AM
Page 10014-02_Ch02.qxd
8/18/06
2
10:28 43
The Fifth Irresistible Force:
Ghosts and Zombies
T
he fth force that creates increased pressures for labor
mobility is rapid and massive shifts in the desired populations of various countries. In short, the current international economic system ignores the variability over time of the
desired populations of nation-states by insisting on the mostly
historically arbitrary but xed borders of the current sovereign nation-states. This lack of labor mobility accounts for the
dramatically poor economic performances that have been
witnessed and is an obvious potential force for greater labor
mobility. To be blunt, there is a signicant possibility that millions, perhaps hundreds of millions, of people are living in
nation-states that because of geographic and technological
shocks to their economies have little or no possibility of sustaining their current populations (much less their projected
future populations) with anything like decent standards of
living.
This chapter rst develops a bit of a framework for analyzing the variability in desired populations and then presents
three pieces of empirical evidence that suggest that variability
in desired populations is in fact quite large.1 This fth force is
1. This chapter draws heavily on my recent paper Boom Towns and
Ghost Countries: Geography, Agglomeration, and Population Mobility
(Pritchett 2004a).
43
10014-02_Ch02.qxd
44
8/18/06
10:28 AM
Page 44
the fifth irresistible force
discussed here in a separate chapter because while the other four forces are well
known, this aspect has been a neglected part of the discussion and requires new
evidence with some elaboration.
What Is the Desired Population of a Region?
The notion of the perfect mobility equilibrium or unconstrained desired
population of a given geographic region is easy to dene: Given the current
and expected future economic (policy, institutional, technological) and political and geographic circumstances, how many people would live in a given spatial territory in the long run if there were perfect mobility? One could dene
the optimal population as the unconstrained desired population with the
best possible policies and institutions (which does not assume that these best
possible policies or institutions are homogenous across countries). This distinction is important because the unconstrained desired population of a
region could change very fast (say, due to a civil war or disastrous economic
policies), even though the optimal population has not changed. In this case,
the obvious solution is to stick to x policies or resolve the conict so that
the desired and optimal populations move closer. But technological shifts
in the world economy can change the optimal populationseven with the best
possible policies and institutions. For instance, once sea transport was possible, the (relative, or perhaps absolute) optimal population of regions that
thrived on overland commerce declined and those near the coast increased.
Changes in desired populations do not create many pressures for labor
mobility if they are small or very gradual. Changes in desired populations
might be small or gradual if either (1) the economic fundamentals of the
desired population do not change or (2) the mobility of goods or other factors
(capital, trade) can compensate for shifts in region-specic labor demand.
Labor mobility is not a big deal for Antarctica because no substantial human
populations ever moved there; its attractiveness for human populations has
not changed. But the classic counterexample is a regional gold rushrst, people do not want to be there; then gold is discovered, and many people want to
be there; and then, when the gold is mined out, people want to leave. The
existence of ghost towns even in prospering countriesplaces that were
once booming and attracting migration that subsequently declined and even
disappearedsuggest that there is variability to optimal populations.2
2. For me, the origin of some of this thinking is that I grew up near Idaho City, which was
once a thriving frontier town (the largest in the Idaho territory) and had a population in 2000
of only 458. Why? Simple. There used to be gold in the river nearby, and now there is not any
commercially exploitable gold.
10014-02_Ch02.qxd
8/18/06
10:28 AM
Page 45
t he fifth irresistible force
45
But even if there are regional shocks, there might not be large variations in
the desired population if the mobility of other factors can compensate. Suppose a region attracts population because it relies on one type of economic
activity and then some natural or economic shock makes that activity no
longer viable. There is no longer any reason for people to be there as opposed
to any other placebut they are there. One possibility is that new activities are
created and resources (capital) ow to that place and people sustain roughly
their same living standards but change their activities. Certainly, in the story
of many of the major cities of the world, the original reason for the citys location has long since ceased to be relevant (for example, fortication, transport
linkages) but the city continues to thrive. Yet there are two other possibilities.
One is that new resources do not ow in and the optimal population falls and
people leave. The other possibility is that the optimal population falls, perhaps
dramatically, but people are not allowed to leave for more attractive locations
due to barriers to labor mobility, and hence all the adjustment to the variability in the optimal population of regions is forced onto real wages and living
standards.
Suppose that a realistic feature of a model of the international or interregional economy are region-specic shocks that produce, even after all
accommodating changes in capital stocks and goods, large persistent changes
in regional labor demand. The simplest possible supplydemand diagram
illustrates the possibilities.
If there are region-specic shocks to long-run labor demand and population mobility is allowed, then the regional supply of labor is elastic in the long
run. In this case, one should observe large variability across regions in the
growth rates of populations and relatively small variability in the interregional
growth of real wages. In this case, large negative region-specic shocks to labor
demand can create ghostsregions that consistently lose population (either
absolutely or just relatively) (gure 2-1).
If there are region-specic shocks to labor demand but population mobility is restricted and hence the regional supply of labor is inelastic, then the
forces will be accommodated with large variability in the growth of wages
(and incomes) across regions but relatively small variability in populations.
The consequence of a distribution of large region-specic changes in labor
demand and restrictions on labor mobility is that there will be regions that
experience large, persistent, positive shocks to labor demand and become
boom towns. But there are also geographic regions that will experience large,
persistent, negative shocks. Because desired (and optimal) populations can
fall much faster than the actual population, this will create situations in which
the actual population will vastly exceed its new desired level:
10014-02_Ch02.qxd
8/18/06
10:28 AM
Page 46
the fifth irresistible force
46
Figure 2-1. How Changes in the Demand for Labor Cause Pressures for
Labor Mobility
Wages
Inelastic labor
supply (mobility
restricted)
Wage fall
in
zombie
Elastic labor supply
(mobility allowed)
Large fall in region
specific labor demand
Population
Population fall in ghost
If the negative shock is large enough and population movements are
allowed, these regions will become actual ghosts.
If the negative shock is large and other regions prevent labor mobility,
then potential ghost countries become unrealized ghosts or zombie countries (zombies are the living dead) because nothing, besides out-migration,
can prevent an extended and permanent fall in wages.
