How do the costs of credit react to monetary policy? This classic question is relevant not only for monetary policymakers scrutinising the impact of their policy actions, but it can also provide guidance for choosing between competing macroeconomic models. Previous evidence on the impact of monetary policy on economic activity provided convincing support to conventional models with nominal rigidities (Christiano et al. 2005). In this column we describe our recent research (Gertler and Karadi 2015), in which we find empirical support for a ‘credit channel’ of monetary policy. In particular, we find that credit costs do respond more to a monetary policy tightening than what would be justified only by higher policy rates and a fair compensation for their altered riskiness. We conclude that this finding supports models where financing frictions play a key role in explaining business cycle fluctuations. Financing frictions are relevant because, for example, without them unconventional policies such as large scale asset purchases could be completely ineffective (Curdia and Woodford 2011, Gertler and Karadi 2011).
We start by setting up a flexible vector autoregression (VAR, Sims 1980) framework, i.e., a system of equations where all the observable variables are explained by their own and all the other variables’ lagged values. The framework proves particularly useful in capturing key dynamic relationships between macroeconomic variables. But further identifying restrictions are necessary to gain insights about the causal effects of monetary policy. The standard way is to impose restrictions on the timing of the impact of a monetary policy shock; one may assume that the monetary policy indicator responds contemporaneously to inflation and real activity, but inflation and real activity only responds with a period lag because of delays in information processing or decision making. This restriction is sufficient to identify a monetary policy surprise and its dynamic business cycle impact in a framework that does not include credit cost variables.
But such timing restrictions are not applicable to models with credit costs because measuring the impact of monetary policy on credit costs is hampered by the problem of simultaneity. In the relevant monthly or quarterly frequencies, not only credit costs respond quickly to monetary policy, but also monetary policymakers can be expected to set interest rates taking credit cost developments into consideration. This implies that the tight correlation between monetary policy indicators – like the federal funds rate – and credit costs – like corporate bond yields – does not imply causation, it might just be a reflection of monetary policy responding to credit costs developments or both reacting to factors omitted from the analysis. As a result, we cannot reasonably restrict the contemporaneous reaction of either variable to the other. Instead, we take a different route in our research. To answer our initial causal question, we utilise exogenous shifts in policy: occasions when monetary policy unexpectedly deviates from its systematic rule. Credit cost responses to these events can help us identify the one-way impact of monetary policy.
We have taken our cue from the finance literature. Previous research of Kuttner (2001) and Gürkaynak et al. (2005) has shown that federal funds futures markets, where participants essentially bet on the future development of the federal funds rate, incorporate new information included in the policy announcements of the Federal Open Market Committee within a very short time span, generally less than 30 minutes. To measure surprises caused by monetary policy announcements, we measure the changes in the federal funds futures rate from ten minutes before an announcements to 20 minutes after the announcement.1 With such a short window around announcements, it is highly unlikely that we include other, unrelated shocks into the measure. We, then, use this ‘high-frequency’ surprise measure as an external instrument in our VAR. The framework we follow (Mertens and Ravn 2013, Stock and Watson 2012) is an extension of the popular instrumental variable method to a dynamic VAR framework. The framework uses the instrument to identify exogenous variation in monetary policy to learn about its causal impact on other observables. The high frequency surprise measure satisfies the requirement to be a valid and strong instrument because it is correlated with the unobserved monetary policy shock, uncorrelated with any other unobserved shock, and explains a sufficiently high variability of our monetary policy indicator variable.
We use US time series data between 1979 and 20122, and include six variables in our baseline VAR specification: real activity is measured by the industrial production, the aggregate price level by the consumer price index, and as mentioned, we use one-year government bond rate as a monetary policy indicator; furthermore, we add three interest rate spread variables reflecting components of credit costs for three significant financial markets.
- The excess bond premium is a variable developed by Gilchrist and Zakrajsek (2012), and, roughly speaking, it is the component of spread between an index of corporate security yields and a similar maturity government bond rate that is left after the component due to default risk is removed. As such, it is possibly interpretable as a pure measure of the spread between yields on private versus public debt that is due to financial market frictions. It is relevant to the cost of long-term credit in the non-farm business sector.
- We also use the spread between conventional mortgage rates and similar maturity Treasury rates that is relevant to the cost of housing finance.
- Finally, we use the commercial paper spread that is relevant to the cost of short term business credit, as well as the cost of financing consumer durables.
Figure 1. Monetary policy shock with corporate and mortgage premia
Figure 1 shows the responses of the variables in the baseline to a representative contractionary monetary policy surprise. In each case, the panels report the estimated impulse responses along with 95% confidence bands. The one year rate increases roughly 20 basis points on impact and then reverts back to trend after roughly a year. Consistent with conventional theory, there is a significant and fairly rapid drop in industrial output that begins after several months and reaches a trough after roughly 18. Similarly consistent with standard theory, the CPI declines steadily, though this decline is not significant. Associated with the output decline is a significant increase, both statistically and quantitatively, in each of the credit spreads. The excess bond premium increases eight basis points on impact and remains at that level for roughly eight months before returning to trend. This increase in the excess bond premium following the monetary tightening is consistent with a credit channel effect on borrowing costs, it cannot be explained simply by an increase in bankruptcy probabilities since default premia have been cleaned off the measure. The mortgage spread increases only two to three points on impact but then increases sharply to seven points above trend after two months. Finally, the commercial paper spread increases roughly five basis points on impact for roughly four to five months. The parallel effects in multiple financial markets suggests that borrowers cannot escape the effects on longer term credit yields by switching to short term since the spread for short-term securities (the commercial spread) increases as well, as Figure 1 shows.
