search

 Basic Lessons from Financial Theory

0 comments

file time: 2008-02-16

filetype:pptx

Click Here To Download...

>  

Risk Management and Value at Risk 

Lecture 20 

Phil Davies

 

The Process of Risk Management 

What are the risks?

How should the risks be measured?

What are the exposures to the risks?

What should be done about the risks identified?

Nothing

Create capital cushion

Hedge

Insurance

Diversify

Limit

 

Risk is Multidimensional 

Portfolio

Concentration  
Risk 

Transaction Risk 

Counterparty

Risk 

Issuer Risk 

Trading Risk 

Gap Risk 

Equity Risk 

Interest Rate Risk 

Currency Risk 

Commodity Risk 

Financial

Risks 

Operational

Risk 

Reputational

Risk 

Business and

strategic risks 

Market Risk 

Credit Risk 

00pecific 
Risk00/b> 

General 
Market

Risk 

Issue Risk 

* For more details, see Chapter-1, 00isk Management00by Crouhy, Galai and Mark

 

Other Types of Risk 

Operational

Hazard

Physical

Strategic

Capital / resource allocation

Industry / competitors

Technological

Databases

Security

Confidential information

Stakeholder 

Legal

Compliance

Regulatory

Human capital

Retention

Training

Reputational

 

Risk Interactions 

We should not look at risks in isolation.

Focusing on market risk may lead to exposure to credit risk or liquidity risk. 

We should try to understand the correlations between risks. 

000the present practice of modeling market risk separately from credit risk, a simplification made for expediency, is certainly questionable in times of extraordinary market stress. Under extreme conditions, discontinuous jumps in market valuations raise the specter of insolvency, and market risk becomes indistinct from credit risk.00/font>  

Speech by Alan Greenspan at the 36th annual conference on bank structure and competition of the Federal Reserve Bank of Chicago, Chicago, Illinois, May 4, 2000.

 

How Do We Measure Risk? 

What risk measure should we be concerned about? 

Systematic risk?

 Volatility?  

In most cases, investment managers get hurt by negative returns  00downside risk.

Large losses can lead investors to shift funds to other managers.

Investment managers who perform poorly get fired.

Hedge funds that perform poorly are liquidated. 

Since we care about downside risk, we must control that risk.

We need a measure of downside risk.

 

Value at Risk (VaR) 

In early 1990s, JPMorgan created a 00orst case scenario00measure for their trading portfolio called Value-at-Risk (VaR). 

VaR measures the dollar loss in value that will be exceeded with a given probability, p, over the next n days.  

For financial institutions n is typically 1 day and p is either 5% or 1%. 

VaR summarizes exposure to downside risk under typical market conditions.

 

Calculating VaR 

Parametric approach:

Choose a distribution, estimate its parameters.

Compute VaR analytically  

Historical Approach:

Let the data tell us the distribution.

The historical approach uses past returns to estimate VaR.

No distributional assumptions are necessary

Concern: It only works if past is a good predictor of future. 

Simulation Approach:

Assume returns have same distribution as in the past.

Compute VaR through simulation using past data.

Monte Carlo Simulations are beyond the scope of this class

 

A random variable follows the standard normal distribution when:

It is normally distributed with an expected value of zero, and standard deviation equal to one.

The probability that a random variable following a standard normal distribution takes on a value of less than -1.65 (-2.33) is 5% (1%) 

Let average daily returns be denoted by m, and the standard deviation of daily returns by s. The VaR is:

 5% 1 day VaR =  (m 001.65s) x Portfolio Value. 

With daily VaRs, the mean is always ignored.  

VaR using the Normal Distribution

 

An Example Using the Normal Distribution 

Fidelity Magellen is an actively managed fund with a net asset value of $47.34bn.  

We want to calculate its 1 day VaR at 5% level using daily data since 2004 for any estimates that we require. 

VaR using the normal distribution

The expected return over one day for Fidelity is 0.0387%

The standard deviation of daily returns is 0.8387%.

The 1 day 5% VaR is 1.65 x 0.008387 x $47.34bn.

VaR =  $0.6551bn 

Interpretation:

There is a 5% chance the firm will lose more than $655m tomorrow.

If the distribution of returns remains constant over time Fidelity Magellen will lose more than $655m 5 days out of 100.

 

VaR Using the Historical Approach 

Use a time series of data for the asset or portfolio. 

