Agent-based modeling for benchmarking banking regulation regimes: Application for the CBDC

Abstract

COVID-19 pandemic challenges the sustainability of the modern financial system. International central bankers claim that banks are solid. They have accumulated significant capital buffers. Those buffers should be further more augmented by 2027 in line with Basel III reforms. However, disregarding such a consecutive rise in the banking capital adequacy requirements, the US financial authorities undertook an unprecedented step. First time in the country history they lowered the reserve requirement to zero at the end of March 2020. Friedrich von Hayek demonstrated the fragility of the modern fractional reserve banking systems. Together with Ludwig von Mises (von Mises, 1978) he was thus able to predict the Great Depression of 1929 and explain its mechanics much in advance. Thus, we wish to utilize the agent-based modeling technique to extend von Hayek’s rationale to the previously unstudied interaction of capital adequacy and reserve requirement regulation. We find that the full reserve requirement regime even without capital adequacy regulation provides more stable financial environment than the existing one. Rise in capital adequacy adds to modern banking sustainability, but it still preserves the system remarkably fragile compared to the full reserve requirement. We also prove that capital adequacy regulation is redundant when the latter environment is in place. We discuss our findings application to the potential Central Bank Digital Currencies regulation.

Keywords

Agent-based model reserve requirement capital adequacy Basel III CBDC

1. Motivation

COVID-19 pandemic has raised the concerns over the stability of the modern banking systems. (World Bank, 2020) made the unprecedented world GDP forecast by the end of 2020 in the mid of the year. It predicted $-$ 5% decline per annum. (Aramonte & Avalos, 2020) claim that the default correlation rose even more than it was during the crisis of 2007–09. As a result, (Véron, 2020) doubts that financial institutions may bear the upcoming realizations of the credit risk. That is why he opts for the delay in the implementation of the new IFRS 9 provisioning methodology. There is an even stronger claim to drastically revise modern banking regulation. For instance, (Pettifor, 2020) asks to strengthen regulation and to create a supranational regulator. This is in addition to the existing since 1974 Basel Committee on the Banking Supervision (BCBS). In order to add financial stimulus to the banking system the prudential authorities of the US (FRS, 2020) and Morocco (BKAM, 2020) have decreased the reserve requirement to zero in March and June 2020, respectively. This is unprecedented for them, though not that novel for the world history. For instance, Canada operates zero reserve requirement since 1990s (Lavoie, 2019).

One may say that reducing the mandatory reserve rate is a logical decision. Banks pay fees for the insured deposits. The fee level is generally the same for the credit institution. However, depending on the jurisdiction, the fee level may vary between the banks depending on their creditworthiness. Generally, the fee does not depend upon the total volume of insured deposits. For more discussion, please, refer to (Shiers, 1994), (Gómez-Fernández-Aguado et al., 2014), (Guizani & Watanabe, 2016).

Same time they have to put aside some funds in the form of the mandatory reserves against the same deposits. Thus, banks are twice penalized. That is why abandoning mandatory reserve may be regarded as a way to decrease burden over banks. However, the principal difference in-between mandatory reserves and fees to the deposit insurer is that the level of fees does not change the total money supply, where as the mandatory reserve ratio does this via the money multiplier.

To sum up, decreasing mandatorily reserves might have acted as a double-stimulus, i.e., to allocate less funds against deposits and to boost the broad money.

Thus, we come to an objective of our research. There seems to be a need for the inventory revision of the existing banking regulation. Right at the creation of the BCBS in 1974 Friedrich von Hayek received the Nobel Prize in economics (von Hayek, 1974). He proved that the fractional reserve banking is prone to regular devastating economic crises. From one side, the 2020 step of the US and Morocco seems to exacerbate the crisis, rather than solve it. From another side, the banking regulation complicated a lot since the times of von Hayek. Perhaps, it did the financial system more stable. We want to find out whether the banking prudential regulation novelties like CAR improved the system stability or not. Ultimately, did the modern regulation rules outpace the full reserve banking or not?

To answer this question, we briefly describe the literature and refer to the key facts about modern prudential banking regulation in Section 2. Then we explain our methodology in Section 3. We collect our findings in Section 4. We compare the modern banking to the full reserve banking. We study the changes in the system stability (fragility) due to changes in CAR and RR regulation. We conclude in Section 5.

2. Literature review

For almost 40 years the academic researchers try to model the endogenous economic cycle starting from (Blatt, 1983), (Haxholdt et al., 1995), (Raybaut, 2014), (Sunaga, 2017), (Colacchio & Davanzati, 2017), (Hasumi et al., 2018) with the most recent paper of (Agliari et al., 2020). In general, those authors extend the Keynesian approach to economic system dynamics. We consider the works by (Minsky, 1982), (Shubik, 1987), (Taleb, 2007) to be adjacent to the mentioned cohort. In essence three of them claim that the financial crises are somewhat extraordinary unpredictable events. They are a sort of sun eclipses. Following their logic, each crisis time may be deemed ‘different’ in (Reinhart & Rogoff, 2009) terms.

