No I meant the region not the music!

The Heart of the Bluegrass

A Population History of Lexington, and its Appalachian Hinterland

Recently, journalist Chris Arnade had a tweetstorm in which he discussed the cultural difference between “Back Row” people/places, where staying close to home is assigned moral worth and value, versus “Front Row” people/places, where, he says, a high degree of residential mobility is an absolute necessity for success. Now, he gets a key set of facts wrong about migration: with the rise of technological, globalized, liberal cosmpolitanism in America, we have seen an associated decline in residential mobility. So it’s just not true that “Front Row” elites are some hyper-migratory group. They migrate less than the parents of “Back Row” people did.

But that’s not my topic for today. As an aside to that thread, Nolan Gray, a Lexington native, tweeted this:

Historical migration commentary about a non-major geographic region? That sounds like a job for me.

So here’s what I want to do. I’ve written about Kentucky migration extensively in the past. Here’s one focused on migration since 1990. Here’s a detailed discussion of the history of slavery in Jessamine County. Here’s an old post mapping some basic migration trends for major KY regions. Here’s a detailed discussion of the Bluegrass region, and another of Inner Appalachia. I’ll repeat some material from those posts, revise some, and do some totally new stuff.

But basically, I want to ask two questions:

  1. Why does Lexington exist and why has it grown historically?
  2. What has Lexington’s relationship to Appalachian Kentucky been?
Streetview image of McConnel Springs.

Lexington’s Early Economic History

Lexington was founded because of a spring. The area was fertile and frontiersmen arrived at McConnell Springs, a good water source that is preserved in a park today. There, they received word of American victory at Lexington in 1775, so they named their settlement Lexington. By 1779, a permanent “blockhouse” or fortified dwelling was built at Bryan Station. By 1789, Transylvania Seminary had been launched, and in 1792 the Kentucky legislature began meeting in Lexington. Below is a map showing, from left to right, the locations of McConnell Springs, Transylvania Seminary, and Bryan Station, overlaid on the modern city.

The downtown grew up largely south of Transylvania University, around the courthouse, some churches, and a number of early taverns. The city had a number of major businesses, but most revolved somehow around high-value services like education, religion, and government, or else providing services to the surrounding slave-based agricultural system. Lexington was a major slave market for the surrounding area.

These services gave Lexington economic strength for a while, and as the regional population increased, fueled by the growth of the slave-labor economy, the city’s population rose as well. In the 1830s there were attempts to build a railroad west to Louisville, but they had failed by 1840. Finally, by 1852, Lexington was connected to Louisville by rail. By 1860, Lexington was the terminal rail connection in central Kentucky. No railroad existed further east, and the only railroads that went further south ran much further west near Bowling Green. Lexington had previously been a major junction for roads like the Maysville Road and the Cumberland Gap Road. Essentially, when settlers came to Kentucky, if they came overland, the central Kentucky region around Lexington was the first really fantastic place to live and farm that they would reach. If they came via the Ohio River, there were more options. And in central Kentucky, Lexington reigned supreme.

But Lexington had another advantage that would come to be more important with time: education. As I said, Transylvania Seminary was founded very early on, and the presence of educational institutions serving the region, and the wider south, was a key factor in early growth. Transylvania University had one of the only medical schools in the south, a large seminary, and the density of local educational establishments, combined with Kentucky’s incredibly robust early democratic traditions of participatory government, earned the city the name, “Athens of the West.” Here’s a map of every higher educational institution that existed in Kentucky in 1860 of which I am aware:

Interesting, right? And note that many of these schools were founded at a time when the towns around them were miniscule; that is, they predate significant urbanization in many cases. We can also look at the Census of 1850, which gives us the number of college students and teachers per county. It doesn’t quite exactly match, to my frustration, but you get the gist of it:

It’s not just Lexington, of course. The trick is that the whole Lexington metro area happens to have had lots of schools. Here it is by 1900:

There are more universities in the east by this point, but most are founded as junior colleges or advanced high schools, or else as missionary institutions, not necessarily full-scale universities. The Lexington area, meanwhile, managed to double its number of institutions and one of those was the University of Kentucky.

So after the Morill Land Grant Act of 1862 transferred the revenues of Federal lands to the state to launch a university, where did they chose to launch it? Why, in Lexington, of course. And why Lexington? Well, a college associated with Transylvania University (then called Kentucky University) had recently folded in 1857. One of its graduates, John Bryan Bowman, wanted to charter a new university under the Disciples of Christ denomination. Through some wrangling and a series of mergers, Kentucky University (now Transylvania) became DoC-affiliated. Bowman was also a disciple and student of Henry Clay, who was affiliated with Transylvania University, and so when Bowman wanted to expand his new Kentucky University under the land grant act, he already knew where to host its students: Henry Clay’s old house, Ashland. With such elite pedigree and connections to Kentucky’s established educational and political elite, Bowman won the day, and the Agricultural School was established as an arm of Kentucky University, though it was spun off as the University of Kentucky eventually, and Kentucky University re-adopted its old name, Transylvania.

The choice to fold a land grant institution into an existing religious university was not politically neutral. It was a product of wartime politics, elite domination of the legislature, expediency, and idiosyncratic individual personalities. It would prove fateful.

Long-Term Economic Trends in Kentucky

To understand where this story is going, you need to understand at least the broad outlines of Kentucky’s historic economic geography. By and large, Kentucky’s historic industrial cores were in the immigrant-dense Ohio River cities like Louisville or the Cincinnati suburbs. Lexington, of course, also developed its share of industry, but never on the scale of the major riparian settlements. Here’s a map of manufacturing wages per capita in 1900, which roughly shows you places where manufacturing was a big thing:

As you can see, while Fayette County definitely had a substantial manufacturing economy for inland Kentucky, it wasn’t anywhere close to the big Ohio River counties like Jefferson, Boyd, McCracken, Kenton, or Campbell. Indeed, Fayette County didn’t even have the highest manufacturing wages per capita in inland Kentucky! Bell County, deep in Appalachia, did. Many other parts of Kentucky had higher manufacturing output.

