In response to my post about Jeff Hummel’s article on slavery, “U.S. Slavery and Economic Thought,” in David R. Henderson, ed., The Concise Encyclopedia of Economics, commenter Warren Platt quoted one part of Jeff’s article and made a lengthy comment. Here’s the part Warren quoted:
Strictly speaking, economists usually and most broadly employ the term “efficiency” as a measure of welfare rather than of output. Thus, while economic historians now agree that antebellum slavery marginally increased the output of cotton and other products, it still could have diminished total welfare. In measuring efficiency, economists have no precise unit to compare the subjective gains and losses from involuntary transfers. But because most coercive transfers in the Old South were from poor slaves to rich slaveholders, to assume unrealistically that such transfers were a wash in which slaveholder gains equaled slave losses is to bias the analysis in favor of slavery. [bold added by Warren.]
Warren wrote:
I guess what’s bugging me is the concept of “efficiency” as deployed in Jeffrey’s encyclopedia entry is that it may be too squishy to be of objective use.
Strictly speaking, economists usually and most broadly employ the term “efficiency” as a measure of welfare rather than of output. Thus, while economic historians now agree that antebellum slavery marginally increased the output of cotton and other products, it still could have diminished total welfare. In measuring efficiency, economists have no precise unit to compare the subjective gains and losses from involuntary transfers. But because most coercive transfers in the Old South were from poor slaves to rich slaveholders, to assume unrealistically that such transfers were a wash in which slaveholder gains equaled slave losses is to bias the analysis in favor of slavery.
I get it that “Slaves were being pushed off their preferred labor-leisure trade off, forcing them to work harder and/or longer than they otherwise would have chosen.” And that “This increased labor input caused output to increase but welfare (economic well-being) to decrease [for the slaves!]
But what about the slaveholders? Couldn’t it be equally argued that the gain in economic well-being exceeded the losses of well-being to the enslaved? The Southern planter class was the closest thing the USA ever had to a European-style aristocracy. Natchez Mississippi in 1860 had the highest percentage of millionaires (who would be worth at least $33 million in today’s dollars) of any town or city in the United States. Could not one argue that such a level of opulence enabled a qualitatively higher, refined lifestyle that far exceeded the welfare losses of the slaves? And that after emancipation, the slaveholders were pushed off their preferred labor-leisure tradeoff and that their loss may have exceeded the gain to the formerly enslaved persons?
If one wanted to put a dollar figure on the gain to economic welfare to the planter class that could perhaps be approximated by the risk and the cost of the war itself. One website says the direct costs to the South were around $3.25 billion. Then of course there was the loss of the value of the slaves themselves. So we’re talking around $7 billion, that would have been worth 100 years of the net economic losses estimated by F&E that only took into account the gain to planters of the extra cotton production, or around 10 years worth of the entire Southern regional income. (Imagine the United States of today expending $230 trillion in a risky & ultimately vain attempt to preserve the current status quo.)
Now, don’t get me wrong: I am not trying to make a case for slavery. I get it that slavery is superbad. But slavery can be sufficiently condemned on moral grounds with no need for economics to bend over backwards to show that the Southern planters were not only immoral but were also stupid. Certainly, Thomas Jefferson was immoral as judged by our 21st century morality; but to say he was stupid? That’s a stretch. [bold in Warren’s original.]
I was pretty sure I knew the answer from having read and edited Jeff’s article. But I was also sure that Jeff would do a better job than I at explaining. Like Jeff, by the way, I had no idea why Warren thought that Jeff was claiming that Thomas Jefferson was stupid.
Jeff has once again stepped up to the plate and responded. Here’s his response:
Warren, yes some slaveholders were very wealthy. But bear in mind that only about one-fourth of white households in the Old South owned any slaves at all, and about half of those owned fewer than five. The South’s per capita income in 1860, including the slaves, was about 20 percent below that of the North. Even if you completely exclude the third of the population enslaved, per capita income in the South was still only about equal to per capita income in the North. Thus the tiny minority of planters who enjoyed a “level of opulence” that “enabled a qualitatively higher, refined lifestyle” were confined to no more than one percent of heads of white households.
