In our recent blog series, we delved into the upcoming implementation of Basel Finalization in the United States, which may significantly increase the capital requirements for large banks. However, it is crucial to understand that funding credit to the U.S. economy with bank capital is generally more expensive than using deposits. As such, higher capital requirements may lead to a reduction in economic activity if banks respond to these increased costs by reducing their assets, such as lending activities or market intermediation.
The effect of higher capital requirements on economic growth has been examined in various academic studies. A recent review of the academic literature by the Basel Committee on Banking Supervision found that a 1-percentage-point increase in capital requirements could potentially reduce annual GDP by up to 16 basis points.[1] These findings suggest that higher capital requirements could have a significant impact on the U.S. economy, highlighting the importance of balancing financial stability and economic growth in regulatory decisions.
In addition, higher capital requirements may have the unintended consequence of incentivizing intermediation activities to move to the shadow banking sector, which is much less regulated and increases the risks nonbanks pose to financial stability. This transition of credit intermediation from banks to nonbanks can also amplify the cyclicality of credit supply, as nonbank lenders often lose access to funding and consequently scale back lending during economic recessions.
However, higher capital requirements can also reduce the likelihood of bank failure. For instance, if a bank receives a negative shock that decreases the value of its assets by 10 percent, it would still remain solvent with a capital ratio of 12 percent, compared to defaulting if it had only started with an 8 percent capital ratio. Nonetheless, the effect of capital requirements on the probability of bank failure decreases as the level of capital increases.
A crucial question arises as to what level of bank capital requirements would make the financial system more stable by offsetting the increase in risk of nonbanks compared to the reduction in the probability of bank failure. This question is particularly relevant in the U.S. context, as illustrated in the solid line in Figure 1, where nonbanks provide approximately 60 percent of the credit to the US economy.[2]
The topic of optimal capital requirements has received attention in academic circles, and Begenau and Landvoigt’s (2022) paper, or “BL” for short, is a welcome addition to the literature.[3] The authors explore the costs and benefits of tighter capital regulation in a financial system where both banks and nonbanks play a role in intermediating credit to firms. What sets BL’s paper apart is its explicit consideration of the interplay between banks and nonbanks, making it the focus of this post.
BL’s paper employs a quantitative general equilibrium model, a useful framework that captures the interactions between different economic actors over time. This model simulates how individuals and firms make decisions based on their preferences, constraints, and expectations of future outcomes, which can have feedback effects on the overall economy. It also helps us understand the long-term implications of policy changes.
The model examined by BL is comprehensive, encompassing both banks and nonbanks involved in the production of goods, a representative household that values consumption utility and deposits on both types of intermediaries, and a government. Banks in the model have insured deposits, but they must pay a fee and meet capital requirements. Nonbanks, however, hold uninsured and risky deposits and are vulnerable to bank runs. It’s important to note that both banks and nonbanks are allowed to fail, resulting in deadweight bankruptcy costs.
While BL’s model provides a rich framework for analyzing the behavior of banks and nonbanks in determining optimal capital requirements for banks, it also has simplifying assumptions that do not reflect reality. For example, it assumes that both banks and nonbanks hold the same type of assets, all deemed risky in the model. Additionally, it presupposes that all bank liabilities are insured, which is not the case in practice. Given these limitations, it is worthwhile to carefully examine these assumptions and their impact on the model’s findings.
When capital requirements are increased in the model proposed by BL, banks are unambiguously forced to reduce the amount of loans they can finance with deposits. This cuts into banks’ profits, since equity is more expensive than deposits, and causes investors to move their funds from banks to nonbanks. Consequently, capital requirements always lead to an increase in the size of shadow banks. However, the effect on the leverage of unregulated banks – and thus their risk of failure – is ambiguous and depends on the model’s calibration. On the one hand, an increase in capital requirements reduces the supply of liquidity from banks and raises demand for shadow bank liabilities as a substitute, which lowers the cost of shadow banks’ liabilities and so leads them to increase their leverage. On the other hand, the expansion of shadow banks increases their funding needs, which raises their debt financing costs, leading them to reduce leverage.
In the paper, the optimal level of bank capital requirements aims to balance the costs of bankruptcy against the provision of liquidity services. A low capital requirement results in high bankruptcy costs (banks and nonbanks) but ample liquidity for households, whereas a high capital requirement lowers bankruptcy costs but reduces the availability of liquidity for households. Based on their model calibration, the optimal capital level that maximizes welfare for both banks and nonbanks is around 16 percent, with shadow banks holding 33 percent of liabilities that are vulnerable to runs.