I want to fast forward here and compare the map above to per capita personal income in 1969 (earliest data I have) and 2015 (most recent). But to do this, I will also include county-level expenditures on farm labor in 1900. Here are personal incomes per capita by county, 1969 and 2015, each represented as a percentage of the statewide average.

As you can see, Kentucky’s personal income per capita is substantially less geographically specific in 2015 than it was in 1969, and 1969 was substantially less so than 1900 manufacturing wages.

Next up, we can compare to output in 1900. One simple way to guess what per capita income may have been in 1900 is to take total expenditures on farm labor by county and add them to total manufacturing wages. This will miss the service sector, investment returns, and farm profits, which are probably substantial, but it’s a reasonable-ish proxy. Here’s a map of that combined manufacturing-and-farm-labor income per capita, again represented as a ratio of state average, on the same scale:

Again, you can see the huge differences here. Some of this is probably just because small farms have less hired labor, so much of the economic product goes unaccounted in wages. But broadly speaking, this map helps us see where the market economy existed in Kentucky. And we can see that it maps onto prosperity in 1969 or 2015 fairly well.

We can then take the Z-score for each county (i.e. how many standard deviations they are from the mean in each year) and compare them. Here’s 1900 vs. 2015, for example.

As you can see, there’s a reasonably good association between 1900 and 2015 incomes. Some counties woefully underperform, like Jefferson, Boyd, or Bell Counties, having much higher relative income in 1900 than they had in 2015. Other counties substantively overperform, like Oldham, Carroll, or Carlisle Counties.

The map below shows the change in income relative to state average from 1900 to 2015.

And that, my friends, is what we call convergence. Poor areas became relatively richer, and rich areas became relatively poorer, as the market economy and modern economic growth spread throughout Kentucky during the 20th century. The biggest gains were observed in eastern Kentucky or in some of the counties in the west. Never let someone tell you the modern market economy exacerbated regional inequalities! It didn’t! It flattened them!

But we can actually step back even further.

Interestingly enough, we have manufacturing labor cost data by county from 1850, similar to wage data. Here it is:

This map looks very similar to the map in 1900. No surprise:

Overall, if we track the indicators that best predict the next time period, in 1850 it’s manufacturing value, in 1900 its combined manufacturing and farm wages, and my only correlate with 2015 income here is 1969 income. The point here is simply to show that these variables are probably reflecting something real about underlying economic prosperity. We can also map the value of all farm and manufacturing output by county in 1850:

Here we see that true economic production per capita was not as spiky as manufacturing was, unsurprisingly. But still, the value of all output in eastern Kentucky was critically low even in 1850. In other words, Eastern Kentucky has been poor since it was first settled. There’s nothing new about Appalachian poverty. It existed before it was popularly discovered after the Civil War.

Although, it should be noted that Democrats swept to power in Kentucky after the Civil War and undid many of the laws past by the Unionist wartime government. Kentucky didn’t ratify the 13th-15th amendments until the 1970s.

But one of the most terrible outcomes of the Civil War was feuding. Kentucky as a battlefield of constant guerrila warfare throughout the war, and most of the famous feuds of Kentucky history reflect the fact that these types of ruralized wars are nasty, intergenerational affairs. Thus, when the Civil War ended at Appomattox, it continued in Kentucky for decades to come. Consider, for example, the actions of General Burbidge, the “Butcher of Kentucky,” whose strategy to crush Confederate guerrilla activities was that whenever an unarmed civilian was murdered, he would haul out four captured Confederate guerrillas from prison and shoot them on the spot. Effective or not, it’s easy to see how this kind of governance didn’t make Kentuckians like each other very much post-war.

The Hatfield-McCoy feud began when West Virginia-based Confederate Home Guards associated with the Hatfields murdered returning Appalachian-Kentuckian Union soldier, Asa McCoy. Other famous feuds, like Stamper-Underwood and Martin-Tolliver, likewise found their origin in Civil War guerrilla combat. While outsiders may see these feuds as symptomatic of cultural backwardness, and certainly cultural factors are a component, they also reflect an unwillingness by the Redeemer government of Kentucky to crack down on violence by former Confederates. In 1899, a German-American elected governor of Kentucky (through corruption and cronyism, but I digress) as assassinated by a state militia sniper under orders from a pretender Republican governor. SCOTUS confirmed the legitimacy of the Democrat and his successor, so the Republican fled to Indiana, and was indicted.

When coal mining moved into the region, it just adopted the local norms of violence, guerrilla combat, and dubiously legitimate law enforcement. These norms had pre-Civil War origins, but were reinforced by the Civil War, and then further reinforced by state governments unwilling or unable to crack down with force and hold local officials accountable. Pro-Confederate mobs of “Regulators” were chastised but not crushed, while Rowzee’s Band of anti-Regulators were quelled with state militia. Kentucky’s tiny counties meant that every powerful family could legalize their personal fiefdom by capturing local offices. And of course, the since the state government was dominated by central and western Kentuckian interests, the state government was not terribly concerned about reinvestment in Appalachia.

19th Century Lexington Population

With that presentation of Kentucky’s regional economic history, let’s look at Lexington’s population growth until 1900. It turns out this task is tricky. Before 1790, Kentucky hadnearly its full present-day territory allocated into just 3 counties: Lincoln, Jefferson, and Fayette. By 1790, the first Census, it had 8 counties. But as counties were formed, they were often carved out of multiple other counties through segmented mergers. And furthermore, the present-day Lincoln County is not even close to the size or location of the original Lincoln County, and the same is true for most KY counties. So, for example, if you assign Bourbon and Woodford Counties to the Lexington metro area, or the Bluegrass region, that would be accurate after 1900. But in 1790, here’s what Kentucky’s county map looked like:

Bourbon County reaches aaaaaallll the way from its present location on the NE border of Fayette County to the Virginia border. And Woodford County reaches all the way from its present location as that narrow split between 1790-Fayette and 1790-Mercer all the way up to the Cincinnati suburbs. In other words, it is very easy to get kind of ridiculous county definitions as long as county-formation continued, as it did until the 1910s. By the 1870s most of the big changes are done and new counties have small effects, but before 1860, tracking these changes is a bear.

