New research proposes to condense policy rates and balance sheet actions into a single implied short-term interest rate. To this end the term premium component of the yield curve is estimated and its compression translated into an equivalent change in short-term interest rates. This implied short-term rate can be deeply negative and allows calculating long time series of the monetary policy stance including times before and after quantitative easing. It is only suitable for large currency areas, however. Indicators of smaller open economies should include the exchange rate as well, as part of an overall monetary conditions index.

Pattipeilohy, Christiaan , Christina Bräuning, Jan Willem van den End and Renske Maas (2017), “Assessing the effective stance of monetary policy: A factor-based approach.”, De Nederlandsche Bank, Working Paper No. 575 November 2017.

The post ties in with SRSV’s lecture on macro trend indicators, particularly the section on creating economic indicators by using theoretical structure.
The below are excerpts from the paper. Emphasis and cursive text have been added.

The basic idea

“Before the [great financial] crisis, the stance of monetary policy was reflected in changes in the central bank’s main policy rate. However, against the backdrop of the effective lower bound…and the wide array of unconventional policy measures…the policy rate no longer accurately reflects the true overall stance of monetary policy…We present a single indicator for the overall effective stance…that combines both its standard interest-rate policies and non-standard balance sheet policies…In absence of balance sheet policies, the calculated indicator coincides with the three-month money market rate…[Else] we use observed long-term yields to construct an implied 3-month rate that is consistent with an unchanged term premium as of the moment that non-standard measures have been put in place.”

“The indicator can be interpreted as an implied short-term interest rate that is not restricted by the effective lower bound…The implied rate… can be thought of as an indicator for the effective stance of monetary policy…[and] provides a good gauge for the identification of non-standard monetary policy shocks.”

“The way we construct our indicator is…similar to the procedure underlying the definition of monetary conditions indices (MCIs), which are computed as the normalized weighted average of a short-term interest rate as well as an exchange rate.”

The gist of the method

Factor analysis is used to extract an expectations and term premium component from fitted yield curve data. Based on this, an implied short-term interest rate is constructed, which reflects how much the short-term rate should have fallen to achieve observed drop in long-term yields, assuming it could not have been caused by a fall in the term premium.”

“We extract an indicator from yield curve data only. In other words, we assume that all information on the monetary stance is priced-in in the term structure of interest rates.”

“Our analysis starts from a simple two-factor term structure model…The yield on a risk-free bond…is determined by the path of the expected short rates over the maturity of the bond, the expectations component, and a term premium component…Term premia compensate for interest rate or duration risk and represent deviations from the expectations hypothesis. In this sense, the term premium component of bond yields can be considered as being orthogonal to (i.e. uncorrelated with) expectations on future short-term interest rates (the ‘expectations component’).

“As the expectations component and the term premium are unobservable, one needs a methodology to extract them from the observed data…We use common factor analysis to extract both components from yield curve data. This is a very simple, a-theoretical and data-driven approach. The key consideration that factor analysis imposes…[is] that the expectations component and the term premium should be uncorrelated.”

“We…transform the yield curve into two time-varying factors (with factor loadings that are fixed over time) that we subsequently map in an economic context…Given the extracted factors, we proceed and construct a short-term rate that assumes balance sheet policies were not in place, and, as a consequence, term premia have remained constant. That is, we aim to assess by how much the 3-month rate should have fallen to achieve the observed drop in long-term yields, assuming it could not have been caused by a fall in the term premium.”

“We assume that balance sheet policies affect yields through the term premium as a consequence of portfolio-rebalancing effects, for example by reducing duration risk to be absorbed by market… By contrast, forward guidance is assumed to impact yields through the expectations component.”

The empirical findings

“The resulting mean implied 3-months rate is shown in [the figure below for the euro area] together with the spectrum of possible rates between the 70% and 90% confidence intervals (i.e. approximately two and three standard deviations around the mean implied rate, respectively)… Since the announcement of quantitative easing by the ECB in January 2015, the mean implied short rate has broadly moved within a range of -5% to -3%.”

“Non-standard monetary policy measures – summarized in the implied rate – have responded significantly to financial variables, suggesting that these instruments were geared to alleviating financial market stress.”

“We estimate a simple VAR model for monetary policy analysis where the implied short-term interest rate is included to identify monetary policy shocks. This experiment confirms that the implied rate performs better in identifying a non-standard monetary policy shocks than policy rates that are used traditionally in monetary policy analysis.”

“The effective stance of monetary policy is influenced to a large extent by monetary policy, but also by other factors…When a central bank engages in non-standard measures to push down longer-term interest rates the link between the formal monetary stance and the effective monetary policy stance may become less strong. The central bank has less perfect control over long-term rates relative to short-rates, since long-term rates can be considered to be dependent on factors not related to monetary policy. This includes, for example, investor risk aversion and exogenous shocks to demand and supply for safe-assets. As a consequence, the effective monetary stance becomes more susceptible to factors unrelated to the central bank policy. This is reflected in increased variability in the implied 3-month rate.”

“[Also] the implied effective stance of monetary policy can differ substantially when applying different methodologies.”