Economies of Education - 10 journal articles review summary
F E A T U R E : N E W E M P I R I C A L A N A L Y S I S I N T H E E C O N O M I C S O F E D U C A T I O N
Paul E. Peterson and Ludger Woessmann
Research in the economics of education has expanded rapidly all over the world in recent years. Scholars have not only inquired into an array of novel topics but they have become increasingly aware of methodological challenges posed by endogenous rela- tionships among the variables of interest. A leading example is research on school choice, because choice necessarily implies selection and the variables that influence selection may be fundamentally intertwined with educational outcomes. To select implies a mutual attraction between chooser and chosen. Given the interaction, it is difficult to pick apart the causal forces of any effects that might be created by the connection. It therefore does not come as a surprise that Coleman et al.�s (1982) classic observational study of student performance in public and private schools in the US set off not just an avalanche of controversy over school choice but a tectonic shift in research design. Observational studies have given way to experimental and quasi- experimental investigations, with the randomised field trial set as the gold standard (Peterson et al., 2003). But the tectonic shift has a range far beyond school choice research. Studies range from private management to teacher compensation policy to peer group relationships to the value of compulsory education. In covering several of these topical themes in education policy and research, the articles collected in this Feature, first presented at a conference jointly organised by CESifo and the Harvard Program on Education Policy and Governance in Munich in 2008, nicely illustrate the range of options for possible identification, comprising experiments in the field and in the laboratory as well as natural experiments identified from observational data.
The first two articles use data from path-breaking randomised field experiments on two pressing topics. Karthik Muralidharan and Venkatesh Sundararaman study the impact of monetary incentives and diagnostic feedback on teacher performance. A randomly selected group of schools in the Indian state of Andhra Pradesh were offered bonus payments to teachers based on their students� learning gains. Results were compared to a random sample of schools that were given the same (but low- stakes) tests to provide teachers with equally detailed diagnostic feedback but no extrinsic incentives and to a sample of schools where no incentives were provided and teacher behaviour remained unevaluated. The authors find that teacher performance pay had a positive impact on student test-score performance – even though observed teacher behaviour hardly differed from the low-stakes tested schools. By contrast, repeated observation of and feedback to teachers had no impact on student test scores when underlying incentives remained unchanged – even though the observed pro- cesses were quite different from non-monitored teachers. These results suggest that it is the teacher incentives rather than pure diagnostic feedback that matter for learning
The Economic Journal, 120 (August), F183–F186. doi: 10.1111/j.1468-0297.2010.02372.x. � The Author(s). Journal compilation � Royal Economic Society 2010. Published by Blackwell Publishing, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA.
[ F183 ]
outcomes and these learning effects come about in a way that is hard to observe in teachers� processes.
To study the effects of school choice, Eric Bettinger, Michael Kremer and Juan E. Saavedra obtained data from a Columbian randomised field trial made possible by a voucher lottery for low-performing students implemented in Bogota during the 1990s. The authors show that students were more likely to finish high school and do better on college entrance examinations if they won the lottery and attended a vocational school, even though the peer characteristics of students at such schools were inferior to those of students at schools attended by non-lottery winners. The design of the study allows the authors to conclude that the benefits from choice cannot be purely redistributive because results cannot be attributed simply to a change in the quality of one�s peer group. Their study also demonstrates certain limitations of even the best experimental research – the inability to disentangle causal impacts of separate ele- ments of an intervention. In Columbia, students lost their vouchers if they did not finish the school year, which provided them with an incentive to focus on their studies. It appears that choice has positive benefits if the student knows academic progress is necessary to keep the option.
While randomised field trials are the most certain way to achieve random assignment into treatment and control groups, they are also expensive, time-consuming and difficult to explain to public officials whose co-operation is generally needed. �How can those assigned to the control group be denied the blessings of the intervention?� they are inclined to ask. In addition, explicit experiments often give rise to Hawthorne effects and their limited scale can miss general-equilibrium effects that may prove important when a policy is implemented throughout the education system. As a consequence, researchers have started to investigate �natural� experiments that approximate the randomised experiment in observational data.