There are three sources of evidence, which together suggest that there are
typically large shifts in the desired populations of regions. Though it is
extremely difficult to separate out which of these are shifts in just an unconstrained desired population (due to remediable factors like policies, or,
optimistically, institutions) and which are shifts in optimal populations,
there is some evidence from comparing regions of countries (which share
many policies and institutions) that some large fraction of the shifts in desired
populations are also shifts in optimal population. These shifts in desired population are accommodated differently depending on the conditions for labor
mobility. The three empirical examples are (1) regions of the United States,
(2) comparisons of within-country versus cross-country variability of popu-
10014-02_Ch02.qxd
8/18/06
10:28 AM
Page 47
t he fifth irresistible force
47
lation and output per person growth rates, and (3) population versus output
variability in history.
One important point, which I stress throughout this chapter, is this
decomposition into changes in desired populations stemming from various
underlying causes. There are changes in desired populations that are due to
differences in income or income growth attributable to policies, politics, or
institutions; and these changes are potentially remediablequickly. Not
every example of economic decline is an example in which population mobility is necessarily an important factor in the solution; it is plausible that a countrys desired population is low, and pressures for outward labor movement
are high, because the country is badly governed (for example, Zaire) or
because of a macroeconomic crisis (for example, Argentina in 2000). Then
xing the problem at the source is obviously a much more attractive policy
than allowing labor mobility. However, here I want to stress that there are
determinants of long-run demand that are beyond the control of policies (or
even institutions, about which there is a debate on how much these can be
purposively altered). It is perfectly plausible that, even with the best policies
and institutions, a region can see its desired population fall by 50 percent or
more due to economic forcesshifts in product demand, agglomeration,
transport costsinteracting with the regions geographic features, and hence
the desired population has fallen because the optimal population has fallen.
This is a much more difcult issue to address.
Evidence of Shifts in Desired Populations:
Regional Populations in the United States
A large country like the United States provides a good laboratory for examining changes in optimal populations. People are completely free to move, so
regions tend toward their unconstrained desired population. Within a large
country like the United States, policies and institutions are held roughly,
though obviously not completely, equal. All U.S. regions have the same monetary policy, the same trade policy, roughly the same legal framework,3 and
similar politics. Nevertheless, U.S. states have had very different rates of population growtha point that is returned to in the next subsection.
3. These are not, of course, precisely equal, as Louisiana has a French style legal system
while all others have an Anglo civil law tradition, and some states are traditionally Democratic while others are traditionally Republican. But the differences are small compared with
other regions (for instance, India, in which some states have had communist parties, other
states have had more conservative parties, and still others have experienced quite personalized policies with state-specic parties organized around a single individual).
10014-02_Ch02.qxd
48
8/18/06
10:28 AM
Page 48
the fifth irresistible force
But state-level data understate the degree of labor mobility. If one moves
from the state down to the county level, one nds counties that were essentially depopulated over the sixty years from 1930 to 1990. For instance,
Slope County, North Dakota, saw its population fall from 4,150 to only 907;
Smith County, Kansas, from 13,545 to 5,078; Huerfano County, Colorado,
from 17,062 to 6,009; and McDowell County, West Virginia, from 90,479
to 35,233.
These are not isolated examples. Even though the United States overall
more than doubled its population from 1930 to 1990, this growth was far from
uniform. An instructive exercise is to assemble groups of counties that may cut
across state boundaries but are contiguous and that are a shape such that it is at
least conceivable that, had history been different, a plausibly shaped country
could have been formed with these boundaries. That is, while we deliberately
gerrymandered the areas to include population-losing counties, we did not
simply cut out cities or make dramatic detours to include this or exclude that
county.
I have assembled ve regions of the United States, which, since I created
them, I will name: Texaklahoma (Northwest Texas and Oklahoma), Heartland (parts of Iowa, Missouri, Kansas, and Nebraska), Deep South (parts of
Arkansas, Mississippi, and Alabama), Pennsylvania Coal and Great Plains
North (parts of Kansas and South Dakota). Even with the constraint of contiguity and (mostly) convexity, one can assemble large territories that have seen
substantial absolute population decline. The Great Plains North is a territory
larger than the United Kingdom, and its population declined 28 percent from
1930 to 1990. Its current population is only a bit more than a third the population it would have been if its population growth had been at the rate of natural increase. The Texaklahoma region is bigger than Bangladesh and is now
only 31 percent the population size it would have been in the absence of outmigration. I use a few counties in the coal-producing region of Pennsylvania
to illustrate that not all these declines are due to the decline of rural and agricultural populationsnatural resource shocks also play a role (table 2-1).
The maps of these regions tell the story. Figures 2-2 through 2-5 show the
county-by-county populations of the states that contain four of the regions
described above. The shades of gray in the gures show counties that, over the
course of sixty years in which the population of the United States doubled,
saw their populations fall in absolute terms. The shading is by the absolute
(not percentage) fall in population: Counties in dark gray lost more than
10,000; medium gray, 5,000 to 10,000; and light gray, 5,000 to 0. Areas with
no shading (plain white) had modest population gains (up to 10,000), while
the striped counties gained more than 10,000 in population.
10014-02_Ch02.qxd
8/18/06
10:28 AM
Page 49
t he fifth irresistible force
49
Table 2-1. Population Change in Assembled Regions, 193090a
U.S. region
Current
population/
Population counterfactual
Population, change,
at rate of
1930
193090
natural
(thousands) (percent)
increase
835.8
36.8
0.31
Heartland
Deep South
1,482.6
1,558.2
34.0
27.9
0.33
0.36
Pennsylvania
Coal
1,182.9
27.9
0.36
Great Plains
North
1,068.0
27.7
0.36
123,202.6
101.9
Texaklahoma
All U.S.
Region
area
(square
miles)
Area per
Countries
capita
of smaller income as
area,with percentage
examples of national
(number)b average
58,403 117 (Nicaragua,
Bangladesh)
59,708
117
36,284 96 (Jordan,
Austria,
Sri Lanka)
2,972 43 (Trinidad
and Tobago,
Mauritius)
100,920 128 (United
Kingdom,
Ghana,
Ecuador)
3,536,278
100.0
92.2
85.2
62.6
84.5
85.4
Source: Pritchett 2004a.
a. A region is a contiguous collection of counties cutting across state borders.
b. Total number of countries considered is 192.
