Problem Note 67918: The QLIM procedure generates a Read Access Violation error when the OVERID option is specified
 |  |  |  |
If the OVERID option is specified in the ENDOGENOUS statement or the MODEL statement in PROC QLIM, the following error messages might be issued in the log:
Please contact technical support and provide them with the following traceback information:
ERROR: Read Access Violation QLIM
Exception occurred at (70B8A011)
Task Traceback
When this error is produced, the overidentification test statistic and p-value are omitted from the output.
To circumvent the problem, specify two separate PROC QLIM steps. In the first PROC QLIM step, remove the OVERID option and obtain parameter estimates and output statistics. In the second PROC QLIM step, compute the overidentification test statistic and p-value as follows:
- Add all of the overidentifying instruments originally specified in the OVERID option to the structural equation.
- Specify the LR option in a TEST statement to test whether all of the added overidentifying instruments in the structural equation are jointly equal to zero.
If an overidentifying instrument is a CLASS variable, then the test should include parameters for all levels of the CLASS variable except for the reference level. The likelihood ratio test statistic and p-value produced in the output are the overidentification test statistic and p-value.
Here is an example that illustrates the circumvention.
Suppose the following PROC QLIM step generates the errors shown above:
model y1 = y2 z1 z2 / discrete overid(y2.z4 y2.y5);
model y2 = z1 z2 z3 z4 z5;
run;
To avoid the error, first run the PROC QLIM step without the OVERID option and obtain parameter estimates and output statistics:
model y1 = y2 z1 z2 / discrete;
model y2 = z1 z2 z3 z4 z5;
run;
Then, to compute the overidentification test, add the overidentifying instruments originally specified in the OVERID option in the structural equation. In this example, the overidentification instruments z4 and z5 are added to the structural equation y1. Specify the LR option in a TEST statement to test the hypothesis that the overidentifying instruments z4 and z5 are jointly equal to zero in the structural equation:
model y1 = y2 z1 z2 z4 z5/ discrete;
model y2 = z1 z2 z3 z4 z5;
test y1.z4 = 0, y1.z5=0 / lr ;
run;
The reported LR test statistic and p-value are the overidentification test statistic and p-value.
Here is an example of the circumvention if the original model includes a CLASS variable. Suppose the overidentifying instrument z4 is a CLASS variable with levels 1, 2, 3, 4, and 5, and level 5 is the reference level with an estimate of 0.
class z4 ;
model y1 = y2 z1 z2 / discrete overid(y2.z4 y2.y5);
model y2 = z1 z2 z3 z4 z5;
run;
As in the previous example, the first PROC QLIM step is as follows:
class z4 ;
model y1 = y2 z1 z2 / discrete;
model y2 = z1 z2 z3 z4 z5;
