Implementing Swiss Monetary Policy: Steering the 3M-Libor with Repo Transactions

Transcription

1 Implementing Swiss Monetary Policy: Steering the 3M-Libor with Repo Transactions THOMAS J. JORDAN* and PETER KUGLER** JEL-Classification: E52, E58 Keywords: Repo Auctions, Interest Rate Targeting, Interest Rate Volatility 1. Introduction After 25 years of monetary targeting, the Swiss National Bank (SNB) adopted a new monetary policy framework at the end of Severe shocks to the demand for central bank money, especially for large denominated bank notes and for reserves held by commercial banks at the SNB, rendered it impossible to use the medium-term target path for the seasonally adjusted monetary base as a guideline for monetary decisions. Since also the demand for the broader money aggregate suffered from an insufficient stability, the SNB decided to abandon monetary targeting. The new framework consists of three elements. The first element is an explicit definition of price stability. The SNB regards price stability as achieved if CPI inflation is below 2 percent. The second element consists of the use of an inflation forecast as the main indicator to guide monetary policy decisions. The third element is a target range for the 3M-Libor as an operational target to implement monetary policy. Money aggregates continue to be important, but they are used as information variable rather than as intermediate targets. As in the old concept, maintaining price stability over the medium term remains the main objective of monetary policy also in the new framework. * Swiss National Bank and University of Bern, Börsenstrasse 15, Postfach, CH-8022 Zürich, Phone: , Thomas.Jordan@snb.ch ** WWZ/University of Basel and Swiss National Bank, Petersgraben 51, CH-4003 Basel, Phone: , Peter.Kugler@unibas.ch This paper benefited from helpful comments by Fleming Würtz, Alfred Günder and Antoine Veyrassat as well as participants of the CFS workshop Implementing Monetary Policy, Frankfurt am Main, November and of the Annual Congress of the Swiss Society of Economics and Statistics, Basel, March 18/ , Basel. Schweizerische Zeitschrift für Volkswirtschaft und Statistik 2004, Vol 140. (3)

2 382 JORDAN / KUGLER The SNB strategy shares some elements with inflation targeting. However, it also differs from it in some important respects. The strategy has no inflation target. Rather, the SNB s concept knows a definition of price stability. The SNB has no obligation to keep inflation under all circumstances and all costs in the range of price stability. Also, the time horizon to bring inflation back in the range of price stability after an inflationary shock is not pre-specified. The SNB analyses each situation individually and decides depending on the current economic conditions. Contrary to countries pursuing an inflation targeting strategy, the SNB has great independence regarding the exact definition of price stability and the policy reaction if inflation is outside the objective. The way monetary policy is implemented within the new framework differs from the procedures of most other central banks. The SNB needs some flexibility in the money market in order to react to exchange rate and other shocks in the very short run without having to declare a change in monetary policy. It thus chose a target range for the 3M-Libor rather than a point target for the overnight rate. This longer-term interest rate is not determined by the SNB directly, but it reflects the conditions in the international Swiss franc money market. The SNB influences these conditions by providing liquidity to the money market through repo transactions. Due to its longer-term character, the 3M-Libor may also be influenced by expectations of the future course of monetary policy. The possibility to steer the 3M-Libor in a sufficiently exact way is thus crucial for the implementation of Swiss monetary policy. The role of the inflation forecast in the SNB policy was studied in a couple of empirical papers (JORDAN and PEYTRIGNET, 2001; JORDAN, KUGLER, LENZ and SAVIOZ, 2002; KUGLER and RICH, 2002; KUGLER and JORDAN, 2004). However, much less empirical work was done with respect to the new operational procedure of SNB. In fact there is only a single study by VEYRASSAT (2001) which considers the influence of repo rates on the 3M-Libor. The purpose of this paper is to fill this gap. It provides an empirical analysis of the effect of the characteristics of the repo auctions on the conditional mean and variance of the Libor of different maturities using daily data from January 2000 to September The remainder of the paper is organized as follows. Section 2 explains the repo auctions of the SNB. In Section 3, we present the result of our econometric analysis and Section 4 concludes.

