For the second time, the Toulouse School of Economics sent a team of Masters and PhD students to compete in the Econometric Game in Amsterdam. Georgios Petropoulos, Daniel Herrera, Juan David Gómez Dávila, Andrés Salamanca Lugo and Eva Raiber represented TSE at this prestigious inter- national competition from 9th to 11th of April.
Econometric Games in Amsterdam
The Econometric Game is organized on an annual basis by the VSAE, the student association for students in Economet- rics, Actuarial Science and Operations research Management at the University of Amsterdam since 1999. It has
evolved into an international three-day competi- tion that gathers teams from prestigious universities from all over the world. The aim is to use economet- ric methods to approach interesting but unusual topics in order to test the students’ ability to apply their theoretical knowledge in unfamiliar areas. Each team consists of five students. This year, 30 univer- sities from across the globe took part in this event, including Oxford, LSE, Cambridge, Harvard, Tilburg and Carlos III Madrid. After being introduced to a topic on the first day, the teams are provided with a case study including a dataset that teams have to use to solve the case study within eight hours and hand in a written report. Ten teams are then selected to participate in the finals on the next day. These teams again receive a case that they have to solve, write a report and make a short presentation, on the basis of which the winning team is chosen.
The Effects of Fiscal Policy on Economic Growth
This year’s topic was the effects of fiscal policy on economic growth. Estimating the effects of fiscal policy on economic output has been a hot topic for academic and political discussion. There is still ongoing de- bate about the size of the fiscal multiplier, especially if the multiplier is more or less important in times of recession. Most stud- ies that address the issue of the effect of fiscal policy have been using linear models, such as Vector Autoregressions and linear- ized DSGE models that do not distinguish between recessions and expansions. There- fore, more recent papers (e.g. Auerbach and Gorodnichenko (2012)) try to trace the dif- ferent impulse responses to changes in fiscal policy during recession and expansion. For example, non-linear switching models can be used for this purpose.
During the opening of the game, the topic was introduced by the case makers, Prof Jörg Breitung (University of Han- nover) and Prof. Matei Demetrescu (University of Bonn). In the afternoon, the students had time to search for literature on the topic as well as on the time series models that were mentioned in the introduction. Since many participants were unfamiliar with advanced time series models, preparation and literature review took many teams until late at night. On the second day, teams received the case with four subsec- tions. First, the teams were supposed to replicate the linear benchmark model of Auerbach and Gorodnichenko (2012)
using the same dataset which included quarterly data of GDP, government spending and taxes for the U.S. from 1948 to 2008. Auerbach and Gorodnichenko (2012) use a structural Vector Autoregression (SVAR) as proposed by Blanchard and Perotti (2002) to trace the effects of shocks in government spending and in taxes on GDP as a benchmark model in order to compare it to a regime-switching model. Using the replicated SVAR results, the teams were then asked to test for mis-specifications, especially those pointing to nonlinearity in the model. Also, the variables that were used in logarithm should be tested for the assumption of cointegration. Last, but very importantly, students were asked to propose alternative identification methods for the SVAR model. So far, identification was based on Blanchard and Perotti (2002), who use institutional information about taxes and assumptions about the timing of fiscal policy changes.
The TSE Team’s Performance
The topic of this year’s Econometric Game was very challenging and not all of us were familiar with advanced time series modeling. However, we managed to address all points of the case in time and handed in a well structured report. For the last question, we proposed long run restriction for identifi- cation for the SVAR model using the assumption that some shocks do not have long run effects on some variables. Yet, we were short in time for the last part of the case and we were not able to provide the motivation for some parts of our analysis as well as we would have liked. As a result, we were not chosen to participate in the finals. Even though we are convinced that our team did a very good job, we acknowl- edge the excellent work that the teams that managed to get into the finals did. It was a surprise to most of the participants that most of the famous universities such as Harvard, LSE and Cambridge did not make it to the finals either. It could be ob- served that the finalists were often from universities with a strong macroeconomic focus.
Forecasting Spain’s Growth Rates
On the third and last day, the finalists were given another case and the non finalists could enjoy Amsterdam through a guided city tour. Afterwards, the Econometric Game congress took place where two presentations were held and the case makers presented the second case. As a surprise to most of the students, the second case was not to further expand the analysis of the first topic but a different and fascinating ques- tion. The teams in the finals were given a time series data set of the Spanish economy including GDP and around 70 other, possible explanatory variables. They were asked to forecast the growth rates for Spain for this year’s four quarters including information from the explanatory variables. Most teams, among those the winners, chose to do a Principal Component Analysis (PCA) and to include these principal components in their forecast. The finalist presented their results at the end of the congress. We want to congratulate the team from Carlos III Madrid that wrote an excellent re- port and presented their re- sults convincingly and fully deserved to win the compe- tition. Carlos III was followed by the University of Copenhagen and the University of Warsaw.
A Rewarding Experience for all Participants
During the Economet- ric Game, we were not only able to work on these interest- ing topics, but we also had time to get to know the city of Amsterdam and the students from the other universities. At night, dinners were organized followed by drinks where par- ticipants could also talk about non-econometric topics. It was a very rewarding experience to compete and converse with what might well be some of the best students in econometrics and economics.
Even though we did not manage to win the competition we, as a team, fostered strong team spirit, motivation, openness and made the most out of this experience. We hope that again next year, a team of TSE will be able to compete and enjoy the atmosphere and working attitude of this event.
We want to thank the faculty, especially Christian Gollier, Jean- Philippe Lesne, Thierry Magnac, Christine Thomas, Marc Ivaldi and Elodie Alet for their continuous support and for financing our participation. We also wish to thank the econometric teachers for offering their help and the BDE for being close to our team and for providing our competition “uniforms”.
For more information on the econometric game, visit www.econometricgame.com.