A substantial shift in the focus of economic research over the last half century has for the most part gone unnamed and unnoticed. This is the turn towards what we call ‘civil society’, including firms as organisations, families, neighbourhood communities, non-governmental organisations, trade unions, social movements, identity groups and other face-to-face settings. Although the term may be unfamiliar to many economists, it dates back to Adam Smith and the Scottish Enlightenment, and it has been variously applied in the other social sciences and philosophy.
Economic interactions in civil society, as we use the term, have in common the fact that relationships are personal and enduring. As a result, identity and other-regarding preferences (for better or worse) are important motives. In civil society, the primary mechanisms guiding economic interactions are neither government regulations nor complete contracts and market-determined prices, but instead hierarchies of private authority and power, social norms, group identity (including hostility to out-groups) and reputation.
The turn towards civil society in economics that we document in our latest work (Bowles et al, 2025) is consistent with two propositions. The first is that much of what we think of as the economy consists of non-market non-state interactions and exchanges under incomplete contracts within the firm as well as in labour, credit, residential housing and other markets. Second, in these and other settings, ethical and other preferences have an important place along with self-interest in explaining behaviour and supporting mutually beneficial exchanges.
This shift in focus has also been reflected in numerous Nobel awards, from Gunnar Myrdal and Friedrich Hayek in 1974 to Daron Acemoglu, Simon Johnson and James Robinson in 2024. The concept is increasingly central within economics research.
Markets, states and civil society
In a study published in Economics Letters, we present civil society along with states and markets as distinct and ideally complementary structures of societal governance. In Figure 1, we treat the pure types of these structures as vertices of a simplex representing a space for economic policies and institutions. Each of these polar cases is characterised in the figure by both a set of rules of the game (by which allocations are implemented) and a set of preferences and social norms (associated with how they contribute to overall societal governance). The coordinates of any point on the simplex sum to one and represent weights giving the relative importance of state, market and civil society (for any particular point in the space, the largest weight is the importance of the closest vertex.)
Figure 1: The space for institutions and policies with ideal-type forms of societal governance at the vertices, their implementation mechanisms and behavioural characteristics
Source: Bowles et al (2025)
We also document the emergence of civil society as an important set of themes in the corpus of papers published in the major economics journals in the UK and the United States between 1900 and 2014 – a total of 27,436 articles (as Figure 2 shows). To represent this shift, we use topic modelling, a probabilistic machine-learning technique that treats a corpus of observed data (the documents in Figure 2) as arising from a hidden data-generating process, the structure of which is to be estimated.
The model provides an answer to the question: what thematic structure is most likely to have – hypothetically – generated the observed data (the distribution of words making up each document in the corpus)?
Figure 2: The corpus: full text of 27,436 research articles by journal and year
Source: Bowles et al (2025)
Our topic modelling on the corpus in Figure 2 generated 100 topics (vectors of words whose weights measure the relative frequency of their usage). The topics are generated by an unsupervised learning algorithm. We named them by hand based on their most heavily weighted terms.
The lens provided by the topic model trained on this corpus records changes in the balance of attention given to themes relating to the state, the market and civil society, grouping subsets of the 100 topics into one ‘metatopic’ at each of the vertices.
Our procedure for assigning a topic to one of the three metatopics was to determine whether the words most heavily weighted in the topic are commonly associated with either the motivations or the mechanisms characteristic of the three vertices. Topics whose primary terms are unrelated to the three systems of societal governance (including those concerning mathematical or econometric methods shown in Figure 3) were not assigned to a vertex. Figure 3 shows the results of this exercise, confirming the growth and decline of mathematical topics and the well-established increased focus on empirical methods over the past few decades.
Figure 3: Mathematical theory and econometrics topics since 1900
Source: Bowles et al (2025)
Notes: The weights shown are the relative frequency with which works in the corpus of each period drew upon the indicated topics. Weights in the earliest period reflect the infancy of mathematical and econometric methods at the time. The topic numbers shown in parentheses are arbitrary.
We also created word clouds showing the frequency of words in the 100 topics and the ten papers drawing most on the terms heavily weighted in each of the three metatopics. Figure 4 gives an idea of their content. For example, important market ‘actors’ – firms, banks, consumers – are prominent in the ‘market’ metatopic.
Figure 4: Word clouds showing the most frequently used terms in the three metatopics respectively from left to right: state, market, civil society
Source: Bowles et al (2024)
Our topic model also allows us to locate any particular text in the state space of Figure 1, including the entire corpus under study for various periods (see Figure 5).
Figure 5: Research in economics: the turn towards civil society themes from the 1970s
Source: Bowles et al (2025)
We see a significant movement in economic research between 1900 and 1970 away from state-related topics towards market topics, even as the economic importance of the state was growing. This shift was slowed, but not reversed, by the war economy of the early 1940s. The period after 1970 witnessed an equally large and uninterrupted shift towards civil society topics, primarily at the expense of market topics.
Figure 6 (upper panel) shows that the shift from the neighbourhood of the state vertex towards the market vertex prior to 1970 was due to dramatic declines in the weight of the topics that we named ‘public regulation’ and ‘international trade institutions’ along with a substantial increase in the weights of two topics concerning market structure.
Figure 6 (lower panel) shows that the shift towards civil society over the last half century was due primarily to a reversal in the earlier increase in the weights of these two market structure topics and an increase in topics concerning asymmetric information, strategic interaction and the behavioural foundations of economics.
Figure 6: The shift from research on state to market topics prior to 1970 (top panel) and the components of the turn toward civil society topics after 1969 (bottom panel)
Source: Bowles et al (2025)
Discussion
The evidence in Figure 3 suggests that the shift from state to market was associated with the increased use of mathematics, especially constrained optimisation. This constituted a change in method – roughly the mathematical formalisation of Alfred Marshall – rather than any novelty in the kinds of economic behaviour under study.
Changes in method – including models of limited information and strategic interaction – along with advances in quantitative methods were also a hallmark of the subsequent shift in research towards civil society themes. But unlike the earlier methodological shift, research after 1970 was also characterised by an enlarged conception of the economic behaviour under study.
Since the late 1960s, this occurred by the application of conventional economic tools to an expanded set of topics – as in the work of Gary Becker on the family and crime – and also through the application of novel insights from other disciplines to the conventional subject matter of economics – as in the work of George Akerlof and co-authors on gift exchange and cognitive dissonance, and of Oliver Hart on authority in the firm and the exercise of power by owners and managers.
