Published by EH.Net (November 2012)

Daniel Berkowitz and Karen B. Clay, The Evolution of a Nation: How Geography and Law Shaped the American States.? Princeton, NJ: Princeton University Press, 2012. x + 234 pp. $39.50 (cloth), ISBN: 978-0-691-13604-2.

Reviewed for EH.Net by Robert A. Margo, Department of Economics, Boston University.

Empirical work on institutions arguably qualifies as one of the most active areas of research in economic history today.? Most of this activity takes its cues from Stan Engerman and the late Ken Sokoloff?s work on endowments and institutions, or from Daron Acemoglu, Simon Johnson and James Robinson?s work on colonial origins.? The majority of recent research has used countries as units of analysis but some researchers have turned to U.S. history for raw material.? Daniel Berkowitz and Karen Clay?s new book fits into this genre.? They focus on two features of the institutional environment ? political competition and judicial independence ? and how these were shaped by initial conditions.? This focus is warranted, they argue, because political competition and judicial independence are positively related to differences in development across countries.

Evolution is divided into seven chapters.? The Introduction (chapter one) sets the stage and summarizes the main arguments, of which there are three.? First, ?elites? have a big effect on initial institutions, and this effect operates primarily through their economic ?homogeneity,? here measured by the extent to which their occupations are similar or different.? Occupational homogeneity, it turn, depends on climate (temperature, precipitation) as well as geographic variation in the costs of engaging in trade, here measured by distance to navigable internal waterways and the ocean.? Second, judicial independence is affected initially by whether or not a state had experience with civil law, even though only state ? Louisiana ? carried civil law forward in its history.? Judicial independence also depends on political competition (and hence indirectly on geography) but not the reverse.? Third, initial levels of political competition and judicial independence persist over time, and (according to the authors) affect current levels of per capita income.

Chapter two classifies the legal origins of the various American states as common or civil, based on a careful assessment of admittedly sparse historical information.? The basic idea draws on the fact that certain areas that later became states ? for example Florida ? were originally under the control of foreign countries with civil law traditions, notably France and Spain.? There are thirteen such states.? Chapter three looks at ?political competition? at the state level and geography.? Political competition is measured by three indices, the most important of which is the ?Ranney? index, which measures differences in party composition in the upper and lower houses of state legislatures.? The index is constructed such that if there are two parties ? Republicans and Democrats ? and both have exactly 50 seats (say), then political competition is at a maximum of 100.? If all the seats are held by one party, political competition is zero.? The authors estimate regressions relating features of the body politic to four geographic indicators ? average temperature, precipitation, distance to navigable internal waterways, and distance to the ocean ? and to their civil origins dummy.?

Chapter four develops state-level measures of the influence of the economic elite on political competition and court independence.? This influence is captured by measures of wealth inequality in 1860 and also by the occupational homogeneity of the elite.? The elite are defined to be adult white men (ages 21 and older) in the top one percent of the wealth distribution in 1860.? Occupational homogeneity is a Herfindahl index of occupational shares, ranging from 0 (complete heterogeneity) to 10,000 (complete homogeneity). The other variable is the percent of total wealth in the state owned by the top one percent.? The authors estimate 2SLS regressions of the Ranney index (average values for different periods after the Civil War) on occupational homogeneity (or the wealth share) using distance to internal water transportation as the instrument.? The effects of occupational homogeneity are negative, somewhat larger than the OLS effects, and statistically significant.? Chapter five presents various regressions of state court characteristics on political competition and legal origins while chapter six does much the same focusing on court expenditures.? Chapter seven wraps up the book by summarizing the results of the previous chapters.? The chapter also contains some OLS regressions of state-level per capita income in 2003.
Evolution is a mixed bag.? On the positive side, Berkowitz and Clay deserve considerable credit for taking up the difficult challenge of applying the ES (Engerman-Sokoloff) and AJR (Acemoglu-Johnson-Robinson) approach to the experience of U.S. states.? Certainly anyone else contemplating something similar will need to study this book very carefully because they will have to grapple with some of the same issues faced by the authors.? The book is timely, well written, and the authors have amassed an interesting body of data.

On the negative side, I am unconvinced by the results.? There are three problems.

First, the ?mechanism? that Berkowitz and Clay adduce was responsible for creating the link between initial conditions and institutions, is vague, certainly when compared with ES or AJR.? In Berkowitz and Clay?s world, climate and geography ?affect? the distribution of elite occupations and wealth but the precise manner that this is supposed to happen is not made clear enough to this reviewer?s satisfaction, certainly not in the sense of a formal model.? I am not sure that it can be.

Second, I have reservations about the econometrics.? Here I have several problems, but I will concentrate in my review on one ? the treatment of region.? The civil origin variable is highly correlated with region, making it potentially difficult to identify the effects of civil origin if regional dummies are included in a regression.? The same is true of the other geographic variables. A natural robustness check, therefore, is to control for region whenever possible in the econometric analysis.?

An objection may be raised to this criticism ? one could say that Berkowitz and Clay are attempting to ?unpack? region by relating it to an underlying set of continuous exogenous variables.? Fair enough, but if the reader is to believe that the variables emphasized by Berkowitz and Clay are the right ones their influence should be evident within region.? If they are not, then either the story is wrong or else the evidence is not sufficient to distinguish among the various alternative hypotheses.? This latter possibility is an important constraint on using geographic variation within the U.S. to evaluate whether institutions matter and anyone working in this area must, in my opinion, confront it head on.? When Berkowitz and Clay estimate what are, in effect, cross section regressions ? for example, Table 4.4 ? they generally do not include regional dummies.? In such cases, I am skeptical about the substantive results because they hinge on the belief that the salient differences across regions are fully captured by Berkowitz and Clay?s geographic variables.? In fairness to the authors, many of the regressions using panel data include state fixed effects; however, these generally do not also control for time effects in the same regression.? When the authors do control for both state and time effects, the results are typically not so good (see, for example, Table 5.1a).?

Third, anyone fond of the AJR method of doing economic history likely will be disappointed by the per capita income regressions in chapter seven.? In AJR?s world, these would be estimated by 2SLS, but Berkowitz and Clay?s are OLS.? I?m not surprised because, if it were me, I would be worried that the instrumental variable used in chapter four, distance to internal waterways, would fail the exclusion restriction.? That means we are left with correlations ? and, taken at face value, correlations that look pretty small (judicial independence, civil origins) or are statistically zero (occupational homogeneity).?

Berkowitz and Clay wish their readers to believe that the diversity of historical experience across U.S. states provides valuable evidence that institutions matter.? After reading this book, however, I concluded that the opposite is a distinct possibility ? other than the obvious case of slavery, about which economic historians have already had much to say, the effects of whatever institutional differences across U.S. states remaining today from initial conditions are probably not worth worrying about or, alternatively, are unlikely to be identified econometrically to a reasonable degree of precision.?

Robert A. Margo is Professor of Economics, Boston University, and Research Associate, National Bureau of Economic Research.? He was recently elected a Fellow of the Cliometric Society.

Copyright (c) 2012 by EH.Net. All rights reserved. This work may be copied for non-profit educational uses if proper credit is given to the author and the list. For other permission, please contact the EH.Net Administrator ( Published by EH.Net (November 2012). All EH.Net reviews are archived at