Abstract
This Selected Issues paper discusses capital and liquidity regulations in Sweden. It recaps the recent debates on capital and liquidity buffers, and discusses a way to consider appropriate levels of capital and liquidity buffers in the case of Sweden. The paper estimates the government’s contingent liabilities from banks by different capital and liquidity levels. Also examined are options for Sweden in case the authorities face constraints to set buffers at their desirable levels.
III. Sweden Monetary Framework: Some Institutional And Operational Considerations1
A. Introduction
1. In January 1993, Sweden’s central bank announced a new monetary policy based on an inflation targeting framework, while allowing the exchange rate to float freely. Under the new framework, the inflation target would apply from 1995 and thereafter.
2. The Swedish model, shared by several other small open economies, including Canada, the United Kingdom, Australia and New Zealand, has proven resilient to severe shocks dealing well with the aftermath of the 1991–1993 Swedish financial crisis, and more recently, with the global financial crisis.
3. When compared to that of its peers, Sweden’s monetary policy performance has been as good or better in controlling inflation and output gap volatility, while avoiding large interest rates gyrations. And controlling for swings in mortgage rates and energy prices—two key but highly volatile determinants of headline consumer price inflation in Sweden—inflation volatility has been practically eliminated by virtue of the inflation targeting system. This is remarkable, especially when measured against Sweden’s own outcomes following the collapse of the Bretton Woods system (Figure 1).
Sweden: International and Historical Comparison of Monetary Policy Outcomes
Citation: IMF Staff Country Reports 2012, 155; 10.5089/9781475506075.002.A003
Sweden: International and Historical Comparison of Monetary Policy Outcomes
Citation: IMF Staff Country Reports 2012, 155; 10.5089/9781475506075.002.A003
Sweden: International and Historical Comparison of Monetary Policy Outcomes
Citation: IMF Staff Country Reports 2012, 155; 10.5089/9781475506075.002.A003
4. The Swedish monetary and financial policy framework has undergone two comprehensive independent reviews by the Riksdag’s Committee on Finance as part of the Riksdag’s work of follow-up and evaluation: the Giavazzi-Mishkin2 review in 2006 (covering the period 1995–2005) followed five years later by the Goodhart-Rochet3 review (covering the period 2005–2010). While the first review focused largely on the monetary policy function of the Riksbank, the second review concentrated on its financial stability function, examining, among other things, whether this has incorporated well lessons from the global financial crisis. The recommendations of the two reviews are summarized in Table 1.
5. This chapter raises additional issues not considered in the reviews, regarding some practical aspects of operating monetary policy in an almost perfectly-transparent fashion. In particular, it focuses on issues related to the publication of the projection of official rates (Section B). It discusses existing dilemmas for the implementation of monetary policy in the context of Sweden’s current institutional and legal structure for financial supervision and regulation (Section C). And finally, it reevaluates the scope for a discussion of measures of resource utilization in the context of monetary policy (Section D).
B. To Publish or Not To Publish: That Is the Question
6. As emphasized by the Goodhart-Rochet review, the Riksbank compares favorably to advanced economy peers in terms of transparency and scores at the top of Eijffinger-Geraats4/Dincer-Eichengreen5 Central Bank Transparency Index. The Riksbank also ranked as the best communicator among the sample of central banks chosen by the authors of a J.P. Morgan (2007) report focusing on monetary policy communication.6,7
7. In the effort to achieve maximum transparency—and in addition to publishing comprehensive forecasts—in 2007 the Riksbank has begun to publish the 3-year-out Executive Board’s expected future path of the repo rate accompanied by a probability distribution. This has several benefits. First, contrary to other central bank forecasts that are made conditional on constant interest rate paths or market rates, the Riksbank forecast is inherently consistent with the Executive Board’s preferred policy path. Second, announcing the entire future path of the policy rate communicates unequivocally to the public, the (at the time of publication) likely direction of monetary policy in the future. However, the Riksbank makes clear that the path is a projection, not a commitment. In turn, this can help anchor the public’s expectations about interest rates by affecting directly long-term interest rates, which can potentially reduce the output gap cost of price stabilization. From June 2007, the Riksbank’s Executive minutes were also attributed.