I am stressing obvious facts about population movements when I point
out three things. First, economic forces have led to the decline of certain
activitieslike farming in the Great Plains, cotton farming in the South, and
coal mining in Pennsylvaniaand that has led to a large population exodus,
particularly from rural areas and small cities.
Second, the ruralurban movement has, almost by denition, tended to
cause small decreases in population in a large geographic area and large
increases in a few concentrated areas (the shaded counties usually contain a
major metropolitan area). This means that geographic regions without sufcient economic force to attract a major city tend to lose population absolutely,
while areas with an urban center have large shifts in population.
Third, even though there were large population losses, this was without
huge losses in absolute or relative income. As seen in table 2-1, even regions
with dramatically declining populations have stayed quite close to the average
10014-02_Ch02.qxd
8/18/06
10:28 AM
Page 50
the fifth irresistible force
50
Figure 2-2. Changes in County Populations in the U.S. Heartland
Region (Selected Counties of Iowa, Missouri, Nebraska, and Kansas)a
Dick inso n
Emme t
Clay
Os ce ol a
Lyon
Pa lo
Al to
Wo rth
Wi nnebag o
Mitche ll
Howa rd
Floy d
Ch kasa w
ic
Alla
make e
Winnes hiek
Ko ss uth
OBr ie n
Siou x
Ke ya Pa ha
Bo yd
Dawe s
Siou x
Kn ox
Ch erry
Sh erid
an
Wa yn e
An telo
pe
Hooker
Thomas
Bl ai e
n
Loup
Ga rfeld
i
Ga rden
Carro ll
Bo on e
Greene
Du buqu e
Jone s
Be nton
Tama
St ory
Dela re
wa
Bu chanan
Jackso n
Linn
Ma rsha ll
Bu rt
Cu ming
Boon e
Mc Ph erso n
Arthur
Va ll y
e
Loga n
Ch eyenne
Clinton
Sh er a n How ard
m
Li c ol n
n
Me rric k
Shelby
Ha rrison
Dodg e
Co lf x
a
Pottawa ttami e
Doug la s
Bu tl r
e
Po lk
Powesh ie k
Jasper
Po lk
Iowa
Johnso n
Sc ott
Ma diso n
Ad air
Ke
okuk
Mahask a
Ma ri n
o
Wa rre n
Wash ington
Sa rp y
Mo ntgo e ry
m
Adam s
Luca s
Cl arke
Un io n
Mo nroe
Jefferson
Wape llo
He nry
Ca ss
Sewa rd
Hami lt n
o
Ca ss
Lo uisa
Sa unders
Yo rk
Ha ll
Bu ffal o
Dalla s
Gu thrie
Mu sc atin e
Mi lls
Dawson
Pe rkin s
Au dobo n
Wa sh ington
Nanc e
Ke ith
Deue l
Platte
Gree le y
Cu ster
Kimbal l
Bl ac k
Hawk
Grundy
Ce dar
Mo rrill
Banner
Cr w ford
a
Clay ton
Bu tler
Ha rdin
Hamilton
We bs ter
Calhou n
Fran klin
Wr ight
Pocahontas
Sa c
Faye tte
Br m er
e
Thurston
Monona
Ma diso n Stanto n
Wh eele r
Id a
Wo odbu ry
Pierce
Bo x Bu tte
Gran t
Humb oldt
Bu en a
Vi st a
Ch er
okee
Dixo n
Dakota
Ro ck
Sc otts Bluff
Plym outh
Ceda r
Ho lt
Brow n
Ce rro
Go rdo
Ha ncoc k
Lancas te r
Fr m ont
e
Pa ge
Ta ylor
Ringgo ld
De ca tur
A noos e Da vis
ppa
Wa yn e
De s
Mo ines
Le e
Va nB uren
Otoe
Fron ti r
e
Haye s
Chas e
Ke arne y
Gosper Ph el s
p
Adam s
Clay
Sa li e
n
Fill
more
Atch ison
Johnso n Nemaha
Dund y
Hi tchcoc k Red Wi ow
ll
Furnas
Ha rl
an
Fran klin We bste r Nu ck olls
Thayer
Pawnee Richardson
Ra wlin s
De catu r
No rt
on
Ph illip s
Sm ith
Graham
Rook s
Osborn e
Jewe ll
Republic
Ma rs ll
ha
Wa sh ington
Nemaha
Brow n Do ni han
p
Sh erid
an
Mi tc ll
he
Li c ol n
n
Gove
Loga n
Treg o
Ellis
Ru ss el l
Sc ot t
Lane
Ne ss
Ru sh
Dick inso n
Rice
Mc Ph erso n
Os ag e
Ma ri
on
Kearny
Hodgeman
Finney
Stanto n
Gran t Ha sk el l
Kiow a
An derson
Sewa rd
Me ad e
Clar k
Coma nche
Ba rber
Wo odso n
Alle n
Bo urbo n
Pe tti s
St . Lo ui s
St . Lo ui s Ci ty
Ga sco na de
Os ag e
Be nton
St .C la ir
Je ff rso n
e
Ma ri s
e
Camd en
Pu lask i
Ce da r
Po lk
Fran klin
Mo rgan
Ve rnon
Greenwoo d
St . Ch arle s
Mo niteau
Hick ory
Ha rv
ey
Ph elps
Craw ford Wa sh ington
Da lla s
De nt
La cled e
Se dgwick
Ir n
o
Ba rt n
o
Kingma n
Ha rper
Co oper
John so n
Mo ntgome ry
Ca ll w ay
a
Wa rre n
Co le
Li n
n
W ilson
Ne osho
Co wley
Gree ne
We bs te r
Ja sper
Wr ight
Ca pe
Gi rardea u
Bo llin ge r
Sh an no n
Wa yn e
Ne wton
Ston e
Mc Do na ld
Pe rry
Re yn olds
Te xa s
Lawren ce
Labe tte Ch erokee
ChautauquaMo ntgome ry
St e.