However, Friedrich von Hayek explained the regular nature of financial crises in the fractional reserve banking world (von Hayek, 1929), (Huerta De Soto, 2006). Friedrich von Hayek said that sometimes the economists lack data on the indicators of interest. However, this should not prevent from mapping the approximately right dependence pattern and not to be obscured by the precisely wrong one (von Hayek, 1974). (Huerta De Soto, 2006) presents the underlying of the von Hayek’s theory. It is the mechanics of the voluntarily saving and the nature of the technical progress. (Huerta De Soto, 2006) uses the example of the Robinson Crusoe who wishes to collect more coconuts. The earlier example belongs to (Roscher, 1854) and thereafter (Bohm-Bawerk, 1886). They consider fishermen who wish to prepare a net to catch more fish. In brief, the voluntarily savings lead to technical progress and the welcomed deflation. However, deflation is not popular. For instance, (Ito & Hoshi, 2020, pp. 169–172) list a set of its drawbacks.

The fractional reserve banking has an attractive feature for politicians. First, (Huerta De Soto, 2006) brings evidence that credit money primarily finances government debt. Using credit money to finance government debt is an alternative to seniorage (printing money to fund government spending). However, both options lead to rise in the broad money. As a result, the inflation is inevitable. Second, the decrease in reserve requirement creates more opportunities for credit money multiplication. Thus, in the shortest run it may create the illusion of the sustainable economic growth. In fact, the growth is temporarily and is followed by an unavoidable and detrimental crisis.

Such appealing features of the fractional reserve banking seem to have assured its popularity in the modern world. That is why, (Huerta De Soto, 2006) brings evidence on the regular crises and banks failures. We may assume that such a sequence led to the financial innovation. Namely, to the development of the prudential regulation – like capital adequacy ratio (CAR) from Basel I – in parallel to the existence of the fractional reserve banking.

The crisis of 2007–09 implied the revisions to the CAR minimum. The finalized Basel III will require up to 18% of the risk-weighted assets compared to around 2% being present during Basel II times on the eve of the crisis, see (Penikas, 2020, p. 93). To sum up, regulators and politicians blamed CAR for the banking system instability without challenging the fractional reserve banking paradigm.

The rise of the CAR minimum is not the only dimension of banking prudential regulation evolution. Andrew Haldane (see (Zhong, 2013)), (Penikas, 2015), (Hernández de Cos, 2019) argue that the recent regulation has definitely become extremely complex. However, this does not evade us from the regulation failures discussed by (Moosa, 2010), (Dewatripont et al., 2010), (Lall, 2012), (Cathcart et al., 2017).

Agent-based models (ABMs) for banking systems were already considered in various papers (Ashraf et al., 2017), (Samitas et al., 2018), (Riccetti et al., 2016), (Poledna et al., 2017), (De Caux et al., 2017), (Wolski & van de Leur, 2016), (Popoyan et al., 2017), (Catullo et al., 2015), (Gabbi et al., 2015), (Chan-Lau, 2017), (Ponomarenko & Sinyakov, 2017). However, current papers have several shortcomings. They might lack realistic yield curve (Chan-Lau, 2017), the prudential regulations are too general (Ponomarenko & Sinyakov, 2017), including the CAR absence (Biondi & Zhou, 2019).

No one previously benchmarked the regulation of the full reserve banking to the fractional one, though (Biondi & Zhou, 2019) approached it closer than the rest of researchers. The rationale is quite simple. They did not exist in history in the same time in the same environment. For instance, (Huerta De Soto, 2006) strongly supports the Bank of Amsterdam practice of full reserve banking in the XVII–XVIII ${}^{\text{th}}$ centuries. However, there was no CAR regulation. On the opposite, when CAR regulation came in, there were no full reserve banking systems. Thus, we lack proper data to econometrically evaluate the so called ‘treatment’ in the form of the prudential banking regulation rules, RR and CAR including. To overcome this challenge, we use the artificial environment.

3. Methodology

Our artificial environment is the agent-based model of the banking system (ABM) from (Ermolova et al., 2021). However, the original ABM was a pure full reserve banking system. The demand (sight) deposits could not be early withdrawn. The liquidity crises were produced from the mere mismatch in the asset and liability maturity profiles.

The agent-based model (ABM) of the banking system reflects the intermediation mechanics of banking. There are deposit and loan flows. The reader may find its parameters in Table 1.

Table 1
Loan and deposit flow parameters

#	Indicator	Units of measure	Deposit	Loan
1	Number of applications	Count	100	300
2	Average application amount	Units of money	5	15
3	Rate	pp. p.a.	3	5
4	Maturity ${}^{*}$	Days	3	5

${}^{*}$ Flat yield curve: rates from the table above do not vary with maturity.

Any deposits application is always accepted. This corresponds to the legal framework of the modern banking systems. It says that a deposit is a public offer. This means that any person bringing money has the right to open a deposit account subject to money laundering checks are well passed. Loans are accepted if there is enough cash and in case the prudential regulations are met. Latter might be in the form of the nonnegative capital or CAR requirement. There is also an option of no CAR regulation at all. We set the initial number of banks in the system and their initial capital. We may allow banks to come in. The general rule is the less the number of banks is the higher the probability of a new bank entry is. Borrowers may default on their loans. Banks may create provisions against those losses. We set the interest rates offered to depositors and borrowers. There is an interbank lending market.