3 Implementing Swiss Monetary Policy The Repo Transactions for Implementing Monetary Policy The SNB has made an enormous effort in the last couple of years to create a highly efficient repo market operating on an electronic trading platform in Switzerland. In a repo transaction, the cash taker sells its own or borrowed securities to the cash provider. At the same time, it is agreed that the cash taker will repurchase securities of the same type and quantity from the cash provider at a later date. The cash taker pays the cash provider interest. From an economic point of view, the repo is thus a secured loan. There is a repo interbank market and a market for repo transactions of the SNB with the commercial banks. The SNB uses the repo platform not only for monetary policy operations but also for providing intraday liquidity to the banks for payment settlement purposes. Since the beginning of 2000, the SNB has been implementing its monetary policy almost exclusively through standardized repo transactions. The SNB aims at adequately providing liquidity to the Swiss money market with repo transactions in order to steer the 3M-Libor. It can prevent an undesirable rise in the 3M- Libor rate by supplying the banks with additional liquidity at lower repo rates. Conversely, by reducing the level of sight deposits or increasing repo rates, the SNB induces an upward interest rate movement. The liquid funds of commercial banks in Swiss francs consist largely of sight deposits held at the National Bank. The commercial banks demand for sight deposits derives mainly from statutory liquidity requirements that have to be fulfilled on average over a monthly maintenance period. Since the introduction of intraday liquidity, the demand for sight deposits stemming from interbank payment transactions has all but ceased, notably in the case of the large banks. The statutory liquidity requirements lead the banking system into a structural liquidity shortage. The SNB executes repo transactions for monetary policy purposes with a daily operation frequency via the electronic repo trading platform. In principle all domestic banks may participate in these repo auctions. Furthermore, foreign banks that are fulfilling some requirements may also be accepted as participants at the auctions. In normal circumstances, the maturity of the repo transactions ranges from one day to a few weeks (overnight, tom next, spot next, 1, 2, and 3 weeks). At 9:00 am on an operation day, the SNB announces the conditions of the intended repo auction (repo rate and maturity on several electronic market information services). The duration of the auction is 10 minutes. During that time, the banks can submit their bids. Immediately after the auction is closed, the SNB decides on the allotment of liquid funds. The announcement of the allotment to the individual banks takes place at about 9:20 am. Regarding the steering of the

4 384 JORDAN / KUGLER 3M-Libor, it is important to note that the SNB defines the repo rate and liquidity allotment before the Libor is fixed. The Libor is usually known at 12:30 pm Swiss time. The SNB executes fixed rate tender auctions. In a fixed rate tender, the SNB specifies in advance the repo interest rate and the participating banks bid the amount of money they want to transact at the fixed rate. The SNB adds all bids together. It then decides how much liquidity it will inject into the market. These funds are then allotted proportionally to the size of the bids after an allotment of a minimum amount to each bank. Until very recently, the SNB almost never allotted the total sum of all bids. It rather determined the amount of the allotted funds based on a forecast of the banks need for liquidity. More recently the SNB started to allot close to 100% of all bids. The SNB has no fix auction calendar. Although it operates almost daily in the market, the SNB is not committed to executing daily auctions. The maturity of the auctions may vary often. Furthermore, the repo rates and the allotment quota may vary as well without being a policy signal. Some information on the development of short-term interest rates in Switzerland and on the SNB repo auctions is provided by the following figures. Figure 1 displays daily data for the 3M-, 1M- and 1W-Libor as well as the SNB target range for the 3M-Libor since the January With the exception of late January and early February 2000 the SNB attempted to keep the 3M-Libor close to the center of the target range. The current range, which is valid since March 2003, goes from 0 to 75 basis points with a target value of 25 basis points. Figure 1 shows that the SNB could always keep the 3M-Libor within the target range and that it was most of the time close to the center of the band although the SNB changed the target range ten times in the last four years. Moreover, since the beginning of 2001 even the Libor rates with the shorter maturities are within the target range. Indeed this suggests the SNB was quite successful in implementing its monetary policy by the executed repo auctions. In particular there were no problems with underbidding and overbidding as they happened at other central banks in particular at the ECB. 1 It is sometimes argued that the good performance of the SNB s operating procedure is caused by the lack of minimum reserve requirements in Switzerland. This argument is not valid since the Swiss commercial banks are subject to liquidity requirements which have essentially the same 1 For a survey of central bank experience with repo auctions and discussion of the problem of underbidding and overbidding the reader is referred to BINDSEIL (2004). Published analyses of these problems include AYUSO and REPULLO (2003), BREITUNG and NAUTZ (2001), GASPAR, PEREZ-QUIROS and SICILIA (2001) among others.

