Financial openness can constrain the monetary autonomy of central banks. This column uses a new dataset to study how central banks use their balance sheets to absorb international monetary shocks since the late 19th century. It documents a systematic expansion of central bank domestic assets following international monetary shocks in financially open economies. By accounting for the role of central bank balance sheets, it sheds light on some of the puzzling behaviour of interest and exchange rates across international monetary regimes throughout history, as well as on recent studies of the global financial cycle.
Can countries have an autonomous monetary policy if they are financially open? This central question of international macroeconomics not only shapes our reading of the history of the international monetary system (Obstfeld and Taylor 2004). It is at the core of current policy debates and academic research on the transmission of US monetary policy to the rest of the world, potentially through its impact on the global financial cycle (Rey 2013, Kalemli-Ozcan 2020).
Our main contribution to this debate is to add a dimension not previously considered in the literature, namely the role of central bank balance sheets, and in particular, domestic assets (i.e. the purchase of securities or loans by the central bank). An expansion of central bank domestic assets pushes the domestic money market rate down and can thus counteract the effect of a rise of the international interest rate.
To test this hypothesis, we have constructed a dataset of monthly exchange rates, interest rates, and central bank balance sheets for 23 countries over the period 1891-2019 (Bazot et al. 2024). Collecting monthly central bank balance sheets for such long periods is a data contribution of its own. Note that the 23 central banks we study constitute essentially the entirety of current central banks which already existed by the late 1930s. This is supplemented with monthly data on industrial production, consumer prices, and stock market indices.
We have two main findings. First, we document a systematic expansion of central bank domestic assets in the face of international monetary shocks in financially open economies from the late 19th century to the present day. This expansion can be observed in both floating and pegged economies; in the latter case, the expansions exceed what is required for the purpose of sterilised foreign exchange interventions.
Second, examining the reaction of central bank balance sheets helps us understand how central banks manage to regain a degree of monetary policy autonomy in a financially globalised world. Even under pegged regimes, the interest rate pass-through remains around 30%. This sheds light on some of the puzzling behaviour of interest and exchange rates across international monetary regimes throughout history, as well as on the recent finding in the literature on the global financial cycle (i.e. Miranda-Agrippino and Rey 2020).
A new dataset of monthly central bank balance sheets for 23 countries from 1891 to the present day
Our study required building a dataset of monthly central bank balance sheets since the late 19th century. Given the variety of central bank operations and accounting rules in history, a challenging task was to standardise the balance sheets and to make them comparable across countries and over time while maintaining an appropriate level of disaggregation. Table 1 reports the main items on which this study focuses.
Table 1 Standardised central bank balance sheet
Assets classified under the broad category 1 are international assets that central banks can draw on to influence the exchange rate. Assets under the broad category 2 are financial operations that can affect the money market rate. We exclude categories 2.4 and 2.5 because they represent long-term loans directly extended to government or (state-owned) financial or non-financial companies. They do not affect interbank credit conditions. In other words, our focus is on the ‘liquid assets’ (either domestic or international) that central banks use at will to influence conditions on the foreign exchange and money markets.
This database is essential for several reasons. First, while Ferguson et al. (2023) have built a dataset of annual balance sheets to study the long-term impact of lender of last resort policies, monthly data are crucial for our exercise. The balance sheet movements we are interested in are unlikely to register in yearly data. In fact, the recent literature on the global transmission of monetary policy shocks finds that the effect on financial variables and exchange rates usually vanishes within six months (Miranda-Agrippino and Rey 2020, Bazot et al. 2019, 2022). Second, the data need to be sufficiently disaggregated, for the reasons discussed above. In most cases, the need for monthly and detailed asset categories required hand collecting previously unused data.
Estimations and main findings
We examine, by means of local projections, the short-term fluctuations in central bank assets, both domestic and international, to exogenous international interest rate shocks, drawing comparisons between different exchange rate regimes and degrees of capital account openness. This requires a good measure of exogenous monetary policy shocks of the leading country whose central bank interest rate influences international financial markets. Otherwise, the variation in the international interest rate could simply be an endogenous reaction to domestic and global economic fluctuations. When available, we use exogenous monetary policy shocks for England (1891-1913) and the US (post-1969) built by other scholars (Lennart 2018, Romer and Romer 2004, Bauer and Swanson 2023). For the remaining period, we have built our own monetary policy shock by using a mix of high-frequency identification (with daily data) and narrative approach, following the approaches of Bauer and Swanson (2023) and Cloyne et al. (2023).
Figure 1 below shows how the international and domestic assets and the interest rate of central banks respond to an exogenous monetary policy shock in the leading country (England before 1931, US thereafter). The figure shows the results for all countries and then distinguishes between floating and pegged economies. We see a positive response of the domestic liquid assets of central banks to an international monetary policy shock in financially open economies. As expected, international reserves react only in pegged economies, which is evidence of foreign exchange interventions. Importantly, the increase in total liquid assets goes beyond what is required by sterilised foreign exchange interventions.
In Bazot et al. (2024), we confirm these main conclusions by providing a battery of robustness checks, with alternative samples, definition of monetary policy shocks and of financial openness, and currency denomination of international reserves (to consider valuation effects). We also discuss at length how the reaction of the domestic assets of central banks shed light on the responses of central bank interest rates. In Figure 1, we see that floating countries manage to keep a stable interest rate but this requires an increase of the domestic assets of the central bank to inject liquidity on the money market rate. In pegged economies with an open capital account, the response of the interest rate to an international shock of 100 basis points is far from perfect: it is less than half of what is predicted by the trilemma of international finance (Obstfeld and Taylor 2004).