run;
Here is the second PROC QLIM step for the circumvention when a CLASS variable is involved:
class z4 ;
model y1 = y2 z1 z2 z4 z5/ discrete ;
model y2 = z1 z2 z3 z4 z5;
test y1.z4_1 = y1.z4_2 = y1.z4_3 = y1.z4_4 =0, y1.z5 = 0 / lr ;
run;
Operating System and Release Information
| Product Family | Product | System | Product Release | SAS Release | ||
| Reported | Fixed* | Reported | Fixed* | |||
| SAS System | SAS/ETS | z/OS | 13.1 | 15.1 | 9.4 TS1M1 | 9.4 TS1M6 |
| Microsoft® Windows® for x64 | 13.1 | 15.1 | 9.4 TS1M1 | 9.4 TS1M6 | ||
| Microsoft Windows 8 Enterprise 32-bit | 13.1 | 15.1 | 9.4 TS1M1 | 9.4 TS1M6 | ||
| Microsoft Windows 8 Enterprise x64 | 13.1 | 15.1 | 9.4 TS1M1 | 9.4 TS1M6 | ||
| Microsoft Windows 8 Pro 32-bit | 13.1 | 15.1 | 9.4 TS1M1 | 9.4 TS1M6 | ||
| Microsoft Windows 8 Pro x64 | 13.1 | 15.1 | 9.4 TS1M1 | 9.4 TS1M6 | ||
| Microsoft Windows 8.1 Enterprise 32-bit | 13.1 | 15.1 | 9.4 TS1M1 | 9.4 TS1M6 | ||
| Microsoft Windows 8.1 Enterprise x64 | 13.1 | 15.1 | 9.4 TS1M1 | 9.4 TS1M6 | ||
| Microsoft Windows 8.1 Pro 32-bit | 13.1 | 15.1 | 9.4 TS1M1 | 9.4 TS1M6 | ||
| Microsoft Windows 8.1 Pro x64 | 13.1 | 15.1 | 9.4 TS1M1 | 9.4 TS1M6 | ||
| Microsoft Windows 10 | 13.1 | 15.1 | 9.4 TS1M1 | 9.4 TS1M6 | ||
| Microsoft Windows Server 2008 | 13.1 | 15.1 | 9.4 TS1M1 | 9.4 TS1M6 | ||
| Microsoft Windows Server 2008 R2 | 13.1 | 15.1 | 9.4 TS1M1 | 9.4 TS1M6 | ||
| Microsoft Windows Server 2008 for x64 | 13.1 | 15.1 | 9.4 TS1M1 | 9.4 TS1M6 | ||
| Microsoft Windows Server 2012 Datacenter | 13.1 | 15.1 | 9.4 TS1M1 | 9.4 TS1M6 | ||
| Microsoft Windows Server 2012 R2 Datacenter | 13.1 | 15.1 | 9.4 TS1M1 | 9.4 TS1M6 | ||
| Microsoft Windows Server 2012 R2 Std | 13.1 | 15.1 | 9.4 TS1M1 | 9.4 TS1M6 | ||
| Microsoft Windows Server 2012 Std | 13.1 | 15.1 | 9.4 TS1M1 | 9.4 TS1M6 | ||
| Windows 7 Enterprise 32 bit | 13.1 | 15.1 | 9.4 TS1M1 | 9.4 TS1M6 | ||
| Windows 7 Enterprise x64 | 13.1 | 15.1 | 9.4 TS1M1 | 9.4 TS1M6 | ||
| Windows 7 Home Premium 32 bit | 13.1 | 15.1 | 9.4 TS1M1 | 9.4 TS1M6 | ||
| Windows 7 Home Premium x64 | 13.1 | 15.1 | 9.4 TS1M1 | 9.4 TS1M6 | ||
| Windows 7 Professional 32 bit | 13.1 | 15.1 | 9.4 TS1M1 | 9.4 TS1M6 | ||
| Windows 7 Professional x64 | 13.1 | 15.1 | 9.4 TS1M1 | 9.4 TS1M6 | ||
| Windows 7 Ultimate 32 bit | 13.1 | 15.1 | 9.4 TS1M1 | 9.4 TS1M6 | ||
| Windows 7 Ultimate x64 | 13.1 | 15.1 | 9.4 TS1M1 | 9.4 TS1M6 | ||
| 64-bit Enabled AIX | 13.1 | 15.1 | 9.4 TS1M1 | 9.4 TS1M6 | ||
| 64-bit Enabled Solaris | 13.1 | 15.1 | 9.4 TS1M1 | 9.4 TS1M6 | ||
| HP-UX IPF | 13.1 | 15.1 | 9.4 TS1M1 | 9.4 TS1M6 | ||
| Linux for x64 | 13.1 | 15.1 | 9.4 TS1M1 | 9.4 TS1M6 | ||
| Solaris for x64 | 13.1 | 15.1 | 9.4 TS1M1 | 9.4 TS1M6 | ||
| Type: | Problem Note |
| Priority: | high |
| Topic: | Analytics ==> Econometrics SAS Reference ==> Procedures ==> QLIM |
| Date Modified: | 2021-11-02 10:20:29 |
| Date Created: | 2021-05-17 16:28:54 |