RDP 2022-05: The Real Effects of Debt Covenants: Evidence from Australia Appendix C: Trends before and after Treatments

Kim Nguyen

October 2022

C.1 Test for parallel trends

To test for parallel trends before the covenants exposure treatment, I estimate the following regression on investment and staff expenses the years before the treatment:

Δ y i;t = α i + β 0 NoDi s i,t−1 * trend+ β 1 Di s i,t−1 *trend+ X ′ i,t−1 μ+ ε i,t

The parameter of interest is the differential time trend coefficients, β 0 − β 1 . The two groups follow parallel trends if the parameter is statistically significantly zero.

Table C1: Test for Parallel Trends between Control and Treatment Groups
Wald test for the difference between time trends
  Time trend coefficients Difference
β 0 − β 1
Control group ( β 0 ) Treatment group ( β 1 )
Investment −0.0065863
(0.0071332)		 −0.0066515
(0.0071293)		 0.0000652
(0.0000284)
Staff expenses −0.0114725
(0.0065714)		 −0.0115148
(0.0065660)		 0.0000423
(0.0000206)

Note: Clustered standard errors at firm level are shown in parentheses.

Sources: Author's calculations; Connect4; Morningstar
Figure C1: Trends in Financial Statistics
Non-financial listed firms, average
Figure C1: Trends in Financial Statistics

Notes: (a) Defined as firms not exposed to covenants in both t – 2 and t – 1.
(b) Defined as firms not exposed to covenants in t – 2, exposed in t – 1 but no breaches in t – 1.

Sources: Author's calculations; Connect4; Morningstar

Figure C2: Financial Statistics and Covenant Breaches
Non-financial listed firms, average
Figure C2: Financial Statistics and Covenant Breaches

Sources: Author's calculations; Connect4; Morningstar