As political economists we are, collectively, concerned with the historical determinants of the distribution of political influence, its institutionalization, and its consequences for efficiency and equality over the medium to long run. Power and its distribution remain the central focus of politico-economic analysis since the very inception of the discipline. Similarly, the dialogue with history and historians is neither new nor easy. Back in 2007, Adam Przeworski asked, rather skeptically, whether “the science of comparative politics” was actually possible.
By definition, history yields censored data, biases the very possibility of measurement, establishes a close link between the effects of any institution of interest and the conditions that facilitated the emergence of the institution in the first place, removes, almost by default, any notion of spatial or temporal independence across units of observation, and, as a result, makes the very notion of a feasible counterfactual hard to conceive. At the time, historical analysis was seen more as a different camp, a refuge for those frustrated with scientific “jargon” (though particularly tolerant to their own), than a potential complement to reconcile our focus on complex, non-manipulable processes and the scientific enterprise. Thanks to the cumulative impact of many great contributions, the terms of the conversation are changing.
What I think identifies historical political economy (HPE) as an approach relative to prior efforts is a three-fold commitment. The first one is theoretical: an explicit effort not to let the description of a process substitute for an argument (often in the form of an endless forest of arrows quickly progressing towards negative degrees of freedom) and, more constructively, to develop an abstract logic about the causal relationships of interest that, subsequently, guides empirical efforts. The second one is methodological: a pledge to both dig deep into historical sources to improve measurement as a theoretically driven exercise and pursue, as far as possible given the data context, compelling research designs. Finally, the third one is a commitment to transgression, transgression of artificial disciplinary boundaries and scholarly prejudices to embrace the diffusion of tools, approaches, models and techniques to maximize access to new data (often in vast amounts) and, critically, to analyze it rigorously.
In analyzing power, the study of the mechanisms that govern the transmission of political influence over time within societies remains a central problem in the field. Do not get me wrong: this very question is in large part the outcome of vast literatures on colonial legacies across several disciplines out of which a long list of potential channels has emerged. These include, non exhaustively, economic and political geography, regulations on migration and factor mobility, choices over enfranchisement, public goods provision, allocation of infrastructure, legal institutions/property rights, and, of course, political institutions. While recent contributions have taken on how the foundational bargains in a regime shape central aspects of its practice and quality (e.g. Albertus and Menaldo 2018), there is large room for work on the analysis of institutional transmission mechanisms of elite advantage from the HPE perspective described above.
In a relatively recent article published by CPS (Economic Geography, Political Inequality and Public Goods in the Original 13 US States) Jeff Jensen and I offered a preliminary analysis of the transmission of elite bias through the system of representation in the United States from 1787 until 1961. The institution of interest in this case was legislative malapportionment at the state level and the outcome differences in public goods provision across states. At the time, we perceived the focus on representative institutions as a transmission mechanism of colonial legacies on political inequality and economic outcomes to be less developed than other channels (property rights, ethnic conflict, direct vs. indirect rule).
During the British North American colonial era (1607-1775), a system of “corporate” representation was established in each of the colonial legislatures, in which the town, parish, or county was the basis of representation. In none of the 13 colonial legislatures was representation apportioned on a “one (white) man, one-vote” basis. The only way to gain the representation of a particular area in the state legislature was to create a new unit (county, parish) and the winners of the system had little incentive to do so. Since the creation of new political units would threaten coastal majorities, coastal elites strategically limited their formation. As people moved westward, the underrepresentation of significant shares of the population increased. An English Royal Proclamation in 1765 forbade the colonial legislatures from reapportioning representation, further fueling tensions. Eventually, with the May 10th Resolutions of 1776, the Second Continental Congress called on each colony to create new governing structures by writing their first sovereign constitutions.
Given this autonomy, elites in each new state had to determine, among other critical choices, whether to maintain colonial institutions that constrained the political power of the poor, especially on the enlarging frontiers. Interestingly, despite the opportunity to institutionally entrench the colonial system of representation in their first sovereign state constitutions, elites in four states—Massachusetts, New Hampshire, New York, and Pennsylvania—implemented a system of representation based on population and regularly-scheduled reapportionment. We classify these four states as proportional-apportionment states (PAS). The remaining nine continued with a corporate basis of representation, which we designate as biased-apportionment states (BAS), in their first sovereign state legislatures. In the paper we approached this choice as an opportunity to evaluate the long run effects of representation systems on political inequality and public goods provision.