5 Implementing Swiss Monetary Policy 385 effect as conventional minimum reserve requirements. However, unlike other central banks, the SNB has a great flexibility with respect to all features of its repo auctions, namely with respect to maturity, frequency and allotment. This flexibility is a conceivable candidate for the seemingly good performance of the SNB in controlling short-term interest rates. Before we turn to an econometric analysis of this issue, we will briefly present some information on the SNB repo auctions since January Figure 2 and 3 display the daily time series for the total allotment ratio and its frequency distribution 2. These figures show that the SNB offered repos on most days in our sample. On average total allotment is equal to 43 percent and in approximately 9 percent of the cases full allotment was provided. Finally, frequency distribution of the share of one-day repos on total allotment displayed in Figure 4 is small (21 percent on average) and in clearly over half of the cases we have no daily one-day repos. Thus, SNB provides most liquidity by repos with maturities of one to three weeks. Figure 1: Short-term Swiss Franc Interest Rates and the Target Range of Monetary Policy Daily Data January 2000 September 2003: Upper Bound (ZBLIB3O ), Lower Bound (ZBLIB3U) of 3M-Libor Target Range 5 a) 3M Libor (LIBOR3MI) 4 ZBLIB3O 3 ZBLIB3U LIBOR3MI The allotment ratio is set to zero when no auction takes place.

6 386 JORDAN / KUGLER 5 b) 1M Libor (LIBOR1MI) 4 ZBLIB3O 3 ZBLIB3U LIBOR1MI ZBLIB3O c) 1W Libor (LIBOR1WI) 3 ZBLIB3U LIBOR1WI

7 Implementing Swiss Monetary Policy Figure 2: Total Allotment Ratio for the SNB Repo Auctions, Daily Data January 2000 to September 2003 ZS Figure 3: Frequency Distribution of Total Allotment Ratio, Daily Data January 2000 to September Series: ZS Sample 2/01/2000 9/23/2003 Observations: 951 Mean Median Maximum Minimum Std. Dev Skewness Kurtosis Jarque-Bera Probability

8 388 JORDAN / KUGLER Figure 4: Frequency Distribution of Share of Allotment to One-day Repos on Total Allotment, Daily Data January 2000 to September Series: ZPDY Sample 2/01/2000 9/23/2003 Observations: 951 Mean Median Maximum Minimum Std. Dev Skewness Kurtosis Jarque-Bera Probability The Effect of Allotment Ratios and Maturity Structure of Repo Auctions on the Adjustment Speed and Volatility of the Libor In this section, we present an econometric analysis of the influence of characteristics of the SNB s repo auctions on the conditional mean and variability of Swiss Franc Libor of different maturities. To this end, we estimate the following basic model for the change in the Libor (r) depending on the deviation from target (r*) and the total allotment ratio (ZS) as well as the maturity structure of the allotments (ZPDY), i. e. the share of the allotment to one-day repos on total allotment. * rt = α+ ( β0 + β1zst + β2zpdyt )( rt 1 rt 1 ) + εt ε t : N( 0, h t ) 2 h = c + aε + bh + c ZS + c ZPDY t 0 t 1 t 1 1 t 2 t The model has an error correction structure in the conditional mean equation with a variable adjustment speed, which depends on the total allotment ratio and