Ge ne vi v e
e
St .
Fr n co is
a
Ma diso n
Da de
Craw ford
El k
Steven s
Linc ol n
Bo on e
He nry
Sumn er
Mo rt
on
Howa rd
Mille r
Reno
Bu tler
Prat t
Sa lin e
Miam i
Co ff y
e
St afford
Fo rd
Fran klin
Chas e
Edwa rd s
Gray
Pi ke
Au drai n
La faye tte
Ba te s
Pawnee
Hami lt n
o
Mo nroe
Ra nd olph
Ra y
Ca ss
Mo rris
Ma ri n
o
Ra lls
Ch arit n
o
Ca rrol l
Lyon
Ba rt
on
Lewis
Sh elby
Ca ldwe ll
Clay
JeffersonLeavenwo rth
Wy ando tte
Jack so n
Shawne e
Ge ar y
Wa ba unse e
Doug la s Johnso n
Sa li e
n
Ells rth
wo
Gree le y Wich ita
Kn ox
Ma co n
Rile y
Ot ta
wa
Wa llace
Ad ai r
Linn
Clin ton
Pl atte
Po ttawa tomi e Jackso n
Clay
Su lliv n
a
Da vies s
De Ka lb
Livi ston
ng
Bu chan an
Thomas
Cl ark
Ha rriso n
Grun dy
Atch is
on
Clou d
Sh er a n
m
Sc hu yler Sc otland
Me rcer
No dawa y
Ho lt
An dr w
e
Ch eyenne
Pu tn
am
Ge ntry
Gage
Jefferson
Wo rth
Ch ristia n
Sc ot t
Ca rter
Do ug la s
Stod dard
Ba rry
Ho we ll
Ta ne y
Oz ark
Oreg on
Mi ssi ss ippi
Bu tler
Ri pley
Ne w Ma drid
Pe mi sc ot
Du nk lin
Source: Pritchett 2004a.
a. Dark gray: lost more than 10,000; medium gray: lost 5,00010,000; light gray: lost 05,000; white:
gained 010,000; striped: gained more than 10,000.
national income (with the exception of the Deep South). These regions and
counties became ghosts, not zombies.
Regions within the United States serve as a thought experiment of what
would happen in a fully globalized worldgeographic units linked with
fully integrated markets for land, capital, goods, and laborand a globalized
world with common policies and economic institutions at that. In such a
world, one can expect that incomes would converge in levels, and, with the
exception of the Deep South, incomes in these created regions are more than
84 percent of the national average. But one can askeven with fully integrated markets with goods and capitalhow much variability is there in
optimal populations? The answer is a lot. Though it may be the case that
population movements were less than they would have been because capital
owed to these regions and goods were mobile, it is still the case that the population shifts within the United States are huge. In particular, they are vastly
larger than the population shifts one sees across the often equally arbitrary
boundaries of countries in the world today.
10014-02_Ch02.qxd
8/18/06
10:28 AM
Page 51
t he fifth irresistible force
51
Figure 2-3. Changes in County Populations in the U.S. Deep South
Region (Selected Counties of Arkansas, Mississippi, and Alabama)a
Ca rroll
Bent on
Boon e
Fu lton
Ba xter
Ma rion
Izard
Ma diso n
Wa shingt on
Craw for d
Fr anklin
Ne wt on
Sear cy
Ston e
Va nB ur en
Johnso n
Gree ne
La wren ce
Craighea d
Independen ce
Clebur ne
Jack so n
Po pe
Co nw ay
Loga n
Seba stian
Faulkner
Ye ll
Po lk
Ga rlan d
Ho t Sp ring
Crittend en
Wo odruff
Lono ke Pr airie
Sa line
Le e
Be nton
De Soto
St . Fr anci s
Pu lask i
Mo ntgomery
Wh ite
Missis si ppi
Poinsett
Cros s
Perr y
Scot t
Cl ay
Rand olph
Shar p
Tate
Tunica
Mars hall
Pano la
Phil lips
Gran t Jeffers on
Ar kans as
Al corn
Ti shomingo
Pren tiss
Lime stone
Colber t
Fran klin
La faye te
La wrence
Ma dison
Mo rgan
Lee Itaw amba
Po ntotoc
Co ahoma Quitman
Da de Ca toos a
Fa nnin
Mu rr ay
Wh itfi el d
Wa lker
Gi lmer
Laud erda le
Tippah
Un ion
Monr oe
Jack son
Ch attoog a Go rd on
Ma rs hall
De Ka lb
Pick ens
To wn s Ra bu n
Un io n
Ha bersha m
Lump ki n Wh ite Step hens
Da ws on
Fl oy dF Ba rtow Ch er ok ee or sy th
Fr ankl in Hart
Hall Ba nk s
Ja ck so n Ma di so n Elbert
Gw in nette Barr ow Clar ke lethor pe
Og
Da llas Cl evelan d Lincoln
Oc on ee
Ha rals on
Wa lton
Bo livar
Desh a
Wilk es Li nc ol n
De Ka lb
Gren ada
Walker
St .C lair Calhoun
Do ugla s Fu lton
Fa yette
Litt le River He mpstea d
Ro ck dale
Clay
La ma r
We bster
Gree neTa liaferro
Co lu mb ia
Cl ay ton
Ne vada Ouac hita
Sunf lowe r Leflore
Ne wt on Mo rg an
Jefferson
Cleb urne Ca rrol l
Drew
Ca lhou n
Montgo mery
Mc Du ffie
Fa ye tte Henr y
Lown de s
Wa rr en
Ch octaw Ok tibb eha
Jasp er Pu tna m
Co we ta
Rich mo nd
Ca rr oll
Br adley
Ta llad ega
Sp aldi ng Bu tts
Miller
Wa shington
Hear d
Ha nc oc k Gl as co ck
Pick ens
Tusc aloosa
Shelby
Columb ia
Ch icot
Clay Ra ndolph
As hley
Un ion
Hu mp hrey s Ho lmes
Burk e
Pi ke
Lama rMo nr oe Jo nes Ba ldwi n Wa sh ingt onJeffer so n
No xubee
At tala
Lafayett e
Wi nston
Trou p Me riwether
Bi bb
