intervention by a management board replacement, and a value of 4 is then assigned to them.
Afterwards, we search for the ACC-year observations with a central ACC intervention by an
agent, and a value of 3 is then assigned to them. Next, we check for the remaining ACC-year
observations those where the chairman of the management board has been replaced and a value of
2 is assigned to them. Finally, we search for the remaining observations for those cases where the
board members changed. The “board change” variable takes the value of 1. After all of this, the
remaining ACC-year observations correspond to non-intervention cases, and have a 0 in our
measure of governance interventions.
As explanatory variables, we used some ACCs performance measures values (Cabo and
Rebelo, 2005), which are quantified in the financial statements, or annual reports:
■ The management of the credit risk reflected by the bad loans variable (X1), measured by
bad loans as percentage of gross credit;
■ Operational costs efficiency by the variables labour costs (X2) and other administrative
costs (X3), both as a function of turnover (interest received + fees + other operational
benefits).
■ As profitability measure we use return on equity7 (X4) measured after taxes.
■ Solvency (X5), measured by the ratio equity to total liabilities, trying to capture the aim of
ACCs to fulfil the capital adequacy requirements of the Bank of Portugal.
Two characteristics are used as control variables: the size of the ACC, measured by the total
assets at the end of the year and the time period (year). Size is often correlated with other
unobserved variables such as asset diversification and managerial skills (Crespi et al., 2004). The
time period variable controls shocks common to all ACCs in a given year.
7 An ACC’s goal is not maximizing profit but, as mentioned earlier, the key-issue for ACCs is the lack of equity.
Therefore, as the growth in equity is fuelled completely by profit retained earnings, ROE is the correct variable
to express the “profitability”.