What is Sensitivity Analysis?
Sensitivity Analysis is a technique used by financial analysts to gauge the impact on the outcomes of a model that may be caused by changes in assumptions or circumstances.
What is a Model?
In finance, a portfolio model is created with the use of mathematical equations that demonstrate the relationships between different financial instruments and markets with each other. These mathematical equations can also tie in concepts such as cash flow with accounting terms such as accrual accounting. The goal of a model is to accurately predict future performance. However, models are never perfect. They tend to be close to perfect in most situations, but not quite accurate enough. It is up to the investor or anyone using the model to decide if they are willing to accept a slight amount of error for some amount of accuracy. Many people may use different models for different purposes, such as risk versus return versus capital versus liquidity versus credit versus accounting versus tax-efficiency. A portfolio model usually looks at all these variables and then determines how much one should invest in each asset class (i.e., equity, debt, real estate, etc.). The model is adjusted over time as well so that the investor can take advantage of changing market conditions. This is where the sensitivity test comes into play.
How Can You Calculate Sensitivities?
There are many ways to calculate sensitivities, but there are two main ways used by investors: proxy testing and simulation. These are explained in greater detail below.
Proxy Testing with a Portfolio Model
For proxy testing, one looks at the historical data to see how sensitive an asset class or market was to one variable versus another. For example, if I wanted to know how sensitive the stock market was to interest rates , I could look at historical data for this relationship . To find this relationship, I would first find the equation of the model that tells me I need to have a certain amount of stock in my portfolio for each % change in interest rates . For example, if my model is correct and I need to have 0.50% of my portfolio for every 1% decrease in interest rates , I would divide 100 by 0.500 to get 5. If there are not enough stocks traded during this time period, you could also use other methods such as different time periods or using different average interest rates.
Once you’ve found your new equation, you can plug it into x = 100/x shows you how sensitive the price of an individual stock is to changing interest rates .
Example of Simulation with a Portfolio Model
For simulation with portfolio models, one uses the model to simulate different variables until it can accurately predict an outcome. This is useful for use in scenario analysis . To do this, one creates a new model that incorporates the variables you want to test. One then runs predictions on this new scenario versus the old scenario to see how effective it is at predicting an outcome.
Example of Simulation with Financial Instruments
For simulation with financial instruments, one uses the model to simulate different variables until it can accurately predict an outcome. This is useful for use in scenario analysis . To do this, one creates a new model that incorporates the variables you want to test.
Types of Sensitivity Analysis
There are four major types of sensitivity analyses: Monte Carlo simulation, Perturbation analysis, Sensitivity Table, and Marginal Analysis. Each has its own advantages and disadvantages so let’s go over them one-by-one.
Monte Carlo simulations (also known as Monte Carlo analysis) is a technique used to find the genetic relationships between items in a multi-level model. It uses random numbers generated from a probability distribution and then runs the model thousands of times to get the probabilities. It can tell you what will happen if we change our assumptions or we change the parameters used in the model and evaluate how sensitive each result would be to different changes.
Perturbation analysis is based on either assuming that parameters are not correlated, or assumptions about independence. Sensitivity analysis examines how sensitive all variables are to changes in assumptions or circumstances. It estimates the impact of changes on the values of variables. There are different approaches, including (1) sensitivity analysis using a probability or sensitivity table, (2) marginal or partial methods of sensitivity analysis, which estimate relationships between future values and current values, and (3) probabilistic sensitivity analysis where probabilities can be calculated for all combinations of estimated parameters.
A Sensitivity Table is a table of numbers used to estimate the effect that changes in parameters have on model outputs. A marginal approach is just that–a marginal approach which focuses on the relationship between the effect of parameter changes on outputs and imputed inputs using imputations to estimate parameter ranges. The imputed inputs are essentially dummy variables related to variation in parameters. The sensitivity analysis approach is taken when there is no information about the exact relationships between the parameters and outputs.
In Sensitivity Analysis in finance, a fully connected model with a continuous time lag is used to simulate a financial strategy using Markov Chain Monte Carlo (MCMC) software. Observations at time t are used to estimate parameters at time t+1, which in turn are used to simulate future outcomes given these parameters. The impact on future outcomes of changing assumptions or parameter values can be found by running the model many times and looking at which results meet certain criteria such as statistical significance or how sensitive they are to changes in assumptions and parameter values. This knowledgeability helps us support our risk management strategies for larger investment portfolios.
Sensitivity analysis can basically be performed to find out how changes in input variables affect the final result. Specifically, one has to measure how uncertainties or randomness in input variables affect the final results. The idea is that if you are not sure about something now, then you should probably take into consideration all possible outcomes so that you know exactly what could happen were things turn out differently than expected. Fintalent, the hiring and collaboration platform for tier-1 M&A and Strategy professionals offers business managers a platform to hire Expert Freelance Financial Analysts to help businesses meet all of their Sensitivity Analysis needs in other to secure their profitable existence.