highest in Switzerland (1.69 percent). Overall, the TE wage premia are found to be fairly stable over time (Boarini and Strauss 2007).
|
The gross employability premium and the conditional probability of employment are estimated in a two-stage approach, which jointly determines employment and participation probability. The two-stage approach enables to correct for a possible selection bias, with the probability of participation being first estimated as a function of a range of individual characteristics and its residual used as a control in the estimation of the employment equation. In particular, a two-stage Probit model is used where the probability of being active is estimated as a function of educational attainment (gender-specific), age (quadratic), gender, marital status, having children and being a discouraged worker because of persistent unemployment; and the probability of being employed is regressed on the same variables with the exception of having children and the region of residence in addition.10
As for Mincerian equations, the two-stage estimation of employment and participation equations is done by country and by year. We find that education increases both the probability of participating in the labour market and of finding a job. In 2001, the estimated conditional probability of employment for an upper-secondary degree holder was around 92 percent for women and 95 percent for men in most countries. With a TE degree the employment probability increases on average by around two-percentage points (Figure 6).11 The largest employability premia (above 3–4 percentage points) are found for men in Italy,12 Poland and Canada, and for women in Hungary, Finland and Sweden. Small (or even negative) effects are found for men in Ireland, the Netherlands, Belgium, Switzerland and France, and for women in Spain, Switzerland, Luxemburg and Italy. As showed in Boarini and Strauss (2007), the employability premia display stronger cyclical sensitivity than the wage premia.
|
The next main components of IRR are fiscal parameters, comprising tax rates, out-of-work-benefits and retirement benefits. Both progressivity and the level of fiscal parameters matter. The additional schooling-related net labour earnings and unemployment benefits can be decomposed into a marginal factor (additional net hourly wage/unemployment benefit, for a given (un)employment level) and an average factor (higher employment probability for a given average net hourly wage/unemployment benefit). The marginal and average tax rates (on labour earnings, unemployment benefits, retirement income) as well as marginal and average out-of-work and retirement replacement rates take as reference the earnings of upper-secondary degree holder. The fiscal parameters are proxied by the rates applying to workers at 100 percent of Average Earning (AE), as defined in various OECD tax and benefit models.13
Among the cost components of the IRR, opportunity costs are defined as the sum of after-tax labour market earnings and after-tax unemployment benefits (respectively weighted by employment and unemployment probabilities) for upper-secondary degree holders. Direct TE costs include tuition fees and other education costs (e.g books), but exclude student living costs. They are measured as the share of (total) annual expenditure per student in TE borne by the private sector and net of possible public subsidies earmarked on tuition fees.14
These various ingredients are put together to compute IRR for several years and for the 21 OECD countries covered (Table 1). We find that IRR vary from over 4 to over 14 percent in 2001. The average return (across both countries and gender) is 8 1/2 percent, which is slightly lower than previous OECD estimates but still substantially higher than long-term real interest rates. The range of returns for women is somewhat wider than for men (from 4.2 to 14.4 percent versus 4.9 to 11.8 percent). By ascending order, Italy, Spain, Sweden, the Netherlands, Germany, Austria, Hungary, Belgium, Greece and Finland have below OECD average returns. In these countries, low IRRs are driven by below average net labour market premia, and not compensated by low direct and/or opportunity costs. Moderate and above OECD average IRR are found in Canada, France, Poland and Denmark, where labour market premia are around the country average. Finally, the United States, Australia, Luxembourg, Switzerland, the United Kingdom, Portugal and Ireland have the highest returns because these countries have the highest wage premia, reinforced either by high employability premia and/or low costs of education. The cross-country cross-time average IRR is found to be slightly above 8 percent both for men and women. IRR vary more across countries than over time. IRR are indeed relatively stable, with the OECD average IRR slightly increasing between 1994 and 2001. The strongest upward trends are observed for Ireland, Portugal and Canada. Conversely, UK displays a downward trend, especially at the end of the observed period.15
|
A sensitivity analysis shows that the main positive drivers of IRR are wage premia, average tax rates and employability premia, while the main negative drivers are marginal tax rates, tuition fees and study duration. Many are, thus, the policy-levers of private returns to TE, and governments may use some of them as to set individual incentives to invest in TE.