Sh arke y
Up so n
Coosa
Gr eene
Je nk in s
Cham bers
Bi in bb
Wilk so n
Ha rris
Chilton
Scre ve n
Ta llap oosa
Ta lbot
Cr awford
John so n
Ha le
Leake Ne shoba
Yazoo
Ke mper
Tw ig gs
Em anuel
Ta ylor
Pe rry
Issaqu ena
Pe ac h
Lee Mu scog ee
El more
Laur ens
Madison
Au tauga
Su mter
Tr eutlen Ca ndler Bull oc h E ffingha m
Ho us ton Bleck le y
Ma rion
Sc ott
Ch atahoo ch ee Sc hl eyMa co n
Wa rr en
Ne wton Laud erda le
Ma con
Da llas
Mo ntgome ry
Hind s
Ev ans
Mo ntgo mery
Pula sk i Do dg e
Russ ell
Ma rengo
Do ol y
Ra nk in
Br ya n
Wh eeler
St ewar t Webs ter Sumter
To om bs Ta ttnall
Ch atha m
Lo wndes
Bullock
Wilc ox
Cl arke
Cr is p
Te lfair
Sm ith Jasper
Ch octa w
Wilcox
Qu itma n
Claibo rne
Je ff
Libe rty
Te rrel l Le e
Ben Hill
Si mpson
Lo ng
Ba rbou r
Da vi s Appl in g
Co piah
Tu rner
Ra ndol ph
Pi ke
Bu tler
Irw in
Je fe rson
Clarke
Cl ay Ca lhou n Do ughe rty
Wa yn e Mc Intosh
Co ffee Ba co n
Wa yne
Wort h Tift
Co vi ngton Jone s
Cren shaw
Henry Ea rly
Mo nroe
La wrence Je fe rson
Pier ce
Bake r
Linc oln
Da vi s
Ad am s Fr anklin
Berr ie n At ki ns on
Da le
Cone cuh
Co fee
Gl yn n
Wash ington
Fo rres t
Mitc hell Co lq ui tt
Miller
Ware Br antley
Co ok
Covington
Marion
Gr eene
La nier
Pike
La mar
Am ite
Ho uston Se mi nole
Pe rr y
Es ca mb ia
Ca md en
Wi lkinson
Gene va
Cl in ch
Wa lthall
Ch ar le ton
Gr ady Th om as Br ooks Lown des
De ca tur
Mo bile
Ec hols
Ge orge
St one
Pe ar l Ri ver
Ba ldwin
Ho ward
Se vier
Pi ke
Clar k
Marion
Ya lobusha
Tall ahatch ie
Ca lhoun Ch icka saw
Ha rris on
Mo nroe
Wi nston
Cher okee
Cullman
Etowah
Bl ount
Po lk
Pa uldi ng Co bb
Jack son
Ha ncoc k
.
Source: Pritchett 2004a.
a. Dark gray: lost more than 10,000; medium gray: lost 5,00010,000; light gray: lost 05,000; white:
gained 010,000; striped: gained more than 10,000.
Adjustment of the Regions of Countries versus
Countries in Output Growth and Population
The second illustration of the variability of desired populations is to show
that the variability of the growth output per worker to the variability of the
growth of population happens exactly as we would expect with large regional
shocks. As illustrated in gure 2-1, with perfect labor mobility, workers and
households will move in response to economic opportunities, and if there
are large geographic shocks to regions that change desired populations
(which, remember, is the combination of shocks and the shock not being
fully accommodated by movements in other factors like capital or by trade)
and the labor market is integrated, then the variability of the growth output
per worker across regions should be relatively small, because regions with
incipient rapid growth should gain population and regions with negative
shocks lose population, while the variability of the growth rate of population
should be large.
10014-02_Ch02.qxd
8/18/06
10:28 AM
Page 52
the fifth irresistible force
52
Figure 2-4. Changes in County Populations in the U.S. Pennsylvania
Coal Region (Selected Counties of Eastern Pennsylvania)a
Erie
McKean
Warren
Forest
Elk
Venango
Mercer
Ca meron
Sullivan
Clinton
Clarion Jefferson
Lawrence
Clearfield
Butler
Armstrong
Beaver
Allegh eny
Cambria
Columbia
Montour
Somers et
Bedford
Perry
Fulton Frank lin
Monroe
Ca rbon
Snyder Northumberland
Schuylkill
Dauphin Le banon
Cumberland
Fa yette
Greene
Union
Huntingdon
Washington
Pike
Luzerne
Mifflin
Juniata
Blair
We stmoreland
Wa yne
Wy oming Lack awanna
Lycoming
Ce ntre
Indiana
Susque hanna
Brad ford
Tioga
Potter
Crawford
Adams
La ncaste r
Yo rk
Northampton
Le high
Berks
Bucks
Montgo mery
Ch este r Philadelphia
Delaware
Source: Pritchett 2004a.
a. Dark gray: lost more than 10,000; medium gray: lost 5,00010,000; light gray: lost 05,000; white:
gained 010,000; striped: gained more than 10,000.
In contrast, if the world is segmented so labor and households cannot move
and there are very different shocks to a geographic regions output potential,
then the adjustment mechanism should be exactly the opposite. One would
expect very little variability in the growth rates of population (because it is primarily determined by rates of natural increase) and enormous variability in
the growth rate of output per person (or worker) as wages fall due to the
geographic-specic productivity shock. This is the natural experiment that the
postwar international system has run, and gures 2-6 and 2-7 show the results.
