Credit models can be broadly separated into comparative pricing models (also known as arbitrage-free models) and econometric models (also known as equilibrium models). In the context of comparative pricing, one tries to derive the price of a new financial instrument from existing instruments in the market. The derivation typically assumes that the credit referenced can be freely traded with little friction (bid/ask spread), high liquidity (daily hedges), and the availability of short positions. Econometric models try to predict default rates based on historical information on default rates for different economic environments and current economic information. Most models proposed in the literature use comparative pricing methodologies (sometime called arbitrage-free pricing). However, long-term buy and hold investors use econometric models.
Econometric models require significant data and sophisticated techniques to analyze the data. The models are typically used to produce buy/sell options and their output may not match observed market prices. However, they provide a useful tool for making decisions on the relative value of various bonds. The key to choosing between these two types of models is whether one is more concerned about estimating intrinsic value (equilibrium models) or value relative to current market prices (arbitrage-free models).
Another important question when implementing a model of credit risk is the technique to be used in the determination of actual prices. The common approaches are closed form (formulaic) solutions, tree frameworks, and Monte Carlo simulations. Closed form solutions are convenient to use and provide quick intuition on important variables, but usually are too simple or too inflexible to be practical in valuing complex securities. Tree frameworks provide more flexibility and are computationally feasible if the problem can be solved with a recombining binomial or trinomial tree. Finally, Monte Carlo simulation provides the most flexibility and is useful for solving non-Markov models (i.e. those that are path-dependent and require a non-recombining tree). However, Monte Carlo analysis is slow and computationally intensive.
Мой блог находят по следующим фразам
0 Responses to “A Review of Valuation Models”