Problem 5 - A Course in Financial Calculus

Question

Recall that we define two-dimensional Brownian motion, \(\left\\{X_{t}\right\\}_{t \geq 0}\), by \(X_{t}=\) \(\left(W_{t}^{1}, W_{t}^{2}\right)\), where \(\left\\{W_{t}^{1}\right\\}_{t \geq 0}\) and \(\left\\{W_{t}^{2}\right\\}_{t \geq 0}\) are independent (one-dimensional) standard Brownian motions. Find the Kolmogorov backward equation for \(\left\\{X_{t}\right\\}_{t \geq 0}\). Repeat your calculation if \(\left\\{W_{t}^{1}\right\\}_{t \geq 0}\) and \(\left\\{W_{t}^{2}\right\\}_{t \geq 0}\) are replaced by correlated Brownian motions, \(\left\\{\tilde{W}_{t}^{1}\right\\}_{t \geq 0}\) and \(\left\\{\tilde{W}_{t}^{1}\right\\}_{t \geq 0}\) with \(\mathbb{E}\left[d \tilde{W}_{t}^{\overline{1}} d \tilde{W}_{t}^{2}\right]=\rho d t\) for some \(-1<\rho<1\)

Step-by-Step Solution

Answer

The Kolmogorov backward equation for independent Brownian motions is \( \frac{\partial u}{\partial t} = \frac{1}{2} (\frac{\partial^2 u}{\partial x^2} + \frac{\partial^2 u}{\partial y^2}) \). For correlated Brownian motions, it is \( \frac{\partial u}{\partial t} = \frac{1}{2} (\frac{\partial^2 u}{\partial x^2} + \frac{\partial^2 u}{\partial y^2} + 2\rho \frac{\partial^2 u}{\partial x \partial y}) \).

Step 1: Understand the Problem Statement

We are given a two-dimensional Brownian motion \( X_t = (W_t^1, W_t^2) \) where \( W_t^1 \) and \( W_t^2 \) are independent standard Brownian motions. We need to derive the Kolmogorov backward equation for this setup, and then re-derive it for correlated Brownian motions.

Step 2: Setup for Independent Brownian Motions

The Kolmogorov backward equation for a stochastic process with Brownian motions involves partial derivatives. If \( u(x, y, t) \) represents the expected value of some function of the state, \( f(X_t) \), then the backward equation for independent Brownian motions involves finding gradient and Laplacian terms.

Step 3: Apply the Kolmogorov Backward Equation

For the independent case, the Kolmogorov backward equation is:\[ \frac{\partial u}{\partial t} = \frac{1}{2} \left( \frac{\partial^2 u}{\partial x^2} + \frac{\partial^2 u}{\partial y^2} \right) \]This equation originates from the notion that the future value depends only on the current value and the diffusion coefficients.

Step 4: Modify for Correlated Brownian Motions

When \( W_t^1 \) and \( W_t^2 \) are replaced by correlated Brownian motions \( \tilde{W}_t^1 \) and \( \tilde{W}_t^2 \) with correlation \( \rho \), the stochastic differential equation is modified. The backward equation then includes a cross-derivative term due to correlation:

Step 5: Write the Backward Equation with Correlation

For correlated Brownian motions, the Kolmogorov backward equation becomes:\[ \frac{\partial u}{\partial t} = \frac{1}{2} \left( \frac{\partial^2 u}{\partial x^2} + \frac{\partial^2 u}{\partial y^2} + 2\rho \frac{\partial^2 u}{\partial x \partial y} \right) \]The term \( 2 \rho \frac{\partial^2 u}{\partial x \partial y} \) accounts for the correlation between the two processes.

Key Concepts

Two-dimensional Brownian motion

In the fascinating world of stochastic processes, two-dimensional Brownian motion is a fundamental concept. It's essentially a generalization of the one-dimensional Brownian motion into two dimensions. This motion can be described by the process \( X_t = (W_t^1, W_t^2) \), where \( W_t^1 \) and \( W_t^2 \) are independent standard Brownian motions.
Each component, \( W_t^1 \) and \( W_t^2 \), represents its own independent random path along the time axis. The two-dimensional aspect comes from combining these two independently fluctuating paths, resulting in a motion that can be visualized as a point wandering randomly on a plane.

A few key characteristics of two-dimensional Brownian motion include:

• Independence: Each component is a one-dimensional Brownian motion that evolves independently of the other.
• Gaussian distribution: At any time \( t \), each component follows a normal distribution with mean zero.
• Continuous paths: The paths are continuous, though they are nowhere differentiable, making the process quite unpredictable.

Understanding these characteristics helps us grasp the behavior of complex stochastic systems in higher dimensions.

Correlated Brownian motions

When two Brownian motions aren't completely independent, but rather exhibit some form of correlation, we deal with correlated Brownian motions.
Imagine two random processes \( \tilde{W}_t^1 \) and \( \tilde{W}_t^2 \) that move in such a way that their changes are somehow linked. This linkage is mathematically expressed through a correlation coefficient \( \rho \), with the constraint \(-1 < \rho < 1\).

The correlation coefficient \( \rho \) indicates how strongly and in what manner the two motions are related:

• \( \rho = 0 \) means the motions are independent.
• \( \rho > 0 \) implies the motions tend to move in the same direction.
• \( \rho < 0 \) shows a tendency to move in opposite directions.

With correlated Brownian motions, the Kolmogorov backward equation must be adjusted to include interaction between the two processes. This results in the inclusion of a mixed partial derivative term, indicating how changes in one variable affect the other. Understanding these interactions in correlated systems is pivotal, especially when modeling phenomena where dependencies between variables exist, such as in financial markets.

Stochastic processes

Stochastic processes are mathematical objects used to model systems that evolve randomly over time. These processes are crucial for understanding and predicting a wide array of phenomena in fields like physics, finance, and biology.
At their core, stochastic processes incorporate elements of randomness and uncertainty, allowing them to model real-world unpredictability.

Key elements of a stochastic process include:

• Random variables: These are indexed by time, providing snapshots of the process at different moments.
• State space: The set of all possible values that the process can take.
• Path or trajectory: A realization of the process over time, showing its evolution.

In the context of our previous discussions, Brownian motion represents a classic example of a stochastic process, characterized by its erratic paths and Gaussian distribution.
The Kolmogorov backward equation, related to these processes, provides a powerful tool for understanding the evolution of probabilities over time.

By understanding stochastic processes, we can delve into models that depict everything from stock prices to the diffusion of particles, offering insight into systems influenced by random factors.