SQL Statement failing in COBOL program

I asked this in the WDSc list, but none of the responses helped me find the
problem. I have the following statement;

Exec SQL
Declare LTRC1 cursor for
with t1 as (Select distinct lnbss, lnbond
from slsfiles/sllnrep
WHERE lnlsts < "P90" )
select
Distinct ZIP_CODE,
PLUS_FOUR,
CAR_RTE_CD,
CAR_RTE_CD,
PLUS_SIX,
PAGE_INFO,
ADDR_LINE1,
ADDR_LINE2,
ADDR_LINE3,
ADDR_LINE4,
BORR_NAME,
BORR_SSN,
RTN_CODES,
LNBOND
from BCTEST/BCSRREP,
SLSFILES/SLBRREP Join T1 on BrBSS = LNBSS
WHERE Dec(BORR_SSN,9,0) = Dec(BRQDVA *100,9,0)
Order by LnBond, Zip_Code
END-EXEC

The select statement works in STRSQL and selects the records I want to
process. When I run it in the program, the open cursor statement completes
normally, but the fetch gets me a SQLSTATE of 22018 with a SQLCODE of -420.
The code book says;

"The character value for the CAST, DECIMAL, FLOAT, or INTEGER scalar
function is invalid."

Every column selected is alpha except the LNBOND which is a packed decimal 3
long in the file. I have the receiving field for the LNBOND field set as a
Signed Packed Decimal 3 long. BRQDVA in the "WHERE" clause is a number in
9,999,999.99 format, and so to compare with the account number (not really
SSN, but was at one time) I have to move decimal to the right like
999,999,999. The BORR_SSN is actually a number, just stored as character.

This statement happens to be in ILE COBOL SQL, but the question is more SQL
based then program based. Also I know that in this example, taken from the
program the constant 'P90' has double quotes, required by COBOL, and not
single as required by STRSQL.

I'm not sure what I am being told by the message.

Any thoughts/Help?

Thanks!

Jim