8. However, publishing an interest rate forecast entails several complications. The Goodhart-Rochet review’s main criticism to the Riksbank’s new forecasting regime relates to the fact that the announcement of the outcome of the voting on the policy rate should be made at the time of the meeting during which the inflation and growth forecasts are finalized, in that these are necessarily produced simultaneously. Or else, the title, function and remit of the MPC meeting—where interest rates are de jure but not de facto determined—should be altered. This critique is not persuasive. Although these are largely decided by then, the final decisions on the final forecast and policy rate paths are actually taken during the monetary policy meeting—in the sense that these are preliminary until then—and so there is no artificial delay between the announcement of the policy rate path and the finalization of the forecast. At the pre-monetary policy meeting the staff briefs the Board on the forecast and their recommendation for monetary policy, and gets feedback from the Board so that it can finalize the forecasts (including the policy rate) coincidentally with the monetary policy meeting.)
9. In fact, there may be more difficult issues related to the publication of the path—some of which but not all mentioned in the Goodhart-Rochet review—to be concerned about.
10. Over the past 4 years—largely coinciding with the period that followed the collapse of Lehman Bros.—Swedish market yield curves and the projection for official rates have deviated substantially, especially for maturities beyond the very short run. Although this gap disappeared in early 2011, it has reappeared more recently (Figure 2).8
Repo Rate Actual and Projected Paths: April 2012
Citation: IMF Staff Country Reports 2012, 155; 10.5089/9781475506075.002.A003
Repo Rate Actual and Projected Paths: April 2012
Citation: IMF Staff Country Reports 2012, 155; 10.5089/9781475506075.002.A003
Repo Rate Actual and Projected Paths: April 2012
Citation: IMF Staff Country Reports 2012, 155; 10.5089/9781475506075.002.A003
11. There are various hypotheses of why market yields tend to show little or no adjustment towards the published path of official rates. These include: (i) problems when extracting market expectations from forward rates, as term premia (especially at a few years’ horizon) are very low, or even negative. Survey expectations are somewhat more in line with the Riksbank’s policy rate path; (ii) a systematically more gloomy read of the outlook on the side of the markets, relative to the Executive Board; (iii) an upward bias in the way in which the Riksbank’s predicts future TCW-weighted policy rates abroad to evolve, which, other things equal, has the property of pushing up artificially the forecasting model’s implied repo rate path (which is one key input of the published official rate projection), causing it to deviate systematically from the market yield curve;9 (iv) for the period when market rates are below the policy rate path, the existence of persistent dissent (documented in the minutes to the monetary policy meetings of the Riksbank’s Executive Board) within the MPC relative to the right course for interest rate. This has possibly undermined the interpretability of an Executive Board’s joint repo rate projection in outer periods.
12. Whatever the cause, the persistence of this unresponsiveness risks eroding the credibility of the Riksbank: it reflects a concrete rejection by markets of the Riksbank’s expectations of its own policy actions. This makes monetary policy less effective, because, over time, it blocks the expectational channel of monetary transmission, which is one of the key ways via which monetary impulses are transmitted in an inflation targeting regime. So far these issues have not compromised the credibility of the Riksbank’s inflation target—expectations remain firmly anchored at the target. But the gap between market expectations of the policy rate and the Riksbank’s projection for it represent an ongoing qualification to the overall credibility of the inflation targeting regime.
13. Against this background, there are various options for development of the new forecasting regime that could be considered simultaneously to alleviate or remove this problem. The Riksbank, for example, could:
Produce inflation and unemployment forecasts conditional, first, on market expectations of the repo rate and then on the Riksbank’s preferred repo rate. This would bring to the fore the differences in outlook and risks—if any—between the markets and the Executive Board and help cross-check the markets’ trust in the forecast assumptions rather than just their trust in the Board’s preferred policy path.