The case illustrates a series of questions that, more broadly, applied projects focusing on the dynamics and transmission of political influence should address:
- What is the origin of the institution? Critically, can we credibly establish that the origin is driven by factors other than the outcome of interest?
Yes. As in the case of the franchise (Nikolova, 2017), economic geography shapes the elite’s choice of malapportionment through its impact on labor scarcity and the importability of slaves. The shift to population- based apportionment takes place where labor scarcity is high and climatic conditions are unsuitable for growing cash crops.
As is evident in the maps above, there only four states at the end of the colonial era which were both sparsely populated (left) and unable to meet labor needs with imported slaves (right): Massachusetts (which at the time included present-day Maine), New Hampshire, New York, and Pennsylvania).
- What are the theoretical expectations about its impact on the distribution of political influence as well as its effect on the outcome of interest?
We expected political inequality to be much higher in BAS than in PAS states and this, in turn, to translate into the capture of public goods for the benefit of overrepresented elites. Persistent political inequality has tangible material consequences.
- Is it possible to trace (empirically) and analytically account for the institution’s continuity over time? Additionally, is it feasible to quantify its persistent effect on the outcome of interest?
Yes. We were able to compute the Relative Representation Indexrepeatedly starting in 1775 (in the colonial legislatures) and then every half-century since 1790 (first post-colonial census) through 1960, documenting persistent differences in political inequality across the two subgroups of states. Figures below shows the relationship between a county’s representation in its state legislature and its distance from the Atlantic coast in 1775, 1790, 1850, and 1950. The first (1775) shows that the bias was present in all states during the colonial era and the choice in the four PAS to change this resulted in immediate change in representation (1790 onwards). This remained all the way until the mid-20thcentury.
We also show that these differences lead to significant differences in state spending per (white)capita on education during the period 1850-1960. Interestingly, political and economic modernization only increased the winners’ incentives to protect their institutional advantage.
- If applicable, what accounts for the ultimate removal of that particular institution and with what effects on the outcome of interest?
In 1964, the US Supreme Court required the apportionment of state legislatures be equalized across districts according to population should have resulted in a significant reallocation of state public resources toward the previously underrepresented counties within each state (Ansolabehere et al. 2002).
Our initial approach leaves much to desire for but it exemplifies the potential payoffs of analyzing and measuring transmission directly over time. Obviously, the logic is not restricted to institutions. In similar spirit and much greater detail, other scholars have analyzed other forms of transmission by focusing on the long run imprint of slavery of political behavior (Acharya et al. 2018).
There is much work ahead in this agenda. As an earlier post of Sean Gailmard argued, it is one thing to establish a relation in one case (even if it was perfectly identified, which in this case is not) and a different one to develop a theory of the transmission of political influence capable of integrating multiple contexts. Looking ahead, while we are gaining a lot of insight from in depth studies on specific instances of political manipulation (Mares and Young 2019, Gingerich 2020), there is much to gain from a more integrated understanding of how elites design their portfolio of strategies to preserve their advantage in different contexts and with what consequences for the nature and stability of political regimes.
What seems increasingly a thing of the past, and I take this to be a sign of progress in the field, is to document a strong correlation between a factor of interest and an outcome hundreds of year later without modeling and measuring directly (often at lower levels of analysis) the various channels driving the association. To the extent that HPE continues to build a cumulative body of evidence on the origins, persistence, and effects of institutional interventions oriented to the preservation of political advantage, we will be closer to a comprehensive theory of the transmission of political influence and perhaps grow more optimistic about the feasibility of a “science of power”.
 To calculate the RRI (Ansolabehere et al. 2002), we identify each state’s electoral la specifying the number of representatives and senators, respectively, apportioned to each county of the original 13 states for each decennial census year. We then divide each county’s apportionment of representatives and senators, respectively, by each county’s adult White male (AWM) population. For each county, we divide this ratio by another: the state’s total number of representatives and senators, separately, divided by the state’s total AWM population. This results in a relative measure of representation for each county-chamber of the (bicameral) state legislature. Any value greater than 1 indicates that the county is overrepresented relative to “fair representation” in the respective chamber of the state legislature. Any value less than 1, therefore, indicates the county is underrepresented.