To capture general-equilibrium effects of private school competition, Martin R. West and Ludger Woessmann devise such a quasi-experimental investigation. It is very dif- ficult to estimate impacts of school choice systems that have been brought to scale over many decades. Yet that is the kind of impact that policy makers would like to know and about which the economic theory of school choice (as presented in Milton Friedman�s (1955) classic essay) theorised. Thus, using international PISA data, the authors estimate long-term competitive effects on a country�s student achievement of the size of the private school sector. To solve endogeneity problems, they use the size of the Catholic population in 1900 as an instrumental variable, exploiting the fact that in the late nineteenth century, Catholic resistance to state schooling led to the emergence of private school sectors that remain until today. Their identifying assumption is that the relationship of Catholic shares in 1900 with student achievement in 2003 is nil except for its impact on the contemporary size of the private sector. Because contemporary Catholics may differ from non-Catholics in their readiness and commitment to learn, the authors control for contemporary religious affiliation – a virtue made possible by the historical character of the instrument. Their results indicate positive achievement effects of larger private school shares. A significant part of the overall impact appears to come from general-equilibrium effects of better performance in public schools exposed to private competition.
F184 [ A U G U S TT H E E C O N O M I C J O U R N A L
� The Author(s). Journal compilation � Royal Economic Society 2010
Because instances where �nature� makes experiments in the education area are not easy to find, Thomas Dohmen and Armin Falk revert to another road of identification: they combine evidence from a laboratory experiment implemented outside education with more descriptive evidence from observational education data. As they point out, in the long run teacher incentive programmes such as the one studied in the first article will not only have a direct effect on teachers� effort, but also the effect of determining which people choose to join the teaching profession. Thus, incentives may not only affect the behaviour of the current teaching force, but they may also alter its future composition. The authors present evidence from a laboratory experiment showing that participants with different personality traits self-select into fixed vs. variable pay schemes when given the choice. Among other results, they find that more productive workers tend to select the variable pay scheme. Descriptive evidence from German field data corroborates patterns of employee self-selection into the public sector in general and into the teaching profession in particular. Teachers tend to be more risk averse, more trusting and less negatively reciprocating. In this descriptive part, it remains an open question whether results are due to selection or learning from participation in the teaching profession. Still, the findings alert us to the possibility that incentives could change the composition of the teaching force in the long run – and this may be either for the better (e.g., attracting more productive people) or for the worse (e.g., attracting less trustful people).
The more explorative nature of descriptive analyses is also evident in the final article, in which Christian Dustmann, Stephen Machin and Uta Schönberg use extraordinarily rich administrative data for all students in English state schools to explore the development of the relative performance of students from different ethnicities from age 5 to 16. They show that the substantial initial achievement gap between ethnic minority students and white British-born students declines or even reverses during the course of compulsory schooling. Exploiting the longitudinal structure of the data, they look into differences in family background, school quality and teacher incentives as three possible underlying factors. The authors identify language as the single most important correlate of the closing of the ethnic test-score gap, but also find roles for poverty, attended schools and implicit incentives to focus attention on students who are at the threshold of passing the requirement underlying school league tables published at the end of secondary school. The study illustrates the illuminating use to which rich observational data can be put and shows the way for future research to investigate the causal nature of the different links.
The five studies, taken together, reveal the breadth of topics covered in current empirical analysis in the economics of education, encompassing such themes as school choice, teacher incentives and ethnicity. Just as Coleman�s explorative observational study initiated a rich literature using different experimental designs to identify the causal relationships in school choice, additional research, using both experimental and observational data, will be needed to obtain identification of causal relationships in other policy-relevant domains. Clearly, the applied economics of education remains a nascent and exciting policy and research arena, with plenty of low-hanging fruit to be plucked.
Harvard University University of Munich and Ifo Institute for Economic Research
2010 ] F185A N A L Y S I S I N T H E E C O N O M I C S O F E D U C A T I O N
� The Author(s). Journal compilation � Royal Economic Society 2010
References Coleman, J.S., Hoffer, T. and Kilgore, S. (1982). High School Achievement: Public, Catholic, and Private Schools
Compared, New York: Basic Books. Friedman, M. (1955). � The role of government in education�, in (R. Solo, ed.), Economics and the Public Interest,
pp. 123–44, New Brunswick, NJ: Rutgers University Press. Peterson, P.E., Howell, W.G., Wolf, P.J. and Campbell, D.E. (2003). �School vouchers: results from random-
ised experiments�, in (C.M. Hoxby, ed.), The Economics of School Choice, pp. 287–341, Chicago: University of Chicago Press.
� The Author(s). Journal compilation � Royal Economic Society 2010
F186 [ A U G U S T 2010]T H E E C O N O M I C J O U R N A L
Copyright of Economic Journal is the property of Wiley-Blackwell and its content may not be copied or
emailed to multiple sites or posted to a listserv without the copyright holder's express written permission.
However, users may print, download, or email articles for individual use.