Condition the forecasts (including the one for the official rate) on TCW-weighted market implied forward rates (adjusted for credit risk and maturity premia). Deviations in assumed and implied foreign forward rates explain a sizeable part of the current misalignment between market yields and thus using this different conditioning assumption would reduce the misalignment, likely reducing the erosion of credibility caused by the existence of a large gap between market yields and Riksbank projections of its own policy rate. This recommendation does not come with the drawback that the forecast becomes less consistent than before, since TCW-weighted forward rates are already determined off-RAMSES II (Version II of the Riksbank’s Aggregate Macromodel for Studies of the Economy in Sweden).
Shorten the minutes of the Executive Board’s monetary policy meeting. Long minutes tend to emphasize the perception of dissent, and risk impairing the clear and effective communication of policy decisions (and rationales thereof). The views of the Executive Board’s minority voters could then be reported in a table containing their growth, inflation and interest rate forecasts. This would help quantify divergencies in the members’ assessment of the best course of action and, thus, help markets evaluate the impact of dissenting votes onto the projection of the official rate. In addition, MPC members could vocalize their views in more depth online with, for example, individual web pages on the Riksbank website; or through speeches; or through individual hearings in front of the Committee on Finance of the Riksdag.
Alternatively, and more drastically, the Riksbank could :
Consider adjusting the form in which it publishes the projected paths for the official rate and, in the extreme case, it could discontinue publication altogether. Alternative forms include shortening the horizon for the projections and/or publishing less frequently. In this context, or in case of discontinuation, the likely direction of future rates could be discussed in the minutes, like in other central banks (for example the Bank of Canada, or the Bank of England) where press releases on interest rate votes are at times accompanied by the expression of a bias for future policy rate decisions. Alongside, the Riksbank could publish an explanatory note describing the interest-rate rule used as a basis for the MPC interest rate discussions, which would help guide a mechanical guesstimate of future levels of the repo rate, given the Executive Board’s published inflation and GDP growth forecasts. Projected repo rate paths could still be produced and shown to the public (for example during hearings of the Executive Board at the Committee on Finance of the Riksdag; or in speeches) but only on an occasional basis. Discontinuation of publication of the projected official rate would not imply changing the conditioning assumption of the inflation and growth forecasts regarding monetary policy, which would remain the Executive Board’s (unpublished) projected official rate path.
14. While it is clear that, in the long run, discontinuing the publication of the projected official rate could halt the potential erosion of credibility of the Riksbank as a result of the misalignment of market yields from its own forecast, such discontinuation potentially carries reputational risks from a retrenchment in transparency that needs to be evaluated against the reputational benefits of ceasing to see the credibility of the Riksbank’s repo rate forecast systematically challenged. Likewise, it is hard to judge whether continuing with the publication of the projected interest rate path but shortening the minutes, better quantifying the impact of dissent and nearing the conditioning assumption on foreign policy rates to markets’ expectations would make market yields more responsive to changes in the projection of official rates. However, since all these measures would deliver additional overall benefits to the clarity and rigor of the Riksbank’s communication, they could be implemented nonetheless especially if the publication of the projection of official rates is retained.
