R16 CME Flashcards
R16
How CME fits into investment process
- Step 1 - Construct IPS
- Step 2 - Formulate CME - Macro (inflation, interest rates, asset classes etc) and Micro expectations (concerning individual assets)
- Step 3 - Asset Allocations (SAA)
R16
7 Step Framework for CME
- Specify the final set of expectations that are needed, including the time horizon to which they apply.
- Research the historical record - what are drivers of past performance?
- Specify the method(s) and/or model(s) that will be used and their information requirements.
- Determine the best sources for information needs.
- Interpret the current investment environment using the selected data and methods, applying experience and judgment.
- Provide the set of expectations that are needed, documenting conclusions (answers step 1)
- Monitor actual outcomes and compare them to expectations, providing feedback to improve the expectations-setting process
R16
Problems of forecasting - 1
Limitations of Economic Data
- Limitations of Economic Data:
- Time lag between collection of data and when it is distributed
- Data can be revised - but revisions are not made at the time they are distributed
- Data definitions and methodologies change over time.
- Data indices are rebased over time - meaning that the specific time period used as the base of the index is changed
R16
Problems of forecasting - 2
Data Measurement Errors and Biases
- Data Measurement Errors and Biases:
- Transcription errors. These are errors in gathering and recording data. Such errors are most serious if they reflect a bias.
- Survivorship bias.
- Appraisal (smoothed) data - Appraised values tend to be less volatile than market-determined values for the identical asset would be. The consequences are 1) the calculated correlations with other assets tend to be smaller in absolute value than the true correlations, and 2) the true standard deviation of the asset is biased downward
R16
Problems of forecasting - 3
The Limitations of Historical Estimates
- The Limitations of Historical Estimates:
- Regime changes - Changes in the technological, political, legal, and regulatory environments, as well as disruptions such as wars and other calamities, can alter risk–return relationships. Gives rise to the statistical problem of nonstationarity (meaning, informally, that different parts of a data series reflect different underlying statistical properties)
- A long data series may be a statistical necessity BUT :
The risk that the data cover multiple regimes increases.
Time series of the required length may not be available.
In order to get data series of the required length, the temptation is to use high-frequency data (weekly or even daily). Data of high frequency are more sensitive to asynchronism across variables. As a result, high-frequency data tend to produce lower correlation estimates.
R16
Problems of forecasting - 4
Ex Post Risk Can Be a Biased Measure of Ex Ante Risk
- Ex Post Risk Can Be a Biased Measure of Ex Ante Risk
Understates risk and overstate returns
R16
Problems of forecasting - 5
Biases in Analysts’ Methods
- Biases in Analysts’ Methods:
- Data-mining bias - introduced by repeatedly “drilling” or searching a dataset until the analyst finds some statistically significant pattern. Such patterns cannot be expected to be of predictive value
- Time-period bias - Research findings are often found to be sensitive to the selection of starting and/or ending dates
R16
Problems of forecasting - 6
The Failure to Account for Conditioning Information
- The Failure to Account for Conditioning Information:
- Prospective returns and risk for an asset as of today are conditional on the specific characteristics of the current marketplace and prospects looking forward
- We should not ignore any relevant information or analysis in formulating expectations. Indeed, the use of unconditional expectations can lead to misperceptions of risk, return, and risk-adjusted return.
R16
Problems of forecasting - 7
Misinterpretation of Correlations
- Misinterpretation of Correlations
- Assumes a linear relationship between variables
- Without the investigation and modeling of underlying linkages, relationships of correlation cannot be used in a predictive model
R16
Problems of forecasting - 8
Psychological Traps (6)
[Applies to forecasters]
RASPOC
Psychological Traps
- Anchoring trap - The tendency of the mind to give disproportionate weight to the first information it receives on a topic. The analyst can try to address this trap by consciously attempting to avoid premature conclusions.
- Status quo trap - The tendency for forecasts to perpetuate recent observations—that is, to predict no change from the recent past. The status quo trap may be overcome with rational analysis used within a decision-making process.
- Confirming evidence trap - The bias that leads individuals to give greater weight to information that supports an existing or preferred point of view than to evidence that contradicts it.
- Overconfidence trap - The tendency of individuals to overestimate the accuracy of their forecasts. To overcome: Examine all evidence with equal rigor. Enlist an independent-minded person to argue against your preferred conclusion or decision. Be honest about your motives. To prevent this trap from undermining the forecasting endeavor is to widen the range of possibilities around the primary target forecast.
- Prudence trap - The tendency to temper forecasts so that they do not appear extreme; the tendency to be overly cautious in forecasting. To avoid the prudence trap, an analyst is again wise to widen the range of possibilities around the target forecast
- Recallability trap - The tendency of forecasts to be overly influenced by events that have left a strong impression on a person’s memory. To minimize the distortions of the recallability trap, analysts should ground their conclusions on objective data and procedures rather than on personal emotions and memories.
R16
Problems of forecasting - 9
Model Uncertainty
- Model Uncertainty:
- Model uncertainty - Uncertainty concerning whether a selected model is correct.
- Input uncertainty - Uncertainty concerning whether the inputs are correct.
- Input uncertainty and model uncertainty in particular often make it hard to confirm the existence of capital market anomalies (inefficiencies)
R16
Tools for formulating CME (5)
- Statistical Tools
- DCF Models
- Risk Premium Model
- Financial Equilibrium Models
- Survey and Panel Methods
R16
Statistical Tools for CME
- Use arthmetic average for single periods, but geometric average for mutliple periods
- Shrinkage Estimator - Estimation that involves taking a weighted average of a historical estimate of a parameter and some other parameter estimate, where the weights reflect the analyst’s relative belief in the estimates. Uses a target covariance matrix - A component of shrinkage estimation; allows the analyst to model factors that are believed to influence the data over periods longer than observed in the historical sample. Used when data set is very small. Reduces the incidence of historical outliers. Can be used for both covariance and mean returns forecasting.
- Time series model - forecasting a variable on the basis of lagged values of the variable being forecast and often lagged values of other selected variables. Useful in developing particularly short-term forecasts for financial and economic variables. Notably applied to estimating near-term volatility
- Volatility Clustering - will high or low volaility persist in short term.
- Multifactor models - When the factors are well chosen, a multifactor model approach may filter out noise. This structure has been found useful for modeling asset returns and covariances among asset returns
R16
DCF Models for CME
- Gordon Growth model
- Have the advantage of being forward-looking.
- They do not address short-run factors such as current supply-and-demand conditions, so practitioners view them as more appropriate for setting long-term rather than short-term expectations
- DCF models can be used for Fixed Income and equity but for fixed income we should use YTM. The YTM calculation makes the strong assumption that as interest payments are received, they can be reinvested at an interest rate that always equals the YTM
R16
DCF Models for CME - Gronold-Kroner
E(Re ) ≈ D/P −(ΔS) + i +g + ΔP/E
- Expected income return: D/P − ΔS.
- Expected dividend yield D/P
- Repurchase yield: − ΔS.
- Expected nominal earnings growth return: i + g.
- Expected repricing return: ΔP/E
- Compounded Annual growth rate or expected rate of return on equity: E(Re)