In such cases, distinction only by educational level would not cover all relevant educational
dimensions. Especially in countries that have a vocational education and training (VET)
system offering high quality training at the secondary level such as Austria, Germany,
Switzerland, or Denmark, distinction by educational level might prove insufficient.
Therefore analyzing educational spillover effects not only by level but also by type of
education is important to catch the whole range of possible spillovers.
This paper contributes to the spillover literaturein two ways. First, we presenta new type
of spillover effects, which we call “reverse spillovers.”We discuss a conceptual background
that establishes differencesin knowledge as the underlyingmechanism for spillovers.Thereby
we define a new type of spillovers that goes in the opposite direction as the “traditional”
spillover.“Traditional”spillovers go from highly educated workers to lower-educated workers
and assume thatlower-educated workershave no additional skills or knowledge that could be
relevantfor higher-educated workers(Moretti, 2004b).We define a new type of spillover thatis
distinctfrom the “traditional”viewon spillovers: reversespillovers. Reverse spillovers occur if
a formallylower-educated workershas skills and knowledgethat is differentfrom but relevant
for a formally higher-educated worker. Thus, we assume that lower-educated workers can
have a different rather than a reduced skill set than higher-educated workers. We extend the
traditional view on spillover effects by arguing that not only the level of education
(i.e. secondaryvs tertiary education) but also theeducational type (i.e. academicvs vocational
education) causes differences in knowledge and thus spillover effects.
Second, we include predictions on the functional form of reverse spillovers in our
hypotheses. In line with Battu et al. (2003), we expect a non-linear relationship. They find a
positive but diminishing return from an increase in the overall educational level of a
workplace. We empirically analyze the spillovers to tertiary education from co-workers with
secondary vocational education like VET.
To test our hypotheses, we use data from the Swiss Earnings Structure Survey (ESS),
a large firm panel that also contains information on worker characteristics. We estimate
Mincer (1974) earnings equations. The ESS is a very good match for our empirical
analysis, as the data set contains information on workers’education and earnings and
allows us to measure education by using educational degrees instead of years of schooling.
In our estimation strategy we consider the potential endogeneity of a firm’s workforce
composition and use two instruments for the number of workers with VET degrees.
Since the traditionof training apprenticesis more widespread in the German-speakingregions
of Switzerland than in the non-German-speaking regions, we use a firm’slocationasan
instrument for the employment of workers with VET degrees. Moreover, we use the number
of higher vocational diplomas awarded in each major region as a second instrument. Higher
vocational diplomas are tertiary degrees that instructors of apprentices often hold. Thus,
the number of higher vocational diplomas reflects the supply of instructors in VET.
Our results show that the effect from an increase in the number of workers with VET
degrees on the productivity of workers with tertiary education is positive but diminishing.
The effect is robust against the inclusion of regional, year, and sector controls, as well as
firm controls. Furthermore, the results remain robust with fixed-effects estimation.
The remainder of the paper is structured as follows. Section 2 presents the theoretical
considerations and derives our hypotheses. Section 3 explains our estimation strategy, and
Section 4 introduces the data set. Section 5 presents our empirical results and robustness
checks. Section 6 concludes.
2.1 Conceptual background
From a classical spillover perspective, educational spillovers occur when higher-educated
workers transfer their knowledge and skills to lower-educated workers and thereby increase