We extend the earlier version of our ABM (Ermolova et al., 2021) to allow for the credit money creation. There are only demand (sight) deposits (there are no term ones). They have infinite maturity. Thus, they are constantly accumulated unless are withdrawn upon demand. This can be done at any point in time. The withdrawal frequency depends in the pre-set value of the PW parameter (probability of withdrawal). It equals zero during normal times and 100% in crises ones. We assume that all the loans are not withdrawn in cash till crisis comes in. This corresponds to parameter $k=$ 100% as for the (Huerta De Soto, 2006, p. 203) notations. When crisis occurs, bank clients run on the banks and withdraw their sight deposits. We keep in mind that most sight deposits in fractional reserve banking are actually funds allocated to the approved loans. This is also a stylized fact about borrower behavior in crisis. Lines of credit are typically sourced to maximum till the limits are reduced or even closed.

Our research strategy is as follows. We set seven banking system types depending on the prudential regulation format, see Table 2. Reserve requirement (RR) in the modern banking system is set to one per cent. This corresponds to the EU practice (https://www.ecb.europa.eu/explainers/tell-me/html/minimum_reserve_req.en.html). The reduction of RR to zero reflects the US and Morocco financial authorities step in 2020.

For now, we wish to study the banking system operation in the extreme scenario. On the one hand, we allow for sight deposits only. From another, in crisis we set the 100% sight deposit early withdrawal probability (PW). This is higher than in (Luck, 2020) case. The case corresponds to the Creditanstalt bank default and consecutive ‘bank run’ (deposit early and massive withdrawal) in Germany in 1931. At that time there was neither CAR, nor any RR requirement according to him. Interestingly, he shows that the volume of sight deposits merely not changed in crisis from February 1931 to November 30, 1931. Nevertheless, the amount of term deposits dropped mostly by 40% during two months from mid-May to mid-July 1931 (Luck, 2020, p. slide 7 out of 17).

However, we consider our situation of 100% bank run to be a feasible one for the presence of another recent historical example. In September 2008 one of the world-famous investment banks The Lehman Brothers went bust. This implied turbulence in both the world financial system, and the US economy. The Lehman bankruptcy implied a wave of bank runs over the US banks. At those day the trend was such that the entire volume of deposits could have been withdrawn in two weeks. This led to the decision to raise the deposit insurance from the USD 100k to USD 250k. This means that 100% deposits withdrawal is a feasible stylized feature of the modern banking crisis. We proceed with it.

Table 2

Considered prudential regulation regimes

#	Banking system acronym	Banking system description	RR	CAR minimum
1	Modern	Typical system for most of the economies today. There are fractional reserve and the minimum capital adequacy requirements. Reserve requirement relates to all deposits (sight and term ones)	1%	10%
2	100% RR	Full reserve banking; there is no CAR regulation. Full reserve is attributed to sight deposits only	100%	n/a
3	100% RR $+$ CAR	Full reserve banking augmented with the CAR requirement	100%	10%
4	Modern $+$ RR	Modern banking system with higher reserve requirement. The reserve requirement is set at the level to equal the entire volume of sight deposits (may deem it to be quasi-full reserve banking)	1%	10%
5	Modern $+$ CAR	Modern banking system with the enhanced resilience via the larger CAR requirement. This corresponds to Basel III novelties	1%	20%
6	Modern – RR	Modern banking when the reserve requirement is set to zero in crisis. This mirrors the US	1% (0%	10%
		financial authorities March 2020 step	in crisis)
7	Free Banking	The banking system that has no regulation at all, i.e., neither the mandatorily reserves (RR), nor the CAR	n/a	n/a

Notes: n/a – not applicable; RR – reserve requirement (per cent of the sight deposit amount to be withhold in cash, not granted as a loan). CAR – capital adequacy ratio (the amount of own funds divided by the risk-weighted assets); own funds stand for initial equity and retained earning subtracted by the provisions (share of loans granted); risk-weighted assets are the product of loan amount and the risk-weight (latter parameter we set equal to the general value of 100%).

We distinguish two crisis periods, see Table 3. The more severe crisis happens when the liquidity risk occurs together with the credit risk realization. This means that the actual default rate on loans exceeds the provision rate three times. During both times there is a sight deposit withdrawal.

Table 3

Crisis parameters

Period type	Crisis type	DR	PW
Normal	Cred $+$ Liq	5%	0%
Normal	Liq	5%	0%
Crisis	Cred $+$ Liq	15%	100%
Crisis	Liq	5%	100%

Note: DR – default rate; PW – probability of sight deposit withdrawal.

We differentiate two system types: closed and open. In the closed system banks can only go bust. New bank may enter the market in the open system only. We allow each banking system to perform for 200 days. A day is an artificial unit of measuring time. In reality, it may be a week, a month, a quarter, a year. First 100 days are normal times. At that time there is no deposit withdrawal. During the last 100 days a crisis occurs.

As a result, we have seven prudential regulation regimes from Table 2, two types of crises and two types of economic systems from Table 3. Thus, overall, we model 28 scenarios, see Table 4.

