Elsevier

Journal of Monetary Economics

Volume 125, January 2022, Pages 132-147
Journal of Monetary Economics

The supply and demand for safe assets

https://doi.org/10.1016/j.jmoneco.2021.07.010Get rights and content

Highlights

  • Safe assets are needed as stores of value and collateral.

  • Safe assets are provided both publicly (eg., government bonds) and privately (eg., ABS).

  • The supply of government bonds discourage information acquisition about the heterogenous underlying qualities of private safe assets, improving their safety.

  • Government bonds crowd out the creation of private safe assets, but crowd in their safety.

  • The optimal supply of government bonds should factor in the dual role for inter- and intratemporal smoothing and their impact on the quantity and safety of private assets.

Abstract

Safe assets are demanded as stores of value (to smooth consumption inter-temporally) and as collateral (to facilitate credit intra-temporally). Some are supplied publicly (government bonds) and some privately (asset-backed securities). Private assets are heterogeneous in quality, and information about their quality reduces their safety properties. We show that government bonds discourage both production of (crowding quantity out) and information about (crowding safety in) private assets. Hence, the optimal supply of government bonds need to take into account their dual roles and their impact on the quantity and informational content of private assets.

Introduction

Assets that display a stable value are considered “safe” and play an essential role in the economy. Agents demand them for two reasons: as stores of value to smooth consumption inter-temporally and as collateral to smooth consumption intra-temporally, for instance by facilitating credit and moving resources from nonproductive to productive agents. The supply of safe assets comes from two sources: public and private. Public safe assets are government bonds, promises that the government can make by relying on its taxation power. Private safe assets (such as asset-backed securities) take the form of promises that private agents can make by transforming otherwise riskier and potentially non-pledgeable private assets.

Issues surrounding safe assets have become increasingly pressing since the transformation of the financial system from a retail-based banking system to a wholesale banking system, which started in the late 1970s and seems permanent.1 In this new financial landscape, the use of U.S. Treasuries as safe assets have become more prevalent. Krishnamurthy and Vissing-Jorgensen (2012 and 2015) show that Treasuries have a convenience yield, arising from their safety property. Gorton et al. (2010), Xie (2012), Sunderam (2015) and Lenel (2020) further document that public and private safe assets are substitutes in terms of quantity: the private sector produces more private safe assets when the supply of Treasuries declines.

This paper highlights that public and private safe assets are also complements in terms of safety. Private assets are not perfect substitutes for public safe assets because they come in heterogeneous and volatile qualities. How closely private assets can substitute for public assets depends on their informational content. If there is no information about the differences and evolution of the assets’ qualities, they can be traded as if they were homogeneous and stable, becoming better substitutes for public assets. When such information is generated and revealed, their safety and their role as collateral and store of value declines, a financial crisis, in the words of Gorton and Ordonez (2014 and 2020b), if information is suddenly and involves large amounts and categories of private assets.

The information content of private assets is not exogenous and depends on the availability of public assets. When the stock of government bonds in the economy is high, agents rely less on the use of private assets to obtain credit and there are fewer incentives to examine their qualities. Safer private assets reduce the consumption variance of agents, in a way completing markets by replicating allocations that would be available with insurance contracts, making government bonds less relevant as safe assets. Indeed, as government bonds not only crowd out private assets but also informational financial frictions, their dual effect on the quantity and safety of private assets affects the extent to which such bonds are needed in the economy. Here we characterize the optimal supply of bonds that take into account these intricate interactions.

To build these ideas and identify the role of each element on the optimal provision of government bonds, we proceed in steps. We first explore the two roles of safe assets as stores of value and as collateral assuming that only the government supplies safe assets by issuing government bonds. Optimal provision should balance a trade-off: providing bonds to serve as collateral in a given period may distort consumption smoothing across periods. The optimal trade-off is achieved when the intertemporal distortion equalizes the value of collateral in relaxing borrowing constraints.

We then introduce the possibility to create private safe assets. At a cost, agents can transform non-pledgeable and perishable goods into pledgeable and non-perishable assets. Non-perishability allows the private asset to be used as a store of value and the pledgeability allows them to be used as collateral. This possibility reduces the need for the government to provide public safe assets in the optimum. In contrast to government bonds, private assets can potentially be of heterogeneous quality. This is relevant because information about such qualities in credit and asset markets introduces dispersion in investment scale and consumption (i.e., ex-ante risk) increasing the demand for bonds by risk averse agents, both as stores of value and as collateral.