|
6 Financing the individual investment in TE |
|---|
|
TopAbstract1 Introduction2 Cross-country differences in...3 Structural and policy...4 Supply side: the...5 Demand side: the...6 Financing the individual...7 Explaining aggregate...8 Policies to enhance...9 Summary and conclusionsReferences |
|---|
If financial markets were perfect, students were risk-neutral and perfectly informed about returns to education, the latter should be perfectly correlated to individuals’ schooling decisions. However a number of these conditions are not fulfilled in many OECD countries, as argued by Barr (2001). On the supply side, the imperfections are mainly related to asymmetric information on students’ abilities and motivation, the uncertainty about their future income and the lack of collateral. On the demand side, students engaging in higher education are not always in possession of all relevant information on future labour market prospects (Romer 2000) and may display some aversion to undertake a risky investment, particularly so if they come from low-income families (Chapman 2005).
For all these reasons, we need to introduce in our analysis a proxy of availability of funding which might explain why individuals do not enroll in tertiary studies even when the private incentives to do so are considerable. The literature has variously dealt with the impact of financial constraints on the accumulation of human capital in different ways. Evidence is available, especially at individual-level (see for instance Cameron and Taber 2000), while few studies explore the impact of financial restrictions on the accumulation of human capital at aggregate level. As to the latter, Benhabib and Spiegel (2000) approximate financial constraints by the Gini Index (at country level), the (under) development of financial markets and the (low) share of banking sector in total assets. They find that only the latter variable robustly explains the accumulation of human capital. For the sake of our analysis, however, it is preferable to build a more specific measure on the affordability of tertiary studies, which accounts for the current financial conditions of access to TE. OECD countries funding systems are extremely heterogeneous with respect to rules, coverage and actual take-up rates of grants/loans. For descriptive purposes, systems can be classified along two dimensions: the target and the composition of funding (see Oliveira Martins et al. 2007 for a detailed discussion of the typology). Funding systems are either targeted on students themselves or on the households where students live. In addition, funding systems can either rely on loans, grants or some other measures (tax-credits, family allowances). Many OECD countries offer several types of funding, but we decided to classify countries according to the predominant funding option. Universal funding systems rely on either grants or loans. The main feature of these systems is that they provide universal funding to students as individuals, i.e. irrespective of family-income conditions. By contrast, family-based funding systems generally offer limited help for studying, and the help is conditional on mean-test at family level. Within individual-based systems, loans and grants schemes are quite diverse, especially with respect to the magnitude of public subsidies available (OECD 2006b).
While it is possible that the composition of funding matters for easing liquidity constraints over and above the overall amount of funding available, it is not possible to build a summary indicator reflecting the differential impact of grants and loans on financial restrictions for accessing TE. Essentially, this is due to the paucity of data on the actual repayments of loans (regardless of statutory rules) and of the sometimes awkward functioning of borrowing schemes, which makes it difficult to conjecture about the real part of subsidy in attributed loans. For the sake of our analysis we thus build an indicator of overall availability of funding, which regards grants and loans in the same way. This indicator is defined as the ratio of the overall costs of TE to the total resources available to students to cover these costs. Costs include tuition fees (average of public and private sector) and an estimate of living costs for students (Table 2). Resources made available to students are the sum of specific funding in the form of grants and loans, an estimate of students’ earnings from part-time work and an estimate of resources available at household level. Students’ earnings are assumed to be equal to 80 percent of the wage of an upper-secondary degree holder (working half-time) and adjusted for youth unemployment rate. Private resources to fund FE are assumed to be equal to the equivalized median household disposable income. The indicator ratio of funding availability (shown in last column of Table 2) ranges from less than 20 percent for Nordic countries to 135 percent for Mexico. While little informative in absolute terms, this indicator allows for meaningful cross-country comparisons of the extent to which the access to TE is likely to be restrained even for someone living in a family with median incomes. Results on this indicator show that financial constraints are the least (most) binding in individual (family)-based funding systems.
|
|
7 Explaining aggregate investment in TE |
|---|
|
TopAbstract1 Introduction2 Cross-country differences in...3 Structural and policy...4 Supply side: the...5 Demand side: the...6 Financing the individual...7 Explaining aggregate...8 Policies to enhance...9 Summary and conclusionsReferences |
|---|
The institutional set-up of TE, the IRR and the indicator on availability of funding to students are now used to explain aggregate graduation patterns in OECD countries. As discussed in the first section, the empirical strategy consists of estimating the investment in TE, without identifying restrictions on the structural determinants. The econometric analysis assumes first that individuals take the decision to invest in TE on the basis of a pre-determined return to tertiary degrees and for a given quantity of tertiary educated graduates. In a second step, the returns to schooling are let to be endogenous, and modelled as depending on the quantity of human capital demanded and supply in labour markets and on labour market institutions [employment protection legislation (EPL), union density, bargaining arrangements, etc.].