9 Implementing Swiss Monetary Policy 389 its maturity structure. These two variables also enter the GARCH conditional variance equation as regressors. By this specification we are able to estimate the effect of the characteristics of the repo auctions on the adjustment speed of the Libor and on its volatility. Note that the current values of ZS and ZPDY are realized three hours before the current Libor is fixed and are thus predetermined with respect to the right hand side variable. The conditional mean equation is augmented with dummies for the dates of changes in the target range 3. Moreover, lagged interest rate terms are included if it was necessary in order to obtain white noise standardized residuals. The ML estimates of the parameters of interest using the daily data plotted in Section 2 are reported in Table 1. Besides the 3M-Libor target rate, we present results for the 1M- and 1W-Libor. The two shorter-term interest rates are included as they are probably less influenced by expectations on the future stance of monetary policy. First let us consider the results obtained with respect to the conditional mean equation. The constant term of the error correction coefficient is always negative and statistically significant. This coefficient seems to be time invariant as the hypothesis that the total allotment ratio and the share of the allotment to one-day repos on total allotment jointly have no influence on the error correction coefficient cannot be rejected at the 5 percent level in all three cases. The correction of the Libor towards the center of the target range is 0.03 per day for the three- and one-month rate but clearly stronger for the oneweek rate (0.07). This finding most likely reflects expectations effects, which are more important for the longer-term rates. By contrast to the conditional mean equation the estimates of the conditional variance equation points to strong volatility effect of the total allotment ratio and the maturity structure of the repo auctions. Besides the highly significant and persistent GARCH effects, we always find a highly significant negative effect of the total allotment ratio. This means that a higher allotment ratio decreases interest rate volatility. If we assume that the relationship detected is stable, this obviously implies that full allotment leads to minimal interest rate volatility. The result with respect to the maturity structure of the repos is less clear cut: More weight on one-day repos (statistically) significantly increases the volatility of the 3M- and 1M-Libor rate but decreases that of the 1W- Libor. There seems to be a volatility trade-off with respect to the maturity structure of the repo auctions: 3 We include a dummy variable for upward and downward changes of the target band, respectively. As we estimate a model for the change in the interest rate first differences of these dummies are included, too. Moreover, it should be mentioned that these dummy variables are not statistically significant in the conditional variance equation.

10 390 JORDAN / KUGLER relying more on the one-day (week) transaction decreases (increases) the volatility of the very short-term interest rates and increases (decreases) the volatility of the longer-term interest rates. Table 1: ML Estimates for the Change in the Libor Rate, Daily data 2000/2 2003/9 * rt = α+ ( β0 + β1zst + β2zpdyt )( rt 1 rt 1 ) + εt ε t : N( 0, h t ) 2 h = c + aε + bh + c ZS + c ZPDY t 0 t 1 t 1 1 t 2 t R β0 β1 β2 c0 3M (0.0129) 1M (0.0115) 1W (0.0222) (0.0165) (0.0118) (0.0299) (0.0175) (0.0144) (0.0268) ( ) ( ) ( ) R a b c1 c2 3M (0.0113) 1M (0.0295) 1W (0.0277) (0.0114) (0.0157) (0.0307) ( ) ( ) ( ) ( ) ( ) ( ) Standard errors in parentheses; estimated coefficients of policy change dummies and lagged interest rate changes are not reported. 4. Conclusion Since the beginning of 2000 the SNB uses fixed-rate repo auctions to control short-term Swiss franc interest rates. The way monetary policy is implemented within the new framework differs from the procedures of most other central banks as the SNB needs some flexibility in the money market in order to react to exchange rate and other shocks in the very short run without having to declare a change in monetary policy. It thus chose a target range for the 3M-Libor rather than a point target for the overnight rate. This longer-term interest rate is not determined by the SNB directly. Rather it reflects the conditions in the international Swiss franc money market that are influenced by

