Projects

Project Leaders: Daniel Mayerhoffer, Daria Tisch

Research Question: The super-rich are notoriously hard to capture by survey studies, as they tend to not respond to survey requests. To study this group, journalistic “rich lists” like the one provided by Forbes present a viable solution. However, since they are not prepared for the needs for statistical data analysis and need to be scraped or manually digitised and need to be enriched with further information, they are still seldom used for scientific purposes.

Data: Our proposed research workshop leverages two already digitised datasets for rich lists – the Forbes world’s billionaires list (2010-2023) and the list of the 500 or 1000 richest Germans by the “manager magazin”. One expert (DT) has already connected the Forbes list to WikiData, which enables us to study not only the time evolution of wealth levels and returns but also the associated sociodemographic and genealogical data. To further enrich our analysis, we will utilize the Bureau van Dijk’s ORBIS firm database to link data on super-rich individuals and families to business activity, constructing a network of interactions among the super-rich.

Methods: We plan to utilize data retrieved from large-scale databases and using web-scraping techniques (incl. the usage of APIs) and employ statistical analysis, and network analysis techniques to delve into the mechanisms of wealth reproduction and the influence of elite networks on this process. Integrating rich lists with data on corporate boards and Wikidata offers plenty of possibilities to study how wealth is perpetuated, and how knowledge flows within elite networks might contribute to this perpetuation, which have so far not been frequently utilized.

Description: In the project, we aim to study comparatively, how wealth is accumulated and transmitted in time an space by comparing different rich lists and billionaires of different social origins and gender. A connection to WikiData will allow us to analyse family linkages and dynastic wealth accumulation. Furthermore, we will link rich lists to firm data and examine if and how the wealth of billionaires translates to centrality on the network of interlocked directorates. The workshop will give hence focus on theme “Mechanisms for the Reproduction of Wealth”: Given the contemporary debate on billionaires and their companies such as X, Alphabet and Amazon displaying disproportionate political and social influence, we hope to address a timely topic and study, how this wealth and power comes about in the first place.

Project Leaders: Zeyad Kelani, Mohamed Emish.

Research Questions:

• Can satellite imagery and machine learning methods accurately predict socioeconomic status at a fine-grained geographic level across diverse urban contexts?
• What visible urban features are most associated with wealth/poverty worldwide?

Data:

• Points of Interest (POIs) data from Open Street Maps
• Buildings shapefiles from Leafmap
• Government census on household assets and wealth at the city level from Central Agency for Public Mobilization and Statistics (CAPMAS).

Description: This project aims to apply cutting-edge Geographic Information Systems (GIS) techniques to predict socio-spatial inequality across the developing world from satellite data. By training machine learning algorithms on diverse urban landscapes, we will evaluate their effectiveness at estimating wealth from spatial features across varied contexts, building on previous work linking satellite imagery to economic well-being measures [1]. Tracing changes over time will reveal the evolving geography of inequality in cities worldwide [2]. Advanced deep learning approaches, such as unsupervised feature extraction, can help accurately classify urban features predictive of poverty [3,4]. Building-level shapes and types of POIs in cities offer a useful complement to understanding ground-level socioeconomic divisions [5]. This research will generate new insights into spatial divides and provide an innovative data science approach to studying urban poverty. The method could be applied to inform policy in any developing city with suitable satellite imagery and survey data.

References:

[1] Jean, N., Burke, M., Xie, M., Davis, W.M., Lobell, D.B. and Ermon, S., 2016. Combining satellite imagery and machine learning to predict poverty. Science, 353(6301), pp.790-794.

[2] Jiang, Y., Li, Z. and Ye, X., 2022. Mapping urban poverty using machine learning and satellite imagery: A case study of Nairobi, Kenya. Habitat International, 118, p.102480.

[3] Jean, N., Wang, S., Samar, A., Azzari, G., Lobell, D. and Ermon, S., 2019. Tile2vec: Unsupervised representation learning for spatially distributed data. Proceedings of the AAAI Conference on Artificial Intelligence, 33(01), pp.3967-3974.

[4] Kohli, D., Sliuzas, R., Kerle, N. and Stein, A., 2012. An ontology of slums for image-based classification. Computers, Environment and Urban Systems, 36(2), pp.154-163.

[5] Suel, E., Polak, J.W., Bennett, J.E. and Ezzati, M., 2019. Measuring social, environmental and health inequalities using deep learning and street imagery. Scientific reports, 9(1), pp.1-10.

