RDP 2019-03: Explaining Monetary Spillovers: The Matrix Reloaded 2. Why Do Monetary Policy Spillovers Occur?

2.1 Spillover Channels

Yield curves can be in influenced by a range of domestic and international factors.^[3] In most financial systems, short-term market rates are dominated by central bank policy actions. Central banks' policy mandates and goals differ across countries, but most respond to macroeconomic conditions (in particular inflation, and often unemployment or the output gap) and, for some, exchange rate considerations matter as well. Central banks' control over long-term rates is usually significantly weaker under most monetary operating systems.^[4] Long-term government bond yields reflect not only current and expected short-term rates, but also various risk premia (such as term premia and in some cases, for example emerging markets, also credit premia). Based on these broad macroeconomic and financial determinants of short and long interest rates, we identify three potential channels through which spillovers can occur from interest rates in an originator economy to those in the recipient economy.

2.1.1 Domestic macroeconomic conditions

Monetary policy announcements (in the originator country) may reveal new information on economic conditions in that country, as suggested by Campbell et al (2012) and Nakamura and Steinsson (2018). This may in turn lead investors to update their expectations of macro conditions in the recipient country due to the various economic linkages between the two economies. Such interlinkages can result from trade flows, or can encompass a range of business and information flows that manifest themselves through co-movements in business cycles (e.g. Kose, Otrok and Whiteman 2003; Baxter and Kouparitsas 2005) and/or inflation dynamics (e.g. Ciccarelli and Mojon 2010; Neely and Rapach 2011).

2.1.2 FX regime

Spillovers can occur via a foreign exchange (FX) channel if a country pegs its exchange rate to that of a larger economy, either formally or implicitly (including arrangements such as a managed or ‘dirty’ float). If it has an open capital account, then the country implementing the peg will need to maintain interest rates close to those of the larger economy in order to avoid exceptionally large capital flows (e.g. Shambaugh 2004).

Changes in interest rates in the larger economy will then be reflected almost mechanically in the yield curve of the smaller economy at least through expectations of the domestic policy interest rates, even if the recipient country's central bank does not respond immediately. In effect, the country pegging its exchange rate virtually ‘outsources’ its monetary policy to the larger economy. Not only will this lead to a co-movement in short-term policy rates, but if the peg is credible and expected to persist, interest rates at all maturities will co-move. Even some countries with notionally flexible exchange rate regimes may want to avoid large exchange rate adjustments against a major currency, for example for trade competitiveness or financial stability reasons, and hence their policy rates may shadow that of the larger economy. Alternatively, they may intervene in the FX market to smooth the bilateral exchange rate. Even if such interventions are sterilised, local bond yields could still be affected through signalling and/or portfolio rebalancing effects.

2.1.3 Bond risk premia and financial conditions

With globally integrated capital markets, movements in term premia (and other possible risk premium components) in a large economy can drive those in other economies. This can occur, for instance, through the portfolio flows of international investors that are active in different countries' bond markets as they seek higher yielding assets, often described as a ‘search for yield’.^[5] Spillover effects can also arise due to the presence of global intermediaries and their relevant risk constraints (e.g. Bruno and Shin 2015; Malamud and Schrimpf 2018).

The intensity of these spillovers will depend on the degree of financial integration between the economies. This type of spillover, in particular if it operates independently of the exchange rate regime, also relates to the ongoing debate on the global financial cycle and the ‘dilemma not trilemma’ conjecture of Rey (2013, 2016). We return to this issue when we discuss the implications of our results.

2.2 Related Literature

This paper relates to several branches of the literature. Various papers examine how foreign asset prices respond to monetary policy shocks, although nearly all only consider interest rate changes by the largest central banks, the US Federal Reserve and/or ECB. Typically extant work also considers only a relatively narrow set of recipient countries (often emerging markets).^[6] A number of papers have documented interest rate spillovers to foreign bonds, notably Gilchrist, Yue and Zakrajšek (2014) and Andersen et al (2007).^[7] While most papers consider spillovers to (longer-term) bond yields, Edwards (2015) and Takáts and Vela (2014) find evidence of spillovers to short-term or policy rates although Devereux and Yetman (2010), Miyajima, Mohanty and Yetman (2014), and Obstfeld (2015) do not.^[8] Others have looked at interest rate spillovers in a broader context, noting there are net economic spillovers, for example Fukuda et al (2013), Ammer et al (2016) and Georgiadis (2016).

