Dr. Lino Wehrheim

11. “Money Can’t Buy Love?” Creating a Historical Sentiment Index for the Berlin Stock Exchange, 1872–1930

Together with Janos Borst and Manuel Burghardt (accepted conference paper at DH 2023).

Financial economists and psychologists agree that collective sentiment plays a crucial role in financial markets (Akerlof and Shiller 2009, Tuckett 2011). In this paper, we present our workflow for creating a daily aspect-based index that captures the sentiment at the Berlin Stock Exchange from 1872 to 1930, a time when Berlin was the key financial market in Germany. This index is based on market reports published every trading day in the Berliner Börsen-Zeitung, which give a verbal description of the sentiment among market participants. Due to daily publication and the long observation period, our corpus consists of about 18,000 market reports. We apply a combination of expert annotation and machine learning. With this newly-created data, we will be able to gain a better understanding of the historical German stock market and, more generally, the nature of financial sentiment itself. How has sentiment developed over time and how is it related to historical events? Did sentiment influence prices and/or trading volume, or vice versa? To answer such questions, we need data on financial sentiment which, to our best knowledge, is not available for the aforementioned case yet. From a DH perspective, this paper covers two novel aspects: First, we focus on both historical and highly domain-specific language, a dual challenge that thus far has rarely been addressed. As there are many similar historical sources, such as reports by chambers of commerce, our solutions will be helpful for future research. Second, we address a characteristic but neglected feature of financial texts that might be relevant also in a broader sentiment analysis context. Particularly, we focus on aspect- and entity-based sentiment analysis.

8. Turn, Turn, Turn. A Digital History of German Historiography, 1950–2019 (2022)

In: Journal of Interdisciplinary History, Vol. 53 No. 3, pp. 471–507. Together with Tobias A. Jopp and Mark Spoerer. https://doi.org/10.1162/jinh_a_01871.

The increasing availability of digital text collections and the corresponding establishment of methods for computer-assisted analysis offer completely new perspectives on historical textual sources and historiographical output. In this paper, we use the possibilities of text mining to investigate the history of post-WWII German historiography. Specifically, we use topic modeling, i.e., a method of automated content analysis, to explore publication trends in eleven leading German-language history journals (around 9,000 original research articles), thereby gaining data-based insights into the history of the discipline. We are particularly interested in the question of whether the various historiographical turns that have been happening in German historiography – and, in fact, in historiography worldwide – so far really brought about deeper conceptual reorientations regarding the way historians’ work and write or whether they merely represent sets of catchy labels to tune up one’s research rhetorically. Based on the topic modeling output, we suggest that some historiographical turns, indeed, were deep turns of the former sort (e.g., the cultural and post-colonial turns), but that a number of turns may be called superficial as they do not show up on deeper levels of the texts analyzed (e.g., the iconic and performative turns).

7. The Sound of Silence. On the (In)Visibility of Economic Experts in German Print Media (2022)

In: Vierteljahrschrift für Sozial- und Wirtschaftsgeschichte, Vol. 109 No. 1, pp. 29–71. https://doi.org/10.25162/vswg-2022-0002.
WP Version available here: LINK

One way for economists to influence society is to shape what Robert Shiller has called “economic stories”. This, in turn, puts the media in their role as professional storytellers in a central position. In this paper, I investigate how economists have been covered by print media since the 1960s: How has the quantitative visibility of economists developed over time? And how can news stories covering economists be characterized in terms of their content and framing? To answer the first question, I provide a comparison of economists’ quantitative media visibility in international newspapers. Exploring the second question, I build on a corpus of more than 12,000 German newspaper articles to conduct a case study on the German Council of Economic Experts, Germany’s most prominent group of economic experts.

6. Cyclicity of Real Estate-Related Trends: Topic Modelling and Sentiment Analysis on German Real Estate News (2021)

In: Journal of European Real Estate Research, Vol. 14 No. 3, pp. 381–400. Together with Franziska Plößl and Tobias Just. https://doi.org/10.1108/JERER-12-2020-0059.

