The Trajectory of a Booming Field: A Look Back at a Decade of HPE Research
Historical political economy (HPE) is a vibrant research field that traverses traditional disciplinary boundaries. In the past ten years, the number of articles in this tradition in top political science journals has more than doubled. In this post, I discuss some patterns and tendencies in the HPE articles in top political science journals that Eugene Finkel, Scott Gehlbach, and I have compiled for a joint project.[1]
First, a quick note on the definition of HPE, which is still subject to some debate. Political economy (PE), for its part, can be understood as the study of political institutions and behavior using game theory or empirical methods of causal inference. PE sometimes also refers simply to work at the intersection of politics and economics, regardless of method (see Sean Gailmard’s discussion on this in the first issue of the JHPE).
Where does HPE end?
But what is history, and when does an empirical case shift from historical to contemporary? Here Sean Gailmard’s definition of historical events as those that “occurred in the past under temporally bounded social or institutional configurations no longer in operation in the place where the event occurred” is particularly useful. This implies that the specific time period will vary from case to case.
For coding purposes, however, we had to set a precise cutoff. We ultimately settled on the year 1945 (including articles where the main explanatory variable was measured in 1945 or earlier). It was an important juncture in world history, coinciding with the end of the Second World War, the onset of the Cold War, and the formation of the Bretton Woods institutions. In much of the world, political regimes and institutional configurations changed. The year 1989 was another contender but seemed more relevant for some cases (esp. post-communist states) than others. 1989 is also a date recent enough for data to be available outside archives and for which a prominent type of HPE research, long-run historical legacies, is less applicable.[2]
Methods, topics, and regional focus of HPE research
HPE as a field developed in parallel with the credibility revolution in the empirical social sciences and benefited from the development of new technological tools such as OCR and GIS. A defining characteristic of HPE studies, in addition to a focus on history, is attention to statistical identification and research design. HPE research often uses quasi-experimental techniques to isolate the causal effect of a specific variable. That often requires the creation of original quantitative datasets and integration of spatial variables, which are another hallmark of HPE research.
The focus on research design, coupled with greater data requirements, has contributed to a marked shift from cross-national comparisons to within-country designs – which make up 67% of the articles we analyzed.
Thematically, HPE work is extremely heterogeneous. The articles we examined range from explaining gender gaps in voter turnout to the legacies of slavery. A keywords-based analysis showed that the top five themes are: colonialism, the state, institutions, democratization, and conflict.
Not all parts of the world are equally represented in current HPE research. A significant share of articles use data from the US (25%) and the UK (11%). Other countries in Western Europe (excluding the UK) account for another 19% of articles, with Germany the most frequent. Articles spanning multiple countries account for 33% from the total. Data availability and archival access, English language, and American politics as a separate subfield likely contribute to this pattern.
HPE studies generally fall into one of three categories: (1) the past as a cause; (2) the past as a setting; and (3) history as a way to understand the present. These categories sometimes overlap and distinguishing them can be a matter of framing. I discuss each category in more detail below.
HPE as a way of understanding the past
First, political scientists draw on historical cases as new way of understanding the past for its own sake. This is the largest category: just under half of all HPE research fits this model. This work reexamines the conclusions of prior research by applying new methods to old data or, more frequently, collecting original quantitative data that allows for more rigorous testing of old theories. Studies of the origins of important institutions, such as the legislature, property rights, proportional representation, feudalism, and territorial states are prominent in this group. Some articles also use panel data spanning many decades to examine structural change in how institutions work.
For instance, in a 2018 CPS article Gary Cox, Jon Fiva, and Daniel Smith reexamine the origins of proportional representation systems (PR) in Europe using roll call data from Norway in 1918-1923. They show that party leaders favored PR because they would gain control over nominations and thus increase party cohesion under this system. Their analysis challenges the dominant explanations of PR adoption as a way for conservative parties to preserve their seat shares following the extension of franchise.
Another example is a 2020 APSR article by Agustina Paglayan, which shows that in much of the world, public primary schooling emerged a century before democratization – and thus challenges conventional explanations of the rise of primary education. Her analysis uses datasets covering 109 countries over 200 years.
History as a setting
Another approach uses the past as a setting to investigate general theoretical issues. The trick is extending the insights from a historical case to contemporary contexts. This can be the case when a historical case offers better data or identification to address the limitations of more contemporary cases. The most successful work in this genre uses theory to motivate the analysis and discusses generalizability of the insights outside the specific historical context. This research requires deep case knowledge, but can appear motivated more by identification than theory.
One example in this tradition is a 2020 JOP article by Francesc Amat, Charles Boix, Jordi Muñoz, and Toni Rodon. They examine the effect of parties, trade unions and their associational networks on turnout among low-income citizens using a unique panel data set of official registers with information on individual voting and demographics of almost 25,000 electors in Barcelona in the 1930s. The granularity of the dataset allows the authors to address ecological inference problems associated with the use of aggregate voting data and with the overestimation of turnout in surveys.
The past as a way of understanding the present
History can also be used to explain variation in present-day outcomes. Papers on the long-run consequences of historical events fall in this category. A growing number of studies analyze the effect of colonial legacies on long-term economic and political outcomes. The seminal work in this literature is the 2001 AER by Acemoglu, Johnson, and Robinson (AJR) on the effect of colonial institutions on contemporary economic growth. Like AJR, the first wave of research has focused on country-level variation.
The recent trend is toward studying the political and economic consequences of subnational variation in the colonial governance and delving into the internal dynamics of colonial rule to gain leverage. For example, in a 2018 APSR article Jenny Guardado exploits the variation in prices at which the Spanish Crown sold colonial provincial governorships in 17th- and 18th-century Peru, which varied due to the timing of European wars, to investigate the effect of governor quality on long-run developmental outcomes.
Another subset of this research examines the enduring legacies of repression and violence – the series of papers on Nazi repression in Germany and Poland fall into this category (see my earlier post on this). Studies on the persistence of cultural norms and values are also a popular subset of this line of work, but are more prevalent in economics than in political science.
The average span of years between cause and outcome in the articles we collected was 222 years. The award for the longest period goes to Andreas Wimmer’s 2016 CPS article that challenges the diversity deficit literature by proposing that ethnic fractionalization and the state’s capacity to provide public goods may both result from a previous history of state formation. Wimmer provides evidence from multiple cross-national analyses that span 50-100 years, but also includes a regression of contemporary ethnic fractionalization on the index of state centralization that starts in 100 BC. In case you are wondering, state centralization predicts ethnic fractionalization some two thousand years later.
Having written several articles in this genre myself, I see exploring the mechanisms of persistence and eliminating the many intervening variables as its main challenge in this type of research. The longer the time period between “treatment” and outcome, the more challenging the interpretation becomes. This type of research has also been challenged on grounds of spatial autocorrelation. For a rundown of main challenges of the “legacies” research, see this post by Alberto Bisin. Also see my earlier post on intergenerational transmission mechanisms that help explain the persistence of cultural norms and values.
We welcome comments on the definition, cutoff date, and the types of HPE research.
For another, more light-hearted take on the types of HPE papers, check out this post by Ali.
—
[1] So far, we coded 166 articles in the HPE tradition in AJPS, APSR, JOP, BJPS, and CPS. We are adding more journals.
[2] But see fascinating research on the legacy of Soviet Communism by Grigore Pop-Eleches and Joshua A. Tucker.