1. Hoyt Bleakley
Univ. of Chicago
Joseph Ferrie
Northwestern Univ.
June 2013
Up from Poverty?
Long-run effects of the 1832 Cherokee
Land Lottery on Wealth
99. Eligibility
l The land in this area was made available to settlers of
European descent by the eviction of the Cherokees
in that area
l Every adult male who had been resident for at least
three years in Georgia by 1832 was eligible to one
draw in this lottery
l Widows and orphans were eligible for two draws, but
we drop them (can’t identify them in 1850)
100. Eligibility, part 2.
l Revolutionary War veterans were also eligible for
multiple draws, but we drop them as well (they
cannot be identified in 1850 when we create
treatment and control groups).
l But most were dead by 1850 anyway
l A group of highwaymen (members of “The Pony
Club”) were also excluded
l The Cherokee Land was surveyed and subdivided
l The lottery commences…
107. Specification Checks
l Is our sample more or less likely to have won a
parcel?
l For at least two counties (Columbia
Oglethorpe), lists of eligibles have survived.
l In our sample, 15.5% are (1/n) matched to the
Smith list. In Columbia Oglethorpe county,
16.0% of the eligible names were drawn
108. Did the non-winners actually register?
l 75.8K registered 1832 from
{WM; 18+ age; 3+yr in Ga}
l 78.0K in 1830 census
{WM; 15+age; currently in Ga}
l 97.2% registration rate
204. Analysis of Outcomes: Georgia Lottery
d variables, as seen in Section 3, casts doubt on t
We return to this issue in Section 8.3 with an alt
equation, which we generally estimate using O
ijk j ijk a k ijkY = T + BX + ⇥ + ⇥ + ⇤
jrest, T , is a binary variable that denotes treatm
aariables are as follows: ⇥ is a set of dummies for
393. Rawls again
ay to vary the equity weight when
an aggregator with a constant elas
¯Uj =
2
4
X
i2Zj
1
Nj
w
⇢ 1
⇢
i
3
5
⇢
⇢ 1
s in our sample, j is an indicator fo
497. Risk and churn
l Compare 1850 to 1860 wealth. Corr = 0.6
0.1.2.3.4
Probabilityof1860WealthNoGreaterThanThreshold
.1 .25 .5 1 2.5 5 10 20 30 50
Natural Log of Total Wealth, 1850
Fraction holding no more than $100 in 1860
95% Conf. Int.
Prob. Density Function in 1850
498. Hoyt Bleakley
Univ. of Chicago
Joseph Ferrie
Northwestern Univ.
December 9, 2012
Up from Poverty?
Long-run effects of the 1832 Cherokee
Land Lottery on Wealth
499. Extensions / Related Work
l Placebo sample from South Carolina
l Relaxing the rank-invariance assumption in the
quantile analysis
l Intergenerational effects
l Importance of initial property allocation
(follow up the place rather than the person)
500. Mapping to returns?
l Rank invariance (quantile regs)
l Compute bounds for weaker assumptions,
using Markov matrix P:
l xt' = xc' P
l 1 = P 1
l 1 P 0
l Expected returns: Py or (P-I)y, net of starting yi
l Linear in P, with bounds.
501. Mapping to returns?
l Rewrite as
l A vec(P) = b
l A, b formed by Kronecker product of Identity
matrix with x’s and 1 vectors.
l 1vec(P)0
l Treatment effect Δ = C’vec(P); C = kron(y’,Im)’
l Compute bounds on elements of Δ
502. Mapping to returns?
l Linear programming
l Assumptions:
1. Bounds only
2. Ex post Pareto dominates to win (|yi)
3. FOSD to win (|yi)
4. Expected Δ positive for all yi
l Feasible?
l Bounds?
503.
504. Shocking Behavior
The Cherokee Land Lottery of 1832
and Later Life Outcomes
Hoyt Bleakley
Univ. of Chicago
Joseph Ferrie
Northwestern Univ.
Spring 2013
505. Land Openings
on the Georgia
Frontier
and
the Coase
Theorem in the
Short- and
Long-Run
Hoyt Bleakley
(Univ. of Chicago)
Joseph Ferrie
(Northwestern Univ.)
Spring 2013
506.
507. Summary
l Land Openings in “Lottery Zone” of Georgia
l Initial Parcel Sizes were...
l Quite sticky for c. 80 years (one for one)
l Unstuck by c. 150 years
l Misallocation cost c. 20% of 1880 land value