Because gures 2-6 and 2-7 are new, they require a bit of explanation, but,
like all great art, it is worth it as this art embodies two features. First, the annual
growth rates of output per capita and of population are on the vertical and horizontal axes. Though software packages that produce graphs rescale the axes
independently so that one cannot visually compare the variability, in this case
I have forced the axes to have exactly the same range. Second, I show the 90th
and 10th percentile boxes of each variable, so that the two vertical lines contain 80 percent of the regions growth in population (because the rightmost
line is the 90th percentile of population growth and the leftmost line is the 10th
percentile). Similarly, for growth of output per capita, the top horizontal line
10014-02_Ch02.qxd
8/18/06
10:28 AM
Page 53
t he fifth irresistible force
53
Figure 2-5. Changes in County Population in the U.S. Great Plains
North Region (Selected Counties of Nebraska and South Dakota)a
Divide
Renville
Bottineau
Burke
Ro lette
Pembina
Cavalier
Towner
Wa lsh
Williams
McHenry
Mountrail
Ramsey
Pierce
Ward
Benson
Grand
Forks
Ne lson
McKe nzie
McLean
Sher idan
Eddy
Wells
Foster
Dunn
Griggs
St eele
Traill
Mercer
Oliver
Billings
Kidder
Burleigh
Golden
Va lley
Stark
Stutsman
Logan
McIntosh
Ransom
La Moure
Dickey
Hettinger
Slope
Cass
Barnes
Morton
Grant
Emmons
Bowman
Sioux
Ad ams
Mc Ph erso n
Ca mp bell
Co rson
Richland
Sargent
Marshall
Br ow n
Ha rding
Pe rkins
Wa lw orth
Edmu nds
Po tter
Faulk
Ro berts
Da y
Grant
De we y
Bu tte
Sp ink
Co dington
Cl ark
Ziebac h
Sull y
De uel
Ha ml in
Ha nd
Hy de
Me ade
Lawr ence
St anley
Ha ak on
Be adle
Hu ghes
Buffalo
Pe nnington
Jo nes
Cu ster
Je rauld
King sb ury
Sa nborn
Br ooki ngs
Lake
Mi ner
Mo ody
Ly ma n
Ja ck so n
Ha ns on
Au rora
Br ule
Da viso n
Mi nn ehaha
Mc Co ok
Mell ette
Fall Rive r
Do uglas
Tripp
Sh annon
Be nnett
To dd
Gr egory
Hu tchins on
Tu rner
Linc oln
Ch arles Mi x
Bo n
Ho mm e
Ke ya Pa ha
Ya nk ton
Un ion
Cl ay
Bo yd
Da we s
Si ou x
Kn ox
Cher ry
Sherid an
Ceda r
Holt
Br ow n
Pi er ce
Box Bu tt e
Antelo pe
Gran t
Sc ot ts Bl uff
Th om as
Hook er
Bl ai ne
Lo up
Garf ield
Ga rd en
Arth ur
Mc Ph erso n
Boon e
Loga n
Va lley
Ch eyen ne
Howa rd
Linc oln
Daws on
Pe rk ins
Co lfax
Th urst on
Bu rt
Cumi ng
Dodg e
Wa sh ingt on
Nanc e
Ke ith
De uel
Pl at te
Gr ee ley
Sher ma n
Cust er
Ki mb all
Wa yn e
Ma diso n St anto n
Whee ler
Mo rrill
Bann er
Di xo n
Dako ta
Ro ck
Bu ffalo
Me rr ick
Po lk
Sa unde rs
Yo rk
Hall
Cl ay
Sewa rd
Fi llmo re
Sa rp y
Sa line
Ha milt on
Adam s
Doug las
Bu tl er
Ca ss
Lanc aste r
Ot oe
Chas e
Haye s
Fr onti er
Go sp er Phel ps
Ke arne y
Johnso n Ne ma ha
Dund y
Hi tc hcoc k
Red Wi llow
Fu rn as
Harl an Fr an klin
We bste r Nuck olls
Thay er
Jefferso n
Ga ge
Pa wn ee Ri ch ards on
Source: Pritchett 2004a.
a. Dark gray: lost more than 10,000; medium gray: lost 5,00010,000; light gray: lost 05,000; white:
gained 010,000; striped: gained more than 10,000.
10014-02_Ch02.qxd
8/18/06
10:28 AM
Page 54
the fifth irresistible force
54
Figure 2-6. Large Shocks, Accommodated with Population Growth in
Large Countries, Per Capita Growth across Non-OECD Countries versus
the United States, Japan, and Canada
(Boxes at 90th/10th percentiles of each variable)
Growth of GDP (or income) per capita
.060
.050
Japan (5590)
.040
.030
Canada (2692)
.020
USA (190090)
.010
.000
Non-OECD countries (5090)
Zero
.015
.030
.005
.005
.015
.025
.035
.045
.055
.065
.075
.085
Growth of population less rate of natural increase
is the 90th percentile of growth while the bottom line is the 10th percentile. If
regions have large regional shocks that lead to nearly equal output per capita
growth but different population growth, then one would expect a long, skinny
horizontal box. Conversely, if there are large regional shocks that are accommodated through wages and output, then there should be a tall, skinny vertical box. With small regional shocks, the boxes should be smaller because there
is less to be accommodated either way.
These gures show exactly what we would expect with large changes in
desired populations regionally but differences in restrictions on labor mobilitylarge countries have long, skinny horizontal boxes (nearly equal economic growth, differing population growth), while the other countries of the
10014-02_Ch02.qxd
8/18/06
10:28 AM
Page 55
t he fifth irresistible force
55
Figure 2-7. Large Shocks, Accommodated with Population Growth in
Large Countries, Per Capita Growth across Non-OECD Countries versus
European Countries
(Boxes at 90th/10th percentiles of each variable)
Growth of GDP (or income) per capita
.060
.050
.040
Spain (5587)
Italy (5090)
.030
France (5090)
.020
UK (5090)
Germany (5090)
.010
.000
Non-OECD countries (5090)
Zero
.015
.030
.005
.005
.015
.025
.035
.045
.055
.065
.075
.085
Growth of population less rate of natural increase
world show tall, skinny boxes (very little population growth difference, huge
differences in economic growth).4 The standard deviation of growth rates of
output per person across countries not belonging to the Organization for
Economic Cooperation and Development (OECD) is 1.9 percent a year. This
is ve to six times larger than the typical standard deviation of output growth
of regions within countries. In contrast, the standard deviation of the growth
of population less the rate of natural increasea proxy for the component of
4. This evidence alone of course does not resolve whether these variations across countries in labor demand are the result of policies (which presumably could be changed), institutions (which might be able to be changed), or geographic or technological shocks (which
cannot be changed).