So what I’ve done is assign KY counties to regions based on their 1790 counties. All Indian Lands are one county as well. In other words, I’m tracking what “original” Fayette County population would have been, 1790–1900. There are some errors where counties grew then shrank again and because there are counties that form across 1790 boundaries, so this method is imperfect, but for line-straddling counties I’ve tried to assign based on where their likely population center was at the time of county formation. Here’s a map of modern counties, as I’ve assigned them to 1790 regions:

Here’s the resulting population trends:

As you can see, Lincoln County is by far the most populous, which is no surprise since it was also by far the biggest. Jefferson County is next, basically tied with Mason, with the population centers of both regions being along the Ohio River. Then comes Nelson County, including Owensboro, along the Ohio River west of Louisville. Then comes Woodford County, representing the central and western Bluegrass. Then comes Indian Territory, predominantly the Jackson Purchase area in the west, and then Bourbon County, including the Eastern Bluegrass and a swathe down to the Virginia border. After that is Madison County in western Appalachia. Finally we get to diminutive Fayette County and Mercer County, the two smallest of the original counties.

Let’s represent this data again, but as a share of state population.

Here we can see the rising prominence of Jefferson County, the sharp increase in Lincoln County population as the western counties were settled, and the relative decline of Fayette and Mercer Counties.

It’s worth pursuing in somewhat greater detail what happened with Fayette County as I define it. Below are the populations of the three counties I include as part of “Original” Fayette County: Fayette, Jessamine, and Clark Counties:

As you can see, population increases occur, except for where boundary changes impact estimates, until about 1830. But the 1830s until the 1850s show population stagnation or decline. I’m not 100% sure why this is, but it’s likely that a core component was the basic problem of plantation slave economies like Kentucky’s: big, efficient plantations buy all the land, and small farmers, who represent most potential population increase, can’t support themselves. Now, if you have a big manufacturing or urban sector, you can compensate for that. But Fayette County didn’t have nearly the manufacturing base of Louisville, especially in 1850. It was the last stop on the railroad, and so had modestly elevated foreign-born population, but nothing to write home about. No surprise then that the major agricultural places like Jessamine County and Clark County declined, while somewhat-more-urban Fayette County eked out stagnation before the Civil War.

But headline population is just one indicator. We can also look at the enslaved share of each of the old counties from 1790–1860. Here’s that data:

Now you see why I’ve been emphasizing Fayette County’s slave economy. It’s because, well, it was a slave economy. And it wasn’t really getting any less so. The Jackson Purchase area (Indian Territory) was becoming more enslaved. Mercer County, adjacent to Fayette County, also had a stable enslaved share, as did the vast farmlands of Lincoln County, but the Fayette County area was really the slave capital of Kentucky, with similar enslavement rates as might be observed deep in the cotton, sugar, or rice belts of states much further south.

The state on the whole was becoming less slave-dominated because of Eastern Kentucky counties like Mason, Madison, and Bourbon, as well as counties that including growing Ohio River immigrant hubs, like Jefferson, Woodford, Nelson, and Mason. It’s no surprise that Mason County, representing the eastern hills that had poor conditions for plantation agriculture and the immigrant-dense Ohio River towns of Ashland, Maysville, and Campbell County, had the lowest enslaved share.

Abraham Lincoln kept the map below in his office, which shows the enslaved share of each county in the US in 1860, with 1860 county-lines.

That shows fairly clearly what was going on. But we can also, inaccurately but perhaps still somewhat informatively, map the enslaved share using current county borders.

Every county in Kentucky had at least one enslaved person in it; nowhere was innocent of the execrable commerce. But some places sure had a lot more. Central Kentucky around Lexington was a particular hotspot for Kentucky’s slave economy, along with southwestern Kentucky. Where we observe far less slavery is in eastern Kentucky. The aree was hilly and populated by poorer settlers, thus plantation agriculture never really got going.

The other major group we’ll want to think about is immigrants. Here’s the immigrant share of the population for the old counties from 1820–1900. Data from before 1850 involves some extrapolations about the size of the naturalized foreign-born population, since the only data we have is about the non-naturalized population.

As you can see, a few places dominate the pre-1900 foreign-born scene. Jefferson County leads, obviously, thanks to the presence of the manufacturing hub at Louisville. Woodford County does well too, thanks to its inclusion of the German neighborhoods across the river from Cincinnati. Mason County likewise exceeds the state average, thanks to German neighborhoods in Campbell County, but also the Ohio River cities of Maysville and Ashland.

Then you get the rest of Kentucky. Fayette County had a comparatively robust foreign-born share for a non-Ohio-River area. And, interestingly, it did not decline as quickly as the state average did at least through 1880. The whole state saw rapid growth in the foreign born population beginning in the 1840s, but especially the 1850s as that growth began to reach even far-inland Madison County and Lincoln County, and places like Lexington, Bourbon, and Mercer Counties.

As before, I can offer you an imperfect map of the distribution of the foreign-born population.

I’ve elected to use the same scale for the foreign-born population to illustrate that the foreigner share was far smaller than the enslaved share throughout the state. The only places where you can even discern the presence of the foreign-born population is along the Ohio River and, ever so dimly, around Fayette County.

