A researcher studying a sample of 250 firms has estimated two alternative regression models for their access to bank loans:
Model 1
Loans = 42.25 -0.47*(X1) + 0.83*(X2) + 0.35*(X3) - 0.71*(X4)
R-squared 0.75
Model 2
Loans = 38.52 -0.37*(X1) + 0.75*(X2) + 0.42*(X3)
R-squared 0.40
Where
Loans: is the ratio of the value of loans a firm has been granted to its total assets
X1: is a variable attaining the value of one if the firm was a single proprietorship and zero otherwise
X2: is the ratio of the value of profits a firm has to its total assets
X3: is the logarithm of its total assets
X4: is a variable attaining the value of one if the firm does not use external auditors, and zero otherwise.
• Which of the two model specifications should the research adopt, and why?
• Discuss the qualitative and quantitative impact of external auditing on firms’ access to bank loans. Suppose you were told that the standard error of the X1 variable in the preferred specification was 0.07, is there any evidence for ownership-type effects on the bank loan market?
• Based on the estimated parameters, what do you conclude about the loan distribution across firms, depending on whether they use external auditors and the type of ownership (assume that both factors enter significantly in the model)?