How Shadow Banking Reshapes the Optimal Mix of Regulation

Kinda Hachem

Decisions that are privately optimal often impose externalities on other agents, giving rise to regulations aimed at implementing socially optimal outcomes. In the banking industry, regulations are particularly heavy, plausibly reflecting a view by regulators that the relevant externalities could culminate in financial crises and destabilize the broader economy. Over time, the toolkit for regulating banks and bank-like institutions has expanded, as has banks’ restructuring of activities into shadow banking to lessen the regulatory burden. This post, based on our recent Staff Report, explores the optimal mix of prudential tools for bank regulators in a wide range of environments.

Our Model 

We start with a model in which banks use short-term liabilities to fund long-term assets. If an episode of market stress occurs, banks experience significant early withdrawals and may have to sell assets to outside investors at a fire-sale price. In anticipation of this, banks can choose to hold some cash in reserve and set a penalty (a “haircut”) that will be imposed on early withdrawals in stress periods. However, these choices take the fire-sale price as a given, engendering an externality: while each bank positions itself to sell fewer assets in the stress state, it fails to recognize that that choice will raise the sale price of assets in that state, enabling other banks to cover a given cash shortfall with fewer sales of their own. As a result, each bank holds less cash and imposes a smaller haircut than would be socially optimal, motivating the regulator to introduce floors on the fraction of bank assets held as cash (non-contingent regulation) and on the haircut that must be applied to withdrawals in stress periods (state-contingent regulation).  

We then expand the model to allow for shadow-banking technologies that banks can employ to sidestep the effects of regulation. One such technology allows banks to invest in more long-term assets without affecting the cash ratios on their balance sheets—for example, by moving funding into an off-balance-sheet vehicle that is outside the regulatory perimeter. The cost to a bank of this non-contingent shadow activity is a monetary incentive to short-term creditors (for example, a higher interest rate) to induce them to make the move. Another technology allows banks to impose less state-contingency on short-term creditors without affecting the contracted haircut—for example, by providing insurance to creditors in the form of credit lines that can be drawn upon in stress periods; outside of such episodes, these credit lines would not be fully recognized as loans on banks’ balance sheets. The cost to a bank of this state-contingent shadow activity is a capital charge once the loan is fully recognized. Since the cost of the state-contingent shadow activity is only incurred in stress periods, it directly increases the amount of money that banks need to raise in such conditions. As a baseline, the regulator is fully informed about the cost parameters of these shadow-banking technologies, but later we allow for the regulator to be less informed than banks about those costs. 

Finally, we introduce a bailout instrument that the regulator can deploy in the stress state to decrease the amount of money that banks need to raise through asset sales. Bailouts come with a direct social cost: diverting resources from the production of a valuable public good. They also involve indirect social costs: by propping up the sale price of assets, bailouts decrease the incentives of banks to hold cash and apply haircuts, which is the traditional moral hazard concern. 

Some Key Theoretical Findings 

Naturally, banks are more likely to undertake shadow-banking activities when they are not prohibitively costly. If these activities were too expensive to offer banks recourse from regulation, our research shows that it would be optimal for the regulator to rely more on state-contingent regulation than non-contingent regulation when the stress state is severe but unlikely, underscoring that the two forms of regulation are not perfect substitutes. 

If instead shadow activities are a feasible option for banks, then that constrains the design of regulation. Though beneficial to banks in the face of binding regulation, shadow activities are socially wasteful, so the regulator never finds it optimal to trigger them. Instead, each regulation is designed such that the marginal cost of any shadow activity exceeds its marginal benefit, ensuring that banks never undertake these activities. Our research finds that this condition will be more difficult to achieve for state-contingent regulation, since the marginal cost of state-contingent shadow activity is only incurred by banks in the stress state and at a price that neglects the externality. 

When the regulator is imperfectly informed about the cost parameters of shadow-banking technologies, it becomes much harder to design regulation that never runs the risk of triggering shadow activities. Instead, such activities may emerge as part of the regulated equilibrium implemented by the constrained optimal policy. We show that state-contingent shadow activity triggers a larger bailout than non-contingent shadow activity would trigger when the regulator cannot commit to a limited-scale bailout before the realization of the stress state. This reflects that the cost of state-contingent shadow activity directly lowers the sale price of assets in the stress state, prompting a larger bailout in the absence of a prior commitment regarding bailout scale.  

A larger bailout is not a panacea, of course, given the direct and indirect social costs noted earlier. Thus, faced with imperfect information and a limited capacity to commit to a smaller bailout, the regulator achieves lower welfare when uncertain about the cost to banks of state-contingent shadow activities than when uncertain about the cost to banks of non-contingent shadow activities. A regulator who ignores or greatly overestimates the cost parameters faced by banks also generates a larger welfare loss when naively using state-contingent regulation than when naively using non-contingent regulation; in addition, the regulator generates an amplification of welfare losses when both types of shadow activities are feasible for banks because the bailout that will be triggered by the circumvention of one form of regulation via shadow activities increases banks’ incentives to engage in shadow activities that circumvent the other form of regulation. 

A Cautionary Empirical Finding 

The emergence of shadow activities in response to non-contingent regulation has been well-documented in economics research since the 2008 financial crisis. In contrast, much less is known about shadow activities that undermine state-contingent regulation. While our contribution is primarily theoretical, we also present some empirical evidence regarding the circumvention of state-contingent regulation. 

The empirical setting we explore is the issuance of contingent convertible bonds, which receive favorable treatment under Basel III in many European countries, and the associated provision of credit lines, proxied from banks’ financial statements, as a potential form of insurance to investors against conversion. We find evidence that banks provide more credit lines when they issue more of these bonds, with price movements suggesting that the lines decrease the degree of state-contingency in the bonds. Thus, the financial stability threat posed by shadow activities extends to state-contingent regulation and should be closely monitored given our theoretical results. 

How to cite this post:
Kinda Hachem, “How Shadow Banking Reshapes the Optimal Mix of Regulation,” Federal Reserve Bank of New York Liberty Street Economics, July 16, 2025, https://libertystreeteconomics.newyorkfed.org/2025/07/how-shadow-banking-reshapes-the-optimal-mix-of-regulation/.