If we add these together, we can get the enslaved-or-foreign share of the population, as follows:

Obviously, slavery dominates this map, but immigration is influential in areas as well. We end up seeing a few little lightly-colored areas in central, northern, or western Kentucky, but not that many. But the entirety of eastern Kentucky manages to have few enslaved people and few foreign-born. It was a place for free native-born people. Interestingly, the eastern counties also tended to have a larger share of domestic migrants. In 1850, over 20% of the population of the eastern counties were born outside of Kentucky but in the US. Just 7% of the population of the central Bluegrass region were domestic inflows, just 15% of the Louisville area, and only about 14% of the Old Lincoln County area. To be clear, this doesn’t mean Appalachia was a hotbed of positive migration: it reflects Appalachia as being later-settled.

It’s worth noting, in 1870, Kentucky’s governor wanted to start a state Immigration and Statistics Office. At the time, most immigration law was managed by states (yep). States had the power to deport, set residency rules, and even had vital roles in naturalization. But the legislature rejected this idea; apparently they didn’t want to promote immigration into Kentucky. I have to say, this decision frustrates me deeply. Not just because more migrants would have made Kentucky richer and more politically powerful, but because having a state statistical office from 1870 onwards would have been a treasure trove of population data.

But when we tie all of this together, we can construct an interesting population history for Old Fayette County. Here, I’ll provide the whole time series for the foreign-born, up to 2015, so you can see the usefulness of using these historically-consistent regional definitions.

This gives us a pretty neat story. Fayette County in the 19th century was a heavily slave-dependent economy, far more so than the state that surrounded it, and it wasn’t really becoming any less so. Meanwhile, it had a below-average foreign-born share despite a meaningful urban hub and above-average manufacturing. But over the course of the late 19th and early 20th centuries, Fayette County’s foreign-born share didn’t fall as fast as the state’s on the whole and, indeed, it eventually began to rise and outpace the rest of the state. Today, as it turns out, the Old Fayette County region has the highest foreign-born share in Kentucky, having surpassed the old Ohio River gateways in the 1960s. This reflects the shift in international migration from old industrial hubs towards educational hubs.

For comparison, here’s Mason County:

As you can see, except for an anomaly 1790, Mason County had a much lower enslaved share of the population, and by 1860 actually had more foreigners than slaves. But Mason County’s trend was the opposite: it transitioned from a high-immigrant area, thanks to Maysville, Ashland, and Campbell County, to an anomalously low immigrant area. A major reason why is, of course, the region has very, very few universities or higher educational institutions. Its industrial profits were never invested in enduring institutional projects. Remember these facts. Keep in mind that northeastern Kentucky was a comparatively immigrant-dense location before the Civil War!

Then we can pivot to Jefferson County, a particularly interesting case.

Jefferson County had elevated levels of both enslaved people and foreigners until 1860, when its enslaved ratio dipped just below the state average thanks to the rapidly-growing foreign-born population inflating the denominator. But unlike Mason or Fayette Counties, Jefferson has maintained an above-average foreign-born share continuously since 1820. Jefferson County had lots of manufacturing early on, and has lots of educational institutions today.

But one factor you may have noticed in all of these charts was the decline in the foreign-born population after 1860. This is odd. Immigration did not slow down appreciably; it continued at a rapid pace throughout the late 19th century. Population growth didn’t slow down; Kentucky actually experienced a population boom, partly thanks to rising coal mining employment in the east, which ought to have been well-suited to immigrant employment. But nonetheless, throughout Kentucky, the foreign-born population fell. It seems possible there were political or social discouragements to foreign-born settlement, though a German-American led the Democratic party in the 1890s and was elected governor…. oh wait, yeah, then he was assassinated.

We can also do the usual treatment of Lexington’s population history I do for pretty much all cities.

Lexington and Fayette County merged in 1974, hence that big break in series. To be honest, the most interesting thing to me is the long-run stability of Lexington’s surrounding counties.

As late as the 1950s some of Lexington’s nearby suburbs and exurbs had declining populations, and it wouldn’t be until after 1960 that the non-Fayette-County portion of the metro area broke its previous population record. That’s fairly remarkable. Lexington pulled huge growth out of an immediate regional environment of stagnation, malaise, and decline.

How did Lexington do it?

The City and the Hinterland

When we look at cities, we tend to look at their downtowns, their hubs, their built environments, the places where urbanity “is” or “exists.” We want to explain the growth or decline of cities by looking at factors within the sphere of urban life. But for much of American history, urban growth patterns have been shaped as much by what’s happening in non-urban spaces as by what’s happening in urban spaces. Migration out of farming areas thanks to rising farm productivity and rising manufacturing wages drove as much of America’s rising urbanization in the 19th and early 20th centuries as did immigration. One reason the south didn’t urbanize as much is people who had freedom of migration (whites) often benefited from an area experiencing curiously high output per person-with-freedom. Slavery locked the “losers” in the system in place, and gave the “winners” returns that discouraged urbanization. And when the kids of “winners” couldn’t inherit, they just moved west to extend the reach of slavery further thanks to cheap American land policies. The lack of slavery and large-scale plantation economies, meanwhile, helped encourage urban migration in the northern states.

In Kentucky, some of this trend was at work. Kentucky’s agriculture is relatively high value per acre, with lots of cash crops like tobacco, or livestock. Plus, Kentucky was comparatively close to big urban markets. The result was that agriculture in Kentucky remained profitable in a way it didn’t in North Dakota, so while North Dakota experienced declines beginning almost as soon as the foreign-born share peaked, Kentucky saw steady growth. Add in the rise of mining, and voila. Here is Kentucky’s population vs selected other states from 1850 to 1940.

While the Dakotas and Kansas grew faster in the early periods, they quickly flatlined. Virginia’s growth rates were similar to Kentucky’s after 1870, but the 1850–1870 period was awful for Virginia, such that Kentucky became more populous than its mother-commonwealth. Meanwhile, West Virginia experienced quite rapid growth.