Of course, all slaveholders were wealthier than their slaves because they owned between $2.7 and $3.7 billion of their slaves’ human capital. But because wealth and income are related yet still distinct measures, that total alone tells us little about the impact of emancipation on incomes, much less welfare. Emancipation did not eliminate this human capital; it just transferred it back to its rightful owners, the slaves. Around four million former slaves became wealthier, while a far smaller number of former slaveholders became a lot poorer. Yet because the freed slaves did not use all of their restored human capital to generate income but instead to consume more leisure, total commodity output initially fell after emancipation. Because the freed slaves were now devoting a portion of their human capital to consumption, the resulting reduction in the monetary value of the freed slave’s stock of human capital is roughly equivalent to the discounted present value of the prior increase in the slaveholders’ income flow from forcing slaves to work harder and longer than they would have freely chosen.
How large was this additional income gain to planters? For one year alone, 1850, Fogel and Engerman estimated it to be $10 million. This is not slaveholders’ total return on their exploited human capital but just the increase from forcing slaves off their preferred labor-leisure trade off. F&E also estimated that the resulting increase in output also generated an additional $14 million to consumers of slave products in the form of reduced prices. Based on the wage premiumthat would have been required to entice free labor to work at the same pace as slaves, F&E concluded that slaves suffered an additional loss of $84 million in potential income. No matter how you slice the gains from the increased output, and even assume all gains went to slaveholders instead of consumers of slave products, you still have a total deadweight loss of $60 million. That is how much the welfare loss to slaves exceeded all the gains from additional output to anyone else in the world.
Based on alternative estimates of the underlying variables offered by other cliometricians, you can end up with total deadweight loss slightly lower or much larger. But none of these alternative calculations comes close to finding that the gains to slaveholders were greater than the welfare loss to slaves. That is why my entry brackets the annual deadweight loss from output inefficiency to the southern economy (ignoring gains to foreign consumers) at between $38 million and $176 million. Unless you drastically reduce the wage premium required to attract free labor to plantation work, it is impossible to get a result in which the slaveholders’ monetary gain exceeded the slaves’ welfare loss. Indeed, if the slaveholders’ gain had in fact been greater, after emancipation planters should have been able to pay former slaves or other workers a wage high enough to continue laboring on plantations. Yet the only place this actually happened to a limited extent was on some particularly lucrative sugar plantations in Louisiana.
As for slaveholders being stupid, nowhere did I suggest that nor does my analysis in any way imply such. Rather it treats planters as hard-nosed, capital-owning businessmen guided by profit under the constant pressure of competition. Only the slave owner willing to employ his slaves in the most productive manner would earn a return sufficient to cover slave prices. But that does not mean that the outcome was efficient in the sense of maximizing welfare. The businessman who successfully lobbies for a lavish government subsidy is certainly not stupid and usually hard working. But under most circumstances, economic analysis demonstrates that the resulting gains to him and his customers are less than the losses imposed on taxpayers, potential competitors, and others.
Given that most planters were already hard working, it is not likely that emancipation significantly altered their preferred allocation between labor and leisure. Even if it did, changing one’s preferred labor-leisure trade off is far different than being violently forced off one’s preferred labor-leisure trade off. People’s preferences can alter all the time due to changing market circumstances, and doing so maximizes their subjective welfare. Surely forcing a teetotaler to pay for and drink alcohol against his or her will is not quite the same as a former teetotaler who has voluntarily decided to buy a drink. In fact, the supply of labor is partly a function of how changing wages cause individuals to change their labor-leisure trade-off. Or, to return to our businessman with a government subsidy financed with taxes, if instead his enterprise was financed with voluntary contributions, then that part of what welfare analysis previously counted as a loss would now become a gain. You likewise would get a different result if the slaves worked on the plantations voluntarily (but then they wouldn’t be slaves).
