Skip to main content
. Author manuscript; available in PMC: 2021 Jan 5.
Published in final edited form as: J Policy Anal Manage. 2018 Nov 5;38(1):65–98. doi: 10.1002/pam.22104

Table 6.

Effects of housing vouchers on labor supply and public assistance receipt (LATE vs. ATE).

LATE (Replicating Jacob and Ludwig 2012) ATE
CM
(1)
IV
(2)
CCM
(3)
Obs.
(4)
IV
(5)
Obs.
(6)
HHH Employed 0.592 −0.036**
(0.009)
0.605 42,358 −0.032**
(0.015)
42,025
HHH Earnings 3291.02 −328.949**
(74.560)
3113.90 42,358 −471.420***
(137.952)
42,025
HHH Earnings>$3220 (FT@$8/hr) 0.404 −0.045**
(0.009)
0.403 42,358 −0.056***
(0.015)
42,025
HHH Earnings conditional on working 5557.98 −227.544**
(80.221)
5128.37 38,628 −435.003***
(144.568)
38,362
HHH Log earnings conditional on working 8.279 −0.073**
(0.018)
8.220 38,628 −0.117***
(0.029)
38,362
HHH Received public assistance 0.46 0.067**
(0.009)
0.552 42,358 0.075***
(0.012)
42,025
HHH Received TANF 0.146 0.017**
(0.004)
0.110 42,358 0.018***
(0.005)
42,025
HHH Received Medicaid 0.4 0.058**
(0.009)
0.484 42,358 0.062***
(0.012)
42,025
HHH Received Food Stamps 0.375 0.076**
(0.008)
0.449 42,358 0.084***
(0.011)
42,025

Notes: Columns 2 through 4 replicate the results from Table 3 of Jacob and Ludwig (2012). Column 5 presents re-weighted estimates, weighting by the inverse of the predicted compliance probability. The difference in sample size between columns 4 and 6 is because we trim observations with less than a 5 percent predicted compliance probability. HHH = household head. CM = control mean. IV = instrumental variables. CCM = control complier mean. See text for discussion of these estimates. We construct standard errors (clustered at the household level) by bootstrapping (N = 200) the entire estimation process to account for variance in the estimated compliance score. All earnings measured in 2007 dollars.

***

Significant at the 1 percent level;

**

5 percent level;

*

10 percent level.