What West Virginia and Kentucky shared in common, of course, was coal, and this period saw the rise of mass-labor coal mining and its attendant population boom. Looking at Kentucky’s old county regions for this period, here’s what their growth trajectories look like, indexed to population in 1850:

The fastest growth is in Madison County and Mason County, two of our eastern Kentucky regions. Then comes Indian Territory, which includes far western Kentucky and a bit of SE KY, and then Jefferson County, a major beneficiary of urbanization. All of the other regions lag behind.

But the coal boom didn’t begin in 1850. Data on coal extraction in the US begins in 1870 and rises to an early peak in 1918. So that’s probably the core period of growth. Let’s look at who grew most from 1870 to 1918.

As we can see, Madison County grew like gangbusters. Indian Territory did too, then Mason County, then Jefferson and Bourbon Counties. Fayette County’s performance is looking better, especially in the later years. Fayette County had a very weak 1870s as it dealt with economic disruption from the end of slavery.

The problem with these counties is that they don’t match precisely only what we mean today when we think about eastern Kentucky. Luckily, by 1870, most counties had reached their modern shape and area, and new county formations have a small population impact. So we can redefine our regions to reflect modern economic regions to look at these historic trends. We’ll use this map:

Using that, we can track 1870–1918 population as follows:

So this one speaks for itself I’d say. Appalachia was a major source of growth. And associated with Appalachia came the Ohio River cities like Louisville, Owensboro, and Paducah, who benefited from access to the flood of coal coming down from the mountains. Cities along the Appalachian periphery also benefited from providing services to Appalachians. Corporate governance was rarely headquartered within Appalachia proper. Usually, cities like Lexington, Louisville, or Cincinnati were the hubs of finance, politics, education, and ultimately the place where profits were reinvested.

The rise of mass employment in Appalachia helped boost incomes, as you saw. It also led to some degree of increase in the foreign-born population. Here’s the foreign-born share for modern economic regions, 1870–1920:

In 1920, Inner Appalachia managed to have an above-average foreign-born share, and 1880–1890 saw rising foreign-born shares for both Inner and Outer Appalachia, even as the rate fell for other regions and the state. That is to say, these were episodes where these regions received hundreds or thousands of immigrants. The absolute numbers are not enormous, and the total share is hardly impressive, but it’s also not nothing.

We can look at a few key counties driving these trends.

These are tell-tale waves of migration by foreigners, though whether it’s immigration by fresh-off-the-boat migrants, or domestic flows of foreign-born who’d been in the US for a while, is less clear.

But the point is, the Appalachian region grew quickly, and not just through fertility. It received thousands of foreigners during the coal boom, and almost certainly tens of thousands of domestic migrants as well. Unfortunately I can’t get state-of-birth-by-county between 1880 and 1930, so I can’t directly inspect the role of migration. Furthermore, readily-available age data isn’t complete or detailed enough to reasonably approximate a residuals-based migration estimate, especially given county subdivision during the period. So all we can really do is look at residual growth rates.

The above chart should give us an idea of what migration patterns looked like, relatively speaking. Fertility and mortality differences matter as well, and it’s likely Appalachia did have higher fertility, but it’s also likely that Appalachia had higher mortality. What we can see is some areas reliably having sub-par growth performance: the Western Pennyroyal, the Bluegrass, Northeastern Kentucky. Others have stronger growth performance, like Louisville, the Eastern Pennyroyal, or the Jackson Purchase. Then we get the big winners, Inner Appalachia, and Outer Appalachia before the 1910s.

Now, look. We know some of this really is migration. We know the foreign-born share spikes coincident with the growth spike in Inner Appalachia, and coincident with the 1880–1890 strength in Outer Appalachia. We don’t know so much about domestic migration, but this should give us a sense of where any migration was favoring Kentucky.

Well, we do have ways to estimate statewide domestic migration. I won’t waste time on the details of that here, but if we add/subtract these excess growth rates (which give a statewide excess growth rate of 0) to the statewide net migration rate, we may actually be able to guesstimate the net migration for each region.

The important thing to recognize here is Kentucky had negative total net migration for this whole period. That makes Appalachia’s positive migration all the more impressive.

And, to be clear, a 60-year period of fairly consistent 2% net migration is extremely robust and impressive.

Inner Appalachia’s migration from 1870–1930 was comparable to that of Texas or Florida today.

It was a region of movers, not, as popular lore portrayed it, a region set in its ways, time-out-of-time, untouched since first settlement. Yes, it was isolated, so it had low migratory churn, i.e. probably low gross rates, just as is the case today. And, being isolated, it was technologically backwards, slow to adopt new lifeways and practices. But it wasn’t actually as demographically or genetically isolated as stereotypes made it seem; certainly not by the 1920s. By the 1920s, Appalachian Kentucky was the beneficiary of radical cultural and demographic alterations as a result of migration.

We can also look directly at Fayette County to see what was happening there:

There were outflows in the 1870s, then inflows in the early 1880s; though note I have this as a dotted line because these county-level estimates get more erratic the smaller a region I define. Then we see migration decline into negative territory, then a rise into positives in the 1920s.

The first thing to make clear is that although Fayette County had negative net migration most likely, it was almost certainly getting positive net migration within the Bluegrass region. That is, Fayette County’s nearby hinterlands were sending population into Fayette County, and into Lexington.

That explains a substantial question about Lexington’s growth. How did Lexington grow while surrounding areas were stagnant?

Simple Answer: Surrounding areas were stagnant BECAUSE Lexington grew.

Lexington was growing via migration at a time when the state and area had negative net migration. And who was leaving?

Weeeeeellllll, lots of people were leaving as the northern states industrialized faster than the southern states, but the real story here is about race. In 1870, Kentucky had 222,000 black people. In 1900, it had 285,000. In 1930, it had 226,000. Even that 1900 figure is actually a lower share of population than in 1870.

Fayette County, meanwhile, had 12,500 blacks in 1870, 15,400 in 1900, and 16,400 in 1900: consistent growth. But in the remainder of the Bluegrass region, the black population went from 51,000 to 55,000 to 32,000.