We then endogenize information acquisition about private assets. We show such information is more likely to be produced when government bonds are scarce relative to the needs for collateral and relative to the average quality of projects to be financed and the average quality of private assets that are used as collateral. Furthermore, information acquisition may be sudden and may involve large amounts and categories of private assets, raising rapidly and drastically the need of bonds in the economy.

After characterizing these forces and interactions in a two-period setting, we study their dynamics in an overlapping generations environment, and make two points. First, we identify conditions under which providing the optimal quantity of bonds is not feasible, a policy trap: government bonds may be too abundant (their benefits do not compensate the distortions that their provision generate), but reducing their supply may induce information production about private assets, reducing safety in the economy and increasing the need for those bonds, making them too scarce. The possibility of facing this trap highlights the difficulty in specifying the optimal amount of government debt without taking into account the stability of private financial markets. The second point is that transitory shocks that reduces the supply of government bonds may increase the incentives to acquire information about private assets persistently, both through the effect of the shock on wealth and on the endogenous creation of private assets.2

Our results characterize the optimal provision of government debt, but they also have applications to all policies, fiscal and monetary, that affect the quantity of bonds that circulate in the economy. Central banks, for instance, conduct monetary policy with open market operations that exchange one kind of money, short-term Treasuries, for another kind of money, cash (numeraire in our setting) or quantitative easing programs that provide liquidity by purchasing long-term Treasuries with cash. Our work argues that, in conducting monetary policy, the Central Bank may inadvertently modify the informational content of private safe assets, inducing changes in credit conditions. Hence, macroprudential policy cannot be separated from monetary policy, contrary to the existing literature; see Svensson (2018) and Bernanke (2018)).

This insight has implications for the tools needed to conduct monetary policy. Standard monetary policies that exchange bonds for cash (numeraire) operate through the size of the central bank’s balance sheet (such as open market operations and quantitative easing), and may not be able to achieve both goals at the same time. If this is not possible, the government needs an additional tool that exchanges government bonds for private assets (not cash), changing the composition of the central bank’s balance sheet (a recent example was the Term Securities Lending Facility, operated by the Fed during the recent Financial Crisis; see Fleming et al. (2009)). We call this additional tool the “Bond Exchange Facility” (BEF), with a plausible implementation discussed in Gorton and Ordonez (2020a). While standard monetary policies should focus on implementing optimal intertemporal smoothing, the BEF is needed to guarantee that the safety of private assets is maintained. It may mean that it is optimal for macroprudential reasons for the Central Bank to maintain a balance sheet of enough size to implement BEF when needed.

We are not the first in recognizing that, on top of facilitating savings, government bonds can also relax borrowing constraints.3 Aiyagari and McGrattan (1998), for instance, study the role of bonds to improve risk-sharing among agents in an incomplete-market setting. They highlight that the cost of providing those bonds comes in the form of i) taxation that distorts incentives and the wealth distribution, and ii) crowding out of private assets. We instead abstract from the sources of distortions (which we introduce in reduced form), and explore more prominently how bonds have the additional role of enhancing the safety of private assets. How well private assets can substitute for bonds to save and to relax borrowing constraints depends on their informational content, and their informational content is determined not only by private assets’ fundamentals, but also by the supply of government bonds.

Other papers explore the intricate relation between the dual role of government bonds and the functioning of markets and the presence of other assets. Kocherlakota (2003) shows that government bonds, when illiquid (positive nominal return), allow agents with different intertemporal marginal rates of substitution not only to store value but also to realize additional intertemporal gains to trade. Infante and Ordonez (2020) study the roles of government bonds for saving and for risk sharing, and show that the effect of aggregate volatility on the extent of insurance in the economy critically depends on the composition of public and private safe assets. In contrast, we study the creation and informational linkages between government bonds and private assets when both can fulfill the two roles as stores of value and as collateral.

