It is a pleasure to be here today to celebrate 20 years of central bank independence for the Bank of England.

I have been tasked with discussing central bank independence in retrospect. In part, this is because a little more than two decades ago, Stan Fischer and I wrote about central bank independence in prospect and posed the question: How Independent Should a Central Bank Be?^[1]

When we asked that question in 1994, central bank independence was all the rage. I wrote my PhD dissertation at MIT on it, under Stan's expert tutelage. A bit further up Mass Ave, Adam Posen was writing his PhD dissertation at Harvard on the very same issue. In New Zealand, the government had established a new framework for independence for the Reserve Bank of New Zealand (RBNZ) in 1989, which granted the RBNZ significant independence, including making the Governor directly accountable for any failure to achieve the price stability objective. The framework for the European Central Bank (ECB) had just been established, which legislated a very high degree of central bank independence. Looking back, that may have been the high-water mark for central bank independence. Though, of course, the Bank of England's independence was still to come, so perhaps that was the apogee!

Fast forward to today and central bank independence is no longer in the ascendancy. Its legitimacy and effectiveness are under scrutiny. Is this ‘the end of the era of central bank independence’?^[2]

The answer to that question is no, in my opinion. The reason why has a lot to do with the answer that Stan and I proposed in answering the question of how independent should a central bank be. We made the important distinction between goal and instrument independence, arguing that central banks should be goal dependent but have instrument independence. That is, the goal(s) that the central bank is pursuing should be determined by the political process. But, once given the goal, the central bank should be able to pursue the goal as it sees fit, with appropriate accountability.

That distinction we drew between the two aspects of independence stands up to scrutiny twenty years on. If anything, I would argue that the experience of the past two decades serves to reinforce our conclusion, with the Bank of England's own experience being a prime example.

In my remarks today, I will first revisit the motivation for central bank independence. I will then ask whether those justifications are still relevant today. Or is central bank independence a solution to yesterday's problems? I will then discuss why central bank independence is under more threat today and ask whether modifications to the model of central bank independence are required.

Three Foundations of Central Bank Independence

First, I would like to revisit the three foundations of the arguments supporting our conclusions on goal dependence and instrument independence. The first foundation was the performance of the Bundesbank. We questioned how effective the Bundesbank had been, in the sense that recessions were at least as costly (in terms of lost output) in Germany than they had been in other countries with less independence. That is, the sacrifice ratio in Germany was not lower, which would have been expected if there had been a material credibility benefit from the Bundesbank's independence.^[3] We cautioned against the cult of central bank independence, that is, the need to avoid excessive concentration on inflation avoidance. Hence the conclusion that a central bank should have goal dependence.

Second, there was the academic literature. Stan and I noted that much of the academic literature focused on rules versus discretion in the context of dynamic inconsistency.^[4] Dynamic inconsistency doesn't have any particular implication for central bank independence. A rules-based monetary policy can be implemented mechanically by an agency with neither goal nor instrument independence. The argument won out in favour of discretion; if not in all of the theoretical literature then absolutely in practice. But both in practice and to a large extent in theory, the solution was in the form of ‘constrained discretion’.^[5] The constraint on discretion is the goal for the central bank set by the government. The central bank has complete discretion as to how best to achieve that goal.

Given discretion, what then is the best institutional framework? Stan and I also noted that Ken Rogoff's conservative central banker could be too inflation averse to be socially optimal.^[6] The optimal degree of central bank independence was ‘chosen by trading off the reduction in mean inflation secured by conservatism against the less than optimal trade-off between inflation and output variability produced by that same conservatism’.^[7] Carl Walsh took up this point in designing a socially optimal incentive contract for central bank governors, and the New Zealanders went a fair way towards operationalising it.^[8]

Third there was the empirical evidence of Alex Cukierman, Alesina and Summers, and Grilli, Masciandaro and Tabellini.^[9] Those papers showed a clear negative correlation between average inflation and central bank independence, and this result was influential in the changes to central bank independence that occurred around this time. Stan and I drilled down empirically into what particular aspects of central bank independence underpinned this correlation and concluded that inflation performance is likely to be better if there is goal dependence and instrument independence.

Before discussing how these foundations stand up to scrutiny two decades later, I will spend some time talking about the overlap between central bank independence and inflation targeting.

The issue of goal dependence raises the question about what are the appropriate goals for a central bank. At the same time as central bank independence was coming to the fore, inflation targeting was in its nascence. Indeed, I attended a conference here at the Bank of England in 1995, organised by Andy (Haldane), on the brand-new topic of inflation targeting.^[10] The New Zealanders had adopted the first formal inflation target in 1992. The Canadians and Swedes followed. The UK had just adopted an inflation target. It is noteworthy that this was a couple of years before the Bank of England received its independence.

