R11: CME Forecast Returns Flashcards
What are the three ways to forecast fixed income returns?
- Equilibrium model
- DCF (usually need spot and interest vol but use YTM instead, assuming CFs fully reinvested on time at YTM)
- Building block (Rf+liquidity prem+credit prem+term prem)
What are the three ways to forecast equity returns?
- Grinold-Kroner: income, growth, repricing. Assumes unlimited time horizon. Expected return.
- Singer & Terhaar. Expected risk premium
- Historical stats
Issues with EMs?
Economic issues
Political and legal issues
Weak corporate governance
Weak accounting
How can you estimate expected returns for real estate? What’s a cap rate?
Cap rate + NOI growth rate - %change cap rate
Risk premium: term premium (dur)+credit premium (cost of tenant debt)+ equity premium (changes in property value)+ liquidity premium
Can use Singer Terhaar but need to consider impact of appraisal on vol, illiquidity premium and local factors
REITs vs private
Sightly higher returns and lower vol. In SR like stocks, in LR like direct
What are the theories that can explain changes in exchange rates?
CURRENT ACCOUNT
- Net trade flows
- PPP
- Competitiveness and sustainability
CAPITAL ACCOUNT (know formula)
- Domestic currency will overshoot and depreciate when Rd > Rf
- Uncovered interest rate parity and hot money
- Portfolio balance
PPP: pros and cons
Pros:
Reasonable predictor over LT if the inflation differential is large and caused mainly by monetary condition
Cons:
Assumes goods tradeable and ignores impact of capital flows on FX rates
Works poorly in ST, more appropriate for LT estimates
When is the current account deficit most likely to be sustainable?
When it’s caused by high investment rather than high domestic spending leading to low domestic saving
What are the negative impacts of hot money?
Limits MP effectiveness, will need sterilisation
ST financing for LT investments increases financial market risk
Exchange rate overshoot, business disruption
What are the two main approaches to estimate a constant variance-covariance matrix? What are the pros and cons for each?
- Sample statistics from historical returns
Sample size needs to be substantially larger than number of assets or some asset classes may look riskless. Also big enough to reduce sampling error. Consistent and unbiased. - Multi-factor models
Lowers sampling error but estimates biased and inconsistent due to inevitable mispecification of model as as sample size increases, doesn’t converge to true matrix.
Can combine through shrinkage estimator
How do you estimate volatility from smoothed returns?
Unsmooth model: Rt=(1-y)rt+yRt-1
True variance=(1+y)/(1-y) × varianceR
How do you estimate time-varying volatility?
Exhibits vol clustering. ARCH. Variance in t based on prior period’s variance plus shock.
Debt to current account, external debt to GDP, fiscal deficit to GDP, debt to GDP. At what level are they concerning?
- More than 200%. Less than 100% ideal
- More than 50%
- More than 4%
- More than 70%