The paper proceeds as follows. In Section 2 we introduce the basic model with no information frictions, with two sources of demand for safe assets – store of value and collateral – and two sources of supply – public bonds and privately-produced assets. In Section 3 we introduce private assets of heterogeneous quality and information frictions, showing that they are only “safe” as long as no information about their quality is produced and revealed. In Section 4 we extend the model to an overlapping generations setting and study dynamics, showing the potential for a policy trap and for the need to combine monetary and macroprudential policies. Section 5 concludes.

Section snippets

Two period model without information frictions

In this section we introduce the two sources of demand for safe assets (store of value and collateral) and the two sources of supply (public and private). First, as a benchmark, we consider a simple setting with only public supply (government bonds) and a single role (store of value). Then, we add the second role (collateral) and show that the two roles interact – the use of bonds as collateral may increase or decrease their needs as store of value. Finally, we introduce assets of homogeneous

Two period model with information frictions

Now we assume that private assets come in two qualities. A fraction p¯ are good assets, with value ZG=κGZ>Z, and the rest are bad, with value ZB=0, with p¯κG=1, so the expected value of private assets is still Z.

Insights from overlapping generations

Above we used a two-period model to highlight how the optimal provision of government bonds should accommodate their dual role as store of value and as collateral and in affecting the creation and (informational) safety of private assets. Here we propose an overlapping-generation (OLG) extension to study the dynamics of private asset creation and safety in response to government bond changes. We convey two messages. First, the optimal policy in steady state may face a policy trap that requires

Conclusion

Economic agents demand safe assets to use as stores of value and as collateral. The main difference between public and private safe assets is that the latter may come in heterogeneous and volatile qualities. Here we show that information about individual assets’ qualities may reduce their safety and make them less useful for both purposes. When information acquisition is a choice, there are conditions (notably the availability of public safe assets) under which such information is not produced

Declaration of Competing Interest

None.

References (32)

  • Angeletos, G.-M., Collard, F., Dellas, H., 2016. Public debt as private liquidity: optimal policy. MIT Working...
  • A. Auerbach et al.

    Dynamic Fiscal Policy

    (1987)
  • R. Barro

    Are government bonds net wealth?

    J. Politic. Econ.

    (1974)
  • Bernanke, B., 2005. The global savings glut and the u.s. current account deficit. The Sandridge Lecture, Virginia...
  • B. Bernanke

    Should monetary policy take into account risks to financial stability?

    Speech in London, May 20, 2014

    (2018)
  • P. Diamond

    National debt in a neoclassical growth model

    Am. Econ. Rev.

    (1965)
  • Cited by (19)

    • The paradox of safe asset creation

      2023, Journal of Economic Theory
    • Wealth, endogenous collateral quality, and financial crises

      2022, Journal of Economic Theory
      Citation Excerpt :

      Our model implies that wealthier economies will have more severe but less frequent crises. In addition, our paper is also related to Gorton and Ordoñez (2022) who also exploit the effect of wealth on the lenders' incentive to acquire information and also find that wealthier lenders have less incentive to produce information. However, they do not study the firms' collateral quality decision in response to that effect.

    • The determination of public debt under both aggregate and idiosyncratic uncertainty

      2022, Journal of Economic Theory
      Citation Excerpt :

      Yet, such a dynamic consideration of the Ramsey planner is missing in steady-state welfare maximization.5 In a model with informational friction, Gorton and Ordoñez (2022) study the role of government debt for intertemporal insurance against aggregate shocks and the role of credit in providing insurance against idiosyncratic shocks, a topic closely related to our work. In contrast to our paper, their work emphasizes the role of information friction in the determination of optimal government debt.

    View all citing articles on Scopus

    We thank Narayana Kocherlakota, Moritz Lenel, Dejanir Silva and an anonymous referee for discussions and very thoughtful feedback. We also thank Fernando Alvarez, Marco Bassetto, Ryan Chahrour, V.V. Chari, Harold Cole, Igor Livshits, Andrew Postlewaite and seminar participants at Wharton, the Philadelphia Fed, the Barcelona GSE 2020 Summer Forum, the 2021 Carnegie-Rochester-NYU Conference on Public Policy, the 2021 SED Meetings and the 2021 NOVA Summer School Conference for comments. The usual waiver of liability applies.

    View full text