You are free to eschew welfare analysis as “squishy,” and it admittedly can be imprecise, but doing so deprives you of a useful and standard economic tool for evaluating government policies. You then can still argue that lavish government subsidies are immoral but not that they are inefficient. [bold and italics in original]
READER COMMENTS
Thomas Lee Hutcheson
Jan 5 2023 at 9:39am
Surely the standard welfare-economics view would be that that the legally constrained labor market transaction would be less efficient than a free market labor transaction? It’s nice to have the numbers, of course.
David Henderson
Jan 5 2023 at 9:57am
Yes on both.
Fazal Majid
Jan 5 2023 at 9:40am
You’re missing the depressing effect on the wages of the 75 to 99% of white Southern men who didn’t own slaves and were competing with them.
I never understood why so few rebelled against the Confederacy like West Virginians or the Free State of Jones, and went on to give their blood for the benefit of an aristocratic elite that was also oppressing them indirectly. I guess nationalism is a potent drug.
Charley Hooper
Jan 5 2023 at 3:52pm
I think part of the answer is that there were multiple reasons for Southerners to fight in the Civil War: to preserve slavery, states rights, the North was being a bully, peer pressure, and nationalism.
The fact that so many white Southerners who didn’t own slaves fought provides some evidence that the reason wasn’t wholly preserving slavery.
Northerners certainly didn’t think they were fighting to end slavery—not that that proves what Southerners were thinking.
Mark Z
Jan 5 2023 at 3:56pm
I think according to Stanley Engerman, areas of the south least sympathetic to the confederacy – like West Virginia – were the Appalachian regions where there wasn’t much cotton cultivation or slavery, and thus where free workers were least competitive with slaves. It would seem that people tend to see their own economic interests as intertwined with those in their region.
Charley Hooper
Jan 5 2023 at 7:22pm
I’m not sure I follow you. Wouldn’t white workers have faced less competition from slaves in areas where there wasn’t much slavery? If so, they might have perceived the lowest cost from the elimination of slavery (no freed slaves would have taken their jobs).
Andrew_FL
Jan 5 2023 at 12:18pm
Specifically regarding the planter elite, IIRC their descendants quickly became ultra wealthy again, without slavery.
Knut P. Heen
Jan 6 2023 at 8:09am
There is something strange about American slavery I don’t understand. From a purely economic perspective, buying a worker (slavery) vs. renting a worker (employee) should be similar to buying a car vs. renting a car. The long-run total cost should be similar (if market prices adjust properly) except that you treat your own car somewhat better than the rented car. Whenever I read about American slavery, I get the impression that slaves were cheaper than an employee doing the same job (slavery was profitable) and that the slaves were treated worse than an employee. I also read somewhere that the price of a slave was about the same as the price of a house (which seems reasonable).
Jon Murphy
Jan 6 2023 at 9:07am
The major difference is slaver is not a voluntary transaction.
Knut P. Heen
Jan 9 2023 at 8:36am
I understand that it was not voluntary on the part of the slave, but slaves were sold from owner to owner many times before eventually reaching their destination.
The price of a slave should be about the same as the present value of the cost savings of having to pay an identical employee over the same years. If the life-time income of an employee has a present value of $2 million, the price of a fresh young slave doing the same job should be around $2 million minus upkeep (food, cloths, shelter, etc.).
Jeff Hummel
Jan 10 2023 at 5:03pm
This analogy is accurate for skilled slaves with respect to the buyer. As I pointed out in my entry, when the trade-off between positive and negative incentives was close, “slave and free labor went head-to-head in competition, as among skilled artisans and in southern factories.” And since slaves of all kinds were hired out, someone hiring a carpenter, for instance, would end up paying the same wage either to a free worker or for a slave. The difference of course, is that the free worker retained his or her entire wage, whereas the slave received only subsistence and whatever additional payments the slaveholder was willing to grant. The remainder of the shave’s wage went to the slaveholder, which was capitalized into the slave’s asset price.
