The following is a review of the Market Risk Measurement and Management principles designed to address the

learning objectives set forth by GARP®. Cross-reference to GARP assigned reading—Tuckman and Serrat, Chapter 9.

READING 16

THE VASICEK AND GAUSS+ MODELS

Study Session 3

EXAM FOCUS
This reading discusses the Vasicek and Gauss+ models, which are common term
structure models used to analyze interest rate dynamics. The Vasicek model is an
effective model for understanding basic interest rate behavior and assisting with
pricing simple investments. Alternatively, the Gauss+ model is a more advanced
approach that is beneficial for proprietary trading and complex investment analysis.
For the exam, understand the structure of the Gauss+ model and how it compares with
the Vasicek model. Also, be familiar with how changes in the short-term, medium-term,
and long-term interest rate factors are computed using the Gauss+ model.

MODULE 16.1: VASICEK MODEL AND GAUSS+ MODEL

Modeling Interest Rates

Having a model of interest rates is important for many applications in finance,
including risk management and the pricing of interest rate securities and derivatives.
The most useful interest rate models provide a process for how the short-term rate
evolves, which ideally explains (within an acceptable margin of error) the current term
structure of interest rates and maintains consistency with implied volatilities.
Typically, an interest rate model will demonstrate how the short-term rate changes as a
function of other variables. For example, suppose we have a model defined as follows:

In this case, the changes in the short-term rate could be driven by the annual volatility
of rate changes and a random factor that is normally distributed with a mean of zero
and a standard deviation of .
With this model, we start with an initial rate, r0, then simulate the rate in the next
period, r1, and continue until there is one complete interest rate path. This process is
then repeated until there are multiple paths, which creates a distribution for the
evolution of the short-term rate. With this information, we can compute averages and
test the ability of the model to price interest rate securities and derivatives.
In practice, we are looking for a term structure model that meets the following
objectives:
Consistent with market pricing of basic fixed-income instruments. For example, can the
model be calibrated to be consistent with observed bond prices?
Consistent with market pricing of derivative instruments. For example, can the model
be calibrated to be consistent with observed prices of interest rate derivatives?
Mirrors tendencies that market participants expect from interest rate movements. For
example, does the model consider mean reversion—and does the model reflect the
volatility structure of rates in the market?

Vasicek Model
The Vasicek model uses a single factor to model how short-term interest rates change
over time. Its main characteristic is mean reversion where the short-term rate (rt)
tends to revert to its long-term average (θ). In other words, a rise is forecasted for
below-average rates, while a decline is forecasted for above-average rates. This
behavior can be modeled with the following equation:

The Vasicek model is useful for simple applications like pricing zero-coupon bonds and
applying basic hedging techniques. The model’s simplicity allows analysts to
understand the basic effects of mean reversion and stochastic shocks on interest rates.
However, it has the following drawbacks:
It fails to replicate complex term structure shapes and real-world volatility patterns.
It fails to account for observed changes in volatility across maturities.
It fails to adequately incorporate macroeconomic trends and monetary policy shifts.
Despite these limitations, the Vasicek model is still a key starting point in
understanding interest rate dynamics, providing a benchmark for advanced models like
the Gauss+ model.

Gauss+ Model
LO 16.a: Describe the structure of the Gauss+ model and discuss the implications
of this structure for the model’s ability to replicate empirically observed interest
rate dynamics.

Compared to the Vasicek model, the Gauss+ model is a more sophisticated multifactor
framework that provides a better representation of real-world interest rate behaviors.
The model consists of three interacting components: the short-term rate (rt), a
medium-term interest rate factor (mt), and a long-term interest rate factor (lt). These
factors affect each other in a cascading manner in the following expressions:

The Gauss+ model is constructed in response to certain interest rate dynamics,

including (1) mean reversion, (2) correlation between factors, (3) consistency with the
volatility term structure, and (4) pricing consistency.
Regarding mean reversion, let’s take a closer look at the first of the three cascade
equations and what it means for rt to differ from mt:

When rt < mt, the right-hand side of the equation suggests that the rate will be
increasing as we approach the medium-term factor. When rt > mt, the rate will be
decreasing as we approach the medium-term factor. When rt = mt, this is the
equilibrium case, and there will be no change to the short-term rate. This process
results in a cascade of convergence where rt is seeking to converge to mt, which is
trying to converge to lt, which is trying to converge to μ. Note that the only fixed
parameter here is μ, while the other factors are dynamically moving targets.
There are economic interpretations to these three rate factors. The short-term rate
represents the policy interest rate controlled by central banks (e.g., the rate set by the
Federal Reserve). The medium-term factor represents influences from the broader
economy, such as business cycles and monetary policy effects. The long-term factor
represents long-term expectations for economic trends, such as inflation and
productivity growth. Accordingly, the short-term rate reacts quickly to the medium-
term factor, the medium-term factor adjusts more gradually to the long-term factor, and
the long-term factor shifts the slowest over time.
Regarding the correlation between factors, the changes in mt and lt are related through
the parameter ρ. Regarding consistency with the volatility term structure, the model can
be calibrated to offer low volatility on the short end, higher volatility in the middle, and
declining volatility on the long end of the interest rate curve, which replicates
historical norms (i.e., hump-shaped volatility). Regarding pricing consistency, the model
offers enough flexibility to be consistent with the pricing of not only simple interest
rate instruments (e.g., bonds and swaps), but also interest rate derivatives, such as
interest rate caps and floors.
Vasicek Model vs. Gauss+ Model
LO 16.b: Compare and contrast the dynamics, features, and applications of the
Vasicek model and the Gauss+ model.

The simplicity of the Vasicek model restricts its ability to model certain aspects of
interest rate dynamics. For example, it only has one volatility parameter; therefore,
volatility is only calibrated to be consistent with one point on the interest rate curve.
Also, it does not offer much flexibility on the shape of the curve, which affects the
ability to accurately match rates to the degree necessary to enable effective hedging. In
addition, it is unable to capture the observable pricing for interest rate derivatives
because of its inability to model more complex volatility structures. These deficiencies
are mitigated in the Gauss+ model.
In many ways, the Gauss+ model can be seen to be an extension of the Vasicek model.
This expansion of the Vasicek model is designed to provide more adherence to market
behaviors and pricing. The ability to price a wider range of instruments and to be
consistent with interest rate dynamics expected by market participants makes the
Gauss+ model preferable to the simpler Vasicek model.
Figure 16.1 summarizes some differences in dynamics, features, and applications
between the Vasicek and Gauss+ models.

Figure 16.1: Differences Between Vasicek and Gauss+ Models

Applying the Gauss+ Model

LO 16.c: Calculate changes in the short-term, medium-term, and long-term
interest rate factors under the Gauss+ model.

It should be noted that the factors m and l are not meant to tie back to any particular
forward rates—unlike r, which is intended to proxy a short-term rate (e.g., fed funds
[overnight] rate). Instead, these two parameters influence the evolution of the short-
term rate. Once we have a distribution of interest rate paths, we can infer what the
forward rates are based on the distributions at each future date.
To demonstrate how the Gauss+ model is applied—specifically, how changes in rt, mt,
and lt are generated—let’s simulate one interest rate path using U.S. Treasury zero-
coupon data from January 2014 to January 2022. The estimated parameters for the
Gauss+ model are as follows:

Notice how the reversion speeds are fastest for the short-term factor (αr) and slowest
for the long-term factor (αl ).

Also, assume that at the start of a period, we have the following factors: rt = 4.50%, mt =
5.25%, and lt = 4.00%. Further, assume that our time step is one month, or 0.083 years
(= 1 / 12).
Step 1: Simulate .

These are random draws from a standard normal distribution multiplied by the square root of
the time. For the sake of argument, suppose that the values generated are +0.55 and –0.95, which
when multiplied by , become 0.1585 and –0.2737, respectively.

Step 2: Plug in the results from Step 1 into the Gauss+ model equations.

Step 3: Calculate the new one-month rate for the three interest rate factors.
Estimating Gauss+ Model Parameters
LO 16.d: Explain how the parameters of the Gauss+ model can be estimated from
empirical data.

The parameters of the Gauss+ model can be estimated using maximum likelihood
techniques. However, there is a simplified approach that employs the following steps.
This procedure is possible when random components are not used to determine mean
reversion speeds.
Step 1: Start by selecting a proxy for the short-term rate (e.g., fed funds rate).

Step 2: Regress the changes in the short-term rate against changes in a medium-term rate (e.g., 2-year
rate) and a longer-term rate (e.g., 10-year rate).

Step 3: Find mean reversion speeds αr, αm, and αl that replicate the sensitivities seen in Step 2. This is
possible because the regression coefficients do not depend on volatility parameters, only on the
mean reversion parameters.

Step 4: From market data, determine the volatility term structure. This can be found with the observed
volatility of rates, or the volatilities used by the market to price interest rate derivatives.

Step 5: Find the volatility and correlation parameters σm, σl, and ρ, such that we can replicate the
volatility term structure in Step 4.

Step 6: Find the long-term reversion level, μ, by minimizing the squared errors of observed yields versus
estimated yields.

Once the model parameters are estimated, the Gauss+ framework can (1) precisely
model rate term structures and their volatilities; (2) account for the rate expectations
and risk premiums reflected in the rate term structure; and (3) provide a robust tool for
pricing, hedging, and analyzing interest rate products.

MODULE QUIZ 16.1

1. Which of the following statements best describes the structure of the Gauss+ model?
A. A single-factor model with constant volatility.
B. A model without any mean reversion or volatility assumptions.
C. A two-factor model with mean reversion to a fixed long-term rate.
D. A cascade model where each factor mean reverts to the next level.
2. The role of the medium-term factor in the Gauss+ model is that the factor:
A. remains constant over time.
B. directly determines the pricing of long-term bonds.
C. reflects economic influences, such as business cycles.
D. represents the risk premium in the interest rate structure.

3. What is a main difference or similarity in the volatility assumption between the Vasicek model and
Gauss+ model?
A. Both models assume constant volatility.
B. Both models have stochastic volatility components.
C. The Vasicek model incorporates stochastic volatility, while the Gauss+ model assumes
deterministic volatility.
 D. The Vasicek model assumes constant volatility, while the Gauss+ model can replicate hump-
shaped volatility term structures.
4. Assume the short-term rate is 5.0% and the medium-term rate is 6.0%. By using the Gauss+ model,
if the corresponding mean reversion speed is 1.3 and the time step is three months, what is the
change in the short-term rate?
A. –0.325%.
B. +0.325%.
C. –3.250%.
D. +3.250%.
5. When using empirical data to estimate parameters for the Gauss+ model, which of the following
parameters is calibrated by minimizing the squared errors of observed yields relative to model-
predicted yields?
A. Long-term target.
B. Volatility parameter.
C. Correlation parameter.
D. Mean reversion speeds.

KEY CONCEPTS

LO 16.a
Compared to the Vasicek model, the Gauss+ model is a more sophisticated model that
provides a better representation of real-world interest rate behaviors. The model
consists of three interacting components: the short-term rate (rt), a medium-term
interest rate factor (mt), and a long-term interest rate factor (lt).
The short-term rate represents the policy interest rate controlled by central banks. The
medium-term factor represents influences from the broader economy, such as business
cycles and monetary policy effects. The long-term factor represents long-term
expectations for economic trends, such as inflation and productivity growth.

LO 16.b
The simplicity of the Vasicek model restricts its ability to model certain aspects of
interest rate dynamics. These restrictions are mitigated with the Gauss+ model. The
ability to price a wider range of instruments as well as the consistency with interest
rate dynamics expected by market participants makes the Gauss+ model preferable to
the simpler Vasicek model.

LO 16.c
The Gauss+ model expands the Vasicek model into a three-factor cascade structure
where the short-term rate mean reverts to the medium-term factor, the medium-term
factor mean reverts to the long-term factor, and the long-term factor mean reverts to a
constant target (μ).
The expressions for these three interest rate factors are as follows:
LO 16.d
The Gauss+ model parameters can be estimated using empirical data on interest rates.
This process involves breaking down estimation into stages to calibrate the model to
observed market dynamics. By using this estimation approach, the Gauss+ model can
accurately reflect empirical market data, becoming a robust tool for traders, analysts,
and policymakers.

ANSWER KEY FOR MODULE QUIZ

Module Quiz 16.1

1. D The Gauss+ model uses a three-factor cascade structure where the short-term rate
(rt) mean reverts to the medium-term factor (mt), which mean reverts to the long-
term factor (lt). (LO 16.a)

2. C The medium-term factor represents broader economic influences like business

cycles and monetary policy effects. It acts as an intermediate driver between the
short-term and long-term interest rate factors, reflecting the model’s cascading
mean-reversion structure. (LO 16.a)
3. D The Vasicek model assumes constant volatility, which limits its ability to reflect
the hump-shaped term structure of volatilities observed in practice. In contrast,
the Gauss+ model incorporates variable volatilities, capturing higher volatility for
intermediate maturities and declining volatility for longer terms. (LO 16.b)
4. B The change in rt is calculated as follows:

Thus, the short-term rate should increase by 0.325%. (LO 16.c)

5. A The long-term reversion level (μ) is calibrated to align the model’s predicted
yields with observed yields, minimizing squared errors across all maturities. (LO
16.d)