Project Leader: Markus Lang
Research Question: How do changes in intellectual property rights and franchising regulation influencewealth inequality within and between countries?

Data:

• WIPO IP statistics data center
• World Inequality Database (WID)
• World Development Indicators (WDI)

Methods:

• Multilevel regression
• Bayesian regression
• Directed Acyclic Graphs (DAGs)
• Simulation

Description: According to a growing body of literature, both changes in intellectual property rights and franchising regulation significantly contribute to increasing wealth inequality in the United States. A central observation in this literature is that highly profitable companies use IP rights to exclude potential competitors from utilizing technical knowledge. This exclusion often leaves potential competitors with no alternative but to become suppliers to more profitable companies, thereby reducing competition through lower labor costs and investments in physical infrastructure. Franchising, in turn, allows highly profitable
companies to similarly outsource labor-intensive activities to smaller, legally independent entities. The latter bear full responsibility for the well-being of workers but do not control the intellectual property on which they operate.
What is currently unknown is the exact impact of changes in intellectual property rights and franchising regulation on wealth inequality in the United States and other countries. The goal of this research project is to estimate the effect of substantial legal changes on wealth inequality using existing country-year intellectual property rights indices.
Building on this, the project aims to simulate how influential changes in franchising regulation could be on wealth inequality. These simulations could serve as a starting point for constructing currently unavailable franchise regulation indices.

Project Leaders: Benjamin Rosche, Alison Schultz

Research question: Do the ultra-wealthy use their political connections to obstruct progressive tax reforms?

Data: For this project, we both collect new data and draw on existing databases to assemble the following
datasets:

1. Educational trajectories of Forbes 400 billionaires and members of Congress who voted on tax-
   related bills between 2000 and 2020.
2. Political donations of Forbes 400 billionaires to members of Congress between 2000 and 2020.
3. Votes of members of Congress for all relevant tax-related bills between 2000 and 2020.

Methods: Methods
To measure whether members of Congress with higher exposure to the ultra-wealthy are more likely to block progressive legislation, we will draw on peer effect modeling (An, Beauville, and Rosche 2022; Bramoulle, Djebbari, and Fortin 2020; Rosche 2022). The peers in our context are ultra-wealthy individuals who potentially influence congress members. We draw on a combination of synthetic control methods and fixed effects estimators to tackle issues surrounding the identification of peer effects with observational data (An 2011, 2015; Hsieh, Lin, and Patacchini 2020; Abadie 2021).

Details here.

Description
Passing progressive tax reforms has proven to be difficult despite their importance for a more even distribution of wealth. Page and Seawright (2023) identify two primary obstacles to progressive tax reform: an institutional bias favoring the status quo and significant political opposition from wealthy individuals. Unsurprisingly, evidence from the U.S. indicates that most multi-millionaires and billionaires are against any kind of progressive tax reform and there is compelling evidence that actual political outcomes tend to align closely with the preferences of the wealthy (Gilens 2005; 2014). What remains unclear, however, is the exact means by which the ultra-wealthy exert their influence.
This project aims to explore one potential mechanism through which the wealthy may obstruct progressive tax reform: their connections to political decision-makers. Specifically, we will investigate whether members of the U.S. Congress are less likely to support progressive tax reforms when they are more strongly connected to the ultra-rich. The proposed research has three primary objectives:
1. Identify connections between ultra-rich individuals and legislators in the U.S. Congress.
2. Identify tax-related bills that were voted on in the U.S. Congress between 2000 and 2020.
3. Examine whether politicians with greater exposure to the ultra-rich are more likely to vote against (in favor of) progressive (regressive) tax-related bills compared to those with less exposure, controlling for factors such as age, party affiliation, and term length.

Details here.

Project Leaders: Hilke Brockmann, Partick Sachweh

Question: Can the self-presentation and public perception of wealth in social media explain why people accept high and rising levels of wealth inequality?

Method & Data: Our analysis will be based on celebrities and other public figures who are identified through Forbes rich lists and their social media posts (TikTok, Instagram). We will connect the posts with responses, analyze the content and sentiments of the texts and visuals, and predict which content provoked what responses. The findings will be cross validated with an online survey experiment that disentangles if certain signals cause positive or negative perceptions of wealth.

