Did southerners favor slavery? Inferences from an analysis of prices in New Orleans, 1805–1860

In a letter to Dr. Richard Price, Thomas Jefferson (August 7, 1785), for example, expressed the hope that the next generation of Virginians would continue the antislavery tradition and set the precedent for peaceful abolition throughout the South. He wrote to a friend discouraged by the pace of reform in the region: “These [young Virginians] have sucked in the principles of liberty, as it were, with their mother’s milk; and it is to them I look with anxiety to turn the fate of this question. Be not therefore discouraged.”

A generation later, Auguste Levasseur, Marquis de Lafayette’s private secretary, recorded (Levasseur 1829, p. 2): “It seems to me that slavery cannot subsist much longer in Virginia: for the principle is condemned by all enlightened men; and when public opinion condemns a principle, its consequences cannot long continue.”

This expression omits the correction or adjustment for female slaves, because of differences in mortality and the prospective value of child-bearing. For much more precise accounts of the range of, and explanations for, slave auction prices in New Orleans; see Fogel and Engerman (1974, pp 73–77) and Kotlikoff (1979). Some of the most important original work on pricing slaves as capital assets much like equities was by Conrad and Meyer (1958) and Evans (Robert 1962). A richer and more accurate model would obviously want to account for these broader considerations. Nevertheless, our basic claim that the lower the probability of emancipation or escape, the more valuable the slave was as an asset would be largely unaltered by complicating the model. As an empirical matter, if we were modeling the price of slaves at the micro-level, we would nevertheless want to take these considerations into account; however, we use the average annual price of field hands sold at auction in our econometric work. We are unaware of systematic changes in the composition of the slaves available for auction (e.g., their gender make-up, average health, or life-expectancy) beginning in the mid-1830s that would bias the results in favor of our hypotheses. We acknowledge the advice and corrections offered by an anonymous referee in presenting the present value, and correcting the example.

It is also interesting to note that after trending upward for around 20 years, slave prices spiked right before the onset of the Civil War. An interesting implication is that if slave owners believed that they were on the threshold of federally enforced emancipation, then the bottom should have fallen out of the market rather than see prices spike. Based simply on inferences from slave prices, it seems that the South believed that the threat of secession was sufficient to end once and for all questions about the persistence of slavery.

How reliable are these data? Aptheker acknowledges that reports of slave insubordination were often distorted during the antebellum period. On the one hand, because of the widespread fear of slave rebelliousness, reports of unrest frequently were exaggerated. However, Aptheker contends that his examination of the available evidence, requiring multiple sources to document whether an event occurred, effectively distinguishes acts of insubordination or panics from the rhetoric of the period. On the other hand, Aptheker argues that Southern elites appear to have engaged in a systematic effort to keep news of reported conspiracies and insurrections secret. However, a secret insurrection would probably not have had much of an effect on slave prices.

We are grateful to the editor for pointing-out that the late 1850s spike in slave prices coincided with the opening of formerly free territories to slavery by the Dred Scott decision.

This effect is clear when we test the hypothesis of joint significance using an F test. We calculated the F statistic in two different ways. First, we tested whether the aggregate regime effect was significantly greater than zero. That is, we tested the following null hypothesis:We calculated F _1,40 = 18.85, which allowed us to reject the null at p < 0.0001. Second, we tested whether the individual coefficients associated with the regime effect were significantly different from zero. That is, we tested the following null hypothesis:

We calculated F _6,40 = 7.89, which again allowed us to reject the null at p < 0.0001.