Table 4

The list of considered scenarios

System type	Crisis type	Free banking	Modern	Modern $+$ CAR	Modern $+$ RR	Modern $-$ RR	100% RR	100% RR $+$ CAR
Closed	Cred $+$ Liq	1	2	3	4	5	6	7
Closed	Liq	8	9	10	11	12	13	14
Open	Cred $+$ Liq	15	16	17	18	19	20	21
Open	Liq	22	23	24	25	26	27	28

We set the following common parameters for any type of scenario:

10 banks at the starting point.

100 units of money (capital) per bank at start.

Loss given default: LGD $=$ 100%

Risk-weight: RW $=$ 100%

Provisions $=$ 5%

To benchmark the various banking systems, we wish to observe the dynamics of the key banking system indicators. Those are the number of banks, amount of loans and capital. Though visual analysis is vivid and appealing, we proceed with statistical verification to provide accurate arguments. For this purpose we compute the volatility of these banking system indicators. We measure the daily changes (returns, growth rates) of indicators of interest and take their standard deviation. The less the standard deviation is, the less volatile and the more stable we deem the system to be.

One may argue that an ABM seems to be a mere simulation model. In some part it is indeed a simulation model. For instance, we preset the loan and deposit loans, i.e., we simulate those. However, ABM has the principal difference to the conventional simulations. We may call an ABM a ‘reloaded simulation with trimming’. From one side, ‘trimming’ means that at each day the prudential regulations are checked for each bank. If one fails to meet a criterion, it goes bust. An agent dies. It gets out of the simulation routine. This means that the next day the simulation relates to a new (a reduced) subset of survived banks. From another side, ‘reloaded’ means that we allow new banks to enter the market when we look at the ‘open banking systems’ (No. 15–28 in Table 4). This means that agents can be born. Thus, each new day the simulation relates to an enhanced subset of banks. Thus, the core of the system (the particular banks) naturally evolves. As applications for loans and deposits are allocated to banks randomly, some banks earn more profit and are able to easier pass the capital regulations. Same time such banks get larger. Now the system dynamics depends mostly upon them when looking at the total volumes of loans, capital etc. The ABM additionally allows banks to interact through the interbank market. However, due to baseline setting of the flat yield curve the differences in maturities are not that large to activate interbank borrowing and lending. Nevertheless, the interbank market is present and certain parameter combination may trigger its activation. To some extent, as we have shown in (Ermolova et al., 2021), this may prolong the existence of the banking system. That is why our ABM has important abovementioned advantages over the mere simulations, though for each individual agent (bank) the simulation might seem quite trivial. This is the major importance why complex systems are studied and various system-wide effects (like credit or systemic risk contagion) induced by simple individual actions are investigated.

4. Findings

We divide our findings in two parts. First, we focus on more vivid differences in visual dynamics of indicators. We will present results for the open banking systems. The crisis would have dual – credit and liquidity – features. The second sections compare the results for all the regimes statistically.

4.1 Visual analysis

Figure 1 presents the evolution of the number of banks. As we may see from Fig. 1, the modern banking systems allows more banks to be created in normal times compared to the full reserve banking. Latter one stands for the red dashed line. Nevertheless, the number of banks drop materially, when crises come in. It seems that around two banks are still alive. However, this fact need clarification. Because of massive bankruptcies in each day, these one or two banks are new. This means that the number of banks drops to zero each day. This implies the highest probability of new bank entry. This is how the open fractional reserve banking system is left with a couple of banks.

Figure 1.

The number of banks dynamics for the various open banking systems.

On opposite, the number of banks is stable in the full reserve banking. Those are always the same banks. Their assets shrink as the clients withdraw funds. However, this does not imply bankruptcies. Banks are still left with their capital.

Figure 2 shows the dynamics of loans. As we expected, the fractional reserve banking creates much more credit money. As a result, the volume of loans grows up to 25 k of artificial money units.

Figure 2.

The overall banking system loans dynamics for the various open banking systems.

Full reserve banking is not that generous in loans. Its volume at best reaches slightly above one thousand of money units (1.2 k, precisely). Please, refer to Annex 1 for granular details per periods. Evidently, the bank defaults lead to loan volume shrinkage in the fractional reserve banking systems. We may also see that the only implication of increasing CAR is limiting the credit money (loans approved and granted) during good times. Higher reserve requirement implies similar impact.

Let us look closer for the loans dynamics within the full reserve banking system. Figure 3 presents respective data series. We may deem such an inspection a sort of our robustness self-check. We may see that the amount of loans granted gradually rises even in the closed system (black colored series). This is the result of the accumulated profit. Larger own funds permit the bank to grant more loans. In the open system (red-colored series) the number of loans rises even more due to the new banks coming in. When crisis occurs from the 101 ${}^{\text{st}}$ step the volume of loans decreases. When the crisis is only of the liquidity nature, the drop in total loan volume is smaller. Whereas the loan volume decrease is much larger, when the liquidity risk is accompanied with the credit risk in crisis. This is the result of the bank losses. We set the default rate in credit and liquidity crisis equal to 15 pp. (please, refer to Annex 1 for details). Same time the interest rate margin is only 2 pp. If the banks raised the loan interest rates, they could earn the profits to offset losses.

Figure 3.

Full reserve banking system loans dynamics.

