• financial ratios
• expected default frequency
• probit models
• naive model
• merton's model
• option pricing
• contingent claims analysis
• credit risk

Probit models, which adopt both accounting and market information, have
become one of the most suitable statistical method in forecasting a firm's default.
Comparing Merton's Structural model with the Naive alternative approach
suggested by Bharath and Schumway (2008), my work shows that the Naive
predictor, in spite of its simple formulation, has a strong forecasting power.
The latter retains Merton's functional form but avoids the most criticised
aspect of Merton's model: the simultaneous calculation of unobservable variables.
Using a dataset of publicly traded Italian firms between 1990 and
2011, a probit estimation shows that Merton's distance to default is not a
sufficient statistic to predict a firm's financial distress. Data analysis underlines
the similarity in distribution of the main components of Merton's and
Naive models and the strong correlation between them. However, Merton's
functional form and market-based approach remain relevant in credit-risk
models.