Things were different in the plantation sector, where most slaves worked. Although slaves were also hired for work in the fields, free workers almost never were. Why? Because planters could not pay wages high enough to entice free workers. Even if a planter was willing to forego any return from the slave’s labor (reducing the slave’s market value from expropriated income to zero), the resulting wage would still not have compensated free workers enough to voluntarily work at the same pace as slaves. That is just another manifestation of plantation slavery forcing workers off their preferred labor-leisure trade off, because in this sector negative incentives were less costly than positive incentives.
Peter
Jan 15 2023 at 10:57pm
I forgot where I read it, been years, but I believe someone did a study on that and on average southern chattel slaves had a longer life expectancy and higher standard of living in the south than their non-slave owning dirt farming free southern white competitor. Slaves, like any other livestock, need to actually be cared for as it helps productivity and they breed whereas dirt farmers, well they could, and did, just starve.
It’s often overlooked that average freeman in the south, ignoring liberty, had it worse off materially speaking. Shorter life, less calories, less education, worse housing, longer hours. That doesn’t justify slavery but they weren’t the bottom of the ring by far and def worse positions to have in the south from pauper to invalid to widow.
Jan Clifford Lester
Jan 8 2023 at 11:00am
It recently occurred to me that there is a philosophical way of explaining the welfare-reducing effects of slavery (among many other possibilities) without resorting to anything like interpersonal comparisons of welfare, whether in subjective or in monetary terms. It seems that it can be done in a purely intrasubjective way. For the sake of argument, we can ignore morality and assume that people are narrowly egoistic. The first thought-experiment question that is then asked is this: would you accept the chance of being a slave-owner (which might be useful for an egoist) for an equiprobable chance of being a slave (which would probably be an appalling possibility for an egoist)? Virtually no one would accept this gamble (there will usually be some tiny minority that opt for anything weird). Hence, even given these assumptions, it is something that virtually everyone would reject in terms of their own welfare. Hence, slavery is not welfare-enhancing in almost universal intrapersonal terms. However, we can make the gamble more realistic. A slave-owner usually owns many slaves. Then, the question is this: would you accept a 1 in N chance of being a slave-owner for a N to 1 chance that you would be a slave? This is even less likely to be chosen. Does this do the job of showing that slavery reduces welfare without resorting to interpersonal comparisons?
Jeff Hummel
Jan 10 2023 at 5:59pm
I agree that this is a sophisticated alternative. And it reinforces the point I made in my entry that treating coercive transfers as a wash, as does standard welfare theory, “is to bias the analysis in favor of slavery.” John Moes, back in a neglected 1960 JPE article that my entry cites, put it another way. Because nearly everyone, has in his words, “a sentimental attachment to his person,” a slave would value a dollar’s worth of wages more highly once free. Indeed, this would have increased the attractiveness to slaves of manumission through self-purchase, had it been fully legal.
Warren Platts
Jan 9 2023 at 12:28pm
I am glad this discussion is continuing — I found it very thought provoking, and after reading Jeff Hummell’s detailed responses I’ve stewed in silence for a few more days. But I think maybe I can add something.
Granted, I kind of forgot what economists think of when they talk about welfare. E.g., the standard model of tariffs is that the tariff increases producers surplus and tax revenue, but the consumer surplus decreases even more, so overall net welfare declines (the difference being the deadweight losses).
I was thinking more of a sort of monumental, Nietzschean sense of welfare where the refinements enabled by the vast agricultural surpluses that accrued to the 1% Übermenschen could arguably override everything else. An interesting econlib article would be an analysis of whether ancient Egyptians would have been better off if they didn’t build all those pyramids & monuments. Certainly, the 3,000 year old monuments are still paying big dividends — the tourism thereby enabled is practically modern Egypt’s biggest sector. And if you were to go to Natchez Mississippi today, the main attractions would be the many monumental mansions and such.
But let us focus on the econometrician’s notion of welfare. I am still puzzled by a couple of things: for one, in the tariff example, output is specifically excluded from the net welfare analysis, and the focus is instead on consumer and producer surplus. Thus an analysis based on consumer & producer surpluses might come up with a different outcome, as I will endeavor to show.
