Abstract
Norges Bank has since 1989 been using a risk index for banks. The purpose of this risk index is to identify potential problem banks, and to obtain a general picture of the health of the banking industry. In 1994 the risk index was reconstructed based on research by Sigbjørn Atle Berg and Barbro Hexeberg. Using the Norwegian bank crisis 1988-1993 as their estimation period they concluded that it would be sufficient to include four indicators in the risk index. The risk index comprising these four indicators has been left unchanged since 1994, while the banking sector has experienced substantial structural changes. Thus, the need to re-estimate the risk index is clearly present. In this paper a logit model is estimated based on observations from the period 2000-2005. In competition with 23 new indicators, none of the four indicators from the current risk index are included in the recommended risk index. This underlines the need to re estimate such a risk index at regular intervals. In order to ensure that the new risk index has good properties during a deeper bank crisis than the one experienced after 2000, the predicting properties of the recommended indicators are also tested on eleven failed banks from the period 1990-93. The new risk index gives strong and early signals well in advance before the crisis culminates in all of the eleven banks. The risk index includes the following six indicators:

1. The capital adequacy ratio
2. Ratio of Residential mortgages to Gross lending
3. An expected loss measure
4. A concentration risk measure
5. The return on assets
6. Norges Bank’s liquidity indicator