This chapter discusses the recent rise of machine learning (ML) techniques, specifically decision trees and Random Forests (RFs). RFs do not pursue causal inference but are powerful techniques for making predictions. RFs are discussed from various angles and compared with regression techniques like Classification and Regression Trees (CART), Random Forests, Synthetic Minority Oversampling TEchniques (SMOTE), Confusion Matrix, and Receiver Operating Characteristic (ROC) curve. As a running example, the value of RFs is illustrated throughout the chapter by focusing on the context of the prediction of high-growth firms (HGFs, also known as “gazelles”) using Swedish data (limited liability firms, 2004–2015). RFs appear to outperform previously used regression techniques when the task is a prediction of HGFs. ML techniques such as RFs have “raised the bar” in terms of the accuracy of prediction tasks. Improved prediction accuracy can be expected to lead to better decision-making by various stakeholders, such as (actual and potential) entrepreneurs, investors, academics, and policy-makers. We contribute to the emergence of a new class of quantitative analysis tools for making predictions that are expected to have superior predictive power in many contexts, including the prediction of HGFs.

In this study, we seek to understand firm behaviour during times of crisis, with a particular focus on family firms in different contexts. We theorize that family control mitigates (i.e. negatively moderates) the relationship between economic crisis and the layoff of employees, resulting in a higher propensity of family firms to retain their employees during a crisis compared to their nonfamily counterparts. Furthermore, taking a closer look at family firms, based on their location, we argue that family firms in rural regions are more likely to adopt measures leading to involuntary job turnover than family firms in urban areas due to a higher sensitivity to the loss of socioemotional wealth following a business closure. Relying on a panel dataset of Swedish private firms active in the period 2004-2012, our study contributes to a better understanding of family firms as employers in different contexts.

Based on a review of 700+ peer-reviewed articles since 1990, identified using text mining methodology and supervised machine learning, we analyze how neo-Schumpeterian growth theorists relate to the entrepreneur-centered view of Schumpeter Mark I and the entrepreneurless framework of Schumpeter Mark II. The literature leans heavily toward Schumpeter Mark II; innovation returns are modeled as following an ex ante known probability distribution. By assuming that innovation outcomes are (probabilistically) deterministic, the entrepreneur becomes redundant. Abstracting from genuine uncertainty, implies that central issues regarding the economic function of the entrepreneur are overlooked such as the roles of proprietary resources, skills, and profits.

Within the field of innovation studies, researchers have identified systematic failures that hamper investment in R&D, innovation, and growth. Accordingly, researchers in this field often seek to provide policy recommendations on how to alleviate these failures. However, previous discussions have often been lacking considerations to the risks of political failures, meaning that policies fail to achieve their stated goals in a systematic manner. In response to this gap, this article aims to illustrate the concept of political failure and its relevance for innovation research. This is done by both discussing how political failure can impact innovation policy and by reviewing the prevalence of any discussions of political failure among top-ranked journals on innovation for the period 2010–2019, a total of 7161 articles. The results show that consideration of political failure is scarce, with a small number of papers that have a substantial analysis of political failures. If the awareness of political failures could be increased, this could lead to better policy recommendations with a more nuanced discussion of the risks and limitations of public policy. 

We map the distribution of High-Growth Firms (HGFs, or “gazelles”) across the dimensions of firm size and firm age using contour plots, where firm size and growth are measured in terms of employees. The analysis is based on Swedish total population data for the period 1990–2016, covering approximately 11, 000, 000 firm-year observations. The results show that the majority of HGFs are small, which partly follows the high representation of small firms in the population. Yet when considering the proportions of HGFs across firm ages and sizes, a distinct feature emerges in that the territory of small, old firms appears to be almost completely deserted—despite their large overall numbers—where firms in this category have the (by far) lowest chances of becoming HGFs.

This paper presents a conceptual framework for analysing the role of information technology in the formation of high-impact entrepreneurship. Entrepreneurial decision-making is contextualised in the setting of competent teams, and where its role in economic growth is modelled as part of a minimum set of actors necessary for the generation of innovative output - so-called collaborative innovation blocs. By departing from a collective of actors, rather than the individual entrepreneur, transactions costs are shown to become central for understanding the antecedents and conditions for high-impact entrepreneurship as core strategic decisions are often based on asymmetric information and bounded rationality. Subsequently, this also implies a central role for information technology in facilitating the processes that precede high-impact entrepreneurship through its ability to bridge or reduce information asymmetries. Based on the presented framework, the development of information technology is hypothesised to particularly favour new entrepreneurs with growth ambitions, new firm entry, and high growth firms by accelerating the creation and allocation of knowledge. 

Family firms account for a substantial share of economic activity and deviate from standard economic assumptions on firm behavior. However, little is known about how these firms are represented in economic theory. This article examines the inclusion of family business in the curricula of economics doctoral programs in the United States and Sweden as well as professors’ and textbook authors’ views and research on family business. Textbooks, articles and course offerings used in doctoral programs are considered to indicate the state of established knowledge in the field. The findings show that family business is not included in the examined curricula. Furthermore, professors and authors do not publish research on family business and generally do not see a need to incorporate it into economic theory. This article concludes that family business is excluded from ‘core’ economic theory due to a lack of paradigmatic pluralism, axiomatic incompatibility, path dependency, institutional bias and data constraints. Lastly, it is speculated that integration of family business theory into standard economic modeling is likely to occur outside prestigious universities due to path dependency in research.

This study investigates the proposition that family firms have comparative employment growth advantages in relation to non-family firms in regions with relatively low population density. This premise is tested across metropolitan, urban and rural regions using total population data on domestically and privately owned, single-plant, non-listed limited liability firms in Sweden. A panel of more than 89,000 firms is followed over a seven-year period from 2004 to 2010. The average family firm is found to grow more slowly than the average non-family firm across the urban-rural context. However, in line with the study's conjecture, these differences are found to decrease across metropolitan, urban and rural regions.

We exploit information on ownership, management and kinship to study the representation of women in top management teams in Swedish family and non-family firms among domiciled limited liability firms over the years 2004 to 2010. The share of female top managers is analysed across listed and non-listed firms as well as across industries. We then estimate the likelihood that a woman is elected into the top management team in family and non-family firms using a probit regression model where we control for firm- A nd individual-level characteristics, including the gender distribution of the firm and kinship relations to existing board members and firm owners. We find that non-listed family firms are more likely to appoint female top managers, whereas we find no differences among listed firms. Moreover, we find that the gender composition and kinship structures of firms influence the appointment of female top managers.

Family firms are often considered characteristically different from non-family firms. However, our understanding of family firms suffers from an inability to identify them in total population data; information is rarely available regarding owners, their kinship, and their involvement in firm governance. We present a method for identifying domiciled family firms using register data; this method offers greater accuracy than previous methods. We apply this method to Swedish data concerning firm ownership, governance, and kinship from 2004 to 2010. We find that the family firm is a significant organizational form, contributing over one third of all employment and gross domestic product (GDP). Family firms are common in most industries and range in size. Furthermore, we find that, compared to private non-family firms, family firms have fewer total assets, employment, and sales and carry higher solidity, although family firms are more profitable. These differences diminish with firm size. We conclude that the term “family firm” includes a large variety of firms, and we call for increased attention to their heterogeneity.