The Collateral Channel: How Real Estate Shocks Affect Corporate Investment

The Collateral Channel: How Real Estate Shocks Affect Corporate Investment

“…These effects 22 In particular, in unreported regressions, we see that most of the drop in the sensitivity comes from adding the control for the Market-to-Book ratio and not from adding Cash. 23 As we explain in our working paper (Chaney, Sraer, and Thesmar 2009), the drop in β once the Market-toBook ratio is controlled for can easily be interpreted in the light of a simple model of investment with collateral constraints. Intuitively, to leave the Market-to-Book ratio unchanged after a positive shock to the value of the firm's real estate assets, there needs to be a negative shock to unobserved productivity.…”

“…In the working paper versionChaney, Sraer, and Thesmar (2009), we develop a simple model of investment under collateral constraint to justify this specification.18 Formally, if the estimated coefficient ˆ β is positive after controlling for the market value of a firm, we can reject the assumption that firms do not face any financing constraint, while if the estimated coefficient ˆ β is negative, we cannot reject this assumption.19 Using only the initial value of real estate in 1993 offers an additional advantage: if a firm discovers a profitable investment opportunity, and if it leases some of its real estate, we may expect that its landlord will try to extract as much rent as possible from this future investment; to escape from this hold-up problem, we may expect this firm to become owner of its real estate exactly when it is about to invest; in such a scenario, we would then see a spurious correlation between the current value of the real estate a firm owns and its investment. We circumvent this problem by using variations in the value of real estate that come only from market prices, and not from the contemporaneous strategy of the firm.…”

Collateral and the Choice Between Bank Debt and Public Debt

“…The underlying assumption is that the majority of a firm's real-estate assets are located either in the same MSA or the same state as its headquarters. As pointed out by Chaney et al (2012), if not all of a firm's real-estate assets are located in its headquarters MSA or state, the empirical methodology tends to overestimate the fraction of a firm's realestate assets that comoves with local real-estate prices and therefore to underestimate the impact of collateral on a firm's financing decisions. The magnitude of the measurement error will depend on the likelihood that a firm's real-estate assets are located outside of the headquarters MSA or state and the time-series correlation of real-estate prices across different MSAs or states during the sample period.…”

“…8 Because real-estate assets are reported at historical cost, the market value of real-estate assets can be very different from the Compustat numbers depending on the time when the assets were purchased and the real-estate price variation from the purchase year to the reporting year. Therefore, my second measure of real-estate ownership is to follow Chaney et al (2012) 8 The drawback is that it introduces measurement errors to the amount of real-estate assets that are exposed to fluctuations in realestate prices. An alternative that reduces the measurement errors is to measure a firm's real-estate assets at the beginning of the year and examine how the shocks to real-estate prices affect its financial decision by the end of the year.…”

“…To address the endogeneity concern that the local real-estate prices are driven by local economic shocks that affect a firm's financing costs, I use the land supply elasticities introduced by Saiz (2010) as an instrument for real-estate prices. (Mian andSufi 2011 andChaney et al 2012 also use this instrument.) Still, even local real-estate cycles not induced by local shocks could have a feedback effect on local economies and perhaps local credit supply.…”