10014-02_Ch02.qxd
8/18/06
10:28 AM
Page 56
the fifth irresistible force
56
population growth due to mobilityis 0.40, which is half the population
growth variability within regions of the United States, Canada, Japan, or
Spain and about that of most European countries.
Adjusting to Shocks, Then versus Now
The nineteenth century was truly an age of mass migration (Hatton and
Williamson 1998), because many of the areas of recent settlement had open
borders with respect to immigrants (at least with certain ethnic and national
origins). It was also an era of rapid reductions in transport costs and shifts
toward freer trade in goods, open capital markets, and massive movements in
capitalthe rst era of globalization. Hence, this period is an interesting
example of the question: How would we expect geographically specic
shocks to be accommodated in a globalizing world? Comparing Ireland to
Bolivia highlights the obvious: that nearly all developing countries with negative shocks have seen their populations continue to expand rapidly, while
when there was freer labor mobility in the international system, labor movements accommodated negative shocks (gures 2-8 and 2-9).
That is, during the entire period of Irelands huge negative shock of the
potato blight and its aftermatha classic example of a region-specic shock
that reduced desired, and likely optimal, population (just as the introduction
Figure 2-8. Changes in Real Wages and Population during the Period
of Accommodating the Shock of the Potato Famine and Its Aftermath in
Ireland, 18101920 a
Relative to 1870=1
1.6
1.4
1.2
1.0
0.8
0.6
0.4
0.2
Population
GDP PC
Wages/UK
1820 1830 1840 1850 1860 1870 1880 1890 1900 1910
Sources: Maddison 2001 for population and GDP per capita; ORourke and Williamson 1999 for real
wages.
a. Index of population, real unskilled urban wages, and GDP per capita, 187071.
10014-02_Ch02.qxd
8/18/06
10:28 AM
Page 57
t he fifth irresistible force
57
Figure 2-9. Changes in Real Wages and Population during the Period of
Accommodating Negative Shocks in Bolivia, 197095 a
1972=1
Population
1.5
1.3
1.1
RGDPPW
0.9
Wages/USA wages
0.7
1975
1980
1985
1990
1995
Sources: Penn World Tables 6.0 for output and population; Rama and Arcetona 2002 for industrial
wages.
a. Index of population, real industial wages relative to the United States, and GDP per capita, 1972 = 1.
of the potato, by lowering the cost of calories per hectare, had raised optimal
population)real wages in Ireland relative to the United Kingdom never fell
and gross domestic product (GDP) per capita never fell.
In contrast, Bolivia had a clear negative shock as well, but one that occurred
in a period in which there was little or no international labor mobility. So,
rather than the shock being accommodated by changes in population while
real wages of Bolivians remained constant (both in Bolivia and elsewhere), real
wages in Bolivia fell spectacularly.
Implications for Labor Mobility
Zambia is a country with a clear narrative. In part, people moved to Zambia,
and to a particular region of Zambia, because you could dig a hole in the
ground and extract something valuable (copper).5 Around that large hole in
the ground, a city developed. Now, the world economy and technological
5. I ike the example of Zambia because as a schoolchild I visited the worlds largest open
pit copper mine, the Bingham Mine outside Salt Lake City. Since the price of copper has fallen,
there have been hard times in the regions near the mine, and the mine has changed ownership
three times as various corporations have gotten into dire nancial straits.
10014-02_Ch02.qxd
58
8/18/06
10:28 AM
Page 58
the fifth irresistible force
conditions have changed such that it is likely the case that the protability of
digging copper out of the ground has been permanently reduced. Zambia is
also landlocked, so exporting manufactures is probably not in the cards. Zambia is not particularly overpopulated in the absolute sense of land/labor
ratios, but if Zambia were a region of a larger, integrated, geographic unit,
then its population would likely be a small fraction of what it is today. The
population of the Pennsylvania coal counties, where mining has shrunk as a
viable economic activity, declined by 30 percent in absolute terms over sixty
years. Zambias population is twice what it was at its peak output per person.
If we assume that Zambias optimal population has fallen by as much as the
regions in the United States30 percentthen Zambias current population
is almost three times higher than its optimal population.6 It is hard to see how
anything other than large sustained migration is going to reverse that.
One should rightly hesitate to declare that any particular territory is simply incapable of supporting its current population at acceptable standards of
living. But, conversely, simply maintaining a ction because it is politically
convenient for industrial countries is no better. I dene potential ghost
countries (which are all, given the lack of population mobility, zombies) as
countries where (1) GDP per capita has fallen by more than 20 percent from
peak to trough (where, for data purposes, the peak must come before 1990,
so recent ghosts are ruled out), and (2) GDP per capita today remains less
than 90 percent of peak GDP. This produces a list of thirty-three countries.
Of this list, I have no way of showing which countries are geographic
ghosts and which are not. In particular, I have no way of knowing which of
these are policy and institutional ghosts and which are geographic ghosts.
That is, it could be that anticipated output fell because of disastrously bad politics or policies, which, if reversed, would cause the area to be enormously
attractivethink of the boom Cuba is going to have when Fidel Castro is
gone, for instance. To document which are geographic ghosts, I would have
to specify and parameterize some particular model of location, which would
require grappling with the thorny issues of increasing returns to scale and the
like. Instead, I will do two calculations, which are hypothetical, and simply
illustrate the consequences of the possibility that these countries are ghosts.
First, because output per person has fallen in all these countries (by denition), I ask the question: If optimal population has received as large a neg-
6. Of course, this assumes that even with best possible policies and institutions, there is
still a large shock to the desired population, which is impossible to prove, because Zambia has
combined bad shocks with not the most sterling track record on the other dimensions.