The purpose of this paper is to identify and analyse the news coverage and sentiment of real estate-related trends in Germany. Trends are considered as being stable and long-term. If the news coverage and sentiment of trends underlies cyclicity, this could impact investors’ behaviour. For instance, in the case of increased reporting on sustainability issues, investors may be inclined to invest more in sustainable buildings, assuming that this is of growing importance to their clients. Hence, investors could expect higher returns when a trend topic goes viral. With the help of topic modelling, incorporating seed words partially generated via word embeddings, almost 170,000 newspaper articles published between 1999 and 2019 by a major German real estate news provider are analysed and assigned to real estate-related trends. This dataset is then analysed based on whether the tone of the news coverage of a specific trend is subject to change. The articles concerning urbanisation and globalisation account for the largest shares of reporting. However, the shares are subject to change over time, both in terms of news coverage and sentiment. In particular, the topic of sustainability illustrates a clearly increasing trend with cyclical movements throughout the examined period. Overall, the digitalisation trend has a highly positive connotation within the analysed articles, while regulation displays the most negative sentiment.

5. Diskurs, Narrativ, Sonderweg, Hitler, Turn. Konjunkturen geschichtswissenschaftlicher Begriffe im „Clio Viewer“ (2021)

In: Historische Zeitschrift, Issue 313, No. 1, pp. 129–154. Together with Tobias A. Jopp and Mark Spoerer.  https://doi.org/10.1515/hzhz-2021-0025

At the suggestion of a (non-representative) Twitter survey among historians, we would like to present in this article a corpus of essays from eleven German historiographical journals and illustrate how the use of various historical terms has changed since 1950. The aim is not to present a “digital conceptual history” of various terms or even to draw conclusions about the development of the discipline. Rather, our aim is to present several terms and their “business cycles” in order to lay the foundation for a more in-depth discussion of digital historical methods and the use of a “Clio Viewer”. We pay special attention to discussing the methodological pitfalls associated with the use of simple keyword searches.

4. Zur Konjunktur des Zählens – oder wie man Quantifizierung quantifiziert. Eine empirische Analyse der Anwendung quantitativer Methoden in der deutschen Geschichtswissenschaft (2020)

In: Historische Zeitschrift, Issue 310, No. 3, pp. 580–621. https://doi.org/10.1515/hzhz-2020-0019. Together with Michael Buchner, Tobias Jopp, and Mark Spoerer. For an English version, see here: LINK

The establishment of social history as ‘historical social science’ during the 1970s added quantitative-statistical methods to historians’ methodological toolbox, complementing existing qualitative-hermeneutical approaches. Since the days of the ‘Bielefeld school’, technological progress has provided ever more powerful and user-friendly software tools principally facilitating quantitative analyses. However, quantitative approaches towards historical research questions seem to be applied mainly in few sub-disciplines such as economic history. A reason for this methodological restraint might be widespread skepticism towards quantitative methods among the many supporters of the ‘new cultural history’. Yet, the question as to how intensively quantitative methods have been used in German-speaking historiography has attracted only little empirical research. We aim at filling this gap by exploring a text corpus consisting of more than 7,600 articles published in ten German historical journals, among others the HZ, and covering the period 1951-2016. Our approach is both qualitative and quantitative in nature, combining the counting of tables and figures with a lexicographical inquiry and an extensive discussion. Our results confirm the hypothesis that the cultural turn largely reversed the growing trend towards application of quantitative methods in many parts of historiography. However, the ‘quantification of quantification’ holds some surprises.

3. What’s in the news? (Erfolgs-) Rezepte für das wissenschaftliche Arbeiten mit digitalisierten Zeitungen (Panel at the DHd conference 2020 in Paderborn)

In: Schöch, Christof. (2020). DHd 2020 Spielräume: Digital Humanities zwischen Modellierung und Interpretation. Konferenzabstracts, pp. 70-72. With Bernhard Liebl, Sarah Oberbichler, Torsten Roeder, Jana Keck, Estelle Bunout, and Marten Düring. LINK

2. Von Wirtschaftsweisen und Topic Models: 50 Jahre ökonomische Expertise aus einer Text Mining Perspektive (2019)

In: Patrick Sahle (Hg.): DHd 2019 Digital Humanities: multimedial & multimodal. Konferenzabstracts. Frankfurt, pp. 240–245. (In German) LINK

The increase in the availability of digital texts presents new challenges and opportunities for research in the fields of contemporary and economic history. The present article examines how automated content analysis using Topic Modelling can be used to study a historically significant corpus: the annual reports of the German Council of Economic Experts, one of the most prominent bodies in economic policy advice. The aim is to show how text mining methods in general and topic models in particular open up new approaches to the historical analysis of the work of the Councisl, often referred to as the “Olympus of Economists”, and how the Council’s more than 50 years of work can be viewed in its entirety. To this end, the topics of all the reports published between 1965 and 2015, a corpus of about seven million words, are presented with the help of a topic model. The results are beeing compared to actual economic developments, and further possibilities of the Topic Modeling approach are illustrated.