Here it all is in percentages:

So Fayette County also saw a decline in the black share, despite a rise in the nominal black population. But broadly speaking, all these areas were experiencing black outflows alongside less severe white outflows, and, by the later part of the period for Fayette County, almost certainly strong net inflows of whites.

Where did those net inflows come from, though?

If Fayette County got its growth 1870–1930 from the Bluegrass hinterland, where did it come from after that?

Modern Population Trends

The chart below shows population for each of our modern economic regions in Kentucky from 1930 to 2016.

First of all, Greater Louisville is very big. But it had a really, really awful 1970s and 1980s, while the Bluegrass region posted solid growth from WWII onwards, and especially outgrew Greater Louisville during those two bad decades. Basically, Louisville got hit by the decline of traditional manufacturing, but since Lexington never had as much traditional manufacturing, its slowdown wasn’t nearly as severe.

But the really fascinating story is Inner Appalachia, of course. The easternmost reaches of Appalachian Kentucky continued to show extremely rapid growth through 1940, then a normal wartime decline, then some good-if-not-spectacular growth until 1950, then holy cow what happened!?

Well, this happened:

Coal mining employment slumped dramatically in Eastern Kentucky. Then when population bounces back in the 1970s and 1980s, it’s because of another boom in coal-mining employment. And now the recent sharp slump in coal mining employment is triggering another wave of sharper population declines; but by now, Appalachia’s population is also graying, experiencing lower fertility, and being ravaged by the opioid epidemic, with the result that even when coal jobs are doing fine, population declines.

The counties of Inner Appalachia have gone from being Kentucky’s #2 most populous region after Greater Louisville from 1936–1950, to being on track to the least populous region by 2020 or just a few years after.

Appalachia’s sharp decline was coincident with another change, however: a sudden increase in Bluegrass-area population. Where did those Appalachians leaving the mines go? Well, many places, but not least among them was Lexington.

One neat thing we can do is, for some geographies and census years, see what share of the population was (1) born in Kentucky and (2) residing in a non-metro county 5 years earlier. We can’t see what non-metro county they were in, but we can see they were in some nonmetro county. Given that we know Appalachia was sending out major outflows and that it was a large population area, we can presume this will somewhat proxy Appalachian migrants in KY metro areas.

Sadly, we don’t have Lexington data in 1940. But we do have Cincinnati and Louisville. And what we can see is that Lexington was higher than them in both years.

Crucially, we can make some adjustments to this data to try and guess how many were truly Appalachian-born. So, for example, we can decompose these groups by the share of the non-metro population generally that was Appalachian, and then adjust for what we think outflow rates were for each major nonmetro region. Then we can assume, perhaps that Lexington receives relatively more Appalachian migrants than Louisville or Cincinnati due to proximity. With a lot of these sort of vague extrapolations, we can guesstimate the Recent-Appalachian-Kentuckian-Migrant population for these three MSAs as being:

So okay. A series of very reasonable assumptions yield a picture of Lexington as being substantially more Appalachian-Kentuckian than Louisville or Cincinnati. In other words, an appreciable amount of Lexington’s migratory inflows are probably Appalachians.

That’s over 4% of total population arriving 1985–1990 from Kentuckian Appalachia. Now, doubtless a fair share of these inflows turn around and become outflows, but given that we’re restricting to nonmetro flows, which tend to be more unidirectional, we can assume that the net flows are a relatively large share of the gross flows. So we might think that Lexington gained between 1–2% population growth in most year-periods as a result of Appalachian outflows up through 1990.

Let’s see what that would mean for Lexington’s population. The chart below shows Lexington MSA’s actual, historic population, and then an alternative series where in each 5-year period its growth rate is 1% lower from 1935–1940 onwards.

Now, to be clear, this is just an approximation. But I think it’s conservative, as Appalachian migrants may have been younger and higher-fertility than Lexington natives, and I have not tried to untangle any differences in fertility, just direct effects of migration.

What Does Appalachian Migration to Lexington Mean?

There are two ways to look at this.

The first perspective, a fundamentally correct view, is that as economic opportunities vanished in Appalachia, Lexington absorbed tens of thousands of Appalachian migrants. Lexington is a comparatively high-income city with a very high level of education, good amenities, and a high standard of living. This is unambiguously a story of individuals escaping poverty.

Furthermore, as I mentioned above, those individuals escaped poverty and did not leave their homes behind. That is, Appalachian income has been converging with Kentucky’s statewide average for decades now, so even as outflows were severe, incomes rose faster in Appalachia than in Lexington, though we haven’t seen full convergence yet.

That’s the story Nolan Gray was asking about.

So let’s try and quantify this. Let’s assume that every migrant from Appalachia to Lexington would have had the average income + 5% in Appalachia, but will obtain the average income -5% in Lexington; in other words, let’s be conservative about effect size. Let’s only analyze from 1969 to 1990 to be consistent with our data. Let’s adjust for inflation using the PCE index and make it all 1990 constant dollars.

When we do that, our very conservative estimate suggests that, over the 22 years of this exercise, this suggests that Appalachian migrants boosted their annual income by about $1,600 per year on average, in real 1990 dollars. That’s about $2,700 in today’s dollars. The net effect was about $2.5 billion more in total earnings from 1969–1990, and somewhere between $100-$150 million in 1990 income added to Lexington’s total personal income. In other words, just the marginal gain Appalachian migrants experienced amounted to something like 1–2.5% of Lexington’s total personal income in 1990. That may sound small at first blush, but then recall that’s just the marginal increase, and it’s in the hundreds of millions of dollars per year.

Now, this is all a very loose, back-of-the-envelope type calculation, a suggestive way to show that Appalachians have been enormous beneficiaries from migration to Lexington, reaping higher incomes and macroeconomically significant increases in economic output. Moving people to more productive places is very good, and Appalachians are a case study.