Secondly, is the question of deadweight losses. I’m not sure it’s proper to say that the slave economy incurred deadweight losses because the market for slave labor was efficient in the free-market sense that the market for slave labor would have cleared (i.e., quantity demanded = quantity supplied). To illustrate what I mean I made a few Marshallian supply & demand diagrams here.
In all diagrams I set the labor supplied under slavery at 100 and the level of “compensation” to the enslaved workers (room & board, clothing, medical & perhaps the occasional cash bonus to go to the circus with when in town) is also set at 100. For simplicity, I set the elasticity of the planters’ demand curve & and the slaves’ preferred supply curve to unitary (i.e., sloped). They are all entirely schematic, yet probably pretty much in the ballpark for what happened.
Figure 1. attempts to depict the situation under slavery. In this case, however, the labor supply curve under slavery would have been for practical purposes virtually perfectly elastic (i.e., flat) because enslaved people are the ultimate price takers — they literally have no choice. As such, the slaves receive zero producer surplus & all welfare consists of the slave holders’ consumer surplus. Certainly, the enslaved get the raw end of that deal.
Figure 2. attempts to depict the situation after emancipation. Fogel and Engerman’s estimate of $84 million in lost wages that the enslaved would have demanded to work at antebellum levels is useful because it tells us about where the labor supply curve resides. The $84 million would amount to an extra $2/month bonus for the average former slave — a not inconsiderable sum, but not a huge increase over the in-kind compensation under slavery. In those days, a dollar a day was the rule of thumb for paying an unskilled laborer. Thus, in round figures, the $2/month would have amounted to on the order of a 10% raise. Thus the supply curve is P = Q + 10.
However, we also know that the labor provided declined by about a third. Moreover, mortality rates for the formerly enslaved increased by 10%, morbidity by 20%, GDP per capita declined by 20%, and unemployment was rife, all implying that the actual compensation received by the formerly enslaved went down.
This would seem to imply that the demand for labor also crashed (hence Demand Curve redrawn as P = 150 – Q resulting in a 30% decline in labor demanded compared to slavery).
But why? It seems that the main “innovation” of slavery was the enforced reliability. The gang labor system that amounted to sort of agricultural assembly line depended on a large crew that would show up on time day after day, year after year, who knew what they were doing, and would do what they were told with minimal complaint. And since working conditions were close to the limit of human endurance, the planters would basically be facing a backwards bending supply curve where people would work for a while to save up a small nest egg & then quit; thus turnover would have been very high, and thus the former plantation system would have simply broke down for the most part.
Meantime, former slaves would either acquire small farms of their own or engage in sharecropping. iow, the mode of agriculture largely reverted to a Northern style of small family farms that employed family labor with inadequate capital on likely tired land, and with no special training in agronomy. Hence the huge agricultural surpluses that supported the Natchez aristocracy who were the nation’s highest per capita concentration of millionaires would have collapsed.
Under these conditions, emancipation causes net welfare for the society as a whole to decline!. Moreover, the loss of planter consumer surplus exceeds the gain in formerly enslaved producer surplus! in contradistinction to Fogel & Engerman. Yet the former slaves are better off, despite a 20% decline in average “compensation.” At least they now enjoy a modest producer surplus that is much higher than zero.
Warren Platts
Jan 9 2023 at 12:34pm
PART II:
But wait. As Jeff pointed out in his other detailed response, there were a lot of moving parts in the post Confederate Southern economy. And perhaps it was also the case, as Jeff wrote above, that “Given that most planters were already hard working, it is not likely that emancipation significantly altered their preferred allocation between labor and leisure.” Thus we must consider an alternative explanation for the decline in labor provided & reduction in wages & employment: monopsony (Figure 3.) Planters choosing to extract fat rents from their newly freed employees would reduce labor output & wages, and account for the apparent high unemployment.
Ironically, this scenario is the best of all: societal net welfare increases by 68%, even though there is a modest deadweight loss. But both the workers are made better off than they were under slavery <em>AND</em> the former slaveholders are also made better off than they were under slavery!