Higher IRR are expected to lead to higher investment in TE while lower availability of funding should have the opposite effect. The institutional set-up of TE sector may shape investment through several channels. Highly incentive-based systems may attract more students, allowing for more and better services. Moreover, highly accountable TE systems may respond better to the positional component of the demand for higher education (notably through rankings and reputation factors). In addition, more flexible systems, such as those where the educational track allows students to opt out from the educational investment more often than rigid system, are potentially more attractive.16 Incentive-based systems may also induce faster completion of studies and lower drop-out.
A number of other factors, for which we need to control for, may influence the graduation ratios. First, structural trends as the increasing labour participation of women or the increasing demand of high-skilled workers may explain graduation ratios. More generally, all structural and cyclical components of return on skills not comprised in the baseline calculation of the IRR could be retained as additional explanatory variables. In the baseline, we adopted a parsimonious specification where graduation ratios are solely determined by IRR, supply conditions, availability of funding, gender effects and output gaps. This specification also controls for common time-dummies and country-specific trends. The analysis is performed in an unbalanced panel using 19 countries17 and gender as the cross-section dimension. The maximum time span covered is 1992–2002, but for several countries only some years are available.
Table 3 shows the results for this specification (column "Baseline") and of a number of alternative specifications (column 1–8), controlling for other explanatory variables. In the preferred specification, the explanatory variables have the expected sign and are significant: graduation ratios increase with IRR,18 flexibility and accountability of TE supply and with availability of funding. Graduation ratios are found to be negatively affected by the output gap, possibly reflecting relatively better employment and income perspectives for non-graduates during periods of strong economic activity (not captured by IRR). As suggested by the effect of the female dummy, graduation ratios are generally higher for women than for men. This seems to be in line with Goldin, Katz and Kuziemko (2006), for which, in the United States, women graduates outnumber men graduates because of changing social patterns and behavioural differences19 across genders and increasing expectations in terms of high-paying careers.
|
In order to test for the sensitivity of these estimates several other specifications are tested. To avoid multi-collinearity, each additional variable is introduced separately. In particular, two other proxies of financial constraints are retained: a categorical dummy for countries with family-based systems (the reference being individual-based system) and the incidence of part-time student work (see OECD Education at a Glance, indicator C4.2a). Relying on family-based financing systems (col. 1) tends to depress graduation ratios, probably because family-based funding systems tend to deliver less generous financial help than individual-funding systems, and possibly because they give less responsibility to students on completing their studies. The coefficient of the students’ part-time work variable has a positive sign (col. 2), suggesting that liquidity constraints are somehow relaxed when students work.20 We also find that countries with high share of students in private universities display higher graduation ratios (col. 3). This may be explained by private universities being more autonomous than public institutions,21 but could also be due to that they sometimes offer easier curricula. Both the Programme for International Student Assessment (PISA)22 mean score and its standard deviation have a positive influence on TE investment (col. 4 and 5). The former effect is not surprising: better-prepared students from the primary and secondary-level are expected to enrol and to complete tertiary studies more often. The impact of PISA standard deviation is less intuitive. One possible interpretation is that countries where pre-tertiary systems are less comprehensive (i.e. less egalitarian) tend also to have more selective TE institutions, which lead to higher graduation ratios. Another explanation could be that countries where PISA scores are distributed more unequally, less able students do not make it to the university or to the completion of studies. Not surprisingly, the size of upper-secondary graduates’ cohort is also positively related to tertiary graduation ratios (col. 6). Graduation ratios depend positively on the share of foreign students (col. 7). This could be due to the positive correlation between ability and mobility, and to related peer effects. Per student expenditure23 as a share of GDP per capita is found to be positively related to graduation ratios (col. 8). In principle, the direction of influence is ambiguous since expenditure per student could capture both the input price of factors24 and input quantities invested in the production of educational services. Since the expected correlation to graduation ratios would be, respectively, negative and positive, the empirical finding would rather support the interpretation of this variable in terms of input quantities.
