Gambling on War: Confidence, Fear, and the Tragedy of the First World War? by Roger L. Ransom (review)

Ransom draws from theories of decision making in social psychology and behavioral economics to explain why political leaders started “a war that nobody wanted, nobody understood, and nobody can forget” (272), and why they did not stop the war earlier once illusions of a quick victory had dissipated. Ransom argues that political and military leaders, faced with uncertainty and driven by fear and overconfidence, gambled on risky strategies that failed and prolonged the suffering without advancing any state interests.

The first chapter lays out the analytical themes underlying the historical narrative. Noting Keynes’ argument that human decisions are often driven by “animal spirits” rather than by a rational calculation of costs and benefits, Ransom emphasizes the role of confidence, fear, and the propensity to gamble in human decision making.¹ He analyzes the propensity to gamble through the lens of “prospect theory” from social psychology (19–24).² The theory posits that people overvalue losses relative to comparable gains. They are risk-averse when faced with choices expected to lead to positive outcomes and risk-acceptant in choices resulting in negative outcomes—especially when the current status quo is perceived negatively because of “sunk costs” from recent losses. Under these conditions, people sometimes take bold gambles in the hope of eliminating their losses but in the process incur even greater losses. Ransom uses this logic to explain a number of key decisions embedded in his historical narrative, in which leaders chose to gamble on continuing the war rather than to accept a negotiated settlement that would leave them nothing to show for the destruction, pain, and suffering. German decision makers, for example, chose to resume unrestricted submarine warfare despite the high risk of American intervention because the status quo was unacceptable and because, as a German admiral argued, “we have no other option” (160). As Ransom concludes in the epilogue, “Politicians and generals took risky gambles to attain . . . victory, which only magnified the cost of the war if the gambles failed” (271).

Gambling on War is a highly readable narrative of the diplomatic background, origins, conduct, and termination of World War I. Ransom gives more attention to the conduct of the war than to its origins, providing particularly good accounts of states’ economic capacities to continue fighting. In this context, however, he might well have offered a more thorough discussion of the presence or absence of advance economic planning for war, and the implications thereof for various interpretations of the causes of the war.

The analytical themes of fear, overconfidence, and gambling—from the calculations of Otto von Bismarck to the Schlieffen plan and the Ludendorff offensive—distinguish this book from others in a crowded field. Previous historians have emphasized these themes, but Ransom’s idea of building upon social-science research allows for a much more rigorous examination of the psychological sources of risk taking and a new way of thinking about decision making leading to the outbreak, escalation, and continuation of the war. Ransom’s summary of the prospect-theoretical analysis of risk taking will not satisfy social psychologists or behavioral economists (Ransom draws less from research about fear and overconfidence), but it is good enough to guide the historical narrative that follows.

The problem lies in the implementation. The historical narrative is only weakly guided by the theoretical framework outlined earlier. In explaining many of the high-risk gambles that continued the war without producing gains, Ransom repeatedly resorts to the overly general concept of “animal spirits” rather than to the more specific and analytically discriminating concepts of loss aversion, sunk costs, and risk-acceptant decision making, which are not prominently mentioned after the first chapter. An analysis guided by these concepts would not be an easy task, given the many difficulties of applying them outside of a controlled experimental setting, but their potential for improving our understanding...