C. Setting Monetary Policy in the Context of Sweden’s Macroprudential Framework
15. The Sveriges Riksbank Act (1988) tasks the Riksbank with—among other goals—the mandate of keeping the payment system stable.10 To this end the Riksbank is allowed to extend discretionary credit to market participants,11 and demand information from financial institutions. 12
16. In the absence of clearer and more extensive official assignments on financial stability, the Riksbank has in practice interpreted its responsibility for promoting stability in the payment system in the broader sense of promoting stability in the financial system, recognizing however that this responsibility is shared with other public bodies, notably the Swedish SFSA (Finansinspektionen), the Swedish National Debt Office (SNDO), and the Ministry of Finance.13
17. However, despite such restrictive de jure arrangements, the Riksbank is de facto a key player in the design of Sweden’s macroprudential policy and both during the global financial crisis and in normal times, it has closely collaborated and provided inputs to other governmental agencies (notably the SFSA) in this area. In January 2012, this relationship was strengthened through the creation of a council between the SFSA and the Riksbank acting as a forum where assessments of risk and questions regarding macro-prudential policy are discussed jointly. The council, which meets twice a year and includes among its members the governor of the Riksbank and the head of the Financial Supervisory Authority, met for the first time on February 24, 2012, but has not changed the two institutions in terms of their responsibility.
18. The limited de jure, but considerable de facto responsibilities of the Riksbank with regard to macroprudential policy and financial stability, mean that it is presently not clear whether past, present and future policy actions in these areas are currently treated as endogenous or exogenous in the MPC’s decisions on monetary policy.
19. This confusion translates into the decision making process of the Riksbank: judging from the minutes of the monetary policy meetings of the Executive Board, it is evident that different MPC members have different attitudes toward what are the MPC financial stability and macroprudential responsibilities, which complicates the Board’s debate about the best course of interest rate policy action. Prominent examples of this are individual MPC’s member views about the role of monetary policy with respect to house price dynamics; and, to a lesser extent, the exchange rate.
20. Since it remains questionable whether the Riksbank should use the policy rate to affect asset prices, or constrain its actions in the pursuit of the primary objective of stabilizing consumer prices to limit potential consequences of these onto asset markets (tasks that should be addressed preferably and primarily through specific macroprudential policy instruments like capital requirements, LCR and NSFR floors, LTVs ceilings—similarly to how it is done, for example, in Canada which is faced by similar asset price dynamics as Sweden), the first best option that could be considered is a separation of roles within the Riksbank with regard to monetary and financial stability policies (see Chapter I of the Selected Issue Paper). More specifically, following the Bank of England model:
A Financial Policy Board (‘FPB’, operating as an interim body modeled on the Executive Board and akin to the Bank of England’s Financial Policy Committee or the Fed’s Office of Financial Stability Policy and Research) could be created to spot unsustainable rises in credit that threatened the financial system. In addition to voicing concern, the FPB would be able to force banks to maintain larger buffers in case of an impending crisis. The Riksbank’s Executive Board would hence be free to use the interest rate to aim at its primary objectives, better aligning members’ policy preferences and thereby facilitating discussions about the best course of policy and the communication thereof to the external world.14
21. As a second best, under the assumption that the institutional status quo is maintained, the Riksbank could:
Explore whether, using Swedish data, there is empirical support for the idea that the policy rate has any relevant effect on financial stability in the banking sector or on risks in the housing- and mortgage market, especially given the already available tools of capital requirements, LCR, NSFR, LTV and other macroprudential instruments.15
Augment its forecasting toolkit to fully embed transmission to and from the financial markets. Inflation and growth forecasts could increasingly be generated by models that fully encompass financial markets and the loop from these and back onto the real economy.16 Although the issue in macroprudential policy is not to fine-tune financial variables, but lies in building up sufficient resilience to disturbances,17 this would allow a more focused debate inside the Board about the potential implications of policy actions on these markets, allowing the Board members to place explicit priors on assumptions, shocks and sensitivities in the model—as well as helping them quantify their preferences for financial stabilization vis-à-vis those for inflation and employment stabilization in the context of internal debates and the minutes. In turn, this could aid both internal and external communication, help the Riksbank take positions vis-à-vis other supervisory and regulatory agencies, and possibly, in the near future, help institutionalize a stronger de jure financial stability role for the Riksbank.