In a second step, the assumption of a pre-determined IRR is relaxed.25 Two issues are considered here. First, returns to investment may fall with a rising number of tertiary graduates. More specifically, an increasing supply of tertiary graduates is likely to put downward pressure on gross wage premium. In turn, a lower wage premium reduces the incentives to invest in TE. Second, labour market policies and institutions influence wage dispersion and hence may also affect incentives to invest in TE. To take into account the potential simultaneity issues, we estimate a system of two equations modelling the investment in TE and the determination of the relative price of skills.26
The first equation is the same than the baseline specification estimated above, while in the second equation gross wage premia are regressed on the log of tertiary graduation ratio, the lagged stock of human capital, EPL for temporary workers, EPL by type of contract (temporary and regular), the degree of bargaining coordination and the trade openness of the economy. Indeed the specification of a wage premia equation follows a standard approach accounting for labour supply and demand forces, as well as labour market policies and institutions (Katz and Autor 1999). The wage level can be seen as the result of a competitive wage plus or minus a deviation due to either labour market institutions or measurement problems.27 The competitive wage is a function of the relative supply of skills (decomposed into the lagged levels of the tertiary and secondary human capital stocks and inflows of tertiary human capital). The relative supply of skills is let to be endogenous in the estimation procedure. Deviations from competitive wages are proxied by a number of labour market institutions; in the specification shown here these are the degree of coordination of wage setting negotiations and EPL. The equation also controls for (exogenous) relative demand shifts arising from trade openness, and include country-specific time trends and common time dummies. The estimation is based on the standard three-stage least square estimator (3SLS). We find that the impact of determinants of graduation ratios is relatively robust to the assumption of exogenous IRR (see the last two columns in Table 3). We also find that the impact of labour market institutions on wage premia turns out to be considerable. EPL for temporary contract is found to have a negative impact on wage premia while the opposite is obtained for EPL for regular workers. Coordination bargaining is found to have a negative impact on wage dispersion, corroborating previous evidence in the field (Barth and Lucifora 2006).
|
8 Policies to enhance TE outcomes |
|---|
|
TopAbstract1 Introduction2 Cross-country differences in...3 Structural and policy...4 Supply side: the...5 Demand side: the...6 Financing the individual...7 Explaining aggregate...8 Policies to enhance...9 Summary and conclusionsReferences |
|---|
Shortage of skills is as a serious handicap for the growth potential of economies. Low efficiency of public expenditure on TE is another issue, considering the large amount of public resources allocated to TE in some countries and the relatively modest outcome of this investment. Last but not least, rising mobility and labour market integration across OECD countries raise new challenges for TE systems, with TE institutions increasingly competing for scholars, students and resources on a global scale. Reforms of TE systems are thus on the current agenda of many OECD countries, with the two-fold objective of making TE systems respond to labour market dynamics of global economies and obtaining the most from the public financial effort in the sector.
With regard to the latter, we have argued in previous sections that TE institutions may not operate in incentive-compatible settings in countries where TE systems are little accountable and autonomous.28 Reforming those systems by putting in place adequate governance settings which allow for an efficient provision of high-quality educational services should be seen as a priority. Getting an efficient TE system might not be enough, however, in a context of increasing search of excellence. Increasing both the quality and quantity of tertiary graduates seems to be hardly feasible without injecting more resources in the system (Teixeira et al. 2004). A widely shared position is that pressure on public budgets makes it difficult to increase public funds. Augmenting the role of the private sector seems viable, not only because in many OECD countries the current level of the private participation is limited but also for the potential efficiency gains that this solution may induce. If more resources are warranted by the household sector, typically through higher tuition fees, students could be made more responsible with respect to completion, quality of learning and fast progression in their studies. Greater reliance on tuition fees may also have positive effect on the supply side: notably tuition fees may transmit signals on the quality of education provided, as well as increase effective competition among universities by pushing institutions to cost-efficient delivery of educational services. Increased financial participation of the industrial sector could be appealing to the extent that it were to allow partnership between firms and universities and improve the matching between the education delivered and skills required by production needs. Finally, equity concerns (Barr 2001; Jacobs and van der Ploeg 2006; Oliveira Martins et al. 2007), such as public spending in TE being regressive and crowding out public resources that could otherwise be earmarked for liquidity-constrained students, suggest that making students paying for education and at the same time introducing appropriate financial help such as income-contingent loans or means-tested grants, could increase both efficiency and equity.