The paper examines two additional scenarios. In the first, banks are solely responsible for intermediating in the economy, and the optimal bank capital requirement is determined to be 15 percent. In the second scenario, shadow banks are not susceptible to runs and therefore hold a larger proportion of intermediated assets (42 percent), resulting in an optimal bank capital requirement of 14 percent. The lower optimal capital requirement in this case is due to the relatively larger size of the shadow banking sector, resulting in diminished benefits from imposing higher capital requirements on banks.
Finally, BL’s model defines bank capital requirements using the Tier 1 capital to assets ratio. However, the model assumes that banks only invest in risky assets, so in this case, total assets are equivalent to risk-weighted assets. Therefore, the relevant capital ratio for this model is similar to the Tier 1 capital risk-based ratio reported by banks. As shown in Figure 2, the current Tier 1 capital risk-based ratio for U.S. banks is nearly 14 percent. If we accept the paper’s conclusions, the optimal capital requirement for U.S. banks would be 16 percent, representing a 2-percentage-point increase from the current Tier 1 capital ratio.
Discussion
In their paper, BL use a quantitative general equilibrium model that employs sophisticated nonlinear methods to solve the model when banks are subject to a capital constraint. However, the authors have made simplifying assumptions to make the model tractable, which have significant implications for determining the optimal capital requirement of banks.
The first limitation of the study is its narrow definition of the shadow banking sector. It mainly focuses on the nonfinancial business sector, failing to account for significant credit provided by nonbanks. The model does not capture the rapid emergence of shadow banks in the U.S. residential lending market, which overstates the size of the banking sector’s role in providing credit to the U.S. economy and the impact of bank failure on welfare.
The second limitation of the paper’s model is its lack of recognition of post-crisis reforms that have significantly influenced bank behavior, making them safer and easier to resolve in bankruptcy. The model assumes that banks only hold risky assets, and all bank liabilities are insured, which is unrealistic. These simplifying assumptions lead to a higher optimal capital requirement than would be the case with more realistic assumptions.
Shadow Banking Sector. To fully understand the impact of nonbanks on the economy, the paper’s calibration of the shadow banking sector should be broadened beyond the “fragile” shadow banks identified by Gallin (2013) to include other nonbank lenders, asset-backed securities and investment vehicles. This narrow definition fails to capture the full impact of nonbanks on the economy, particularly the migration of bank-intermediated credit to nonbanks through the mortgage market and other nonbank lenders. A broader definition is necessary to fully appreciate the impact of nonbanks on the economy.
A larger nonbank sector would reduce the benefits of increasing capital requirements since the cost of bank bankruptcies on aggregate consumption would be smaller due to the smaller role banks play in the economy. For example, the paper’s analysis shows that the optimal capital requirements decrease from 16 to 14 percent when the size of the shadow banking sector increases from 33 to 42 percent.
Additionally, nonbank lenders tend to lose access to funding and thus contract lending significantly more than banks during periods of financial turmoil. This phenomenon has been observed in nonbank lending to small businesses (Ben-David and others 2022) and in the syndicated loan market (Aldasoro and others 2023), where nonbank lenders were unable to fund loans due to rising financing constraints. It has also been shown that shadow banks were less likely to provide a suspension of household debt payments during the COVID-19 pandemic (Cherry and others 2021). Moreover, the Financial Stability Board’s holistic review of the March 2020 market turmoil noted that the sale of U.S. Treasuries by leveraged nonbank entities exacerbated market dysfunction during the pandemic.
Post-Crisis Regulations. The paper by BL has two limitations in its treatment of banks’ balance sheets. First, the model assumes that banks only hold risky assets, which is not realistic. Since the implementation of Basel III liquidity requirements, banks are required to maintain a significant portion of high-quality liquid assets, which can be easily liquidated in a financial crisis. This reduces the potential magnitude of a bank’s losses during a crisis and the likelihood of the bank’s failure. Panel A of Figure 3 shows that banks hold approximately 25 percent of their balance sheet in extremely liquid assets such as reserve balances, U.S. Treasury securities, and agency MBS guaranteed by the GSEs.
Second, the paper assumes that all bank liabilities are insured deposits. However, only about half of deposits are insured, and large banks hold a significant amount of long-term debt that can be converted to equity in the event of a bank’s failure. This long-term debt is available to absorb losses in resolution, thus reducing the bankruptcy costs assumed in the model’s calibration. Panel B of Figure 3 shows that the eight U.S. global systemically important banks have a total of 16% of long-term debt relative to risk-weighted assets, yielding a total of 30 percent in total loss-absorbing capacity.
