Lecture 7 Default predictions Flashcards
Why do we need to estimate default likelihoods who cares?
- Default prediction is relevant in practice in many settings
o Investors
o Banks/financial institutions
o Suppliers/customers
o Sovereigns - Perception of default likelihoods feedsb ack into the ability of a firm to acquire funding, funding costs and to conduct business
Throughout historiy what have been the ways to predict default likeliness?
- Default prediction is relevant in practice in many settings
o Investors
o Banks/financial institutions
o Suppliers/customers
o Sovereigns - Perception of default likelihoods feedsb ack into the ability of a firm to acquire funding, funding costs and to conduct business
What are the most famous credit risk models?
- Moddys credit ratings (1909)
- Altman’s z score (1968) and ohlsons O-score (1980)
- KMV model (moody’s EDF: late 1980s) is practical implementation of the merton model
- Campbelll et al (2008)
What are the main challenges in the estimation of default probabilities
- Samples are small not that many firms go default
- Data limitations (firms do not share their statements any more after bankruptcy)
- Definition of default events
- Horizon of default probability estimation
- Relevant benchmark group (your model will be calibrated on this if your benchmark are only highly profitable healthy firms it can easily pick out firms that are likely to go bankrupt, this is harder if it also includes highly levered firms that did not go banktupt.o
According to S&P what is a credit rating?
- S&P: a forward-looking opinion about the credit-worthiness of an obligor with respect to a specific financial obligation
- The opinion reflects standard and poor’s view on the obligors capacity and willingness to meet its financial commitments as they come due
what are features of credit ratings?
- Contains public information as well as information obtained from management interviews
- Rating-through-the-cycle
- Legal implications for several large institutional market participants, more oversight for example if low rating
What bonds do credit ratings usually cover?
publicly traded and large non publically traded firms
DO credit ratings change frequently?
no, usually once per quarter due to new statements
How does the Altman’s z score work?
- Uses multiple discriminant analysis (MDA) to classify firms into bankrupt and non bankrupt firms
- Two step estimation:
o Identify variables that help to classify bankrupt and non bankrupt companies among a set of potential predictors
o Determine weight (parameter of selected predictors
So a probit model for bankruptcy. And which variables are likely to lead to a bankruptcyw
what are predictors of the altmans z score for bankruptcy?
- Working capital/total asset -> liquidity
- Retained earning/total asset -> profitability
- Ebit/total asset -> operating efficiency
- Mv of equity/bv of total liabilities -> leverage
- Sales/total asssets -> asset turnover
Threshhold values of the altman z score
Z > 2.675 non distressed firm
1.81 < Z < 2.675 distressed firms
Z < 1.81 highly distressed firms
Explain the merton model
- Refinement of the binominal tree model but with continuous time
The value of a firm now can be seen as a stochastic model
dVt=μVtdt+σVtdWt
Here’s an explanation of the variables and the function:
1. dVt: This represents the infinitesimal change in the firm’s value over an infinitesimal time interval ttt.
2. Vt This is the value of the firm at time ttt.
3. μ: This is the drift term, which represents the expected rate of return of the firm’s value. It indicates the average rate at which the firm’s value grows over time.
4. σ: This is the volatility term, which represents the standard deviation of the firm’s returns. It measures the uncertainty or risk associated with the firm’s value.
5. dt: This is an infinitesimal increment of time.
6. dWtd: This is an infinitesimal increment of a Wiener process (also known as Brownian motion), which introduces a stochastic (random) component to the firm’s value. The Wiener process represents the random fluctuations in the firm’s value.
What does the merton model give as an ouput?
A probability distribution of certain outcomes,
What is th Moody’s KMV model?
- Goal: expected default frequency (EDF)
- Starts with merton model
CDF is the cumulative distribution function of the standard normal distribution that came from the merton model - INtuitino: how many standard deviations is the company away from a default
What is the campbell model?
Used a lot by acadamics not in practice:
Campbell model
- Logit regression model taking into account all years of the firms finanicals and include a dummy for when it went bankrupt
- It goes like firm A t-3 t-2 t-1 t and at t the dummy becomes 1 Firm b t-3 t-2 t-2 t dummy does not become 1 now it can take into account things that changes over time leading up to the bankruptcy