Multivariate Regression of Interest Rate Derivatives on Exposure Independent Variable Sign Model 1 Coef. (t-stat) Model 2 Coef. (t-stat) Model 3 Coef. (t-stat) Model 4 Coef. (t-stat) Model 5 Coef. (t-stat) Intercept +/-
.154
(4.55)***
.173
(5.14)***
.095
(0.83)
.166
(4.17)***
.066
(1.24)
NVIR -
-.139
(-1.82)**
-.047
(2.12)**
-.226
(-1.67)**
-.140
(-1.79)**
-.004
(-0.03)
Float -
-.071
(-2.86)***
-.069
(-2.87)***
-.060
(-0.90)
-.049
(-1.80)**
-.114
(-2.98)***
BM Ratio
-
-.015
(-2.06)**
-.020
(-2.81)***
-.005
(-0.19)
-.017
(-2.28)**
-.010
(-0.86)
ICR -
-.016
(-3.69)***
-.016
(-3.97)***
-.015
(-1.95)**
-.019
(-4.04)***
-.012
(-2.16)**
LSales +
.018
(3.69)***
.020
(4.33)***
.024
(1.60)*
.018
(3.45)***
.022
(3.23)***
Foreign +
.076
(2.43)***
.075
(2.39)***
.053
(0.68)
.083
(3.13)***
.151
(3.22)***
Quick +
.021
(1.46)*
.024
(1.63)*
.022
(0.80)
-.001
(-0.03)
.029
(1.66)**
F Ratio
11.03***
12.58***
3.09**
10.46***
8.84***
R2
.1889 .1956 .1723 .2019 .1260
Observations
235 235 47 190 235
Model 3 is for 2004 with continuous hedge variable. Model 4 is for users only with continuous hedge variable. Model 5 is the same as model 1
but the dependent variables uses the raw exposure coefficient. Exposure is the dependent variable and measured as absolute value of the
coefficient from the Augmented Capital Asset Pricing Model as described in Equation 1. Float is the ratio of floating rate debt/total long-term
debt. LSales is measured as the log of Sales. NVIR is the total notional value of interest rate derivatives used for hedging purposes (excluding
foreign currency derivatives and speculative derivatives). BM Ratio is the book-to-market ratio measured as the book value of equity/market
value of equity. The book-to-market ratio is winsorized for two extreme observations. ICR represents the interest coverage ratio measured as
income before depreciation/interest expense. Foreign is the ratio of foreign sales/total sales. Quick is the sum of cash, short-term investments
and accounts receivables/current liabilities. In Model 2, NVIR is a dummy variable that is equal to one if the firm is a derivative user and zero
otherwise. Model 3 is a cross-sectional regression for 2004 only. Model 4 is a cross-sectional regression for derivative users only. Model 5 is
shown for comparative purposes and uses the raw exposure coefficient from Equation 1 as the dependent variable.
The results show a significant negative relation between the use of interest rate derivatives and interest rate
exposure at the 5% level in Models 1-4. Lodging firms use derivatives to reduce exposure and not to speculate or
increase exposure as part of their overall risk management program, consistent with prior research. This is an
important finding even after controlling for other risk exposures given that few studies that have investigated the
effect of interest rate hedging on exposure and provided evidence of a significant risk reduction (Guay, 1998;
Allayannis & Ofek, 2001). Exposure from liabilities as measured by floating rate debt is significantly negative in
almost all models except Model 3 which is limited by a smaller sample size. In response to concerns about interest
rates over the sample period, derivative users reduced their floating rate debt exposure by altering the exposure from
a floating rate to a fixed rate through the use of interest rate swaps or interest rate caps. The descriptive statistics on
floating debt and the correlation between derivatives and floating debt from the descriptive analysis provide support
for this finding. Consistent with prior research, firms with high growth opportunities (low book-to-market ratios)
face higher exposure because of the potential for underinvestment costs. The interest coverage ratio is significantly
negative suggesting that firms with lower coverage ratios face greater exposure from a higher potential for financial
distress costs if they are unable to service their debt. Firm size and the foreign sales ratio are also significant and
positively related to exposure implying that large firms and firms with foreign income face greater exposure from
13
For an explanation of the differences between the models, please refer to the table 4 footnotes