The Wilkie Stochastic Investment Model: Structure & Application

The Philosophy of the Wilkie Model

Before diving into the equations, it's essential to understand the motivation behind the Wilkie model. As our last lesson showed, models like GBM are elegant but can feel disconnected from the underlying economy. The Wilkie model takes the opposite approach. It is built on the premise that long-term investment variables are driven by economic fundamentals and are interconnected.

The model was specifically designed for the needs of actuaries in life insurance companies and pension funds, whose liabilities stretch decades into the future. For these purposes, short-term price fluctuations are less important than the long-term relationships between inflation, interest rates, and asset returns.

To get a sense of the model's purpose from its creator, A.D. Wilkie, let's turn to his original 1986 paper.

As you've just read, the Wilkie model is a discrete-time model (using annual steps) that prioritizes economic plausibility over mathematical tractability alone. It identifies inflation as the fundamental economic force that drives the other financial variables in the model. This leads to a "cascade" structure, which is fundamentally different from the single-asset focus of GBM.

The "Cascade" Structure

The Wilkie model is a system of interconnected time series models, which should feel familiar given your background in statistics. Instead of modeling multiple variables simultaneously in a complex vector autoregression (VAR) model, Wilkie simplified the structure into a "cascade," where one variable's output feeds into the next model down the line.

The diagram below illustrates this flow of influence.

Let's walk through this cascade, examining the key components. For each step, the model uses a time series equation (often a variant of an autoregressive AR(1) process) to link the variable to its own past and to the variables above it in the cascade. Each equation also includes a random "white noise" term, which provides the stochastic element for simulations.

1. Inflation (Consumer Price Index):
The model starts with the force of inflation, $I(t)=\ln (Q(t)/Q(t-1))$, where $Q(t)$ is the price index. This is modeled as a first-order autoregressive, or AR(1), process.

In simple terms, this means the inflation rate for a given year is expected to be a weighted average of its long-term mean (${\mu }_Q$) and the previous year's inflation rate, plus a random shock (${ϵ}_Q(t)$). This structure creates mean-reverting behaviour for inflation.

2. Share Dividend Yield:
The logarithm of the dividend yield, $\ln Y(t)$, is modeled as being dependent on the current rate of inflation, $I(t)$, plus its own autoregressive component.

where $Y_N(t)$ is another AR(1) process. The economic intuition is that high and volatile inflation can increase market uncertainty, depressing share prices and thus increasing the dividend yield (Dividend/Price).

3. Share Dividends:
The growth in share dividends, $\ln (D(t)/D(t-1))$, is modeled to depend on a lagged, smoothed version of inflation. A key feature is that the model is constructed with "unit gain," meaning a 1% permanent increase in inflation will eventually lead to a 1% increase in dividends, ensuring dividends keep pace with prices in the long run.

4. Share Prices:
Crucially, share prices are not modeled directly. This is a major departure from GBM. Instead, the share price index $P(t)$ is a derived quantity, calculated from the modeled dividend index $D(t)$ and the modeled dividend yield $Y(t)$:

This enforces an economic link between price, dividend, and yield at all times.

5. Long-Term Bond Yields (Consols Yield):
The model for the yield on long-term government bonds, $C(t)$, is based on the Fisher effect, which you may recall from economics:
Nominal Interest Rate ≈ Real Interest Rate + Expected Inflation
The Wilkie model formalizes this by modeling the bond yield as the sum of two components:

• An expected inflation component, which is an exponentially weighted moving average of past inflation.
• A real yield component, which is itself modeled as an AR(1) process, mean-reverting to a long-term average real rate (e.g., 3%).

6. Short-Term Interest Rates and Other Extensions:
The original model included these four series. Later versions, described in Wilkie's 1995 paper, extended the cascade to include models for short-term interest rates (often modeled as a mean-reverting spread to the long-term rate), property yields, and even exchange rates.

The following resource provides an excellent, concise summary of the equations and structure for each component.

Application: Simulating Economic Futures

Having established the structure, the next question is: how is it used? Unlike the Black-Scholes formula, which gives a single price, the Wilkie model is used for Monte Carlo simulation.

Starting with a set of initial economic conditions (current inflation, yields, etc.), the model steps forward year by year. In each year, a random number is generated for each of the "white noise" terms in the equations, producing a single, plausible path for the economy over, say, the next 50 years. By repeating this process thousands of times, the model generates a rich distribution of possible futures, often visualized using "fan charts".

These simulations are the primary application of the model. They provide the basis for answering critical long-term questions in actuarial science. The final section of Wilkie's original paper is dedicated to this topic.

As you saw, the applications are varied and deeply rooted in the long-term business of insurance and pensions. Here are some of the key uses:

• Solvency and Capital Assessment: An insurer or pension fund can test whether its current assets and future contribution/premium plans are sufficient to meet its long-term liabilities across thousands of simulated economic futures. This helps determine if the company holds enough capital to withstand adverse scenarios.
• Strategic Asset Allocation: It can be used to test dynamic investment strategies. For example, an actuary can program a rule ("rebalance the portfolio to 60% equities and 40% bonds each year") and see how that strategy performs across the full range of simulations.
• Product Pricing and Design: When designing a new long-term insurance policy (like an annuity or an endowment), the model can be used to simulate the investment returns that will back the policy. This helps in setting a premium that is both competitive and has a high probability of being sufficient.
• Valuing Guarantees: Many retirement and insurance products contain complex guarantees (e.g., "your investment will grow by at least 2% per year"). The Wilkie model can be used to estimate the cost and risk of these guarantees, which might only bite in poor economic scenarios.

Conclusion

Today we have explored the Wilkie model, a cornerstone of long-term actuarial financial modeling. In contrast to the continuous-time, single-asset GBM, the Wilkie model is a discrete-time, multi-asset model built on a foundation of economic relationships. It is a practical tool designed not for pricing, but for understanding long-term risk.

Key Takeaways:

• The Wilkie model is a stochastic investment model for long-term actuarial simulations, using annual time steps.
• Its structure is a "cascade" where inflation drives other economic variables like share yields, dividends, and bond yields.
• It is built on a series of interconnected time series equations (largely AR(1) processes), giving it economic interpretability.
• Share prices are not modeled directly but are derived from modeled dividends and yields.
• Its primary application is through Monte Carlo simulation to generate thousands of future economic scenarios for long-term risk management, solvency testing, and strategic asset allocation.

Having now covered both the theoretical elegance of GBM and the practical, economic focus of the Wilkie model, you have a solid understanding of two distinct approaches to modeling asset returns.

In our next lesson, we will pivot from modeling the underlying assets themselves to the world of derivatives. We will begin Module 6, "Fundamentals of Derivative Securities," by defining the basic building blocks of this world: forwards, futures, and options.