Agent-based model of the Russian banking system: Calibration for maturity, interest rate spread, credit risk, and capital regulation

ABSTRACT

The Basel III regulation raised the minimum capital requirements for banks. However, its implementation may not have reduced systemic risks. Academicians investigating optimal banking regulations do not have a consensus on whether to increase or decrease capital requirements. Here, we use the agent-based approach to study capital regulation and its implications on the evolution of the banking system. We chose the Russian banking system to proxy key model parameters. We find that lower capital requirements imply higher financial stability than the Basel III regime, where the regulator requires banks to have capital over 10% of its risk-weighted assets’ amount. However, the regulatory rule to merely have a non-negative capital is the simplest solution that best fits heterogeneous economies. It produces the highest ratio of capital to assets, the least number of bank bankruptcies, and the lowest demand of banks to enter the interbank market to cover liquidity problems for all systems.

KEYWORDS:

• Agent-based model
• capital regulation
• capital adequacy ratio
• Basel

Acknowledgments

The work was conducted in the International Laboratory of Decision Choice and Analysis (DeCAn Lab), Department of Applied Economics of the National Research University Higher School of Economics; Laboratory of Mathematical Modeling of Complex Systems, P.N. Lebedev Physical Institute.

We thank three anonymous reviewers, editors-in-chief and the associate-editors for insightful comments.

The opinions expressed herein are solely those of the authors and do not necessarily reflect those of the affiliated institutions.

Disclosure statement

There are no conflicts of interest to disclose.

Notes

1. For more information on agent-based modelling and computational economics, we recommend referring to Tesfatsion (2006) and to the following website: http://www2.econ.iastate.edu/tesfatsi/aintro.htm.

2. Hereafter: [1] – Ashraf et al. (2017), [2] – Liu et al. (2017), [3] – Samitas et al. (2018), [4] – Riccetti et al. (2016), [5] – Poledna et al. (2017), [6] – De Caux et al. (2017), [7] – Wolski and van de Leur (2016), [8] – Popoyan et al. (2017), [9] – Catullo et al. (2015), [10] – Xu et al. (2016), [11] – (Gabbi et al. (2015), [12] – Chan-Lau (2017) ABM* – the proposed agent-based model in the current paper.

3. A tax on systemic risk is proportional to the contribution of a bank to a systemic risk level (Poledna et al., 2017). The tax on systemic risk is more effective than Global Systemically Important Banks’ (G-SIBs) surcharges. The latter changes the loss distribution only slightly. We need to note that this concept is too vague because the centrality measures, namely a map of the preferred transaction (for flows, see more details in Leonidov and Rumyantsev (2016), see also, in a different context, Kireyev and Leonidov (2019)), are associated with systemic risk that is unobservable, hence unmeasurable and non-verifiable.

4. General regulation of system: https://www.cbr.ru/PSystem/system_p/.

5. URL: https://www.ecb.europa.eu/pub/pdf/targetar/ecb.targetar2018.en.pdf.

6. We choose the intermediary approach in modelling, that is, a bank performs only the function of transforming deposits into loans. Colwell and Davis (1992) recognise five main approaches to defining a bank’s functions, such as the production approach, intermediation, asset, user cost, and value-added approaches. Approaches other than the intermediary one require the introduction of alternative costs and benefits in the model economy to show that a bank is looking for the best option with respect to alternative costs, thereby earning a profit and considering deposits (or products other than loans) as a measure of its output. Since we assume the homogeneity of depositors and borrowers, the intermediary approach is sufficient for our simulations.

7. The value of 600 banks is the approximate number of banks in the Russian banking system as of 01.01.2017 and 10 banks is the number of systemically important credit institutions in Russia at the time of the paper preparation.

8. Hereafter: [1] – (Ashraf et al., 2017), [2] – (Liu et al., 2017), [3] – ((Samitas et al., 2018), [4] – (Riccetti et al., 2016), [5] – (Poledna et al., 2017), [6] – (De Caux et al., 2017), [7] – (Wolski & van de Leur, 2016), [8] – (Popoyan et al., 2017), [9] – (Catullo et al., 2015), [10] – (Xu et al., 2016), [11] – (Gabbi et al., 2015), [12] – (Chan-Lau, 2017), ABM* – the proposed ABM in the current paper.

9. 17 model days correspond to one real year. We make a single run with the length of 2500 days. Thus, in fact we have 2500/17 ~ 150 simulations per scenarios. Therefore we eliminate the error of discussing some particular (perhaps, odd or extreme realisation).

10. The value of 4% default rate corresponds to the historical average value for the Russian corporate bond market (Cbond) for 2009–2017.

Additional information

Funding

This work was supported by the Fundamental Research Program of the National Research University Higher School of Economics (NRU HSE) under the state subsidy ‘5-100ʹ program for the leading universities of the Russian Federation. The work was also supported by the Russian Academy of Sciences Presidium program “Fundamental aspects of economic security”.