Published by EH.NET (September 2009)

Dora L. Costa and Matthew E. Kahn, Heroes and Cowards: The Social Face of War. Princeton, NJ: Princeton University Press, 2008. xxvi + 315 pp. $28 (hardcover), ISBN: 978-0-691-13704-9.

Reviewed for EH.NET by Marianne Wanamaker, Department of Economics, University of Tennessee.

One of the first things I learned as an aspiring economist was the universal truth of ?diminishing marginal returns.? The more one utilizes any individual input, the lower the marginal return. Yet, a fantastic counterexample is present in the research ?input? that is Robert Fogel?s Union Army Sample. For unfamiliar readers, the Center for Population Economics at the University of Chicago has spent countless hours collecting military, pension, health, and Census data on more than 35,000 white male veterans of the Union Army. A similar, though smaller, project has done the same for U.S. Colored Troops.

The volume of research that has emanated from this single input source is astounding. Heroes and Cowards: The Social Face of War by Dora L. Costa (UCLA) and Matthew E. Kahn (UCLA) is the latest addition to this long line of research output, and I, for one, see no sign of diminishing value.

Costa and Kahn?s main question is how an individual?s surrounding peer group affects behavior. In the Union Army context, the military company served as the individual?s peer group. The authors use data on troop desertion, POW camp survival, post-war geographic and occupational mobility, and post-war educational attainment as outcome measures to show that military company composition had a significant impact on each of these outcome variables.

The advantage of using the Fogel Sample to analyze peer effects, Costa and Kahn point out, is two-fold. First, the assignment of individuals to companies was quasi-random such that peers were not necessarily self-selected and companies displayed varying degrees of heterogeneity. Second, the stakes were high. Continuous physical hardship and frequent death meant that the soldiers? situation was quite different from that in modern-day laboratory experiments attempting to measure peer effects and social interactions. (These experiments have been criticized because the stakes in them are low and behavior may be different with higher valued outcomes.)

Chapters 1 and 2 are introductory. After a brief literature review and a survey of the data, Costa and Kahn present case studies of nine Union Army soldiers. The experiences of the nine soldiers differed greatly, and the authors use these differences to motivate their research questions. (For instance ?Why were Adams French, James Monroe Rich, and Robert Rich such loyal soldiers?? to motivate an inquiry of desertion behavior.) These nine experiences are woven into the subsequent six chapters lending a richer context to the analysis.

Chapter 3 details the determinants of troop enlistment and the formation of military companies. The main conclusion here is that there was enough random variation in the personal characteristics of companymen to allow for unbiased peer-effects analysis. Military companies, then and now, hold a substantial amount of internal diversity in terms of place of birth, socioeconomic status, and age. The authors use this fact to identify the effects of company heterogeneity on soldiers? behavior and outcomes.

In Chapter 4, the authors use desertion as a negative outcome to assess why soldiers chose to fight. They find that, in addition to individual characteristics, company composition mattered. Company heterogeneity in age, place of birth, and occupation all led to increased probabilities of soldier desertion. Individual characteristics (i.e., personal ideology, age, place of birth and occupation) and other company characteristics (i.e., morale and leadership) still mattered, but company diversity was the main determinant of desertion. Interestingly, this was true for both white and black soldiers.

Chapter 5 is devoted to the relationship between survival and cohesion. The combat death toll for Civil War soldiers was high, and it would be enlightening to discern whether company cohesion affected the probability of death. But self-selection into the front line (cohesive companies tended to be where the fighting was most intense while less cohesive companies lost stragglers to the rear) makes identification difficult. Rather, Chapter 5 is devoted not to combat situations but to the survival probabilities of the unfortunate soldiers who found themselves in the Andersonville POW camp. In a reprise of their 2007 American Economic Review article, Costa and Kahn examine how a soldier?s access to his ?friends? (fellow companymen) affected his chance of surviving a camp where the death rate was 40 percent. While the presence of company friends increased the probability of survival, the average effect was small and dominated by the effects of age, camp size, the individual?s rank and his height. Interestingly, survival probabilities increased substantially with the presence of soldiers of the same ethnicity among German, French, and Irish soldiers.

Chapters 6 and 7 focus on the post-war experience of soldiers, both white and black. For white soldiers, the authors are concerned with the effect desertion had on post-war geographic and occupational mobility. Were deserters ?punished? by their local community in some measurable way? Were they ostracized or forced to migrate? For black soldiers, the questions are similar but focus more closely on the soldiers? peer group while in service. Did a diverse peer group encourage geographic migration? And what were the biggest influences on black soldiers? literacy?

For both white and black soldiers, social contexts mattered. White deserters were more likely to move away from their hometowns, and the effect was bigger for hometowns with a larger number of war supporters. These results are robust to instruments for desertion probability. White deserters were also slightly more likely to experience downward occupational mobility. Black soldiers who served in diverse companies (measured by the place of birth) were more likely to migrate following the war. Company makeup mattered in another important way as well. Those with a larger fraction of free blacks and with abolitionist officers were associated with higher literacy rates among former slave soldiers.

A simple criticism of this book is that its appeal to a more general audience may leave the economic history community less satisfied. This is especially true for scholars familiar with the relevant prior work of Costa and Kahn. Most of the econometric detail of their original research, including model specifications, standard errors and confidence intervals, r-squared statistics, sensitivity tests, and IV results, have been replaced with a number of literature reviews and background information in order to place the research in a wider context. For example, the authors include a history of statistical analysis of treatment effects and a history of prisoner survival mechanisms. Although this may appeal to readers interested in a social history of war, it does not serve the book?s main purpose of using the special circumstances surrounding the Civil War as a natural experiment in social interactions.

As a result, this book is a good introduction for economic historians unfamiliar with Costa and Kahn?s prior work. For those who are, it provides a larger context into which that research can be placed.

Marianne Wanamaker is an assistant professor of economics at the University of Tennessee. She can be reached at