Private Safe-Asset Supply and Financial Instability

​

This paper studies the private supply of safe assets by banks and evaluates its implications for financial stability. Banks originate loans, improve loan quality through hidden effort, and create safe assets by issuing debt backed by the safe payoffs from their own loans and a diversified pool of loans from all banks. The interaction between banks' effort and diversification decisions determines their safe-asset supply. In the context of incomplete markets, a free-rider problem arises: individual banks fail to internalize how their effort influences the ability to generate safe assets through diversification, as this depends on the collective effort of all banks. This market failure creates a novel inefficiency, that worsens as the scarcity of safe assets increases, leading to a backward-bending safe-asset supply curve. The public provision of safe assets helps mitigate this inefficiency by reducing their scarcity, but it cannot fully resolve it. Moreover, the impact on the total private supply of safe assets is ambiguous: public safe assets reduce incentives for diversification (crowding-out effect), which in turn increases banks' incentives to exert effort (crowding-in effect).​

​

Safe assets, capital flows, and macroeconomic outcomes

(with Dmitry Kuvshinov, Björn Richter, and Victoria Vanasco)​

​

What is the sectoral composition of the market for safety, and does it matter for economic stability? To address these questions, we construct a novel dataset of sectoral safe asset positions in 24 advanced economies since 1980. We document that the ratio of safe to total financial assets has remained stable in most countries, despite considerable growth in gross and net safe-asset positions relative to GDP. We find that fluctuations in safe asset positions are mainly driven by the financial and the foreign sectors, with the real economy playing a muted role, indicating that financials in advanced economies have been increasingly intermediating safety within and across borders. We conclude by showing that increases in safe asset demand by foreigners – or its counterpart, the supply by financials,– are associated with expansions in domestic risky credit and lower subsequent output growth.

​

A Theory of Bank Liquidity Requirements 

(with Charles W. Calomiris, Florian Heider and Marie Hoerova)

​​

We develop an asset-side theory of bank liquidity requirements, which focuses on the risk-management gains from requiring cash reserves. The key role of cash in the bank is to attenuate the banker's moral hazard.  Because cash is observable and riskless, its value does not depend on the banker's risk management effort. Greater cash holding improves bank incentives to manage risk in the remaining, non-cash, portfolio of risky assets. Because cash is ring-fenced from moral hazard, it serves as a form of collateral to bank creditors. We allow cash to be generated either initially from the funding of the bank or subsequently by selling risky assets in the market. But buyers of risky assets have limited aggregate resources, which generates a fire-sale discount on risky assets sold by the bank to generate cash. The fire-sale externality leads to a wedge between privately-optimal and socially-optimal liquidity holdings, requiring regulation of bank cash holdings to address the externality. More equity capital ex ante also improves risk management incentives but cash can be complementary to equity because it can be generated in the bad state, at the time when it is hard or impossible for a bank to raise more equity. Our theory has several implications for the design of liquidity regulation that are absent from existing regulation.

​

Understanding the Interplay between Safety and Liquidity 

​

The literature often takes for granted that safe assets are liquid, and even if this premise builds on a strong empirical foundation, assuming this relation always holds ignores important financial fragilities. I empirically assess the interplay between the safety and liquidity characteristics of an asset, by using COMPUSTAT data and bond level data provided by Refinitv to construct measures of default risk and liquidity at the bond level. Following the corporate bankruptcy literature, I use two measures of default risk: (i) Alltman’s Z score which is an index constructed using balance sheet data, and (ii) Merton’s probability of default, to calculate the market value of assets by viewing the observed equity price as a call option on the unobserved market value of the entire firm. Regarding the liquidity measure, I use the standard bid-ask spread. I study how the correlation between these two measures evolve over time, and with economic and institutional conditions. 

​