Let us look at the contribution of the CAR regulation to the full reserve banking system. It is marked in red color. Figure 4 shows a closed banking system. The full reserve requirement is more limiting than CAR. Thus, adding CAR does not change the lending pattern. This holds true for both open banking system and in crisis time. The crisis nature brings the only difference in loan volume dynamics.

Figure 4.

CAR does neither limit, not make the open full reserve banking system more stable.

To sum up, we observed that modern banking regulatory regime does not overcome full reserve banking one in crisis times, though the former overpasses the latter much in normal times. Enhancing CAR and RR does not significantly changes the system stability in crisis. It remains fragile. To some extent similarly adding CAR to the full reserve banking regulation does not change the pattern. Thus, we saw that the full reserve banking outpaces the modern one. We see that the von Hayek’s logic holds true in recent times when CAR regulation developed much wider than in Great Depression times. Nevertheless, we wish to obtain a more robust evidence. That is why we proceed with the banking system indicators’ volatility analysis.

4.2 Volatility analysis

We present Tables 5 and 6 with the volatility estimates for the number of banks and the total lending volume, respectively. Columns 5–11 stand for various banking systems. Rows 3–6 contain results only for the normal times from day 1 to 100. Rows 7–10 contain estimates for the following crisis period from day 101 to 200. Last rows 11–14 state the estimates for the pooled sample from day 1 to day 200.

We distinguish crisis types in column 4 and system openness type in column 3. We grouped the banking systems by similarity. First, there comes a free banking one. We did not present figures with it because of high volatility in crisis times. We look at several modifications of the fractional reserve banking systems. Their names start with ‘modern’. At the last two columns we bring two variants of the full reserve banking: without CAR requirement and with it.

Table 5
Volatility (st.dev.) of the banking system indicator: number of banks

Notes: Volatility is the standard deviation of daily returns (growth rates) of the indicator of interest; red color stands for the largest value for visibility; yellow – for the medium range values; no fill – for the smallest ones.

Table 6

Volatility (st.dev.) of the banking system indicator: overall banking system loans

Each cell has the volatility estimate for the chosen indicator, banking system, time period, and crisis type (if applicable). Volatility is the standard deviation of the following return series. Suppose we have a variable $X(t)$ , where ‘ $t$ ’ stands for the day count. Then the daily return equals to $r(t)=(X(t)/X(t-1))-1$ . Then we compute the standard deviation with respect to $r(t)$ . Thus the values in tables are relative ones and we may compare them with each other. The lower the volatility is, the more stable the system with the respective prudential regulation regime is. For the visibility we used traffic lights color coding. The red color stands for the highest volatility values, whereas the cells with no fill correspond to its lowest one.

The least volatile dynamics of the number of banks is in the full reserve banking systems, see columns 10 and 11 of the Table 5. It is comparable to alternative banking systems only in good times for the closed system, rows 3 and 4. We observe the most material unwelcomed dynamics during crisis times in closed banking system, rows 7, 8 and columns 5–9. All the banks disappear. Formally, as we already discussed, the number of banks jumps around two banks in the open system, rows 9–10. We remember that each day those are new banks as they cannot survive for longer. It is worth mentioning that higher CAR implies higher stability (lower volatility) of the bank number dynamics in crisis, rows 7, 8 in column 7.

Average results still show that modern banking demonstrates higher volatility in the number of banks dynamics, rows 11–14. Thus, we should prefer the full reserve banking to the modern one if we wish to have more stable banking environment in terms of the bank number.

When we look at Table 6 for the volatility of loan volume dynamics comparison, we come to mostly similar findings. The free banking system delivers the most volatile lending patterns, see column 5. Modern banking with extra CAR requirements (like those in Basel III) is less volatile both in normal and crisis times, see rows 3–8 in column 7. Adjusting reserve requirement downwards – as the US and Morocco did in 2020 – does not principally impact the lending volume dynamics. It is still larger than in the full reserve banking one.

When reviewing both Tables 5 and 6, we cannot see material differences in system stability due to adding CAR to the full reserve banking mode, see columns 10, 11.

Now we may evidence that the full reserve banking is not only more stable than the fractional one as von Hayek already portrayed. It is also more stable than the fractional regime with current and prospective (enlarged) CAR requirement.

5. Conclusion and discussion

Our research suggests that the financial system is more stable when the CAR regulation is in place compared to the situation when such regulation is not present. It limits the lending growth in good times and results in lower volatility in crisis ones. However, CAR regulation neither in its current form, nor in the extended one as proposed in Basel III does not make the banking system anywhere near the stability level reached by the full reserve banking system.

We do not have blind believe that the full reserve banking can prevent its agents from losses or defaults. Actually, we embed the possibility of such defaults in our agent-based model. As a result and a self-check we observe that when the credit crisis comes in, banks start losing their capital. We did not allow them to offset losses by raising lending rates or lowering the deposit ones. Former may have negative externality of adverse selecting the credit-unworthy borrowers. We focused on the baseline case. This means that if the credit crisis continued, the capital of our banks could get in to the negative territory at Figs 2–4.

