We document that mortgaged homebuyers pay an 11% premium relative to all-
cash buyers in residential real estate transactions. This premium far exceeds the 3%
premium implied by a realistically calibrated model of rational home sellers with trans-
action frictions. We obtain similar results from various estimators (e.g., repeat-sales,
instrumental-variable, matching, semi-structural), novel data on non-accepted offers,
and an experimental survey of U.S. homeowners. Experimental evidence suggests that
pessimistic priors and uncertainty driven by inexperienced sellers best explain the puz-
zle of 8% (11% − 3%). Our findings matter economically, as all-cash purchases account
for one-third of all U.S. home purchases over 1980-2017.
Consider a home seller with offers from two competing buyers: one is mortgage-financed,
and the other is all-cash. Define the mortgage-cash premium as the expected difference in
log prices between the two offers. In the absence of frictions, the mortgage-cash premium
ought to be zero (Modigliani and Miller 1958). More realistically, the premium ought to be
positive to compensate sellers for frictions in the mortgage origination process, namely risk
of transaction failure and a longer time to close. We find that the mortgage-cash premium
averages 11% over the past 40 years. In dollar terms, a seller would be indifferent between
a $500,000 all-cash offer, the average in our sample, and a mortgaged offer that is $55,000
larger. The magnitude of this premium far exceeds reasonable compensation for transaction
frictions implied by traditional economic logic. In policy terms, U.S. taxpayers subsidize
$8 trillion of mortgages to promote homeownership (Federal Reserve 2019); reducing the
mortgage-cash premium would enable a smaller subsidy to accomplish the same goal.
