Is the character of SEC comment letters relevant to recipients?: empirical evidence of constraining allowance for doubtful accounts