Description: All societies must justify inequality in wealth (Piketty & Goldhammer, 2020). Democracies are particularly challenged if resources and ownership is amassed by very few people, as this could undercut the ideal of an evenly distributed political power among the electorate. During the last decades, many scholars detected rising economic inequalities in many (rich) democracies (Atkinson, 2015; Piketty, 2014; Savage, 2021),. Particularly the super wealthy experienced extraordinary growth trajectories.
However, the public outcry of the 99% has been short-lived – for example during the Occupy-Wallstreet movement in the aftermath of the global financial crisis in 2008 – or has never materialized in higher inheritance or wealth taxes. Many researchers are surprised at the reluctance of many people to tax the rich more heavily and call for political action (https://www.tax-the-rich.eu/). This project asks: To what extent can the self-presentation and public perception of wealth in social media explain why people accept high and rising levels of wealth inequality?
The sociologist Georg Simmel (1957) stated that the fashionable lifestyle of the leading social classes serves two functions – it sets an adorable trend which lower classes imitate and thereby equalizes the gap between the have and the have-nots. At the same time fashion creates persistent change by which the richer people can segregate from their poorer fellows. We will test if this trickling down hypothesis is still true and if fashion creates continuous positive response. The later hypothesis points to the rising awareness that the rich have an enormous environmental footprint and contribute disproportionality to climate change (Chancel, 2022). 

References:
Atkinson, A. B. (2015). Inequality: What can be done? Harvard University Press.
Chancel, L. (2022). Global carbon inequality over 1990–2019. Nature Sustainability, 5(11), 931–938. https://doi.org/10.1038/s41893-022-00955-z
Piketty, T. (2014). Capital in the twenty-first century. The Belknap Press of Harvard University Press.
Piketty, T., & Goldhammer, A. (2020). Capital and ideology. Harvard University Press.
Savage, M. (2021). The return of inequality: Social change and the weight of the past. Harvard University Press.
Simmel, G. (1957). Fashion. American Journal of Sociology, 62(6), 541–558.

Project Leaders: Jan Schulz-Gebhard, Jan Lorenz

Question: How do stochastic growth and wealth redistribution effects growth and inequality of total wealth?

Data:

• World Inequality Database (wid.world)
• Distributional Wealth Accounts (European Central Bank)
• Datasets on the Super Rich e.g. by Forbes, manager magazin or the Netzwerk Steuergerechtigkeit

Methods:

• Agent-Based and Computer Simulations
• Stochastic Differential Equations/Itô Calculus
• Calibration and Validation using empirical estimates, e.g., for return volatility from Stock Market Indexes

Description:

An often-neglected mechanism to create extremely unequal wealth distributions is the cumulative effect of stochastic growth events where each event multiplicatively affects individual wealth. These types of processes are known to converge in distribution with lognormal densities characterized by very few rich and a majority with comparably little wealth. Furthermore, wealth inequality increases monotonously over time. Consequently, a surprising phenomenon can be observed in simulations of a finite population of individuals: In the long run, the combined wealth of all individuals may decline although their expected wealth increases in every stochastic event. That means the wealth of every individual increases on average every time step, but in the long run, the total societal wealth declines and vanishes (while inequality still increases). This combination of the collective ruin of all agents that individually exhibit positive expected growth rates features prominently in the emerging field of ergodicity economics that distinguishes between the ensemble-average growth rate (of the cross-section of agents at a specific point in time) and the time-average growth rate (of an agent through time).

For multiplicative growth regimes, both growth rates diverge, which begs the question of which kinds of mechanisms can be employed to let the time-average growth rate approach the ensemble-average growth rate and avoid the collective ruin of agents. Sharing and pooling of resources (Peters and Adamou, 2022) or redistributive taxation (Lorenz et al., 2013) have both been shown to achieve this, providing a fundamental dynamic efficiency argument for taxation in multiplicative growth regimes. Yet, building on these results, there remain several potential avenues to further study these redistribution mechanisms in more detail: Theoretically, the stochastic growth process could be augmented by featuring type- and scale-dependence that were shown to be necessary to fit the convergence trajectories of empirical wealth distributions (Gabaix et al., 2016) and were documented in administrative data (Bach et al., 2020). A further theoretical possibility would be to investigate coupled growth processes with differing network topologies representing these couplings or to introduce other tax-efficiency mechanisms featuring, e.g., a constant elasticity functional form generating taxation Laffer curves (Mayerhoffer and Schulz, 2023). Finally, looking at wealth mobility for different stochastic processes to gauge the effects of redistribution on inequality of opportunity might also enhance our understanding of the potential efficiency gains from redistributive taxation. On the empirical side, a further extension would be to bring the simulation models to the data and estimate the implied amount of redistribution based on the methodology introduced by Berman et al. (2020). They find negative redistribution and thus an unstable multiplicative regime for the US. Extending their methodology to Europe, e.g. using data from the World Inequality Database or the ECB’s newly released Distributional Wealth Accounts with time series going back to 2009 would thus help to establish if the US is an outlier in this case or if the systemic instability is indeed a common phenomenon. Since the super-rich constitute a distinct group, using the empirical results of the group led by Daria Tisch and Daniel Mayerhoffer at the same summer school might also provide helpful data to aid calibration and validation.