Against this background, the empirical findings of the previous section suggest some avenues for reform. Accordingly, we present suggestive policy simulations on the increase of flexibility and accountability of supply, tuition fees and availability of funding for students. The first simulation consists of benchmarking the summary indicator of flexibility and accountability of supply side on the best performer in the sample (Australia). Indeed, catching-up with the Australian system would require ambitious reforms for a number of OECD countries. The objective here is not to suggest drastic policy changes but rather to give a flavour of the room of high-potential returns to policy action in the field of TE. This simulation suggests that countries such as Greece, Germany and France would benefit the most from reforming the supply of TE (Figure 7).
|
In the second simulation, tuition fees are increased to the sample mean plus two standard deviations (around 4,000 US$ at PPPs). This policy shock is considerable in all countries where tuition fees are nil (e.g Nordic countries, Greece, etc.) and negligible in countries which already feature high tuition fees (Australia, United States). The increase in fees has a two-fold negative effect (Figure 8). The first operates through a fall in the IRR (as direct costs go up), while the second, much stronger effect, works via stronger liquidity constraints (assuming unchanged individual-funding systems). The cumulated negative effect can be large in absolute terms. This result suggests that an increase in tuition fees may call for other flanking policies.29 Given that the main effect relates to increased liquidity constraints among possible compensating policies, a natural candidate is the development of individual financing. Indeed, countries introducing or raising tuition fees have usually taken simultaneous action in this field, as in Australia, New Zealand and the United Kingdom.
|
To assess the beneficial effect of flanking policies to the greater reliance on tuition fees that TE system may opt for, we simulate the impact of aligning the ratio of costs to financing resources (Table 2 above) to the minimum in the sample. The impact ranges from nearly 1.5 percentage points in Portugal and Spain to virtually zero in Denmark and Finland (Figure 9). The simulation results are clearly more relevant in the case of family-based systems, where liquidity constraints are likely to be more binding. However, insofar as reforms of universal funding systems involve use of tuition fees, easing liquidity constraints will have a positive impact in those systems, too.
|
|
9 Summary and conclusions |
|---|
|
TopAbstract1 Introduction2 Cross-country differences in...3 Structural and policy...4 Supply side: the...5 Demand side: the...6 Financing the individual...7 Explaining aggregate...8 Policies to enhance...9 Summary and conclusionsReferences |
|---|
In this work we have explored the main determinants of investment in TE and focused on policy-levers of these determinants. OECD countries’ TE systems have been shown to differ across many dimensions: in terms of the quantitative outcome (graduation ratios), the gender-field composition, the degree of flexibility and accountability in supplying educational services, the private returns to tertiary studies and the financial conditions of access to TE. It has also been shown that graduation ratios are shaped by institutional settings of both TE systems and labour markets, as well as by the availability of financial help to students and other structural factors like the increasing participation of women in TE. In this context, we have simulated the potential impact of some reforms in the TE sector. The results suggest that, depending on the existing institutional conditions, there is a strong potential to increase graduation ratios by introduction/strengthening of flexibility and accountability in TE supply. In order to increase the amount of resources injected in the system without having a detrimental effect on the graduation ratios, the increase of tuition fees requires the introduction/extension of individual funding to TE students. Overall, the different TE indicators and their estimated relationships provided in this article can provide some guidance for reforming TE systems in many OECD countries. Future research in this field is needed to cover some of the aspects that were not sufficiently tackled by our analysis, as the role of public versus private funding and management of TE institutions, as well the possible complementarity between policy actions. It would be also of great importance to look at the quality of the investment of human capital, since a quantitative perspective as the one adopted by this article is necessarily restrictive. In this respect, distinguishing between graduation ratios across fields or across different educational programmes would be relevant and worthwhile. Finally, the analysis of the institutional framework of TE supply could be further enhanced by giving consideration to the extent and shape of market competition in the sector.
