“…However, despite our expectations, CASHFX (significant at 1%) did not present a positive relationship with HA, maybe considering that the cash effect of the exchange gains and losses is not directly linked to the accrued losses on the income statements and, consequently, on OCI's balances. In addition, ROE was not significant in our model, also despite our expectations, mainly because we were not able to isolate the exchange rate effect (gains and losses), which is a limitation already discussed by the supporting theory (RICHIE et al, 2006;AABO et al, 2015).…”