Far from “brain drain,” this is a story of Lexington engaged in the herculean task of heaving Appalachian migrants up into the middle class.

Here we need to remember how to read stories of deprivation about Appalachia. It was never rich or prosperous. There were no good-old-days. In some sense, Appalachia is living through the best days it ever had right now. There is no empirically justifiable story to tell where Appalachian Kentucky, once some kind of economic peer, got “left behind.” Far from it, per capita incomes have risen faster there since 1900 than almost anywhere else in Kentucky! And the reason those incomes have risen is because Appalachia is becoming more integrated with higher-productivity places like Lexington, exchanging goods, services, knowledge, people, and capital.

But, if it’s so good…why does it seem so bad?

Well, there’s another way to look at this.

UK provides data on county recruitment patterns. In 2016, about 1,800 freshmen came from the EKY coalfields. The average net price in 2014–15 for in-state students with incomes under $30,000, which I will assume is a good barometer for Appalachian students, is about $12,000. Now, some of that is housing. Let’s assume that net cost for education is $4,000 per Appalachian student per year, probably a much-too-low estimate. That means 1,800 x $4,000, or about $7.2 million in Appalachian tuition payments to UK per year. This, of course, is not including all the other universities in or near Lexington.

We can see in ACS data which counties gain/lose from Fayette County, which could be associated with collegiate migration.

Many counties have different records, but most of Appalachia is not positive. In fairness, most of Kentucky isn’t positive, and there are some very large margins of error on this data. Note that Pike County, which is green in the far east, has a rapidly growing university.

It would be tempting to treat this as an income question, but it’s actually not. It’s an asset question. Higher education is almost uniformly paid for through debt or else through parental assets; that is, students/parents reduce their present financial net wealth (either through directly spending out of savings, or adding liabilities like loans) in order to obtain what they hope will be a more productive asset, education.

Parental assets are tied to previous location of the student. If we assume even $500/yr in parental financial aid per year is occurring, either explicit in cash payments or implicit by cosigning loans, then we’re recognizing an asset transfer. It seems very likely actual average parental expenditures are higher than that. At $500/yr/student, that’s about $900,000 per year in wealth that we take out of Appalachian equity, savings, retirement, etc, and drop down into University of Kentucky worker salaries. Now, that’s not an enormous amount. But then you have to do the math and keep in mind that any given time 4 cohorts will be at UK at a time, so that’s actually more like $3.6 million per year for all current UK undergrads. That’s about 1% of UK’s tuition revenue being accounted for simply by loading up debt in or stripping savings out of Appalachia. Furthermore, this is just UK: the Lexington region is a mega-dense cluster of universities, making it one of the most degree-dense metro areas in the nation (fun fact!). UK is on the cheap end for the region.

Now, no big deal, this is just how education works. Except, it is a big deal when it’s public education. Imagine if your local primary school were three counties over; how many of the schoolteachers would still live in your neighborhood? And what if all the high schools for 4 or 5 counties clustered all in one town? Would we call it good governance if all the high schools in a multi-county region were clustered in the richest town in that region? Would we think that the economies arising from concentration are worth the obvious diseconomies involved in such a spatial arrangement of education?

Education is big business. The political choice about where to locate universities and how to fund them is important. The University of Kentucky is where it is because a Disciples of Christ figure in the 1860s with a lot of political pull wanted his religious university to get Federal funding. Areas without access to capital, especially the all-important family-and-friends financial capital that fuels first-stage entrepreneurship, will not experience as much economic growth. Thus anything that induces a poor area to strip its financial assets and load up liabilities in order to purchase something that will not benefit the local area is likely to reduce that location’s future level of income. Now if graduates return to productive careers that deploy the assets acquired through education, it’s no loss. Indeed, it’s probably a net positive investment! We want people to be educated! But degree-holder return to Appalachia is extremely low both because (1) many outmigrants actively dislike the idea of return and (2) even those who desire to return cannot find suitable work.

To be clear, this isn’t about brain drain. Brain drain is good: it means people are experiencing economic uplift and realizing their full potential. This is about fiscal drain. You cannot expect a place to get rich if you strip their financial assets out every generation, then hand them income-based welfare that, by the way, imposes restrictions on asset accumulation!

Now, you might think Kentucky can fix this by subsidizing tuition for people in asset-poor localities.

But not so fast! First of all, more subsidies on the intensive margin could rectify the problem, but if more subsidies alter the extensive margin (i.e. more students), then asset-stripping may get worse!

Plus, there’s a question of how you’d pay for it. When you combine state and local tax rates, Kentucky already has fairly high taxes. Its marginal income tax rates in local areas with taxes are among the highest in the nation and kick in at pretty low income levels, resulting in the 10th highest state and local income tax collections in the nation, as a share of state personal income. Now, Kentucky’s sales tax rate is about average or a bit on the low side, and its base is fairly narrow, thus sales tax collections as a share of goods consumption are 41st in the nation. Estimating property taxes is trickier, but friends who’ve moved cross-border into Indiana, Ohio, and Tennessee all suggested their property taxes fell after leaving Kentucky. Kentucky’s corporate income tax collections are top-10 in the nation compared to gross operating surpluses by state. Alcohol taxes are among the highest in the nation. State death taxes still exist as well. And on top of all that, the state gets coal severance taxes! In other words, it’s not like Kentucky has a huge amount of unused fiscal slack to pay for bigger subsidies for education.

Kentucky’s per capita tax collections are on the low side because it is a poor state, not because taxes are low.

But let’s return to those coal severance taxes. Since the 1970s, Kentucky has collected a severance tax on coal which has amounted, at various times, to as much as 10% or more of general revenues (though it’s low these days). In its early days, virtually no severance revenue was shared with coal-producing counties. The money all went to Frankfurt, where it then helped subsidize general government activities, not least of all university tuition… for kids in coal producing places to leave home… and it did nothing to help pay the Pigouvian costs of coal mining.

