Professional Documents
Culture Documents
Murthy, Volatility
Murthy, Volatility
Murthy, Volatility
Shashidhar Murthy
Indian Institute of Management Bangalore
1
What we do & Why
• ARCH, GARCH-type econometric models for
volatility are common
2
Motivation
• We add to above with model of firm’s investment choice and offer simple,
intuitive, economic rationale. Derive endogenous ARCH-type behaviour
3
Related literature
• A firm has (1) installed capital, and (2) options to add capital in
future
– Firm value = value of (1) + value of (2)
– Returns , and variance, depend on “mix” of (1) & (2)
5
Model - Structure
• Typical firm; partial equilibrium (given SDF)
• Installed capital yields output in a “AK-like”
fashion
• Firm adds to capital over time
• Each investment “project”: fixed size or
indivisible; irreversible
• Resulting “Options to invest” induces non-linear
response of firm value to shocks
• Firm value P(t) = PV of future output from (1)
Assets-in-place & (2) Growth Options
6
Projects (indivisible & irreversible)
At each date j, a project arrives. Investment i. If take project, capital depreciates
deterministically: 𝐾𝑗 ሺ𝑡ሻ = 𝑖𝜋 𝑡−𝑗
Productivity of project j is IID. Projection on aggregate priced risk factor (i.e. SDF)
𝐴𝑗 ሺ𝑡ሻ = 𝜇𝑗 + 𝛽𝑗 ሾ1 − 𝜈 ሺ𝑡ሻሿ+ 𝜀𝑗 ሺ𝑡ሻ; 𝑡 ≥ 𝑗 + 1; {𝜀𝑗 ሺ𝑡ሻ} IID
𝑚 (𝑡+𝑠) 𝑠
Stochastic discount factor: = 𝑅−1 ς 𝑢=1 𝜈 ሺ𝑡 + 𝑢ሻ; {𝜈ሺ𝑡ሻ} IID; 𝐸 ሾ𝜈 ሺ𝑡ሻሿ= 1
𝑚 (𝑡)
7
Value & Returns of each Project
Value of output at t+1:
𝑚(𝑡 + 1)
𝐸𝑡 𝐴𝑗 ሺ𝑡 + 1ሻ𝐾𝑗 ሺ𝑡 + 1ሻ൨
𝑚(𝑡)
= 𝑅 −1 𝐾𝑗 ሺ𝑡 + 1ሻ𝐸𝑡 ൣ𝜈 ሺ𝑡 + 1ሻ𝐴𝑗 ሺ𝑡 + 1ሻ൧
8
Options – exercise & value
Optimal investment decision: take project i.f.f. 𝑉𝑗 ሺ𝑗 ሻ > 𝑖
Value of Asset-In-Place:
𝑃𝑎 ሺ𝑡 + 1ሻ = σ 𝑡𝑗 =0 𝜉 ሺ𝑗ሻ𝑉𝑗 ሺ𝑡 + 1ሻ + 𝜉 ሺ𝑡 + 1ሻ𝑉𝑡+1 ሺ𝑡 + 1ሻ
= 𝜋𝑃𝑎 ሺ𝑡ሻ + (2), given same depreciation factor 𝜋 for all projects.
(2) = 𝑉𝑡+1 ሺ𝑡 + 1ሻ ∗ 𝐼{𝑉𝑡+1 ሺ𝑡 + 1ሻ > 𝑖}; where 𝐼{. } = event { new investment}
= 𝜋𝛼𝐸[𝜆ሺ𝑡 + 1ሻ𝐼{𝜆(𝑡 + 1) > 𝑖/𝜋𝛼}] + 𝜂(𝑡 + 1), uncond mean + orthog error
12
Dynamics of Firm Variables (2)
Lemma 5: {𝑃𝑎 ሺ𝑡ሻ} and firm value ሼ𝑃ሺ𝑡ሻሽ have endogenous linear and
Markov and, in particular, AR-1 transition dynamics. Transition
probabilities converge; a stationary distribution exists with finite
moments. .
13
Dynamics – Role of Growth Options
𝑤 ሺ𝑡 ሻ
𝑤ሺ𝑡 + 1ሻ = 𝑚 + 𝜂 ሺ𝑡+1ሻ
𝜋 + ഥ𝑔 ൨𝑤 ሺ𝑡 ሻ
𝑃
14
Dynamics of Firm Return Variance
Firm risk firm varies with the relative importance of growth options:
Commonplace intuition: risk of option on an asset >= risk of asset. Suggests 𝜎𝑎𝑔 ≥ 𝜎𝑎2 natural restriction.
15
Main result
Theorem 1: Restrict the underlying exogenous shocks’ distributions such
that 𝜎𝑎𝑔 ≥ 𝜎𝑎2 . Then the firm return variance 𝜎𝑡2 is strictly increasing in
𝑤ሺ𝑡ሻ, the relative importance of growth options in firm value. Using
Lemma 6, it follows that the firm return variance process is non-IID and, in
particular, 1st-order Markov. Furthermore {𝜎𝑡2 } possesses a limiting
stationary distribution.
16
Time-series properties of return variances
17
Conclusions
• Model with simple non-linear structure and IID shocks produces
stochastic variances with time-series patterns
• Also consistent with evidence (old) that small firm returns and
(recent) firms with growth options are more volatile
• Extensions?
– When Pure Scaling assumptions are not met, 2 endogenous state
variables needed. I.e. Past and current variances = sufficient statistic?
– General equilibrium? 18