This short video descibes how to set up and run the model.

If you do not have Java 5 or higher installed on your computer, please download and install it first.

Run the Bank Avalanche Model of Systemic Risk


The recent global financial crisis has led economists to seek a better understanding of how the interconnectedness of the global financial system can lead to systemic risk.

In our model of the financial system, individual agents make optimal decisions for themselves without considering the impact their decisions have for the system as a whole. Taken in aggregate, the resiliency of the entire financial system becomes weaker or stronger based on the actions of the individual agents.

We measure system resiliency by the size and frequency of bank avalanches. A bank avalanche occurs when two or more banks fail at the same time. By studying this simplified model, we hope to gain some insights into what causes systemic risk, and how it can be mitigated in the real financial system.

Some questions we ask include:

The Agents

The model has three types of agents:

  1. Non-financial transactors (NFTs), representing commercial enterprises not in the financial industry
  2. Banks
  3. An AIG-like insurance company.

The agents exchange financial claims with each other according to the following set of rules:

  1. NFTs take the following actions:
    • Make deposits into banks
    • Apply for a loan with a specific bank
    • Repay an existing loan (if they have a loan)
    • Default on an existing loan (if they have a loan).

  2. Banks can take the following actions:
    • Take deposits from NFTs
    • Ensure sufficient deposits are on hand to meet reserve requirements
    • Make loans to NFTs with money from deposits
    • Make interbank loans to other banks in the network if deposits exceed reserves
    • Accept interbank loans if the bank lacks sufficient deposits to provide loans to applicants.

  3. The AIG-like insurance company is connected to all banks. It issues insurance in the form of credit default swaps (CDS) to the banks. Every period:
    • Banks sell CDS to the insurance company to insure their loans to NFTs.
    • Banks purchase CDS from the insurance company.

The simulation ends and restarts when all agents in the system disappear. As the simulations run, the model generates histograms of bank life span, bank profitability, bank avalanche size, and bank avalanches per simulation.

Run the Bank Avalanche Model of Systemic Risk

More Information

For more information, please read our recent seminar presentation.