D. Should the Riksbank Evaluate and Use Measures of Resource Utilization ?
22. During the period 1997–2011, average CPI inflation (the measure of inflation targeted by the Riksbank at 2 percent announced in 1993 and applying from 1995) has undershot the target by 0.6 percentage points on average. There are at least four possible reasons of why this might have happened. One possibility is that the Riksbank may have conducted monetary policy in an asymmetric manner, de facto favoring lower inflation outcomes to higher ones. Alternatively, despite symmetric stabilization preferences, the Bank may have systematically overestimated potential unemployment. (This is reminiscent of the argument of Orphanides that the Fed caused too high inflation in the 1970s by underestimating potential unemployment).18 Or more generally, the Riksbank may have systematically overestimated inflationary pressures, one story being that imported inflation systematically came in below expectations and growth in productivity higher than expected. Last but not least, (at least since the Great Recession) the pass-through from the repo rate to other rates, notably the mortgage rate, may have weakened, meaning that easy monetary conditions were not reflected de facto in an easier monetary stance, and implying a stronger disinflationary effect of policy than desired.
23. In either case, a persistent undershooting of the inflation target raises concerns because it can be associated with large, unnecessary unemployment costs, which—under the Sveriges Riksbank Act (1988) and its preparatory works—should be acted against conditional on achieving the inflation target. The Government Bill proposing the Riksbank Act states that the Riksbank should, without neglecting the price stability target, support the objectives of the general economic policy with the purpose of attaining sustainable growth and high employment.
Sweden: Inflation Rates
(Y/Y percent change)
Citation: IMF Staff Country Reports 2012, 155; 10.5089/9781475506075.002.A003
Sources: Haver Analytics and Fund staff calculations.Sweden: Inflation Rates
(Y/Y percent change)
Citation: IMF Staff Country Reports 2012, 155; 10.5089/9781475506075.002.A003
Sources: Haver Analytics and Fund staff calculations.Sweden: Inflation Rates
(Y/Y percent change)
Citation: IMF Staff Country Reports 2012, 155; 10.5089/9781475506075.002.A003
Sources: Haver Analytics and Fund staff calculations.24. This concern is particularly germane in Sweden today, given the national debate about the reasons of the persistence in unemployment following the global financial crisis,19 and the debate on what measures could be employed to bring unemployment back to its long-run equilibrium level.
25. To address this issue the Riksbank could:
Derive estimates of the short-run sustainable unemployment level. The gap between unemployment and the short-run equilibrium measure can be a useful indicator of inflationary pressures, and therefore these are more relevant for monetary policy than long-run estimates. The short-run and the long-run measures of equilibrium unemployment rate can be very different, due to persistence phenomena in the labor market. Eventually the short-run equilibrium converges to the long-run equilibrium.
Employ its own estimates of the long-run and short-run unemployment equilibria to help guide the internal debate on policy setting, alongside forecasts from the main model. The Riksbank could use these estimates as inputs to estimated Phillips curves or could jointly estimate measures of equilibrium unemployment with consumer and price inflation to provide additional and alternative inflation forecasts. This is a common practice in other advanced economy central banks that target inflation and use a DSGE model for forecasting (e.g. the Bank of England, the Bank of Canada and the Reserve Bank of New Zealand). Augmenting information from the main model with information from smaller models may help avoid blind spots in the determinants of inflation, leading to better policy outcomes. In this sense, information from long- run estimates of equilibrium unemployment and Phillips curves based on unemployment gaps could be used to guide judgment in the interpretation of forecast results of the main model, and to apply skews to the inflation and GDP growth probabilistic distributions (also dubbed “fancharts”) used to set rates and then presented to the public in the Monetary Policy Report.
Prepared by Nicoletta Batini (nbatini@imf.org).
Giavazzi, F. and F. S. Mishkin (2006), An evaluation of the Riksbank’s monetary policy 1995- 2005, Rapporter från Riksdagen 2006/07: RFR 1, Finansutskottet, Stockholm.
Goodhart, C. and J-C. Rochet (2011) Evaluation of the Riksbank’s monetary policy and work with financial stability 2005–2010. Rapporter från Riksdagen 2010/11: RFR5. Finansutskottet, Stockholm.