11 Implementing Swiss Monetary Policy 391 repo transactions with a very high flexibility with respect to frequency, allotment and maturity. This paper provides an empirical analysis of the effects of the SNB s operating procedure on the adjustment speed and the volatility of Swiss franc Libor of different maturities. More precisely it presents econometric estimates of the effect of two characteristics of the repo auctions on the conditional mean and variance of the Libor using daily data from January 2000 to September The results obtained indicate that total allotment ratio and the share of allotment to one-day repos on total allotment have no influence on the adjustment speed of the Libor to the center of the target range. However, the total allotment ratio has a clear and highly significant negative influence on the volatility of short-term rates. If we assume that the relationship detected is stable, this obviously implies that full allotment leads to minimal interest rate volatility. The result with respect to the maturity structure of the repos is less clear cut: More weight on one-day repos (statistically) significantly increases the volatility of the 3M- and the 1M-Libor rate but decreases that of the 1W-Libor. References AYUSO, J. and REPULLO, R. (2003), A Model of the Open Market Operations of the European Central Bank, Economic Journal. 113 (490), pp BINDSEIL, U. (2004), Over- and Underbidding in Central Bank Open Market Operations Conducted as Fixed Rate Tender, ZEI Working Paper B BREITUNG, J. and NAUTZ, D. (2001), The Empirical Performance of the ECB s Repo Auctions: Evidence from Aggregate and Individual Bidding Data, Journal of International Money and Finance 20 (6), pp GASPAR, V., PEREZ-QUIROS, G. and SICILIA, J. (2001), The ECB Monetary Policy Strategy and the Money Market, International Journal of Finance and Economics 6, pp JORDAN, T. J., KUGLER, P., LENZ, C. and SAVIOZ, M. R. (2002). Inflationsprognosen mit vektorautoregressiven Modellen, Quartalsheft der Schweizerischen Nationalbank 1/2002, pp JORDAN, T. J. and PEYTRIGNET, M. (2001), Die Inflationsprognose der Schweizerischen Nationalbank, Quartalsheft der Schweizerischen Nationalbank 19(2), pp KUGLER, P. and JORDAN, T. J. (2004), Vector Autoregressions and the Analysis of Monetary Policy Interventions: The Swiss Case, Schweizerische Zeitschrift für Volkswirtschaft und Statistik 140(1), pp

12 392 JORDAN / KUGLER KUGLER, P. and RICH, G. (2002), Monetary Policy under Low Interest Rates: The Experience of Switzerland in the late 1970 s, Schweizerische Zeitschrift für Volkswirtschaft und Statistik 138 (3), pp VEYRASSAT, A. (2001), Steuerung der Geldmarktsätze durch die Schweizerische Nationalbank, Financial Markets and Portfolio Management 15 (3), pp SUMMARY This paper provides an empirical analysis of the effects of the SNB s operating procedure on the adjustment speed and the volatility of Swiss franc Libor of different maturities. More precisely it presents econometric estimates of the effect of the two characteristics of the repo auctions on the conditional mean and variance of the Libor using daily data from January 2000 to September The results obtained indicate that the total allotment ratio and the share of the allotment to one-day repos on total allotment have no influence on the adjustment speed of the Libor to the center of the target range. However, the allotment ratio has a clear and highly significant negative influence on the volatility of shortterm rates. ZUSAMMENFASSUNG Diese Arbeit enthält einen empirische Analyse der Auswirkungen der Operationsprozeduren der SNB auf die Anpassungsgeschwindigkeit und die Volatilität von Schweizerfranken Libor mit verschiedenen Laufzeit. Dazu wurde ein ökonometrische Modell für das bedingte Mittel und die bedingte Varianz des Libors, das die Charakteristika der Repoauktionen berücksichtigt, mit Tagesdaten von Januar 2000 bis September 2003 geschätzt. Die erzielten Ergebnisse weisen darauf hin, dass der Zuteilungssatz und die Laufzeit der Repos keinen Einfluss auf die Anpassungsgeschwindigkeit des Libors an den zentralen Wert des Zielbandes haben. Hingegen wurde ein hoch signifikanter negativer Einfluss auf die Zinssatzvolatilität gefunden.

13 Implementing Swiss Monetary Policy 393 RÉSUMÉ Ce papier donne une analyse empirique de la technique de gestion de la BNS sur la vitesse de réaction et la volatilité du Libor. Pour achever cette tâche les auteurs estiment à base de données quotidiennes un modèle économétrique pour la moyenne et la variance conditionelle en tenant compte des caractéristiques opérationelles des ventes des pensions de titres. Les résultats montrent que ni la durée des pensions de titres ni les taux d adjudication ont un effet sur la vitesse de réaction du Libor à son taux central de sa marge de fluctuation. En même temps on trouve un effet important et significatif sur la volatilité des taux d intérêts.