What remains important to us is whether adding CAR regulation to the full reserve banking pays off. Though we do not neglect credit risk, we do not see any value in CAR regime. The full reserve turns out to be more limiting than the CAR minimum. We suspect that our claim might seem arrogant as many countries use CAR regulation. However, we already refereed to the cases of the regulation failures. In fact, this is an illusion that CAR together with deposit insurance system might assure the system stability. It never can because the deposit insurance system cannot guarantee for all the credit money created from nothing. This is why the Austrian economic school is not favored. When you have full reserve banking, you do not need neither capital, nor liquidity regulation, nor a deposit insurance system.

Our findings are relevant for the evolving discussion of the Central Bank Digital Currencies (CBDC), or stablecoins. Practitioners suggest that CBDC should have the very same prudential treatment like ordinary bank assets, i.e., there should be capital and liquidity requirements against CBDC, there should be deposit insurance system for CBDC-funded accounts. However, they still allow for the fractional reserves against CBDC. Such views one might have heard at the Ban of England Chief Economists’ Conference held virtually on July 12–14, 2021. Such prudential treatment does not solve the problem. If the CBDC’s are fractionally reserved, this means that the system stays as fragile as it was. However, when we come to a true stable coin like that of the Amsterdam florin of 16 ${}^{\text{th}}$ –18 ${}^{\text{th}}$ centuries, we have won twice. First, we have a stable banking system. Remember that the Amsterdam’s economy well by-passed the tulip bubble of 1630s. Yes, there were individual losses from betting against mostly no-value item. However, economy-wise it was no detrimental. Second, full reserve banking does not require neither banking regulation, nor deposit insurance systems. Thus, we may free up a lot of taxpayers’ money. However, many white-collars will lose their currently important jobs. Yes, we agree that our statement looks like an appeal that ‘the King has no clothes’. But we are unaware of how to draw the policy-makers attention. If we truly wish a solid financial system, we need full reserve banking. Thus, our findings are now about applying century-old ideas, but recalling the best of the humankind history achievements. By the way, such arguments were supported by a Nobel Prize for Friedrich von Hayek in 1974. Interestingly, (Huerta De Soto, 2006) presents the facts how von Hayek explain why Friedman was wrong, while Friedman could not prove von Hayek to be wrong. However, Friedman also received a Nobel Prize in Economics two years later, in 1976.

Full reserve banking requires two principal changes in our minds. First, we need to adapt to paying for the deposit, and not receiving an interest on it. We recommend referring to (Huerta De Soto, 2006) for a discussion on the origin of the ‘bank deposit’ item in the modern Civil Codes. Simply saying, it is an illegal contract as two parties treat it differently. A bank treats it as a debt (rent)-like contract, while a bank client – as a storage one. However, we always pay for the storage. The important thing today is that the appearance of the ‘bank deposit’ item made an option disappear from our menu list. Nowadays, we cannot find a purely storage account, and not a safe box. We hear a lot recently the discussion over the negative interest rates. This is not the payment we are talking about. The negative interest rates are the implication of the huge amount of the credit money credited. Those press the demand for financial assets and thus lead to negative rates. However, we are in need of pure negative rates when people consciously pay for the deposited funds. If one wishes to risk and earn money, well, please, invest into the debt contract with particular maturity. When such a person has no right to early withdraw, we are safe from liquidity crunches. Then we do not need Central Banks as lenders of last resorts. They may just print money.

Some people may be concerned with von Hayek’s ideas on the multiplicity of currencies. However, why should we be limited to a single currency? If you are afraid of fraud issues for any private currency, you should not just use it. Stick with the sovereign one. For comparison, remember a case when you visited an unknown city or country with quite different tastes. At some point in time, you might have dared trying local food and most of the time, it occurred tasty. Same time happens with money. People will gradually switch to the preferred currency. Their preferences would be a much more informative reflection of the issue creditworthiness, than the currently existing credit ratings.

Second, such a change to full reserve banking might happen with the change in the views of the government. (Huerta De Soto, 2006) explains that they were the ones who legitimized ‘bank deposit’ so that banks could buy new government debts using part of newly created credit money. We may only hope that there will be governments that do not use the credit money to largely expand government expenses. When this happens, we may have a financially stable world with permanent deflation. We should not be frightened of it as true deflation – not the one in Japan coming from the excessively created credit money – implies switch to capital-intensive technologies, increase in real wages and permanent technological progress.

Footnotes

Acknowledgments

We are grateful to two anonymous reviewers for the interesting comments that provoked additional discussion. We reflected it in the manuscript.

Opinions expressed do not necessarily reflect those of the affiliated institutions.

Annex 1. Zoomed In Dynamics During Normal and Crisis Times.

Figure 5.

The overall banking system loans dynamics for the various open banking systems.

References

Agliari

Böhm

, & Pecora

(2020). Endogenous cycles from income diversity, capital ownership, and differential savings. Chaos, Solitions and Fractals, 130. doi: 10.1016/j.chaos.2019.109435.

Aramonte

, & Avalos

(2020, July 01). Corporate credit markets after the initial pandemic shock. Retrieved from BIS Bulletin No 26: https://www.bis.org/publ/bisbull26.htm.

Ashraf

Gershman

, & Howitt

(2017). Banks, market organization, and macroeconomic performance: An agent-based computational analysis. Journal of Economic Behavior & Organization, 135, 143-180. doi: 10.1016/j.jebo.2016.12.023.