1. Economic History Goes Digital: Topic Modeling the Journal of Economic History (2019)

In: Cliometrica 13 (1), pp. 83–125. https://doi.org/10.1007/s11698-018-0171-7

Digitization and computer science have established a completely new set of methods with which to analyze large collections of texts. One of these methods is particularly promising for economic historians: topic models, i.e., statistical algorithms that automatically infer the content from large collections of texts. In this article, I present an introduction to topic modeling and give an initial review of the research using topic models. I illustrate their capacity by applying them to 2675 articles published in the Journal of Economic History between 1941 and 2016. By comparing the results to traditional research on the JEH and to recent studies on the cliometric revolution, I aim to demonstrate how topic models can enrich economic historians’ methodological toolboxes.

Extracting Textual Data from Historical Newspaper Scans and its Challenges for “Guerilla-Projects” 

In: Regensburg Economic and Social History (RESH) – Discussion Paper Series. Together with Bernhard Liebl and Manuel Burghardt.

In 2022, it is a common place that digital historical newspapers (DHN) have become increasingly available. Despite the undeniable progress in the supply of DHN and the methods to perform rigorous quantitative analysis, however, working with DHN still poses various pitfalls, especially when scholars use data provided by third parties, such as libraries or commercial providers. Reporting from a current project, we want to share our experiences and communicate the various problems we faced while working with DHN. After a short project summary, we present the main problems that we faced in our project and that we think might also be relevant for other scholars, particularly those who work in small research groups. We arrange these problems according to an archetype workflow, which is divided into the three steps of corpus acquisition, corpus evaluation, and corpus preparation. By raising some red flags, we want to call attention to what we think common DHN related problems, to raise awareness for potential pitfalls, and, this way, to provide some guidelines for scholars who consider using DHN for their research.

More Than a Feeling. Introducing an NLP-Based Media Sentiment Index for the Berlin Stock Exchange, 1872–1930 

Together with Janos Borst, Bernhard Liebl, Manuel Burghardt and Mark Spoerer.

Collective emotions or sentiment can have substantial impact on financial markets. In this paper, we present newly created data that reflects the sentiment at the Berlin Stock Exchange between 1872 and 1930 based on daily market reports published in the Berliner Börsen-Zeitung. Using a workflow of optical character recognition, layout detection, machine learning and natural language processing, we create daily aspect-based sentiment data. Although text classification may be regarded as a standard task in some fields, it poses several technical, conceptional, and historiographical challenges in the domain studied in this paper, that is, historical financial news documents. To contextualize this project and to get a hold on the somewhat hard-to-grasp conception of market sentiment, we first summarize the literature on this topic. Second, we present our workflow for extracting the data from raw newspaper images, providing transparency about our efforts to maximize data reliability. Particularly, we highlight the most important domain-specific challenges, from source-specific issues to classification models. Third, we evaluate the data both in a technical and historiographical sense and propose future applications.

• Robert J. Shiller: Narrative Wirtschaft. Wie Geschichten die Wirtschaft beeinflussen – ein revolutionärer Ansatz, in: Vierteljahrschrift für Sozial- und Wirtschaftsgeschichte 2021/3, p. 428–29 LINK
• Maximilian Kutzner: Marktwirtschaft schreiben. Das Wirtschaftsressort der Frankfurter Allgemeinen Zeitung 1949 bis 1992, in: Vierteljahrschrift für Sozial- und Wirtschaftsgeschichte 2021/2, p. 280–81 LINK
• Bertram Schefold: Die Bedeutung des ökonomischen Wissens für Wohlfahrt und wirtschaftliches Wachstum in der Geschichte, in: Vierteljahrschrift für Sozial- und Wirtschaftsgeschichte 2019/3, p. 421–22 LINK