# The Importance of Safe Assets

A theme you often hear among bloggers, but a bit less so in seminars, is the idea that the supply of and demand for safe assets matter. David Beckworth is one such blogger who talks about this, but critics often find it hard to think about the macroeconomy in these terms since the role of money has been marginalized within the New Keynesian wing of macroeconomics. I say this because David’s intuitive explanation of safe asset equilibrium seems to be a cross between New Keynesian intuition and Old Monetarist intuition. He is trying to communicate his message to what is essentially the mainstream of the discipline, but by emphasizing something that isn’t generally in their models.

Along these lines, I was happy to stumble upon this paper by Caballero, Farhi, and Gourinchas. In my view this paper is quite similar to David’s views regarding safe assets and monetary policy and so I thought it might be interesting to outline the basic model in the paper and talk about the mechanisms for monetary policy.

The model is a modified version of an IS-LM model. The one modification to the model is a supply and demand condition for safe assets. Formally, the model consists of the following three equations:

$y - \bar{y} = -\delta (r - \bar{r}) - \delta_s (r^s - \bar{r}^s)$
$r^s = \max[\hat{r}^s + \phi(y - \bar{y}), 0]$
$s = \psi_y y + \psi_s r^s - \psi_{\Delta} (r - r^s)$

where $y$ is output, $r$ is the risky interest rate, $r^s$ is the rate on safe assets, $\hat{r}^s$ is the target interest rate, $s$ is the supply of safe assets, $\bar{y}$ is the natural rate of output, $\bar{r}$ is the natural risky interest rate, and $\bar{r}^s$ is the natural safe interest rate, and the greek letters are parameters. Inflation is assumed to be zero such that there is no difference between real and nominal interest rates.

This framework is a familiar IS-LM framework with the first equation is an IS equation, the second equation is a Taylor Rule subject to a zero lower bound, and the third equation determines the safe asset equilibrium.

The best interpretation of the safe asset equilibrium, as they describe it in the paper, is in terms of the flow of safe assets. According to this view, the flow demand for safe assets is a function of output, the rate of return on safe assets, and the risk premium ($r - r^s$). Thus the supply of safe assets, in this interpretation, is the net increase in the supply of safe assets.

Given that setup, let’s see what the model can tell us.

The first assumption that they make is that the supply of safe assets is unresponsive to the risk premium. In other words, in terms of the model, $\psi_{\Delta} = 0$. Given that many safe assets are exogenously supplied, this seems like a reasonable assumption.

Now, let’s think about the determination of the natural rate of interest. If the central bank sets the interest rate on safe assets equal to the natural rate, then output will be equal to potential (essentially by definition). It then follows from the IS equation that the risky interest rate is also equal to the natural risky interest rate. But how does one determine the natural interest rate?

Consider the equilibrium condition for safe assets. The interest rate on safe assets is the rate that exists when output is equal to potential. From the safe asset equilibrium condition it follows that

$\bar{r}^s = {{s - \psi_y \bar{y}}\over{\psi_s}}$

The central bank then needs to set $r^s = \hat{r}^s = \bar{r}^s$.

However, suppose that the net increase in the supply of safe assets is not high enough to keep up with the demand for new safe assets. In particular, suppose that the net increase in the supply of safe assets is so low that

$s < \psi_y \bar{y}$

In this scenario, the natural interest rate would be negative. However, from the Taylor rule, the market rate of interest is subject to a zero lower bound. As a result, the central bank cannot set the interest rate low enough to clear the market for safe assets. So what happens? Well, the central bank sets the safe interest rate as low as it can go $r^s = 0$. Which implies that output is pinned down by the net increase in the supply of safe assets:

$y = {{s}\over{\psi_y}}$

It then follows that $r > \bar{r}$. In other words, the risky interest rate is “too high” and the risk premium rises. But since the risky rate of interest is higher than the natural risky rate, the IS equation implies that output must fall in order to reduce the demand for safe assets and restore equilibrium.

The policy implication is that to escape this scenario, one needs to increase the supply of safe assets. By increasing the supply of safe assets, this increases output toward potential and thereby reduces the risk premium.

As the authors note, early attempts at quantitative easing in the United States did exactly what the model would prescribe because they swapped the risky assets in the market for safe assets. Fiscal stimulus can also help, but not through any sort of production done by the public sector, but because it increases the supply of safe assets (Treasuries).

### 4 responses to “The Importance of Safe Assets”

1. Now if you compare that to most car engines which are 25%-30% environment friendly, you see
that the RTA96-C is inexperienced (after a trend).

2. Steven J. Kelly

Reblogged this on The International Political Economy Hub and commented:
Here is a good primer from Josh Hendrickson of the University of Mississippi on the mechanics of global safe asset shortage—a recurring topic in my articles/posts.