Biondi

, & Zhou

(2019). Interbank credit and the moneymanufacturing process: a systemic perspective on financial stability. Journal of Economic Interaction and Coordination, 437-468. doi: 10.1007/s11403-018-0230-y.

BKAM. (2020, June 16). Monetary Policy Report No. 55. Retrieved from Central Bank of Morocco Website: http://www.bkam.ma/en/content/download/707707/8204070/RPM-Q2-2020-ENG.pdf.

Blatt

J.M.

(1983). Economic policy and endogenous cycle. Journal of Post Keynesian Economics, V(4), 635-647.

Bohm-Bawerk

E.v.

(1886). Grundzuge der Theorie des wirtschaftlichen Guterwerths.

Cathcart

El-jahel

, & Jabbour

(2017). Basel II: An engine without brakes. Journal of Banking Regulation, 18(4), 359-74.

Catullo

Gallegati

, & Palesrini

(2015). Towards a credit network based early warning indicator for crises. Journal of Economic Dynamics and Control, 50, 78-97. doi: 10.1016/j.jedc.2014.08.011.

10.

Chan-Lau

(2017). ABBA: An Agent-Based Model of the Banking System. IMF Working Paper No. 17/136, 1-33.

11.

Colacchio

, & Davanzati

G.F.

(2017). Endogenous money, increasing returns and economic growth: Nicholas Kaldor’s contribution. Structural Change and Economic Dynamics, 41, 79-85.

12.

De Caux

McGroarty

, & Brede

(2017). The evolution of risk and bailout strategy in banking systems. Physica A: Statistical Mechanics and its Applications, 468, 109-118. doi: 10.1016/j.physa.2016.10.005.

13.

Dewatripont

Rochet

J.-C.

, & Tirole

(2010). Balancing the banks. Global lessons from the financial crisis. Princeton and Oxford: Princeton University Press.

14.

Ermolova

Leonidov

Nechitaylo

Penikas

Pilnik

, & Serebryannikova

(2021). Agent-based model of the Russian banking system: Calibration for the maturity, interest rate spread, credit risk, and capital regulation. Journal of Simulation, 15(1-2), 82-92. doi: 10.1080/17477778.2020.1774430.

15.

FRS. (2020, March 15). Federal Reserve Actions to Support the Flow of Credit to Households and Businesses. Retrieved February 10, 2021, from Board of the Governors of the Federal Reserve System: https://www.federalreserve.gov/newsevents/pressreleases/monetary20200315b.htm.

16.

Gabbi

Iori

Jafarey

, & Porter

(2015). Financial regulations and bank credit to the real economy. Journal of Economic Dynamics and Control, 50, 117-143. doi: 10.1016/j.jedc.2014.07.002.

17.

Gómez-Fernández-Aguado

Partal-Ureña

, & Trujillo-Ponce

(2014). Moving toward risk-based deposit insurance premiums in the European Union: The case of Spain. Applied Economics Letters, 46(13), 1547-1564.

18.

Guizani

, & Watanabe

(2016). The effects of public capital infusions on banks’ risk-shifting to thedeposit insurance system in Japan. Journal of Financial Stability, 26, 15-30.

19.

Hasumi

Iiboshi

, & Nakamura

(2018). Trends, cycles and lost decades: Decomposition from a DSGE model with endogenous growth. Japan and the World Economy, 46, 9-28.

20.

Haxholdt

Kampmann

Mosekilde

, & Sterman

J.D.

(1995). Mode-locking and entrainment of endogenous economic cycles. System Dynamics Review, 11(3), 177-198.

21.

Hernández de Cos

(2019, October 01). Post-Basel III: time for evaluation. Retrieved from Keynote address by the Chairman of the Basel Committee on Banking Supervision and Governor of the Bank of Spain, at the 14th ASBA-BCBS-FSI High-level Meeting on Global and Regional Supervisory Priorities, Lima: https://www.bis.org/speeches/sp191024.htm.

22.

Huerta De Soto

(2006). Money, Bank Credit, and Economic Cycles (2nd, translated from Spanish ed.). Auburn, Alabama: Mises Institute.

23.

Ito

, & Hoshi

(2020). The Japanese Economy. Cambridge, Massachusetts: The MIT Press.

24.

Lall

(2012). From failure to failure: The politics of international banking regulation. Review of International Political Economy, 19(4), 609-638.

25.

Lavoie

(2019). A System with Zero Reserves and with Clearing Outside of the Central Bank: The Canadian Case. Review of Political Economy. doi: 10.1080/09538259.2019.1616922.

26.

Luck

(2020). Micro-evidence from a system-wide financial meltdown: The German Crisis of 1931. 2nd Biennial Conference “Financial Stability and Regulation” October 22, 2020. Online via Webex: Banca d’Italia and Bocconi University.

27.

Minsky

(1982). Can “It” Happen Again?. New York: Routledge.

28.

Moosa

(2010). Basel II as a casualty of the global financial crisis. Journal of Banking Regulation, 11(2), 95-114.

29.