Projects:

• Agent-based modeling / mathematical analysis to describe and better understand the phenomenology extending the basic mechanisms along the outlined ways
• Compare the micromechanisms to real-world mechanisms/calibrate the model using empirical data on wealth distributions (especially for the European case)

References:
Bach, L., Calvet, L. E., & Sodini, P. (2020). Rich pickings? Risk, return, and skill in household wealth. American Economic Review, 110(9), 2703-2747.
Berman, Y., Peters, O., & Adamou, A. (2020). Wealth inequality and the ergodic hypothesis: Evidence from the United States. Journal of Income Distribution, forthcoming.
Gabaix, X., Lasry, J. M., Lions, P. L., & Moll, B. (2016). The dynamics of inequality. Econometrica, 84(6), 2071-2111.
Lorenz, J., Paetzel, F., & Schweitzer, F. (2013). Redistribution spurs growth by using a portfolio effect on risky human capital. PloS one, 8(2), e54904.
Mayerhoffer, D. M., & Schulz, J. (2023). Social Segregation, Misperceptions, and Emergent Cyclical Choice Patterns, BERG Working Paper Series, 186.
Peters, O., & Adamou, A. (2022). The ergodicity solution of the cooperation puzzle. Philosophical Transactions of the Royal Society A, 380(2227), 20200425.

Project Leaders: Johannes Hjellbrekke, Mike Savage

Research Question: How can we use cross-sectional and longitudinal survey data to better understand the class structure of societies?

Data:

• National Income Dynamics Study of South Africa
• Ghanaian Socio-Economic Panel data

Method: Multiple Correspondence Analysis (MCA)

Project Description:

Understanding both wealth and income dynamics is incomplete and inadequate without an understanding of the class structure of societies. These groups shape and are shaped by political decisions, policies, and the economy within which they are located.

Multiple Correspondence Analysis is a sophisticated statistical technique that transcends traditional data analysis methods, offering unique insights into complex datasets. Whether you’re interested in economic trends, sociological phenomena, or the intricate nuances of inequality, MCA provides a dynamic lens to unravel patterns and relationships within your data.

Our expert lectures will guide you through the intricacies of using MCA, showcasing its practical applications with a compelling case study from South Africa, and an in-progress example from Ghana. MCA becomes a valuable tool for both educators and learners alike.

This project represents an opportunity to enhance your analytical toolkit and elevate your understanding of how data can be used in economics and sociology. Join us at the summer school lecture on MCA to learn more about this method for unravelling the complexities of societal and economic phenomena. 

Project Leader: Aseel Addawood

Research Question: What are the socioeconomic and behavioral factors that influence wealth creation, and how can predictive analytics aid in understanding and leveraging these factors for optimized investment decisions?

Data: The project will utilize historical financial data, economic indicators, market trends, alongside socioeconomic and behavioral data to analyze the factors influencing wealth creation.

Methods: The project will utilize machine learning algorithms and predictive modeling techniques to analyze and identify the socioeconomic and behavioral factors that influence wealth creation. These methods aim to provide valuable insights into how predictive analytics can enhance investment strategies for wealth creation.

Description: The project aims to uncover and understand the socioeconomic and behavioral factors that play a crucial role in wealth creation through the application of predictive analytics. By analyzing historical financial data, and economic indicators, and incorporating socioeconomic and behavioral data, the project will shed light on how predictive modeling can assist in making informed investment decisions that optimize wealth creation for investors.

Project Leader: Francesco Smaldone

Research Question: The project aims to develop a data-driven approach for profiling employability skills in the field of data science and provide insights into the skills demanded by the evolving job market.

Data: Job advertisements for data scientists from various sources, with a focus on the US market.

Methods: Text Analysis, Topic Modeling, Skill Correlation Analysis, Network Analysis

Description: Detailed Description