The ‘great inventions’ view of productivity growth ascribes the excellent growth from 1920 to 1970 in the US to a handful of advances, and suggests that today poor productivity performance is driven by a lack of breakthrough discoveries. This column argues instead that the development of an effective governmental infrastructure in the 19th century accounted for a major part of US technological progress and prominence in this period. Infrastructure design thus appears to have the power to reinvigorate technological progress.
By Daron Acemoglu, Jacob Moscona and James A Robinson*
Robert Gordon’s new book, The Rise and Fall of American Growth, argues that rapid technological progress in the US economy between 1920 and 1970 was a result of the availability of ‘great inventions’ that had the potential to drastically change the way individuals lived their lives (Gordon 2016, p. 2). Present day economic growth is slower because inventions that have the transformative power of electricity and the internal combustion engine are no longer emerging.
This perspective runs counter to approaches that emphasise how the pace and direction of technological change respond to incentives and opportunities, often shaped by the institutional environment and policy decisions (see Acemoglu 2009 for an overview).
- Patents, property rights and functioning judicial institutions, for instance, allow individuals to reap the rewards of their investments and new ideas;
- Educational institutions and policies and a legal environment ensuring lack of discrimination against specific groups are also key for opportunities in business as well as in innovation to be open to most individuals in society; and
- Subsidies for research and development or tax credits are important for innovation incentives as well.
These factors have not only been shown to be important in general, but also to have played a critical role in 19th century American innovation (e.g. Sokoloff 1988, Khan 2005).
In a recent working paper, we empirically tests the hypothesis that the US government’s infrastructural capacity helped drive innovation during the 19th century (Acemoglu et al. 2016). Our results suggest that, notwithstanding the view that the American state was weak in the 19th century, a major part of the explanation for US technological progress and prominence is the way in which the US developed an effective state.
State capacity and the Post Office
We measure state capacity by making use of the fact that during this period one of the most widespread and instrumental federal institutions was the post office, established by the Post Office Act of 1792. By 1816, 69% of the federal civilian workforce were postmasters and by 1841 the figure had grown to 79% (John 1995, p. 4). According to John (1895), “[F]or the vast majority of Americans the postal system was the federal government” (p. 4, italics in original).
The scale and significance of the post office during the 19th century was noted by its contemporary observers as well. In his famous travels through the US, Alexis de Tocqueville wrote, “There is an astonishing circulation of letters and newspapers among these savage woods… I do not think that in the most enlightened districts of France there is an intellectual movement either so rapid or on such a scale as in this wilderness” (de Tocqueville 1969, p. 283). In 1852, the New York Times described the post office as the “mighty arm of the civil government” (John 1995, p. 10). In the spirit of the empirical approach in Acemoglu et al. (2015), we use the number of post offices in a county as a proxy for the general infrastructural power and presence of the state, and argue that it was this state presence – not just good timing, randomness, or external factors – that made 19th century innovation and patenting feasible and desirable.
Using historical records compiled by the US Postmaster General, we determined how many post offices were in each US county for several years between 1804 and 1899.1 As a measure of county-level innovative activity, we use the number of patents granted to inventors living in the county (these data are presented in Akcgit et al. 2013).2 There are several reasons for expecting the number of post offices to impact the number of patent grants. First, post offices facilitated the spread of ideas and knowledge. Second, more prosaically, the presence of a post office made patenting much easier, in part because patent applications could be submitted by mail free of postage (Khan 2005, p. 59). Third, the presence a post office is indicative of – and thus the proxy for – the presence and functionality of the state in the area. This expanded state capacity may have meant greater access to legal services and regulation, or greater security of other forms of property rights, all of which are essential conditions for modern innovative activity.
We find a significant correlation between a history of state presence – using the number of post offices as a proxy – and patenting in US counties. We show that the correlation holds either using a sample of the 935 US counties that had been established by 1830, or using a sample to which counties are added as they were established between 1830 and 1890, ultimately reaching 2,644 in total.3 This relationship is not only statistically significant, but also economically meaningful. Our results suggest that the opening of a post office in a county that did not previously have a post office or patents on average increased the number of patents by 0.18 in the long run.
We subject these results to a series of robustness checks. In addition to county and year fixed effects included in all regressions, we control flexibly for a broad range of initial county characteristics – the fraction of the population that were slaves in 1860, the fraction of the adult population that was literate in 1850, and the values of farm and manufacturing output relative to population in 1850. Despite the inclusion of these 36 controls, the relationship between post offices and patenting remains highly significant. We also include county-level linear trends to check that differential county-level trends do not explain our results.
One concern with this initial set of results might be that they are confounded by the possibility that post offices were built in counties that already had more patenting activity. Though we cannot fully rule out such reverse causality concerns, we find no statistically or economically significant correlation between patenting and the number of post offices in a county in future years. This suggests that post offices led to patenting and not the other way around. Historical evidence also suggests that post offices were established for a range of idiosyncratic reasons during the 19th century, making it unlikely that reverse causality is driving the association. John (1995) [A8] notes that pressure for the state’s services from certain segments of society “guaranteed that the postal network would expand rapidly into the trans-Appalachian West well in advance of commercial demand” (p. 44-5). In his early history of the US Post Office, Cushing (1893)[A9] wrote that post offices were often established in US territories before the territories were formally settled:
“The establishment of post offices in Oklahoma and in other regions recently opened has often been in advance of actual settlement. Before Oklahoma counties were named they were called by the Department A, B, C, D, E, etc. … Postmasters were appointed upon recommendations of the delegate from Oklahoma and of Senators Plumb, Paddock, and Manderson” (p. 286).
In this context, it seems improbable that post office construction followed patenting activity.
Taken together – while we do not establish unambiguously that the post office and greater state capacity caused an increase in patenting – our results highlight an intriguing correlation and suggest that the infrastructural capacity of the US state played an important role in sustaining 19th century innovation and technological change. In the current economic climate in which pessimism about US economic growth prospects is common, we present a more optimistic historical narrative in which government policy and institutional design have the power to support technological progress.
*About the authors:
Daron Acemoglu, Professor of Applied Economics, MIT
Jacob Moscona, PhD candidate in Economics, MIT
James A Robinson, University Professor, University of Chicago Harris School of Public Policy
Acemoglu, D. (2009), Introduction to Modern Economic Growth, Princeton University Press
Acemoglu, D., C. Garcia-Jimeno and J. Robinson (2015) “State Capacity and Economic Development”, American Economic Review, 105 (8), 2364-2409
De Tocqueville, A. (1969), Democracy in America, ed. J P Mayer, Garden City, Doubleday & Co.
Gordon, R. J. (2016), The Rise and Fall of American Growth, forthcoming, Princeton University Press.
 The years for which we were able to obtain county-level post office data are 1804, 1811, 1819, 1830, 1837, 1846, 1850, 1855, 1867, 1870, 1879, 1891, and 1899. For the years before 1879, we used United States Post Office Department publications titled List of the Post Offices in the United States (in some years, the publication was referred to as Table of Post Offices in the United States). In 1874, the federal government began publishing post office information more systematically in a publication titled The United States Postal Guide, which is digitised only for some years. This publication is our source for the years 1879, 1891, and 1899.
 We are grateful to Tom Nicholas for sharing these data.
 The former results in a balanced panel of counties while the latter is an unbalanced panel in which counties are included in the data only for the years after their establishment.