Penikas

(2015). History of banking regulation as developed by the Basel Committee on banking supervision in 1974–2014 (brief overview). Financial Stability Journal, 28(5), 9-47.

30.

Penikas

(2020). History of the Basel internal-ratings-based (IRB) credit risk regulation. Model Assisted Statistics and Applications, 15, 81-98.

31.

Pettifor

(2020). Rebuild the ramshackle global financial system. Economic researchers neglect the role of financialization in global. Nature, 582, 461.

32.

Poledna

Bochmann

, & Thurner

(2017). Basel III capital surcharges for G-SIBs are far less effective in managing systemic risk in comparison to network-based, systemic risk-dependent financial transaction taxes. Journal of Economic Dynamics and Control, 77, 230-246. doi: 10.1016/j.jedc.2017.02.004.

33.

Ponomarenko

, & Sinyakov

(2017, July). Strengthening Banking Supervision Implications for the Banking System Structure: Implications from the Agent-Based Modeling [in Russian]. Retrieved from Bank of Russia Working Paper Series No. 19: http://www.cbr.ru/content/document/file/16720/wp_19.pdf.

34.

Popoyan

Napoletano

, & Roventini

(2017). Taming macroeconomic instability: Monetary and macro-prudential policy interactions in an agent-based model. Journal of Economic Behavior & Organization, 134, 117-140. doi: 10.1016/j.jebo.2016.12.017.

35.

Raybaut

(2014). Toward a non-linear theory of economic fluctuations: Allais’s contribution to endogenous business cycle theory in the 1950s. European Journal on the History of Economic Thought, 21(5), 899-919.

36.

Reinhart

, & Rogoff

(2009). This Time is Different: Eight Centuries of Financial Folly. Princeton University Press.

37.

Riccetti

Russo

, & Gallegati

(2016). Financialisation and crisis in an agent based macroeconomic model. Economic Modelling, 52(A), 162-172. doi: 10.1016/j.econmod.2014.11.028.

38.

Roscher

(1854). Die Grundlagen der National-Oekonomi. Stuttgart: Gottascher Berlag.

39.

Samitas

Polyzos

, & Siriopoulos

(2018). Brexit and financial stability: An agent-based simulation. Economic Modelling, 69, 181-192. doi: 10.1016/j.econmod.2017.09.019.

40.

Shiers

A.F.

(1994). Deposit insurance and banking system risk some empirical evidence. The Quarterly Review of Economics and Finance, 34(4), 347-361.

41.

Shubik

(1987). A Game-Theoretic Approach to Political Economy. The MIT Press.

42.

Sunaga

(2017). Endogenous growth cycles with financial intermediaries and entrepreneurial innovation. Journal of Macroeconomics, 53, 191-206.

43.

Taleb

N.N.

(2007). The Black Swan. The Impact of the Highly Improbable. New York: Random House.

44.

Véron

(2020, March 25). Banks in the COVID-19 turmoil: Capital relief is welcome, supervisory forbearance is not. Retrieved from Peterrson Institute for International Economics (PIIE): https://www.piie.com/blogs/realtime-economic-issues-watch/banks-covid-19-turmoil-capital-relief-welcome-supervisory.

45.

von Hayek

F.A.

(1929). Geldtheorie und Konjunkturtheorie. Beitrage zur Konjunkturforschung, vol. 1, ed. Vienna: Springer; Osterreichisches Institue fur Konjunkturforschung.

46.

von Hayek

F.A.

(1974, December 11). Prize Lecture. Retrieved from NobelPrize.org. Nobel Media AB 2020: https://www.nobelprize.org/prizes/economic-sciences/1974/hayek/lecture/.

47.

von Mises

(1978). The Causes of the Economic Crisis And Other Essays Before and After the Great Depression. (J. Percy L. Greaves, Ed.) Auburn, Alabama: Ludwig von Mises Institute. Retrieved from https://cdn.mises.org/The%20Causes%20of%20the%20Economic%20Crisis,%20and%20Other%20Essays%20Before%20and%20After%20the%20Great%20Depression_2.pdf?token=V8MuCp_y.

48.

Wolski

, & van de Leur

(2016). Interbank loans, collateral and modern monetary policy. ECB Working Paper No. 1959.

49.

World Bank. (2020, June). Pandemic, Recession: The Global Economy in Crisis. Retrieved from Global Economic Prospects: https://www.worldbank.org/en/publication/global-economic-prospects.

50.

Zhong

(2013, December 20). Andrew Haldane: The Banker Who Cried ‘Simplicity’. Regulators need to watch for the next thunderstorm, not catch every raindrop, says the man who has changed the world financial debate. Retrieved from https://www.wsj.com/articles/SB10001424052702304858104579263190940438738.

Agent-based modeling for benchmarking banking regulation regimes: Application for the CBDC

Abstract

Keywords

1. Motivation

2. Literature review

3. Methodology

Table 1 Loan and deposit flow parameters

4.1 Visual analysis

Table 5 Volatility (st.dev.) of the banking system indicator: number of banks

Footnotes

Acknowledgments

Annex 1. Zoomed In Dynamics During Normal and Crisis Times.

References

Table 1
Loan and deposit flow parameters

Table 5
Volatility (st.dev.) of the banking system indicator: number of banks