Economic Scenario Generator
Contents
Definition
An Economic Scenario Generator (ESG) refers to a mathematical model (and its computer implementation) that simulates possible future paths of economic and financial market variables[1].
Usage
Economic Scenario Generators are used among others by financial institutions to assess the impact of such scenarios on the firm’s asset and liabilities. The term ESG is sometimes used to denote the entirety of a simulation framework. It is more accurate to limit the scope of the term to the modelling components explicitly related to the simulation of economic and financial market variables and not firm specific / portfolio (asset or liability) specific models. Such auxiliary specific models are typically implemented via satellite models.
Specification
The main attributes of ESG's are
- The scope of the economic domain being modelled (e.g. jurisdictions / geographic regions)
- The economic / financial variables covered: Those range from exclusively market variables when modelling market risk scenarios, to a variety of macro-economic variables when modelling the entire economy
- The granularity of the calculation: ranging from instrument level to aggregate portfolio level
- The temporal sampling and Risk Horizon used: ranging from a day/week for Market Risk simulations to decades in the case of insurance or pension fund liabilities
- The type of model / methodology used (e.g. the degree of reliance on historical data or the integration of economic concepts)
Usage
Economic Scenario Generators are used for a variety of purposes across different field of activity / industries
Insurance
The specific expression Economic Scenario Generator (ESG) commonly refers to approaches used for the evaluation of capital requirements of insurers.
Pension Funds
Similar to Insurance Pension Funds perform very long simulations of assets versus liabilities
Banking
ESG frameworks are also used by various business lines in the banking industry. The two predominant use cases are:
- Commercial Banks perform Monte Carlo Simulation of Credit Portfolios (Credit Value-at-Risk, Economic Capital)
- Investment Banks perform Value-at-Risk Simulations of Market Risk
The role in Macro-Prudential Stress Testing
Since the Financial Crisis, regulatory stress testing has de-facto extended the risk horizon used in bank stress testing exercises, thereby bringing insurance ESG and bank capital stress testing conceptually closer
ESG Methodologies
Monte Carlo simulation is employed to perform thousands of simulations to provide a distribution for the metrics of interest. The different realisations for the variables modelled constitute “economic scenarios”. This constitutes effectively a generalization and automatization of individual Scenario Analysis.
By jointly simulating the different economic variables, joint return distributions for multiple assets and liabilities can be obtained. The respective distributions feed into the financial institution's asset-liability models, permitting an assessment of a potentially large number of different sources of risk to the firm.[2]
Parameterization Aspects
- selecting appropriate steady-state levels
- determining appropriate values for initial conditions,
- identifying stylized facts that must be reproduced by the ESG
- controlling the mean reversion of economic variables
Real World versus Market-Consistent Scenarios
- Market-consistent scenarios enforce certain mathematical relationships within and among financial instruments via no-arbitrage conditions.
- Real-world scenarios are concerned with forward-looking potential paths of economic variables and their potential influences.
Advantages
Monte Carlo simulation provides a number of conceptual advantages over the discrete scenario based analysis more commonly used in Stress Testing
- Assignment of (model based) probabilities to scenarios: Results show not only what could happen but how likely each outcome
- Completeness: All possible solvency (or other relevant metric) states are included (in principle). No "hidden cliff" effects
- Usability: Consistent allocation of risk that is not sensitive to scenarios
Issues and Challenges
- Poorly tested / underpowered simulations may be subject to sampling noise and instability of outcomes
- There is significant Model Risk, both in the construction of economic scenario models and in the evaluation of portfolio impact per scenario. The uncertainty grows the longer the time period considered
Implementations
Open Source implementations of Economic Scenario Generators are available in
- The R package ESGtoolkit
- The IMF Growth-at-Risk model
- (Lower level) econometric models (e.g Vector Autoregression) included in Statsmodels