|
Acknowledgments |
|---|
We would like to thank Jorgen Elmeskov for useful comments, Laura Vartia, Jens Arnold and Clarice Saadi for assistance, as well as Paulo Santiago, Thomas Veko and Eric Charbonnier from the OECD Directorate for Education for comments and data support. The views expressed here are those of the authors and do not necessarily represent those of the OECD or its Member countries.
|
Footnotes |
|---|
OECD Economics Department, 2 rue André-Pascal, 75775 Paris Cedex 16, France. Corresponding authors are Romina Boarini, e-mail: romina.boarini{at}oecd.org and Joaquim Oliveira Martins, e-mail: joaquim.oliveira{at}oecd.org. Hubert Strauss was previously at the OECD Economics Department and is currently at the European Investment Bank.
1 ISCED-5 includes Tertiary-type A programmes and the more vocationally oriented Tertiary-type B programmes. ISCED-6 refers to advanced research qualifications, such as PhDs (OECD 2004b). 
2 See Oliveira et al. (2007) for details about the construction of this variable.
3 See Jacobs and van der Ploeg (2006). The authors also observe that universities have managed to take in larger cohorts of students, which would suggest that supply is fairly elastic.
4 Confidence intervals obtained by random choice of the weights used to aggregate low-level indicators into the sub-category. For details, see Oliveira et al. (2007).
5 For example, higher education institutions in the United States are subject to evaluation by bond rating firms that review and assess the credit-worthiness of institutions, a feature that is not reflected in the summary indicator above.
6 See Boarini and Strauss 2007, for a detailed presentation of the formula used to compute IRR.
7 The main assumptions behind the computation of the stream of benefits and costs associated to education are: (i) the wage premium is an increasing and time-invariant function of schooling; (ii) the experience premium is constant across schooling levels; it is supposed to be a function of potential experience rather than actual years of employment and to grow at a constant rate over time; (iii) the employment probability is an increasing and time-invariant function of schooling; (iv) individuals receive out-of-work benefits if unemployed and pay taxes on either labour income or unemployment benefits. Benefits and taxes are constant over the life cycle; (v) the number of working hours and the length of working life are the same across levels of schooling; (vi) there is no part-time student work.
8 The following household surveys were used: ECHP for 14 European countries; BHPS for the United Kingdom; HILDA for Australia; CPS for the United States; SLID and CNEF for Canada and CHER for Hungary, Poland and Switzerland.
9 See Strauss and de la Maisonneuve 2007, for more details on the construction and interpretation of those variables.
10 See Boarini and Strauss (2007) for more details on the specification and the construction of the variables.
11 This increase is computed as the difference between the estimated employment probabilities for tertiary and upper-secondary graduates, using the coefficients β3 and β5 estimated in the equations above. In this calculation, the other variables are fixed at a reference level (corresponding to a single prime-age individual without children).
12 Employment probabilities refer to the average man/woman for all countries except Italy, where these probabilities are calculated for a woman/man coming from middle-income regions (mostly central regions). This isolates the impact of education on employment probabilities from the impact of idiosyncratic labour market conditions. In fact, Italy is the country where the regional characteristics of the reference individual matter the most for the marginal effect of schooling on the employment probability. For other countries, the marginal effects were computed without specifying the region of residence.
13 Average Earnings are defined according to the new definition of the Average Worker (see OECD 2004a), covering a broad set of industries than before and also including non-manual employees.
14 See Boarini and Strauss 2007, for a discussion of this measure of costs and of alternative measures.
15 For the interested reader, sensitivity analysis and robustness tests on the IRRSs can be found in Boarini and Strauss (2007).
16 See Heckman, Lochner and Todd 2005 for an application of the Option Value Model to the demand for higher education.
17 This includes all countries for which the IRRs were available except Luxembourg and Poland, where the STE indicator is not available.
18 In a robustness check, the basic components of IRR have been entered as separate regressors. We found that graduation ratios increase with wage premia and average tax rates and decrease with marginal tax rates. This is line with the sensitivity analysis of IRR carried out in Boarini and Strauss (2007). The coefficients of the other variables are robust to this specification.
