［2008］Topic 60: Model Risk相关习题 corporate, identify, discuss, expect, least

nayizhenfeng

AIM 1: Define model risk and identify and discuss sources of model risk.

1、One would generally expect model risk to be least for a model that prices:

A) put options.
 
B) callable corporate bonds.
 
C) Treasury bonds.
 
D) barrier options.

nayizhenfeng

The correct answer is C

Less model risk is expected when pricing less complex securities.

nayizhenfeng

2、Model risk is least affected by which of the following:

A) the predictive ability of the model.
 
B) the inclusion of all critical inputs.
 
C) the accurate estimation of model parameters.
 
D) the accuracy of the assumptions underlying the model.

nayizhenfeng

The correct answer is D

In many cases, accurate model assumptions are not critical.

nayizhenfeng

3、Which of the following sources of model risk suggest a poor match of the model to the risk?

Incorrect model specification.
Incorrect model application.
Implementation risk.
Programming errors.
A) I and III only.
 
B) I and II only.
 
C) I, II, and III only.
 
D) II and IV only.
 

nayizhenfeng

The correct answer is B

Incorrect model specification is driven by:

misspecification of the underlying stochastic process,
omission of important risk factors, or
the relationship between variables is misspecified.

nayizhenfeng

4、Which of the following sources of model risk results from making a mistake in implementing the model?

Implementation risk.
Calibration error.
Programming errors.
Data problems.
A) I, II, and III only.
 
B) II and III only.
 
C) I, III, and IV only.
 
D) III and IV only.

nayizhenfeng

The correct answer is B

The correct answer is BCalibration and programming errors are mistakes. Implementation risk and Data problems may represent mistakes, but also may represent a choice between several acceptable alternatives.

nayizhenfeng

5、Which of the following is NOT a source of model risk?

A) Widespread implementation of models across business units.
 
B) Minimal rounding errors from algorithms.
 
C) Failure to recalibrate models.
 
D) Non-synchronous data.

nayizhenfeng

The correct answer is B

Minimal rounding errors from algorithms is an attractive feature of a good model.