Our paper is also related to the recent literature on the international impact of QE. Many papers have found spillovers from the Federal Reserve asset purchases, including Neely (2010), Wright (2012), Bauer and Neely (2014), Rogers, Scotti and Wright (2016) and Fratzscher, Lo Duca and Straub (2018).^[9] In comparison with conventional monetary policies, Curcuru et al (2018) found that QE did not exert greater international spillovers. Other studies have also found that other major central banks' QE policies also triggered spillovers; Rogers, Scotti and Wright (2014) and Chen et al (2016) show that Fed, Bank of England and ECB unconventional policies affected foreign bond yields, although QE by the Bank of Japan did not. In contrast, Fratzscher, Lo Duca and Straub (2016) find that unconventional policies by the ECB had negligible effects on other countries' yields.

Some papers have gone beyond documenting international interest rate spillovers, and attempt to explain them. Two papers have a similar objective to ours. Hausman and Wongswan (2011) look at the effect of FOMC announcement surprises on short and long interest rates (for 20 countries). They use a fairly small number of explanatory variables to model the cross-section of responses, though, and study a sample period that ends before the financial crisis.^[10] Bowman, Londono and Sapriza (2015) examine what variables relate to the intensity of US unconventional monetary policy spillovers to emerging market sovereign yields, but they do not consider spillovers to advanced economies and focus on QE.^[11] The cross-section of responsiveness is modelled in a panel data framework with a broad set of country-specific controls. A number of other papers have found the intensity of spillovers to relate to various specific factors, including Shah (2018) (the level of interest rates), Aizenman, Chinn and Ito (2016) and MacDonald (2017) (degree of integration), Mishra et al (2014) and Ahmed, Coulibaly and Zlate (2017) (economic fundamentals for emerging market economies), Jotikasthira, Le and Lundblad (2015) (risk compensation) and Ehrmann and Fratzscher (2005) (monetary union).^[12]

Our paper improves upon this existing work by precisely identifying interest rate spillovers from a broader set of central banks (seven major advanced economies), not just the Federal Reserve, for both short- and long-term interest rates. A key feature of our work is to consider the full matrix of spillovers to a plethora of advanced and emerging market economies. This approach is sensible given the dense network structure of financial claims connecting different economies highlighted in Shin (2017). Crucially, we then put some structure on the transmission of spillovers by using a comprehensive dataset covering bilateral and aggregate economic and financial linkages. The goal of these empirical tests is to assess through which channels spillovers occur.

Footnotes

See Diebold, Piazzesi and Rudebusch (2005), Gürkaynak and Wright (2012), or Dahlquist and Hasseltoft (2013) for examples. [3]

A notable exception is the Bank of Japan has been implementing a target for long-term bond yields since 2016 based on flexible asset purchases, labelled ‘yield curve control’. [4]

This channel also relates to the risk-taking channel of monetary policy, as coined by Adrian and Shin (2010) and Borio and Zhu (2012). Bekaert, Hoerova and Lo Duca (2013) find that US monetary policy (measured via changes in policy rates) affects variance risk premiums based on the VIX, a common gauge for the global price of risk. [5]

Some papers also look at the spillovers to exchange rates or foreign equities, such as Wongswan (2006, 2009), Kim and Nguyen (2009), Ammer, Vega and Wongswan (2010) and Brusa, Savor and Wilson (2018). [6]

Other earlier contributions include Ehrmann and Fratzscher (2003), Forbes and Chinn (2004), Faust et al (2007), Craine and Martin (2008), and Ehrmann and Fratzscher (2009) for equity markets. [7]

While most papers typically use daily (and sometimes intra data), some others have looked at spillovers to foreign interest rates, or other asset prices, with lower frequency VARs combining monthly or quarterly macro data. In some cases, these papers impose a Taylor rule to attempt to separate common shocks from spillovers, which makes strong assumptions about the suitability of the Taylor rule for identification of spillovers, see for example Bredin, Hyde and Reilly (2010), Fukuda et al (2013), Hofmann and Takáts (2015), Han and Wei (2016) and Dedola, Rivolta and Stracca (2017). [8]

This literature builds on studies finding that QE compressed domestic long-term yields, for the United States see Gagnon et al (2011), Krishnamurthy and Vissing-Jorgensen (2011) and Swanson (2016), and also Christensen and Rudebusch (2012) for the United States and the United Kingdom, and Krishnamurthy, Nagel and Vissing-Jorgensen (2017) for the ECB. [9]

The variables they consider are: trade/GDP, trade with US/GDP, exports to US/GDP, share of equities owned by US, share of equity foreigners can own, total stock of bank lending form US/GDP, exchange rate regime, size of equity market/GDP. [10]

They find smaller spillovers for stock prices and exchange rates. [11]

Other studies have examined how the spillovers to equities and exchange rates in emerging markets relate to economic fundamentals, such as Aizenman, Binici and Hutchison (2016). [12]