The results of the estimates and the “variable deletion tests” which have determined the final and
“parsimonious “ specification are reported in the appendix. The final “parsimonious” specification yields
the estimates of Table 2.
TABLE 2
FINAL “PARSIMONIOUS” SPECIFICATION
EQUATION FOR THE SPREAD BETWEEN INTEREST RATES ON BANK OVERDRAFT AND
MONETARY POLICY RATE (INTEREST RATE ON LOANS BY THE BANK OF ITALY)
The data employed for the estimates cover the period 1990 QI - 1998 QIV
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
D.P.D. RESULTS
LEVELS IV
Number of firms: 54 Sample period is 1991 QII to 1998 QI
Observations: 1512 Degrees of freedom: 1497
Dependent variable is: ta_d
Instruments used are:
CONST ta-d(-2) so_u(-1) so_u(-2) tdif(-4) tdif(-5) c3 c4
a1 a3 a4 a5 a6 IND DUMS
ONE-STEP ESTIMATES WITH ROBUST TEST STATISTICS
|
Wald test |
of joint significance: |
2230.202567 |
df = |
12 | |
|
Wald test |
- jt sig of |
ind dums: |
1.571201 |
df = |
2 |
|
Var |
Coef |
Std. Error |
T-Stat |
P-Value | |
|
CONST |
1.302663 |
1.159927 |
1.123056 |
0.261414 | |
|
ta_d(-1) |
0.748006 |
0.124735 |
5.996774 |
0.000000 | |
|
so_u |
5.543778 |
6.650326 |
0.833610 |
0.404501 | |
|
so_u(-1) |
-5.534232 |
6.553838 |
-0.844426 |
0.398431 | |
|
tdif(-3) |
-0.958507 |
1.332951 |
-0.719086 |
0.472088 | |
|
tdif(-4) |
0.750205 |
0.976935 |
0.767917 |
0.442537 | |
|
c3 |
0.057856 |
1.302231 |
0.044429 |
0.964563 | |
|
c4 |
0.072287 |
1.034254 |
0.069893 |
0.944279 | |
|
a1 |
-0.032614 |
0.085680 |
-0.380652 |
0.703462 | |
|
a3 |
0.125009 |
0.094511 |
1.322695 |
0.185937 | |
|
a4 |
0.249281 |
0.109245 |
2.281853 |
0.022498 | |
|
a5 |
0.393597 |
0.210871 |
1.866527 |
0.061968 | |
|
a6 |
0.303455 |
0.241046 |
1.258909 |
0.208063 | |
|
I2 |
-0.247605 |
0.198526 |
-1.247216 |
0.212318 | |
|
I3 |
-0.392050 |
0.325261 |
-1.205340 |
0.228072 | |
Robust test for first-order serial correlation: -2.878 [ 54 ]
Robust test for second-order serial correlation: -0.386 [ 54 ]
The variables referred to the overdue bank credits (“so_u” and “so_u(t-1)”) are only significant at
a level of confidence of around 60%, while “tdiff(-3)” and “tdif(-4)” (capturing the bargaining power of
the banks) are only significant at a level of confidence of 53% and 56% respectively, although they were
much more significant in the “general unrestricted model”, which also did not support a linear restriction
23