Escolar Documentos
Profissional Documentos
Cultura Documentos
Journal/Year SSRN/2012
Authors Yiwei Dou, Ole-Kristian Hope, and Wayne, B.Thomas
Purpose of
Relationship-specific investments include investments made to support the unique
research
transactions between the firm and its suppliers or customers.
Methodology To decompose total income smoothing, we first run regression (2) for all firms in
country
c and industry i to estimate b3Pc,i and b3Lc,i. Then within each country with at least
20 industries we estimate the following regression for ISAGG:
ISAGGc,i = a0c + a1Pcb3Pc,i + a1Lcb3Lc,i + uc,i.
The predicted value of ISAGGc,i for industry i in country c proxies for the informational
component (IS_INFO), and the residual, ui, proxies for the garbling component
(IS_GAR) of income smoothing.
IS_INFOc,i = 0c + 1Pcb3Pc,i + 1Lcb3Lc,i
IS_GARc,i = uc,i
Estimate the following model to test the interaction effect on the IS_INFO and
Consistent with H2, we expect 3 > 0 for the IS_INFO component. Provide tests of
whether variations in market efficiency across countries affect these tests in Section VI.
Variable
Interaction of country-level legal quality (WEAK) and industry-level relationship-
specificity (RELSPEC) on income smoothing (ISAGG) in country c and industry i
with the following model:
Standard control variables for income smoothing: firm size, leverage, growth,
profitability, age of the firm, book-to-market ratio, and investment intensity
Sample and Period 1992 to 2004. Three years of additional data through 2007 to compute the
periods informational and garbling components of income smoothing. All accounting data are
from Compustat North America and Compustat Global.
Data from 39 countries and a maximum of 370 six-digit NAICS industries with data
missing for some industries in some countries.
Results
That firms that both reside in countries with weak legal quality and operate in high
relationship-specific industries tend to smooth their reported earnings more. In
addition, this relation is driven mainly by the informational role of income smoothing
Comments Although we include numerous controls and subject our results to additional
robustness tests, our findings are subject to certain caveats. First, we acknowledge
that the statistical associations we document do not necessarily establish causal
relations. Second, as we obtain our data from Compustat Global, which tends to
contain large firms, our results do not necessarily generalize to smaller firms around
the world. Third, despite our attempts to assure that measurement error in the income-
smoothing measure is not driving the results, we cannot rule out measurement error as
an alternative explanation for the results.