Calculate returns. 

Sort the returns from the lowest to the highest.

Find the 5th percentile.

 If you have 100 daily returns, it will be observation 5. 

VaR is calculated by multiplying the 5th percentile return by the portfolio/asset value.

 

The Historical Approach in Excel 

In Excel we do not need to sort the returns data to find the 5th percentile. The PERCENTILE function will calculate the 5th percentile directly:

=PERCENTILE(H3:H982,0.05) 

Given the 5th percentile value we can calculate the historical 5% 1 day VaR:

= -0.01362 x $47.34bn

= $0.64477bn

 

Comparing The Two Approaches  

The parametric 5% 1 day VaR is $655m, while the historic 5% 1 day VaR is $645m.

Both approaches provide similar answers.

       This is not always the case.  

Large differences between parametric and historic VaRs can arise if asset returns are not normally distributed.

 - Some asset distributions have fat tails (Excess Kurtosis) and may be skewed to the left or right. 

How do we deal with derivatives?

The return of derivatives are not generally normal. The payoffs are also non-linear. 

Using historical data assumes that the past is a good predictor of the future.

 

The Portfolio Problem 

Suppose now we have many different assets with normally distributed returns. Portfolio variance is: 
 

To estimate portfolio variance, we need estimates of the variance-covariance matrix for all portfolio assets.  

In 1994, JP Morgan made a method to estimate VaR, called Riskmetrics漏, publicly available through a risk management consulting company.

For information, see Riskmetrics.com 

Riskmetrics provides data on volatilities and correlation coefficients for more than 750,000 assets.

 

Does Anyone Really Use VaR? 

Goldman Sachs 10K report for 2006 (Available on Class Website)

 00e use historical data to estimate our VaR and, to better reflect current asset volatilities, we generally weight historical data to give greater importance to more recent observations.00/font> 

The table below is the 1 day 5% VaR for Goldman Sachs for the last 3 years: 
 
 
 
 
 
 
 

Over the past 3 years Goldman has increased its risk levels by 50%. Equity risks have more than doubled in the past year. 

The risk management section of the 10K statement is pages 88 00104. Goldman Sachs provide a good discussion of the issues and problems involved in assessing the risk of their firm. 

Risk Categories   VaR ($m)   2006 2005 2004 Interest rates 49 37 36 Equity prices 72 34 32 Currency rates 21 17 20 Commodity prices 30 26 20 Diversification effect -71 -44 -41 Total 101 70 67  

Goldman Sachs001 Day VaR 

How good is the VaR measure? 

As part of Goldman Sachs00overall risk control process, daily trading net revenues are compared with VaR calculated as of the end of the prior business day. Trading losses incurred on a single day exceeded their 5% one-day VaR on three occasions during 2006.

 

Problems with Goldman00 VaR 

Goldman Sachs highlight several important issues related to VaR estimation:

Shortfalls on a single day can exceed reported VaR by significant amounts.  

 Shortfalls can also accumulate over a longer time horizon such as a number of consecutive trading days. 

Given its reliance on historical data, VaR is most effective in estimating risk exposures in markets in which there are no sudden fundamental changes or shifts in market conditions.

 An inherent limitation of VaR is that the distribution of past changes in market risk factors may not produce accurate predictions of future market risk.  

 Different VaR methodologies and distributional assumptions could produce a materially different VaR.

VaR calculated for a one-day time horizon does not fully capture the market risk of positions that cannot be liquidated or offset with hedges within one day.

 

VaR in Reality 

When you choose a risk measure, you are likely to decrease the risks you measure but increase the risks you do not measure! 

Risk managers are concerned about black holes

Large losses that risk measures do not pick up.

VaR creates its own black holes 00low probability payoffs with huge losses do not affect VaR. 

For example: The recent events in the credit markets that have greatly affected US Banks involved 23 standard deviation movements in returns. Such an event should only occur once every 1000 years or so. VaR would not capture this type of risk. 

When 10Ks for US Banks are published for 2007 it will be very interesting to see how their VaR risk measures performed. 

 

1

 

2

 

3

 

4

 

5

 

6

 

7

 

8

 

9

 

10

 

11

 

12

 

13

 

14

 

15

 

16

 

17

 

18

   download Basic Lessons from Financial Theory

Responses to Basic Lessons from Financial Theory

It's no comment...

 

Your Name:
Your Email:
Your Talk: