Interest Rate Models

This booklet is prepared to give you an overview of interest rate models. The coming booklets
will discuss each interest rate model in detail.

Short rates
Short rates refer to the interest rates on short-term financial instruments, typically those with
maturities of one year or less.

Long rates
Long rates refer to the interest rates applicable to long-term financial instruments, typically
with maturities extending several years into the future.

What is term structure of Interest rate?

The term structure of interest rates, also known as the yield curve, represents the relationship
between interest rates (or yields) and different maturities or terms to maturity for a particular
type of debt security, typically government bonds. It provides insight into future interest rate
changes and economic activity. The term structure can take various shapes:

Kshitij Anand
 1. Normal (Upward-Sloping) Yield Curve: Longer-term interest rates are higher than
shorter-term rates. This shape is typical in a healthy economy where investors expect
future economic growth and potentially higher inflation.
2. Inverted (Downward-Sloping) Yield Curve: Shorter-term rates are higher than longer-
term rates. This is often seen as a predictor of economic recession, as it indicates that
investors expect lower interest rates in the future due to slowing economic activity.
3. Flat Yield Curve: Short- and long-term rates are very close to each other, indicating
uncertainty about future economic conditions.
4. Humped (Bell-Shaped) Yield Curve: Medium-term rates are higher than both short- and
long-term rates. This is less common and can occur during transitions in economic
cycles.

What are interest rate models?

Future interest rates are an important concern for us. They help in pricing bonds, interest rates
derivatives and many such products. Therefore, we need to predict how future interest rates
will be. Interest rate models are the mathematical description of how interest rate changes
over time. These models allow us to simulate the movement and evolution of interest rates.

One factor Interest Rate Models Vs. Multi-factor Interest Rate Models
One-factor model assumes that short-term interest rates (also known as the short rate) follow a
certain statistical process, and that other interest rates in the term structure (i.e., rates for
different maturities) are related to the short rate.

One-Factor Models
1. Short Rate as the Driver

• In one-factor models, the short rate is considered the primary driver of all other
interest rates.
• Other interest rates are not randomly determined; once the short rate is specified,
arbitrage arguments are used to determine the rates for all other maturities.
2. Limitations

• While one-factor models are conceptually simple and easy to implement, they have
limitations. For example, they may not fully capture the complex dynamics of interest
rates, especially during periods of market stress or structural shifts.

Kshitij Anand
Multi-Factor Models
1. Two-Factor Model

• Multi-factor models, such as the two-factor model, introduce additional factors

beyond the short rate to explain interest rate movements.
• A common two-factor model includes a long-term rate as the second factor, which
helps capture trends in interest rates over longer periods.
2. Practical Challenges

• However, in practice, one-factor models are often preferred over multi-factor models
due to the complexity and difficulty of applying multi-factor models.
• Estimating multiple factors accurately can be challenging, and the added complexity
may not always lead to significantly better predictions.

Why are One-Factor Models used?

1. High Correlation:

• One reason for the prevalence of one-factor models is the high correlation between rate
changes for different maturities.
• This suggests that changes in short-term rates often affect long-term rates in a similar
manner, supporting the use of a single factor to explain interest rate movements.
2. Empirical Evidence:

• Empirical studies have also shown that a significant portion of changes in the yield
curve (the graph of interest rates across different maturities) can be explained by a
simple level shift in interest rates, which aligns with the assumptions of one-factor
models.
Till now, we discussed interest rates, interest rates models, short rates, long rates and one factor
model Vs Multi factor model. Now, all interest rate models are classified into major class: -
Equilibrium model and No-arbitrage Models.

Equilibrium models and Arbitrage-free Models

The distinction between arbitrage-free and equilibrium models lies in how they approach the
term structure of interest rates and the underlying assumptions about market behavior.
1. Equilibrium Models:

• Equilibrium models are concerned with explaining the term structure of interest rates
based on economic fundamentals and market participants' behaviors.
• These models often start with assumptions about how investors and borrowers behave
in response to interest rate changes, supply and demand dynamics in the bond market,
and other factors that influence interest rates.

