This paper presents a dynamic multi-factor model using macroeconomic and latent risk factors for loan portfolio credit loss distribution. The latent risk factors are grouped based on three levels, global risk factor across the whole commercial banking system, parent-sectoral risk factors for the intermediate aggregate level and individual sectoral risk factors. For the purpose of estimation, the dynamic multi-factor model is presented as a state space model. For U.S. banking system, we nd that the global and sector-wide frailty risk factors and their spillovers aects the variation of loan defaults in addition to the real GDP growth rate. The credit loss distribution of loan portfolio over a specic risk horizon is generated by Monte Carlo simulation given on the estimated parameters and risk factors and the economic capital is measured using a quantile estimator. Given economic capital estimate, we show the risk contribution of individual sectors and risk factors by combining the Hoeding decomposition with Euler capital allocation rule. The total credit loss is represented as the additive sum of marginal losses due to loan sectors, risk factors and all possible interaction terms.