In a typical say 1 week bond repo deal , as i understand , typically X units of bonds are exchanged for P.X cash at spot date (P=dirty price of bond) , and in 1 week , X units of bonds are returned, and cash of P.(1 + repo*T).X is paid.
BUT , sometimes , a haircut / "overcollateralisation" is done , and so actually at inception 1.05X units are exchanged for P.X cash , and at maturity 1.05Xunits are rturned for P.(1+repo*T).1.05X (like in Lehmans repo105 deal)
What i dont understand here is a) why is it always that it's OVERcollateralisation thats done - and its never UNDercollateralisation? after all , in the repo deal , it can be thought of as an exchange of 2 loans, 1 party borrowing cash and posting bond collateral and the other party borrowing bonds and posting cash collateral, and which ever party is riskier should put up more collateral, so if the party that borrows bonds is riskier, then the deal should be UNDercollateralised - ie P.X cash should be exchanged for 0.95 X units of bonds.
b) the repo rate must depend on the degree of overcollateralisation , ie on how off market the spot deal is , since for the whole repo deal to be fair , the forward deal must be sufficiently out of the-money to compensate for how much the spot deal is in the money. so , when repo rates are reported , why is the degree of collateralisation not indicated? perhaps the answer is that everyone knows the degree of collateralisation, and that it does not change much between dealers , and doesnt have much impact on the rate ?