Ariel Gustavo Letti (p), Mauricio Bittencourt (p)

This work considers exclusively teachers working for more than five years continuosly in 5º grade classes of brazilian public schools. It evaluates the impact of teacher undergraduate courses in their students’ proficiency, in Portuguese and Mathematics.
The gap in the brazilian teachers formation is a reality and presents strong regional aspects. Recent brazilian federal public policies focus in the reduction of this gap and regional inequality. One example is the program PARFOR, that has the objective of estimulating and fomenting the suply of high education (free and with quality) to teachers in activity in the basic educational public system.
In 2015 there were more than 50 thousand teachers participating and more than 12 thousand undergraduate teachers for this program (the majority from the north and northeast).
There is no econometric evaluation to this specific program in Brazil. Thus, this work has this as its objective. In that sense, it uses the difference-in-difference technique with quantile regression. The expected result is that the teacher undergraduate degree is positively related to the students’ proficiency.
It was used ‘Prova Brasil’ (brazilian student test) data from students and teachers (149 without undergraduate degree and 133 with recent undergraduate degree). For each teacher it was calculated the mean proficiency of their students in the year 2009 and in the year 2013, to Portuguese and Mathematics.
Regarding the results, the proficiency calculated presented dispersion between 125 and 300 points, with means around 200 points.
Both proficiencies presented a positive trend in the period, 10.05 and 6.13, respectively (both statistically significant, p < 0.05).
Comparing the difference in the trends between the two groups of teachers, there seems to be a negative difference to the recently graduated group (but not statistcally significant, p > 0.05).
Considering the differences in trends among the proficiency quantiles, the results on the differences are not statistically significant (p > 0.05).
Finally, taking all into consideration, for a teacher working for more than 5 years, having an undergraduate degree has no significant effect on the students’ proficiency, for instance in the first years after the undergraduate course.

Eckhardt Bode (p), Stephan Brunow

Motivated by compelling microeconometric evidence by James Heckman and others who show that labor market outcomes such as occupational choice or wages are affected not only by workers’ educational attainment and work experience but additionally by workers’ personality, this paper seeks to improve measurement of aggregate regional endowments of human capital by complementing the traditional measures of regional human capital, educational attainment and age composition, by new measures related to personality. Unlike statistical information on education, which is available from public data sources for the entire workforce, statistical information on workers’ personalities is available only from rather small-scale worker or population surveys. In these surveys, personality is typically measured by the Big-Five personality traits—openness, conscientiousness, agreeableness, neuroticism and extraversion—constructed from individual ratings of around 20 questions on a Likert scale. One way to map this survey-based information to the aggregate regional level would be using the average rating of all respondents from a specific region for each personality trait as a proxy of personality traits in the region on aggregate. This method requires assuming that the surveyed sample of workers is representative of the workforce in each region, however, which it is frequently not. To avoid this assumption, the paper develops a new method to map micro-level, survey-based data into aggregate, region-level data. It does this mapping by means of an occupational choice model, essentially following Richard Florida who suggest measuring regional creativity, or, for that matter, human capital, by means of the occupational composition of the regional economy. The paper first estimates a microeconomic occupational choice model from German survey data to establish the country-wide average relationship between occupational choice and education as well as the Big Five personality traits. It then uses the inverse of this relationship to estimate, separately for each region, the aggregate endowment with the Big Five personality traits from the observed educational and occupational compositions of the region’s workforce. The paper assesses the practical relevance of these new measures of regional human capital by testing them against the traditional measures in the context of a standard regional growth model.

Justyna Danielewicz, Maciej Turała (p)

The lower secondary schools which appeared in the Polish educational system in September 1999 are going to be abolished at the end of the 2016/2017 school year. The reform has caused heated debate with arguments on both sides. The objective of this paper is to assess the degree in which the lower secondary schools contribute to decreasing the level of spatial differentiation of attained learning outcomes between children attending schools in a chosen region of Poland. This is one of most frequently forwarded arguments against the planned reform.
The specific objectives of the paper include:
• to present the spatial differentiation of attained learning outcomes (as reflected by the results of school leaving examination) by graduates of primary schools in 1999 and 2013;
• to present the spatial differentiation of attained learning outcomes (as reflected by the results of school leaving examination) by graduates of lower secondary schools in 2002 and 2016;
• to analyse and assess the change of the impact of lower secondary schools on spatial differentiation of attained learning outcomes between 2002 and 2016.
The years of analysis were chosen so as to capture the impact of lower secondary schools in the first and last year of their functioning. This will allow verifying whether the impact of lower secondary schools became more significant over time.
In general, the paper aims to verify the following hypotheses:
• the spatial differentiation of attained learning outcomes is smaller for lower secondary schools than for primary schools;
• the spatial differentiation of attained learning outcomes for primary schools is not changing over time;
• the spatial differentiation of attained learning outcomes for lower secondary schools decreases over time.

Gunther Maier (p)

Congress organizers typically define a set of themes to which authors assign their submissions. These themes are intended to structure the programme and to generate sessions where similar and related papers are presented. However, this approach has some potential problems: Do the predefined themes cover the whole area of regional science? Do they limit the integration of new and arising themes? Do all submitters choose the "correct" theme for their paper? Does the content presented at the congresses evolve over time?
These are some of the questions that we will answer for the series of ERSA congresses. We use publicly available information (titles and abstracts) about contributions to ERSA congresses of recent years as input and apply text mining methods. Based on term frequencies, these methods - conceptually related to principal component analysis - allow us to to extract the main dimensions of the content of the material. We identify these "implicit themes" for a number of congresses, present their evolution over the years, and compare them to the predefined themes of the congress programs. In a second step, we show how papers cluster around these implicit themes and to what extent this clustering confirms the assignment to the predefined themes and to what extent it differs. In a third step, we associate the papers with their authors in order to see, who are the individuals who drive certain themes or the development of new themes.