ifoHack 2023

For the first time, the ifo Institute is organizing its very own hackathon – the ifoHack 2023.

The ifo Institute combines sharp economic reasoning and expert knowledge in empirical methods; using current advances in computational methods and open-source tools to leverage both yields enormous potential to address policy-oriented economic and social questions. For example, participants will build a predictive business climate model based on microdata provided by the ifo Institute and present its operating value via a web interface.

Our partners KfW Bank (Kreditanstalt für Wiederaufbau) and Deutsches Zentrum für Luft- und Raumfahrt (German Aerospace Center) will also bring exciting challenges such as using Natural Language Processing for extracting sentiment from financial newspaper articles, predicting user behavior, or using spatial data for predicting real estate prices and will offer exclusive insights into their research activities.

Challenges

1. ifo "KT" Main challenge

In the ifo Business Survey every month approximately 9,000 German firms are surveyed on their current situation (status quo), recent developments (ex-post), and their plans and expectations for the near future (ex-ante).

Build a model which predicts the categorical answers of firms on various questions in the next 12 months. This model can be based on past answers and/or firm-specific information in the micro data of the survey (e.g., firm-size, region, sector).

2. KfW challenge

KfW is the promotional bank of the Federal Republic of Germany. The transformation of the economy and society to improve economic, ecological and social living conditions worldwide is KfW's primary objective. Our credit analysts face the challenging task of assessing the credit risk – i.e. the possibility of a borrower defaulting on his debt – of a diverse cast of counterparties. As such they need to stay on top of current and past news developments on a global level.

Can you help them get a grip on the news?

You are provided with meta information on a set of entity articles published after 01.03.2022. Analyze the data in order to provide useful insights. Predict the likelihood of an article being read by a credit analyst.

3. DLR challenge

   This spatial ifoHack challenge aims to investigate the relationship between land prices and spatial factors that characterize living environments, in large and diverse cities in Germany, like Berlin, Bremen, Dresden, Frankfurt am Main and Köln. By exploring information extracted from satellite images and other relevant spatial datasets, we aim to guide participants into exploring non-conventional data sources, to answer relevant economic questions related to housing markets and spatial inequalities. Are land prices positively associated with a greater availability of spatial determinants of well-being, such as green and open spaces, or walkable and bicycle paths? Is proximity to amenities, educational, leisure or work opportunities a factor that drives higher land prices? Is there a spatial segregation effect in terms of land prices and specific population subgroups? These are only some of the questions we aim to explore together at the ifoHack. By getting to know the right spatial data and methods, formulating new ideas that fit your specific research interests will be only one short step away.

4. ifoGPT

   Use a text corpus of past ifo press releases (use the one provided by us or collect your own) to develop a large language (e.g. transformer) model such as GPT or BERT and provide an intuitive ML-frontend to automatically generate a press release based on keywords such as "positive", "manufacturing sector" and "macroeconomy".

Agenda