This scenario, however, raises a number of questions. Number one: If monopsony is so great, why didn’t the planters employ monopsony power under slavery. Answer: commodity markets are more efficient in the market clearing sense than are free labor markets. After all, whoever heard of an unemployed slave? In addition, at least some slave states legislated a sort of “minimum wage” for enslaved people that mandated X number of pounds of meat (bacon in practice) and corn meal per worker per week, clothes, shoes and so on. After emancipation, the former slaveholders would have been freed from these regulations.
But if planters were actually better off in terms of producer surplus under free labor + monopsony, then why didn’t they voluntarily end slavery? Certainly, institutional inertia would have played a role, but the enslaved themselves represented a huge asset that could be used for collateral. So voluntary emancipation would have caused a massive reduction in wealth unless the federal government bought out all the slaves the way the British government had done decades earlier. That never happened.
Secondly, why wouldn’t competition among planters force them to pay market wages? There are several possible reasons. Collusion, for one: the planters were the paradigmatic Old Boy’s Club, and there would have been little love lost between the planters and their former “unfaithful” slaves. There could have been an informal “maximum wage” mandate. Keeping the Black population as poor as possible would also reduce their political power.
Turnover would also have been <em>much</em> higher than under slavery, so lower wages would be viewed as a sort of compensation for the turnover rate. Higher wages wouldn’t necessarily slow down turnover & might actually increase it.
Finally, the number of small, family farms did in fact proliferate. The laborers between small family farms and large plantations would have been interchangable. Thus any (small) efficiency wages paid by planters would be relative to the alternative wages barely afforded by low productivity family farms.
In conclusion, it seems possible to do an analysis of welfare efficiency based on the standard quantities used in the analysis of a tariff (consumer & producer surplus) rather than output & lost wages à la Fogel & Engelman. Moreover, two different models can account for the empirical 30% loss in labor supplied & a 20% loss in wages, and yet come up with radically different estimates of net welfare. One, based on collapsed demand for labor in fact shows an overall net welfare loss under emancipation; indeed the gain to the planters under slavery exceeds the enslaved’s loss. Alternatively, one can posit planters exerting monopsonist power that greatly increases societal net welfare not only for the former slaves, but for the former slaveholders.
Does this comparative analysis indicate that welfare analysis is “squishy”? For the Civil War era, probably yes. It was so long ago and the data then was sketchy to begin with. However, in today’s world where there is an army of government statisticians gathering data at BLS, Census Bureau, etc. probably not. One thing I will give Trump credit for was his drive to assess the economic costs of regulations, old and new. (Cf. Casey Mulligan’s <em>You’re Hired!</em> where he calculates that Trump’s deregulations saved citizens $500 billion per year in economic costs.)
Warren Platts
Jan 9 2023 at 12:38pm
As for Thomas Jefferson, he died deeply in debt, and his beloved Monticello (including some 300 enslaved people) had to be auctioned off. So maybe he wasn’t such a genius after all! 😉
Jeff Hummel
Jan 10 2023 at 11:57pm
Warren, unfortunately a detailed consideration of your analysis would require an inappropriately long reply for general readers, and this entire discussion will soon be below this webpage’s fold, if it isn’t already. But since you seem to have an intense interest in this topic, I’ll leave it at referring you to my revised dissertation available online at the Social Science Research Network here. (The link is also provided at the end of my summary encyclopedia entry in the list of further readings.) There if you are interested you can find my own demand and supply graphs, along with relevant equations, exhaustive coverage of the various debates about slavery among economists up until 2012, and extensive hard historical data.
Warren Platts
Jan 13 2023 at 12:51pm
WoW! Thank you Jeff! Downloaded! It is a fascinating topic, to be sure. The charts I made were purely schematic to illustrate possibilities rather than realities. But I think the Figure 2 shows that if the crash in Demand is large enough to the point where “market wages” are below subsistence levels (i.e., when the choice is a healthy, long life under slavery or serfdom versus starvation under freedom, then slavery probably maximizes net welfare. Such conditions might have been the case in ancient Egypt or maybe Medieval Europe, but certainly not in 19th century America.
Comments are closed.