10-1 capital market expectations:
framework and macro considerations
1 introduction
2 framework and challenges
a framework for developing
capital market expectations
challenges in forecasting
fundamental law fo investing: the uncertainty of the future
capital market expectations CME:
expectations concerning the risk and return prospects of asset classes
an essential input to formulating a strategic asset allocations
asset allocation - the primary determinant fo long-run portfolio performance
to ensure internal consistency arcoss asset classes (cross-sectional consistency)
and over various time horizons (intertemporal consistency)
→poor risk-return characteristics over any horizon
→distort the connection between decisons and horizon
disciplined
approach
1 specify the set of expectations needed including time horizon apply
2 research the historical record
3 specify the mehods and / or models used and infor requirements
4 determine the best sources for infor needs
5 interpret the current investment enviroment
6 provide the set of expectatiosn needed, documenting the conclusions
7 monitor actual outcomes and compare with expectiontations, providing feedback to improve the expectations-settign process
good forecasts
unbiased objective and well researched
efficient - min the size of forecast errors
internally consistent both cross-sectionally and intertemporally
limitations of econmic data
data measurement errors and biases
the time lag
one or more official revisions to initial data values
definitions and calculation methods change too
periodically re-base
transcription errors
survivorship bias
appraisal (smoothed) data
→measured volatilities biased downward
and correlations with other assets tend to be understated
the limitatiosn of historical estimates
changes in regime → nonstationarity - different parts of a data series reflect different underlying statistical properties
→should use only that part fo the time series that appears relevant to the present
should use the longest data hisoory that reasonable assuance of stationarity
higher-frequency data - improve the precision of the sample variances, covariances and correlations
not improve the precision of the sample mean
many variables are considered - large num of observations may be statistical mecessity
as frequency increases, the likelihood increases that data may be asynchronous across variables
→distorts correlations and induces lead-lag relationships
normally distributed
→skewness and fat tails
ex post risk can be a biased mesasure of ex ante risk
peso problem: looking backward, likely to underestimate ex ante risk and overestimate ex ante anticipated returns
biases in analysts' methods
data-mining bias: such patterns cannot be expected to have predictive value
time-period bias: relates to results that are period specific
to avoid: able to provide an econimic rationale / to examine the forecasting relationship out of sample
the failure to account for conditioning infor
unconditional forecasts: dilute infor by averaging over environments, lead to misperception of prospective risk and return
misinterpretation of correlations
negligible measured correlation may reflect a strong but nonlinear relationship
psychological biases
anchoring bias
status quo bias
confirmation bias
overconfidence bias
prudence bias
abailability bias
model uncertainty: whether structurally and / or conceptually correct
- parameter uncertainty:parameters invariably estimated with error
- input uncertainty: whether the inputs are correct