Does the Great Mortgaging Crowd Out Small Firms?

Mortgage lending, bank capital and the size distribution of firms

Claire Borrelli

Economic History, Sciences Po

2026-10-11

Motivation: the Great Mortgaging

Average of 11 countries, 1960–2019: firms 59% → 38% of bank credit, mortgages 38% → 62%. Over the century: mortgages about 30% around 1900, about 60% today. JST data; Jordà, Schularick & Taylor 2016

Question: did this shift crowd out lending to small, bank-dependent firms, while large firms turned to bond markets?

A tension in the literature

US counties: Chakraborty et al. (2018)

  • 1988–2006, banks in strong housing markets
  • More mortgages, less commercial lending
  • Their borrowers invest less, especially constrained firms and banks

Spain: Martín, Moral-Benito, Schmitz (2021)

  • Housing boom first reduces credit to non-housing firms
  • Later stimulates it: the boom raises bank net worth
  • Crowding-out large but temporary, absorbed by the end of the boom

So is crowding-out only a short-run effect? The aggregate effect can fade while its composition does not: who loses credit, and do they come back?

My idea: firm size

Bank with limited capital

A euro of capital must earn the same in both uses:

\[R^f_t - R^d_t = \frac{\omega^f}{\omega^m}\,\big(R^m_t - R^d_t\big)\]

Mortgages pay more ⇒ the bank asks more from firms.

\(R^f, R^m\): firm and mortgage rates; \(R^d\): funding cost; \(\omega^f, \omega^m\): capital charged per euro lent (Bâle I: 1 and 0.5).

Short run (capital fixed)

Firms pay a higher rate: small firms invest less or exit. Large firms escape through bonds.

Long run (capital adjusts)

Crowding-out in levels fades (as in Spain), but firms stay structurally disadvantaged, and small firms that exited may not come back.

Usable data

Source What it gives Coverage
County Business Patterns Establishments and jobs by size class and sector County, yearly
CRA Small-business loans by bank and county From 1996
HMDA Mortgage lending by bank From the 1990s
FDIC Summary of Deposits Where each bank has branches From 1994
Call Reports Bank balance sheets, capital Quarterly, long run
FHFA + Saiz (2010) House prices + housing supply elasticity County / metro
JST Macrohistory Mortgage and non-mortgage credit, 17 countries Since 1870

All public, already collected.

Identification: the Saiz instrument

Push banks toward mortgages for reasons unrelated to local firms

Housing booms in the other counties where a bank has branches, predicted by how hard it is to build there (Saiz 2010):

\[\text{Expo}_{c} = \sum_{\text{banks } b} \underbrace{s_{b,c}}_{\text{share in } c} \;\sum_{j\neq c} \underbrace{w_{b,j}\,\widehat{\Delta HPI}_{j}}_{\text{booms elsewhere}}\]

Step 1: banks

More mortgages, fewer small-business loans? Stronger when capital is low?

Step 2: firms, 1 to 15 years

Small vs large establishments in the same county. Does the gap close, as in Spain, or persist?

References

  • Adelino, M., Schoar, A. & Severino, F. (2013). House Prices, Collateral and Self-Employment. NBER Working Paper 18868.
  • Chakraborty, I., Goldstein, I. & MacKinlay, A. (2018). Housing Price Booms and Crowding-Out Effects in Bank Lending. Review of Financial Studies, 31(7), 2806–2853.
  • Jordà, Ò., Schularick, M. & Taylor, A. M. (2016). The Great Mortgaging: Housing Finance, Crises and Business Cycles. Economic Policy, 31(85), 107–152.
  • Jordà, Ò., Schularick, M. & Taylor, A. M. (2017). Macrofinancial History and the New Business Cycle Facts. NBER Macroeconomics Annual, 31, 213–263. (JST Macrohistory Database)
  • Martín, A., Moral-Benito, E. & Schmitz, T. (2021). The Financial Transmission of Housing Booms: Evidence from Spain. American Economic Review, 111(3), 1013–1053.
  • Mian, A., Sufi, A. & Verner, E. (2017). How Do Credit Supply Shocks Affect the Real Economy? Evidence from the United States in the 1980s. NBER Working Paper 23802.
  • Saiz, A. (2010). The Geographic Determinants of Housing Supply. Quarterly Journal of Economics, 125(3), 1253–1296.

Backup

Backup: the bank’s problem

\[\max_{L^m_t,L^f_t,S_t,D_t}\; R^m_t L^m_t + R^f_t L^f_t + R^s_t S_t - R^d_t D_t\] \[\text{s.t.}\quad L^m_t + L^f_t + S_t = N_t + D_t,\qquad \omega^m L^m_t + \omega^f L^f_t \le \phi N_t\]

\[\frac{R^m_t - R^d_t}{\omega^m} = \lambda_t = \frac{R^f_t - R^d_t}{\omega^f} \qquad\qquad dL^f_t = -\frac{\omega^m}{\omega^f}\,dL^m_t \quad (N_t \text{ fixed})\]

Long run (\(N_t\) chosen, cost \(R^e_t\)):

\[\lambda_t = \frac{R^e_t - R^d_t}{\phi} \quad\Rightarrow\quad R^f_t = R^d_t + \frac{\omega^f}{\phi}\big(R^e_t - R^d_t\big)\]

The extended version adds default risk, collateral value (house prices) and information costs.

Backup: step 2 regression

\[Y_{c,s,k,t+h} - Y_{c,s,k,t-1} = \beta_h\,\text{Expo}_{c,t}\times\text{Small}_k + \theta_h\,\Delta HPI_{c,t}\times\text{Small}_k + \delta_{c,t} + \mu_{s,k,t} + \varepsilon\]

  • \(c\) county, \(s\) sector, \(k\) size class, \(h = 1,\dots,15\) years
  • Prediction: \(\beta_h < 0\); persistence tells short run from long run
  • Local house prices control for the home-collateral channel (Adelino, Schoar & Severino 2013)
  • Robustness: tradable sectors only (manufacturing)

Backup: larger project

  • Bâle I (1988): mortgages charged half the capital of firm loans, a historical shock to \(\omega^m/\omega^f\)
  • Back to the long-run trend across JST countries: land vs productive capital
  • Model extension: rising cost of capital, or exit with entry costs, to make the effect persist