Eijffinger, S. C. and Geraats, P. M. (2006). ‘How transparent are central banks?’, European Journal of Political Economy 22(1), 1–21.
Dincer, N. N. and Eichengreen, B. (2007). ‘Central bank transparency: Where, why, and with what effects?’, NBER Working Paper 13003.
The sample includes the Federal Reserve System, the European Central Bank, the Bank of Japan, the Bank of England, Norges Bank, the Bank of Canada, the Reserve Bank of Australia, the Reserve Bank of New Zealand and the Swiss National Bank.
J.P. Morgan Research (2007). “Central Bank communication hits diminishing marginal returns”, May 11, 2007.
Initially market rates were above the policy rate path. But lately market rates have been below the policy rate path. These two episodes are very different.
The TCW-forecast is made using models (GPM) and judgment, including information from forward rates. The staff has consistently made the judgment that the market rate has been too low, and that a reasonable forecast is higher than forward rates.
Sveriges Riksbank Act (1988), Chapter 1, Article 2, third paragraph.
Sveriges Riksbank Act (1988), Chapter 6, Article 7, second paragraph.
Sveriges Riksbank Act (1988), Chapter 6, Article 9.
Apart from direct responsibility, there are basically three forms of central bank involvement in banking supervision and macroprudential regulation that are not mutually exclusive. First, central banks can carry out specific supervisory tasks such as monitoring categories of risks incurred by financial institutions and conducting audits at financial institutions (e.g. Germany and Austria). Second, central banks can participate in the supervisory boards and/or management Boards of the supervisory agencies, thereby contributing to the management of the agency in question. In the European Union, for example, this type of involvement is rather common in almost all cases. Third, central banks may share resources (e.g. information technology, staff and administration) with the supervisory agency (e.g. France and Ireland). The SFSA and the Riksbank have concluded a Memorandum of Understanding, which includes similar features as that between the UK authorities, namely sharing information, avoiding duplication of labor is stated, while foreseeing to reach an agreement on the collecting institution and on data communication in cases where the SFSA and the Riksbank need the same information, apart from explicit arrangements for the secondment of staff.
There is already a separate decision procedure for the Executive Board for the Financial Stability Report with its concrete macroprudential recommendations and the meetings with the Council with the SFSA. However, as evidenced by the minutes of monetary policy meetings, this is not sufficient to keep the monetary and financial stability functions separate in practice.
The Riksbank is conducting substantive analysis on the link between monetary policy and financial stability with the primary aim of answering questions like: (i) how does bank capital affects the dynamics of the economy and the monetary transmission mechanism ?; (ii) what are the long-term macroeconomic effects of an increase in capital requirements ?; (iii) what are the interactions between monetary policy and macroprudential policy ? (iv) how does they the transition to a steady state with higher average capital requirements work and does this transition depend on monetary policy ? (v) may a too protracted monetary expansion lead to a credit boom and an increased risk of a crash ? This work however has not been completed and thus the results are not publicly available.
The Riksbank has begun more intense analytical work in this direction following the global financial crisis. As a result of this, the Riksbank current forecasting model incorporates some financial frictions following the Christiano-Motto-Rostagno (2009) version of Bernanke-Gertler-Gilchrist (1999), and so allows for a time-varying corporate spread. The corporate spread is also typically closely correlated with other spreads (e.g. mortgage spreads), so the model can at least potentially capture some of the effects of financial turmoil.
Also, the policy horizon in macro-prudential policy is much longer than in monetary policy.
For a discussion of the potential costs in terms of higher unemployment of undershooting the inflation target, see Svensson, L. (2012), “The Possible Unemployment Cost of Average Inflation below a Credible Target”, mimeo.
See, for example, the government’s 2012 Spring Fiscal Policy Bill, and the NIER (2010) study “The Persistent Labour-Market Effects of the Financial Crisis”, Working Paper No. 117.