It is important to note that (as argued in Bazot et al. 2024) there would be no reason for the domestic assets of the central banks to react if uncovered interest rate parity (UIP) held. If UIP holds, in a floating exchange rate regime, all the adjustment works through the exchange rate, with no movement in the domestic money market rate and required reaction of central bank domestic assets. In a credible peg, if UIP holds, the target zone model (Krugman 1991) applies. And sterilised foreign exchange interventions are effective only if the UIP does not hold (Gabaix and Maggiori 2015).
Figure 1 Responses of central bank domestic liquid assets (in blue) and international liquid assets (in red) to an international shock, full sample (1891-2019), financially open economies
Note: Panel local projections using instrumental variable including 6 lags. Response to a change in the policy rate of the main central bank (BoE for 1891-1913 and 1924-1931, Fed discount rate for 1947-1971, Fed fund rate for 1973-2007, Fed shadow rate for 2007-2019) instrumented by a composite shock based on Lennard (2018) for 1891-1913, our own shock for 1924-1931 and 1947-1971, Romer and Romer (2004) for 1973-1987 and Bauer and Swanson (2023) for 1988-2019. Capital control classification is based on the Ilzetzki et al. (2019) financial openness index post 1947. Exchange rate classification from Ilzetzki et al. (2019). The responses of both domestic and international portfolios are in 12-month variation. The set of local projections also includes the domestic policy rate, the exchange rate, the world business cycle, monthly dummies, a time trend, and country fixed effects. Standard errors are clustered at country level. Error bands correspond to the 68% confidence (lower panels) and 90% confidence (lower and upper panels) intervals.
Historical analysis
Our data and methodology allow us to explore the role of central bank balance sheets in different countries and historical periods, such as the classical gold standard (first globalisation), the interwar gold exchange standard, the Bretton Woods system and the post-1990 second globalisation (for advanced economies or emerging markets). Two examples are given below.
During the gold standard before 1913, there was no restriction on capital flows and countries were pegged to gold. In Figure 2, we see a significant negative response of the international assets and a positive (stronger and more persistent) response of the domestic assets of central banks to a shock on the Bank of England interest rate. This helped to maintain the exchange rate within narrow exchange rate bands (gold points) and to avoid an overly large increase of the central bank policy rate (that does not exceed 20 basis points despite a 100 basis point rise of the English rate). To use Keynes’s famous words, central banks did not play the ‘rules of the game’. Instead of increasing their policy rate by 100 basis points and decreasing the money supply, they provided an elastic currency to the money market, creating money to increase their domestic assets.
Figure 2 Responses of domestic and international assets of the central bank and other financial variables during the classical gold standard (core countries), 1891-1913
Note: Panel local projections including 3 lags. Responses to the exogenous BoE policy rate shock of Lennard (2018). The responses of both domestic and international portfolios are in 12-month variation. The set of local projections also includes the UK business cycle, the UK stock market index, monthly dummies, a time trend, and country fixed effects. Error bands correspond to the 68% and 90% confidence intervals.
It is striking to see a similar increase in the domestic assets of central banks in floating exchange rate regimes today. In theory, the exchange rate absorbs all the international monetary shocks. The exchange rate indeed depreciates by 1% after a 100 basis point shock in the US interest rate. But – consistent with the existence of a global financial cycle and a UIP premium (Kalemli-Ozcan 2020, Miranda-Agrippino and Rey 2020) – this is not enough to insulate the domestic money market. The central bank thus injects liquidity to maintain a stable money market rate, in line with its policy rate. After two months, we even see that central banks decrease their policy rate as they react to the negative spillovers on output (as already observed by Miranda-Agrippino and Rey 2020). This is accompanied by the continuous increase in the domestic assets of the central bank. Central banks would find it difficult to transmit their rate changes to the domestic money market without injecting liquidity into the latter. In the gold standard, short-term changes to the central bank’s balance sheet rounded off the corners of the trilemma. They now play another role: that of managing the dilemma.
Figure 3 Second globalisation, floating countries without capital control, advanced economies, 1994-2019
Note: Panel local projections including 6 lags. Response to the exogenous Fed policy rate shock of Bauer and Swanson (2023). The responses of both domestic and international portfolios are in 12-month variation. The international portfolio valuation is set in special drawing rights. The set of local projections also includes the global real activity index, the countries’ industrial production indices, the countries’ price indices, monthly dummies, a time trend, and country fixed effects. Error bands correspond to the 68% and 90% confidence intervals.
Conclusion
Our analysis introduces central bank balance sheets (and especially domestic assets) into the literature on the trilemma, the global financial cycle, and the international transmission of monetary policy shocks. Our results pave the way for further research into why some central banks are more effective than others at stabilising the money market. From a more general point of view, our historical study sheds light on the raison d’être of central banks and explains why these institutions have become ubiquitous. By providing ‘elastic currency’ – to use an expression with a long history in central banking jargon stretching back at least to the US Federal Reserve founding charter of 1913 – central banks have demonstrated their ability to act as a cushion, or shock absorber, between the national economy and international financial markets.
Source : VOXeu