Today, in its infinite and magisterial generosity, the government in Frankfurt permits the coal-producing counties to keep half of the revenues of the tax.

To be clear, this is economically ludicrous. The whole economic rationale for extra taxes on natural resources is for the revenues to be used to offset the damage done by extractrion, not to induce additional asset transfers out of the region!

The economic rational for a severance tax is that extractive resources have externalities that harm people living near places of extraction or consumption, and that the harms endure after the coal jobs leave. The vast majority of that harm for coal is located in the physical places where coal mining occurs via destroyed scenic aesthetics (i.e. lost tourism dollars), polluted waterways (increased utility costs, diminished health), strain on roads and infrastructure (fiscal stress), and the classic problem of the resource curse. Damages from emissions are best handled by taxes at point of consumption not extraction, and, in reality, are quite small on a per-damaged-person basis compared to the damage from extraction. The only economically rational way to invest this kind of Pigouvian revenue is in same-local-area institutions that will endure long after coal is finished.

Indeed, coal mining may have positive externalities for non-mining parts of Kentucky thanks to cross-subsidies that exist within the regulated energy sector. In other words, probably the soundest economic case can be made for coal-producing counties keeping 100% of severance revenues, not 50%, and there may be some reason to think that a more than 100% return would be appropriate.

The point is, these places shoulder an unfair share of the cost of government to begin with because they suffer the extraction costs of coal while receiving a lesser share of the revenues intended to offset those costs. Reallocating spending would help some, but a simpler solution is to just let them have the money they need to fix the problems of local industry. Don’t use coal money to subsidize Appalachian kids going to UK: use it to make Appalachian community colleges into full-service universities with good programs and generous aid.

Coal severance money is one of the few serious pools of economic development cash in the Eastern Coalfields. If you doubled the annual flow going to Eastern Kentucky and allowed it to be used for a wider variety of development activities (such as trying to attract satellite university campuses or funding other educational institutions aggressively, or upgrading community colleges to full 4-year-universities), much of the harm of Kentucky’s current centralization of higher education could be counteracted. Demographic decline could be slowed, local hubs of knowledge, expertise, and diversified economic activity could grow, and Appalachia could offset some of its pervasive outflows with what it really lacks: migratory inflows.

Appalachia Needs Lexington’s Inflows

Positive migratory inflow-shocks have good economic effects. Furthermore, outflows from a region tend to be less volatile than inflows, because it takes a lot to get people to leave, but once leaving, small factors can impact where they decide to go.

So what do Kentucky state inflows and outflows look like? Weeeeelll….

Migration is very different even across very nearby counties. High-migration counties usually have universities, military bases, big industry-level booms/busts, or high urbanization. But specifically high inflows almost always mean some kind of good thing for a region: jobs, schools, base expansions, mining booms, good neighborhoods, low crime, etc. Crucially, high inflows occur because of good things happening, but then create additional good things. They drive up demand for local goods and services, and especially housing, and good results follow. Areas of persistently low inflows, meanwhile, experience a persistent absence of positive demand shocks.

Many people will be surprised to see that the lowest-outflow counties are the Appalachian ones. No surprise: they’re really just low migration generally! Much of this is demographic, of course: they’re older, so move less. Controlling for age, non-university Appalachian counties do lose young people. But the solution isn’t to try and stop them from leaving; it’s to try and attract other young people, and give leavers a reason to return. This means having amenities. That means getting real investment in the region, it means promoting cultural particularity and uniqueness, it means ripping out corrupt and inefficient governance root and branch (possibly consolidating governmental units), it means building institutions, not just industrial parks, with economic development money. But that takes a big financial commitment, which means the region needs money… if only they got full and fair access to coal severance taxes and were encouraged to use it for lasting institutional investments!

I should note here that the standard deviation of inflow rates is 2.46% vs. a straight county average inflow rate of 6%, whereas the standard deviation of outflow rates is 2.18% vs. average of 6.26%. In other words, despite typical outflow rates being higher at the county level, the amount of variation is lower. This is just another manifestation of the fundamental rule that outflow rates are less volatile than inflow rates.


Lexington benefits from persistently high inflows. With the rising importance and cost of the educational sector, this benefit will amplify in coming years. Lexington will continue to be Kentucky’s superstar. Much of that status is deserved, and it yields real benefits for residents and the region on the whole. Appalachian uplift, historically, has largely been a product of integration with Lexington, Louisville, and Cincinnati, opening the less-developed markets of Appalachia to modernization and commercialization. Nonetheless, Appalachian Kentuckians have witnessed the rapid decline of their coal-boom communities. This decline cannot be reversed, and Lexington’s ability to absorb migrants productively is an economic success story. However, with better investment of resources coal communities deserve to get access to, substantial improvements in current conditions could be made. Lexington’s success is not all come by honestly, being in part the product of decades of post-Civil War governments systematically exploiting unionist Appalachia, a lucky draw for university placement, and the systematic misallocation of revenues from resource extraction.

Check out my Podcast about the history of American migration.

If you like this post and want to see more research like it, I’d love for you to share it on Twitter or Facebook. Or, just as valuable for me, you can click the recommend button at the bottom of the page. Thanks!

Follow me on Twitter to keep up with what I’m writing and reading. Follow my Medium Collection at In a State of Migration if you want updates when I write new posts. And if you’re writing about migration too, feel free to submit a post to the collection!

I’m a native of Wilmore, Kentucky, a graduate of Transylvania University, and also the George Washington University’s Elliott School. My real job is as an economist at USDA’s Foreign Agricultural Service, where I analyze and forecast cotton market conditions. I’m married to a kickass Kentucky woman named Ruth.

My posts are not endorsed by and do not in any way represent the opinions of the United States government or any branch, department, agency, or division of it. My writing represents exclusively my own opinions. I did not receive any financial support or remuneration from any party for this research.