Long term asset creates more revenue, however it is riskier in a liquidity sense. Our question is: How does a liquidity constrained bank make decisions between profitability and liquidity? We present a computable DSGE model of banks optimal portfolio choices under liquidity constraints. Our theory predicts that liquidation plays an important role in a bank's portfolio model. Even though liquidation is an off-equilibrium phenomenon, banks can have rich loan portfolios due to the possibility of liquidation. Liquidity condition is a key factor in banks portfolio. In a moderate liquidity situation, a bank can lend more profitable longer term loans, however, if a shock in liquidity is expected, then the bank lends more loans in short term. According to the liquidity conditions, the bank can have medium term loans which are different from other previous literature. In addition, we extend our model to the bank's securities business where the bank's debts are largely short term deposit. Our theory predicts that the bank securities business produces a chasm between a real liquidity of economy and market liquidity. Banks can have more liquidity by selling their securitized loans, and as our model already pointed out, a good liquidity condition makes the bank have more profitable but less liquid long term loans. As a consequence, long term loans are accumulated with this securitization, simply because a long term loan gives higher revenue. Any market turbulence can invoke a problem in economy wide liquidity.