Figure 2. Monetary policy shock: Response of term premia and excess premia
We now turn to the question of what factors underlie the movement in the credit costs. To address this issue, Figure 2 presents a decomposition of the responses of various government and private interest rates to the monetary policy tightening. We add each government bond rate to our baseline VAR one at a time so as not to overburden the framework with multiple closely correlated variables. Each of the top four panels reports the response of a government bond rate along with the response of its respective term premium. We calculate the latter by subtracting from the nominal yield a ‘risk-free’ yield calculated from the estimated response of the federal funds rate.
As the figure shows, for the two-year, five-year and ten-year maturities, virtually all the rate increase is due to the term premium, with no impact of the path of expected short rates. Even for the one-year rate, most of the movement, roughly 80%, is due to a term premium effect. That the strong term premium effects arise is not surprising given the behaviour of short rates. Short rates quickly revert to trend following a tight money shock due to the weakening of the economy. It is true that the term premium effects dissipate quickly and are not statistically significant for the ten year government bond rate. Nonetheless, they are significant on impact for the five year rate, and at least three to four months for the one- and two-year rates.
The bottom two panels of Figure 2 report a similar decomposition of the response of private credit costs, except allowing for a component due to the movement in credit spreads. In particular, each panel reports the response of the combined excess premium for each security along with the overall response of the rate. It is first worth noting that the responses of the corporate bond rates and mortgages are relatively substantial. A roughly 20 basis point increase in the one-year government bond rate leads to a roughly 15 basis point increase in the corporate bond rate and seven basis point increase in the mortgage rate. In each case, virtually all the movement in rates is due the excess premium, defined as the sum of the credit spread and the term premium.
Using high-frequency monetary policy surprises as external instruments, we find that:
- Unanticipated monetary tightening produces a significant drop in real activity and a modest insignificant drop in the price level (in line with conventional models of monetary transmission).
But we also obtain some results that are inconsistent with the standard model. In particular:
- Monetary policy responses typically produce ‘modest’ movements in short rates that lead to ‘large’ movements in credit costs.
- The large movements in credit costs are mainly due to the reaction of both term premia and credit spreads.
The baseline model of the transmission mechanisms abstracts from both these considerations.
We conclude that to account for the overall response of credit costs, it may be necessary to amend the baseline macroeconomic models to account for movements in term premia and credit spreads. In related research, we provide an example of a framework (Gertler and Karadi 2013) in which monetary tightening can produce increase in both term premia and credit spreads, broadly consistent with the evidence in this paper.
Disclaimer: The views presented here are those of the authors and do not necessarily reflect the official views of the ECB or the Eurosystem.
Christiano, L J, M Eichenbaum, and C L Evans (2005), “Nominal rigidities and the dynamic effects of a shock to monetary policy”, The Journal of Political Economy, Vol.113 (1): 1-45.
Curdia, V, and M Woodford (2011), “The central-bank balance sheet as an instrument of monetarypolicy”, Journal of Monetary Economics, Vol. 58(1): 54-79.
Gertler, M, and P Karadi (2011), "A model of unconventional monetary policy", Journal of Monetary Economics, Vol. 58(1): 17-34.
Gertler, M, and P Karadi (2013), "QE 1 vs. 2 vs. 3...: A framework for analyzing large-scale asset purchases as a monetary policy tool", International Journal of Central Banking, Vol. 9(1): 5-53.
Gertler, M and P Karadi (2015), “Monetary policy surprises, credit costs, and economic activity”, American Economic Journal: Macroeconomics, Vol. 7(1): 44-76.
Gilchrist, S and E Zakrajsek (2012), “Credit spreads and business cycle fluctuations”, The American Economic Review, Vol. 102(4): 1692-1720.
Gurkaynak, R S, B Sack, and E T Swanson (2005), “Do actions speak louder than words? the response of asset prices to monetary policy actions and statements”, International Journal of Central Banking, Vol. 1(1): 55-93.
Kuttner, K N (2001), “Monetary policy surprises and interest rates: Evidence from the fed funds futures market”, Journal of Monetary Economics, Vol. 47(3): 523-44.
Mertens, K, and M O Ravn (2013), “The dynamic effects of personal and corporate income tax changes in the United States”, The American Economic Review, Vol. 103(4): 1212-47.
Sims, C (1980), “Macroeconomics and reality”, Econometrica, Vol. 48(1): 1-48.
Stock, J H and M W Watson (2012), “Disentangling the channels of the 2007/09 recession”, Brookings Papers on Economic Activity, Spring 2012.
1 In particular, we are using the surprises to the 3 month ahead federal funds rate expectations.
2 The results are robust to excluding the financial crisis period after the Lehman bankruptcy.