Kshitij Anand
 • The model then derives a term structure of interest rates that is consistent with these
assumptions and the broader economic environment.
Therefore, equilibrium models try to find term structure by studying the factors that may
influence the interest rates. Hence, Term structure is an output of Equilibrium models
Examples of Equilibrium Models: -
Some popular equilibrium models used in finance to describe the term structure of interest rates
include the Vasicek model, the Cox-Ingersoll-Ross (CIR) model. These models are often used
to estimate the evolution of interest rates over time based on certain assumptions. Here are the
equations for these models:

A. Vasicek Model:
The Vasicek model assumes that interest rates follow a normal distribution and that the short
rate follows a mean-reverting process. The short rate 𝑟𝑡 is given by:

𝑑𝑟(𝑡) = 𝑎(𝑏 − 𝑟(𝑡)) 𝑑𝑡 + 𝜎 𝑑𝑊(𝑡)

Where:

• a is the speed of reversion towards the mean.

• b is the long-term mean of the short rate.
• 𝜎 is the volatility of the short rate.
• dW(t) is a Wiener process (Brownian motion).

B. Cox-Ingersoll-Ross (CIR) Model:

The CIR model is an extension of the Vasicek model that prevents negative interest rates. The
short rate 𝑟𝑡 in the CIR model is given by:

𝑑𝑟(𝑡) = 𝑎(𝑏 − 𝑟(𝑡)) 𝑑𝑡 + 𝜎√𝑟(𝑡) 𝑑𝑊(𝑡)

Where the parameters have the same meanings as in the Vasicek model.

Let’s explore other class of Interest rate models now: -

2. Arbitrage-Free Models:

• In arbitrage-free models, the focus is on ensuring that there are no opportunities for
risk-free profits (arbitrage) by trading based on the model's values and observed
market prices.
• These models typically start with observed market prices of benchmark instruments
and then use a random process for interest rates to derive a term structure of interest
rates that matches these prices.

Kshitij Anand
 • The key point is that arbitrage-free models do not explain or derive the initial term
structure of interest rates. Instead, they take it as given and work backward to find a
model that is consistent with these rates.
Consider No-arbitrage models as something like you already have the term structure (so term
structure here is an input). Then, a random process is presumed to generate the term structure
of interest rates, incorporating a drift term and volatility for interest rate fluctuations. Using
this process and an assumed value for the drift term parameter, a computational method is
applied to derive the term structure of interest rates (known as the spot rate curve). This
procedure aims to ensure that the model's valuation aligns with observed market prices for
benchmark instruments.
In essence, arbitrage-free models are focused on matching observed market prices, while
equilibrium models are focused on explaining the term structure based on underlying economic
factors. Arbitrage-free models take the initial term structure as given, while equilibrium models
seek to explain why the term structure takes a certain shape.
Examples of no-arbitrage models include Hull-White Model, Ho-Lee Model, HJM Model
among other.
A. Hull White Model
The Hull-White model extends the Vasicek model by making the mean reversion
parameter time-dependent. The short rate r(t) at time t is given by:
𝑟(𝑡) = (θ(t) − 𝑎 ∗ 𝑟(𝑡)) 𝑑𝑡 + 𝜎 𝑑𝑊(𝑡)

Where:

• a is the speed of reversion towards the mean.

• θ(t) is the time-dependent mean reversion level.
• 𝜎 is the volatility of the short rate.
• dW(t) is a Wiener process (Brownian motion).

B. Ho-Lee Model
The Ho-Lee model assumes that the short rate is normally distributed and follows a
driftless process. The short rate 𝑟𝑡 in the Ho-Lee model is given by:

𝑟(𝑡) = θ(t)𝑑𝑡 + 𝜎 𝑑𝑊(𝑡)

How to interpret whether an equation belongs to equilibrium model or arbitrage free model?

• Any model that has time dependent parameter (say theta for example) is an arbitrage
free model. This is because at that point of time, it tries to match with the market prices
and hence its parameters are different, hence time dependent.

Kshitij Anand
We will be seeing all these models and other models that I have not shown in example
individually in great detail. We will start with intuition and then we will move onto equation,
calibration, python code for all models. Follow me on LinkedIn to continued learning!

Kshitij Anand