10014-02_Ch02.qxd
8/18/06
10:28 AM
Page 59
t he fifth irresistible force
59
ative shock relative to its peak in this country as it has in the counterfactual
[see three options below], then what is the ratio of the postshock population
to the current population? The three counterfactual scenarios are What if
the population in country Y has fallen relative to its population at peak GDP
per capita by as much as the actual population
fell peak to trough in Ireland in the nineteenth century (53 percent)?
fell between 1930 and 1990 in three regions of the United States (Deep
South, Great Plains North, Pennsylvania Coal) (28 percent)?
rose only as fast as the bottom 10th percentile of population growth in
regions of the eight OECD countries in table 2-1 (0.01 percent a year)?
This is obviously not proof of the changes in the desired populations of
the countries, but just a matter of exploring the implications of plausible
counterfactual scenarios. In all these regions, GDP per capita rose substantially while populations fell. In the countries, GDP per capita fell while populations rose. It is at least plausible that these simply represent different
adjustments to similar-sized shocks to geographic-specic maximal incomes,
pushing the adjustment either into wages and capital stocks or into population movement.
Second, I ask the question: If the elasticity of GDP per person with respect
to population is negative 0.4, by how much would population have to fall in
order to
restore previous peak GDP per capita, or
move GDP per capita to the level it would be had it grown at 2 percent
a year since the peak (roughly the world average growth rate, hence just avoiding divergence)?
Table 2-2 shows ghosts that I believe are hard-core ghosts, in that they are
optimal population ghosts, not just desired population ghosts, for three reasons (actually, to keep the technical terminology clear, these ghosts are currently embodied as zombies because of population restrictions but would be
ghosts with labor mobility). First, the decline is more likely geographic than
policy or institutional. Though none of these countries has terric policies or
institutions, they are not the Zaires of the world that have resource abundance
but are political or institutional ghosts. Second, all these countries are landlocked, which makes the substitution into other industries more difcult.
Third, they all have small populations (less than 20 million), which suggests
that, in a locational equilibrium with population mobility, there might not be
sufcient population for even one large city to serve as a growth pole, in which
case the declines in desired population might be even more dramatic than
those in the table because of the agglomeration effects.
Zambia
CAF zone
Niger
Chad
Rwanda
Bolivia
Romania
0.59
0.44
0.50
0.50
0.75
0.87
0.74
10,089
3,603
10,832
7,694
8,508
8,329
22,435
Current
population
18
27
17
30
33
33
54
25
37
23
41
45
44
74
35
51
32
57
63
62
103
OECD
lagging
regionsb
(percent)
36
24
29
29
55
72
54
Previous
peak GDP
per capita
0.4 (percent)
14
11
11
17
30
34
34
GDP per capita
implying 2%
annual growth
since peak
(no divergence)
0.4 (percent)
Source: Authors calculations.
a. Potential hard-core ghosts.
b. Average of p10 of population growth (0.01 percent per annum growth).
CAF = African Financial Community Franc; GDP = gross domestic product; OECD = Organization for Economic Cooperation and Development.
1964
1970
1963
1979
1981
1978
1986
Country
or region
Ratio
GDPpc2000/
GDPpcpeak
U.S. ghost
regions
28% fall
from 1930
to 1990
(percent)
. . . the labor force fell to
restore GDP per capita to
X assuming an elasticity
of output per person to
population of 0.4
10:28 AM
Ireland
48% fall
from 1841
to 1926
(percent)
. . . the shock was as large
as the realized population
changes in the following
three cases:
Ratios of the population to the current actual population if . . .
8/18/06
Year
of peak
GDP per
capita
(GDPpc)
Table 2-2. How Large Is the Ghosthood?a
10014-02_Ch02.qxd
Page 60
10014-02_Ch02.qxd
8/18/06
10:28 AM
Page 61
t he fifth irresistible force
61
Because I began with Zambia, let me use it to illustrate both the very simple way the ve scenarios work and the results. Zambias GDP per capita
peaked in 1964 when its population was 3.5 million. Today, its GDP per capita
is only 59 percent of the peak, and the population is 10 million. If Zambias
population had fallen from its 1964 level by as much as Irelands actual population (48 percent), then its population today would be only 1.86 million
18 percent of its current level. If Zambias population had fallen from its 1964
level by as much as population has fallen in three of the ghost regions in the
United States (28 percent), then its population would only be 2.52 million
25 percent of its current level. If Zambias population had grown at the
0.01 percent of the 10th percentile in population growth regions of the eight
OECD countries, its population today would be about what it was in 1964,
3.52 millionbut that is only 35 percent of its current level.
The two output scenarios provide similarly striking ratios. Under the simple assumptions made about population and output per person, population
would have to fall to 14 percent of its current level to raise GDP per person to
the level of a nondivergent trend. This is consistent with a negative shock
roughly the magnitude of Irelands. To raise output per person just to its previous peak, the populations would have to fall to 36 percent of their current
levels.
I am aware of how striking these numbers are. But it is not implausible that
the optimal population of the Sahel (for example, Niger, Chad) has fallen by
as much as the optimal population of the Great Plains North counties of the
United States. That is, there is nothing of any particular Afro-pessimism in
this; this is not about the culture or politics of Africa any more than it is about
the culture or politics of Iowa or North Dakota (which are quite good). If this
is so, then, if population mobility were not constrained, three out of every
four people would leave Niger, and this might only be enough to restore output to its level of 1963. With the simple assumed elasticities, Chad, just to
return to its previous peak (1979) GDP per capita, would require that seven
of every ten people leave.
Conclusion
One force for increased population mobility is that many countries in the
world have experienced large negative shocks, such that, even with the best
possible responses in policies and institutions, the optimal population has
fallen signicantly. In the current international system, these people are
trapped. A helpful way of thinking about desired populations is the following: There are 10 million people in the Sahelian country of Niger; if there were
10014-02_Ch02.qxd
62
8/18/06
10:28 AM
Page 62
the fifth irresistible force
globally free labor mobility and only 1 million lived in Niger now, how many
people would move there? Though some people might say that this creates a
case for more aid or freer trade, it is hard to believe that if people moved out
of Kansas because farming was no longer an attractive opportunity, then the
best that can be done for the people of Niger or Chad is that they get slightly
more assistance and slightly better prices for the items they grow. The fth
irresistible force for labor mobility is changes over time in the optimal populations of regions as economic opportunities change.