A more accurate calibration of the model would need to account for these regulatory differences between banks and nonbanks, and their respective bankruptcy costs. Incorporating the higher share of safe assets and long-term debt that can be converted into equity in the calibration of the model would result in a lower optimal capital requirement for U.S. banks.
Conclusions
To summarize, the paper by BL provides valuable insights on the impact of capital requirements on credit intermediation activity migration to nonbanks. However, the paper has limitations that need to be acknowledged to fully understand its implications for bank capital requirements. While quantitative general equilibrium models, like the one used in the paper, are useful in assessing the qualitative impact of policy changes on economic agents, they do not account for many important factors, making it challenging to rely on their quantitative findings.
This blog post highlights two main limitations of the analysis that the model does not capture effectively: (i) the size and impact of the shadow banking sector during economic recessions, and (ii) the stylized format of banks’ balance sheets and the absence of consideration for the significant amount of safe assets that banks hold in their balance sheets, as well as the long-term debt that can be converted to equity upon bank failure. By accounting for these factors, the model’s quantitative findings could be adjusted, providing a more accurate optimal capital requirement for U.S. banks.
[1] At the current GDP levels, this translates to an annual loss of around $40 billion and a cumulative loss of over $1.2 trillion over a 30-year period.
[2] The chart plots the ratio of nonbank assets to the sum of nonbank and bank assets. The definition of nonbank assets follows the one used in Michael S. Barr speech “Why Bank Capital Matters,” December 1, 2022. Available at
https://www.federalreserve.gov/newsevents/speech/barr20221201a.htm
[3] Examples of other quantitative general equilibrium models in the context of setting optimal capital requirements include Van den Heuvel (2008), Martinez-Miera and Suarez (2014), Clerc and others (2015), Begenau (2020), Elenev and others (2021), and Nguyen (2015). There are other approaches used to estimate the optimal level of capital requirements that do not rely on quantitative general equilibrium models, and one such approach is used by the Basel Committee on Banking Supervision in their 2010 LEI study.
References Cited
Aldasoro, Iñaki, Sebastian Doerr and Haonan Zhou, “Non-bank lending during crises,” BIS Working Papers, No. 1074, February 2023. (link)
Basel Committee on Banking Supervision, “An assessment of the long-term economic impact of stronger capital and liquidity requirements,” August 2010. (link)
Basel Committee on Banking Supervision, “The costs and benefits of bank capital – a review of the literature,” June 2019. (link)
Begenau, J. (2020). “Capital requirements, risk choice, and liquidity provision in a business-cycle model,” Journal of Financial Economics 136 (2): 355–378 (link)
Begenau, J. and Tim Landvoigt, “Financial Regulation in a Quantitative model of the Modern Banking System,” The Review of Economic Studies, Volume 89, Issue 4, July 2022, Pages 1748–1784. (link)
Ben-David, Itzhak, Mark Johnson, and René Stulz, “Why Did Small Business FinTech Lending Dry Up During the COVID-19 Crisis? NBER Working Paper No. 29205, November 2022. (link)
Buchak, G., G. Matvos, T. Piskorski, and A. Seru, “Fintech, regulatory arbitrage, and the rise of shadow banks,” Journal of Financial Economics, Volume 130, Issue 3, December 2018, Pages 453-483. (link)
Cherry, Susan, Erica Jiang, Gregor Matvos, Tomasz Piskorski, and Amit Seru, “Government and Private Household Debt Relief During Covid-19,” Brookings Papers on Economic Activity, Fall 2021. (link)
Clerc L., Derviz A., Mendicino C., Moyen S., Nikolov K., Stracca L., Suarez J., & Vardoulakis, A.P. (2015). “Capital Regulation in a Macroeconomic Model with Three Layers of Default.” International Journal of Central Banking 11 (3): 9–63. (link)
Elenev, V., Landvoigt, T., & Van Nieuwerburgh, S. (2021). “A Macroeconomic Model with Financially Constrained Producers and Intermediaries.” Econometrica 89 (3): 1361–1418. (link)
Financial Stability Board, “Holistic Review of the March Market Turmoil,” November 2020. (link)
Gallin, J. “Shadow Banking and the Funding of the Nonfinancial Sector,” FEDS Working Paper No. 2013-50. (link)
Nguyen, T. T. (2015). “Bank Capital Requirements: A Quantitative Analysis.” Fisher College of Business Working Paper No. 2015-03-14. (link)
Martinez-Miera, D., & Suarez J. (2014). “Banks’ Endogenous Systemic Risk Raking,” Manuscript. (link)
Van den Heuvel, S. J. (2008). “The welfare cost of bank capital requirements.” Journal of Monetary Economics 55 (2): 298–320. (link)