19 For example, Goldin, Katz and Kuziemko (2006) report that grades K-12 boys have higher incidence of behavioural problems; teenage boys display higher (self-reported) incidence of arrests and school suspensions than teenage girls; girls spend more time doing homework than boys.
20 On top of relaxing liquidity constraints, student part-time work leads also to higher IRR, as shown in Boarini and Strauss (2007).
22 PISA (Programme for International Student Assessment) is an international study conducted by the OECD which measures how well young adults of age 15 are prepared for possible later studies or direct entry into the labour market.
23 This includes both private and public spending.
24 The factor price usually includes the cost of academic staff and infrastructures, but this price should also include subsidies paid to students as a result of the customer-input technology used in the production of human capital (Winston 1999).
25 This approach is further developed in Boarini, Nicoletti and Oliveira Martins (2008, forthcoming).
26 The main identifying assumptions are: (i) the relative supply of skills is proxied by the relative stock of human capital in the population, i.e. the labour market participation is assumed to be constant across skills; (ii) there is no lag in the transmission of the price signals, i.e. wage premia and graduation ratios are determined contemporaneously; (iii) the structural relationship between IRR and wage premia is not specified.
27 Measurement problems arise from differences in non-wage compensation across jobs.
28 The article does not provide a direct test of the importance of public financing of tertiary education, though this dimension is partly captured by the financial constraint indicator which measures the extent to which public resources are made available to finance tertiary studies for a representative student in the economy. While we do not attempt to provide an assessment of cost-effectiveness of public spending on tertiary education, our finding such that graduation ratios are affected by institutional set-up of supply suggests that efficiency-enhancing reforms in this sector may be adopted in the form of higher autonomy and accountability of education supply.
29 A simulation where financial help to students is increased together with tuition fees (not shown here) suggests that graduation ratios would still decrease but by much less than in the case where no flanking policies are put in place. However, the effect of higher tuition fees is not zero because of the way the financial constraints indicator is constructed. An increase of tuition fees in the numerator matched by an equal increase of resources in the denominator would still entail an increase of the overall ratio. This could be interpreted as the fact that education cost would rise comparatively to other items, notably living costs, and this would discourage demand for tertiary education. However, if the additional financial help to students were made strictly conditional on paying the additional tuition fees, this should not affect decisions of (potential) students and both flows would be best interpreted as a zero addition to (net) costs, i.e. the indicator would remain unchanged. The financial constraints indicator in this article does not apply a net cost concept due to the lack of data on the breakdown of financial help to students into grants and loans, earmarked for tuition fees or living costs.
|
References |
|---|
|
TopAbstract1 Introduction2 Cross-country differences in...3 Structural and policy...4 Supply side: the...5 Demand side: the...6 Financing the individual...7 Explaining aggregate...8 Policies to enhance...9 Summary and conclusionsReferences |
|---|
-
Aghion P, Cohen E. Education et Croissance. (2004) La Documentation Francaise: Paris.
Barr N. The Welfare State as Piggy Bank: Information Risk, Uncertainty and the Role of the State (2001) Oxford: Oxford University Press.
Barth E, Lucifora C. "Wage Dispersion, Markets and Institutions: The Effects of the Boom in Education on the Wage Structure". (2006) IZA Discussion Paper No. 2181.
Becker GS. "Human Capital and the Personal Distribution of Income: An Analytical Approach". (1967).
Benhabib J, Spiegel MM. The Role of Financial Development in Growth and Investment. Journal of Economic Growth (2000) 5:341–260.[CrossRef][ISI]
Boarini R, Strauss H. "The Private Internal Rates of Return to Tertiary Education: New Estimates for 21 OECD Countries". (2007) OECD Economics Department Working Papers No. 591.
Boarini R, Nicoletti G, Oliveira Martins J. "Investment in Tertiary Education: Does Labour Market Matter?" (2008) OECD Economics Department Working Papers forthcoming.
Cameron S, Taber C. "Borrowing Constraints and the Returns to Schooling". (2000) NBER Working Paper No. 7761.
Card D. Estimating the Return to Schooling: Progress on Some Persistent Econometric Problems. Econometrica (2001) 69(5):1127–60.[CrossRef][ISI]
Card D, Lemieux T. "Drop-out and Enrollment Trends in the Post-War Period: What Went Wrong in the 1970s?" (2000) NBER Working Paper No. 7658.