Table 8

Robustness Checks

School TypePanel A: Multinomial Logit Estimates
Renters Only
(1)		Controlling for 1997 Home Price
(2)		Including Non-attenders
(3)		Using Top 50 Public
(4)
Flagship/top 50 public2.52e−9
(3.90e−9)		0.0023**
(0.0011)		0.0014**
(0.0005)		0.0024*
(0.0014)
Four-year private0.0002
(0.0002)		−0.0014
(0.0028)		0.0012
(0.0016)		0.0001
(0.0018)
Community college−0.0025
(0.0021)		−0.0069**
(0.0028)		−0.0060*
(0.0035)		−0.0054*
(0.0030)
No college attendance  −0.0095**
(0.0033)		 
Dependent VariablePanel B: School Resource Estimates
Renters Only
(i)		Controlling for 1997 Home Price
(ii)		State Fixed Effects
(iii)		1992–94 Average Quality Measures
(iv)
25th percentile math SAT0.476
(0.448)		0.778**
(0.363)		0.729**
(0.338)		 
75th percentile math SAT0.796
(0.482)		0.676**
(0.320)		0.523*
(0.316)		 
Faculty-student ratio0.0003
(0.0002)		0.0003**
(0.0001)		0.0003**
(0.0001)		0.0002*
(0.0001)
Expenditures per student63.487
(53.404)		235.958**
(80.308)		295.864**
(98.210)		224.699**
(74.304)
Instructional expenditures per student14.059
(15.257)		53.992**
(17.731)		63.598**
(22.875)		82.371**
(25.503)
Graduation rate0.0007
(0.0013)		0.0011
(0.0009)		0.0005
(0.0015)		 
Tuition−14.369
(30.553)		80.922**
(30.565)		78.436**
(36.818)		16.379
(35.371)

  • Notes: All estimates in Panel A include state fixed effects and all estimates in Panel B except those in Column 3 contain MSA fixed effects. All estimates include age in 1997 fixed effects as well as controls for all covariates used throughout the analysis and are weighed by sampling weights. Estimates in Columns 2–4 include only homeowners. Housing price changes are real housing price changes over the four years prior to students turning 18 predicted by the conventional mortgage housing price index. In Column 1, the main independent variable is the four-year percent change in home prices at the MSA-level, and the sample is restricted to renters in 1997. Standard errors clustered at the MSA-level are in parentheses:

  • ↵** indicates significance at the 5 percent level and

  • ↵* indicates significance at the 10 percent level.