Andy wasn't sure whether to invite the RBA as we had put in place a ‘soft’ version of inflation targeting around 1993, in marked contrast to the (then) hair-shirted framework of the New Zealanders and the Canadians. Graciously he erred on the side of inclusiveness in inviting Glenn Stevens and me to talk about the Australian model. That model now looks very much like the standard flexible inflation-targeting framework that is practiced here in the UK and elsewhere, including both NZ and Canada.

Inflation targeting and central bank independence are sometimes conflated given their similar birth dates in a number of countries. In part this is because both can been seen as a response to the inflation experience of the 1970s and 1980s. The problem at hand was how best to establish and maintain a low inflation environment while delivering the desired macroeconomic outcomes.

But the nature of the underlying issue that the two were intended to address was quite different. The political business cycle was one of the primary motivations for central bank independence. The temptation to run the economy too hot in the lead-up to an election argued for the allocation of monetary policy management of the economic cycle to an agency insulated from the political cycle.

While one might argue that the political business cycle is less of a consideration today in many countries, its current abeyance should not be assumed to be a permanent state of affairs. Moreover, it still very much remains a live issue in a number of emerging markets.

In contrast, inflation targeting arose in part because other monetary frameworks had seemingly proven unable to deliver the appropriate degree of inflation control. As Governor Bouey of the Bank of Canada famously said, ‘We didn’t abandon monetary targets, they abandoned us'. In that regard, Mervyn King describes inflation targeting as a coping strategy,^[11] coping with the inability of other monetary policy frameworks to deliver the required outcomes.

Central banks in many cases didn't have a lot of public credibility in terms of macroeconomic policymaking. In the Bank of England's case, it wasn't the monetary policy decision-maker, so the institution itself didn't have any track record of monetary policymaking. So in the UK, as in NZ, Canada and Australia, there was a need to build that track record. A track record requires a track. Inflation targeting provided that track. Here was a goal over which the central bank believes it has a sizeable degree of influence. The inflation-targeting framework provides a benchmark against which the central bank can be held to account in close to real time. If successful, then credibility is likely to build through time.

The Foundations for Central Bank Independence in Retrospect

Ultimately the proof of the central bank independence pudding is in the eating. Has central bank independence delivered? The subsequent two decades have clearly seen low inflation. Somewhat ironically, the problem today is that inflation is too low, which has contributed to the claim that central bank independence, and particularly inflation targeting, is the solution to yesterday's problem. Certainly, up until 2007, the low inflation outcomes did not come at the cost of inferior outcomes in terms of output or employment. This was Mervyn's ‘nice’ decade.

But were these ‘nice’ outcomes causation or correlation? It is difficult to disentangle from other influences, including the emergence of inflation targeting as the operating framework for many central banks. There is also the integration of the Chinese economy into the global economy and the significant disinflationary impulse that came with that.

Ed Balls and his co-authors have revisited the empirical findings of Stan's and my paper and show that they stand up with the addition of 20 years of data.^[12] They demonstrate that the negative correlation remains between the instrument independence of the central bank and inflation.

There is also a strong test case in the recent experience in the UK. The period between mid 2010 and mid 2012 saw a prolonged period of inflation above target. Yet inflation expectations remained well anchored. This was reflected in subdued wage outcomes, leading to a significant decline in real wages.

This event seems to me to demonstrate that the credibility of the Bank of England in terms of inflation control was well entrenched. Could such a conjuncture have been achieved in earlier periods in the UK? The experience of earlier decades suggests that it is highly likely that inflation expectations would have gone up with inflation. How much this can be attributable to central bank independence or the inflation target is difficult to disentangle. But the fact that the Bank of England had the independence to pursue the inflation target set by the Chancellor must count for something. The fact that the target was not adjusted upward to accommodate the rise in inflation is also consequential. The Bank of England did not have a political incentive to achieve surprise inflation, Barro-Gordon style.

That's the good news. The economic outcomes of the 1990s and 2000s in terms of low inflation, low unemployment and lower economic volatility are at least partly attributable to central bank independence. The political business cycle has been largely absent from discussions for some time now, particularly in the advanced economies.

In the end, though, this assessment mostly relies on assertion, rather than empirical proof, particularly for the advanced economies. The unknowable counterfactual as to what would have happened in a world of less central bank independence remains just that, unknowable. The emerging economies provide a more fertile ground for empirical validation, although China stands as a very large counterexample.

Central Bank Independence under Threat?

But there is also the bad news. Why is central bank independence under pressure today?

In summarising the economic outcomes over the past two decades, I stopped short of the financial crisis and its consequences. Clearly the output performance since the crisis has been inferior (labour market outcomes less so), although inflation has generally been low and stable. In some cases, the concern has been that inflation has been too low. The UK is not obviously one of these cases, given that the average inflation rate over the past decade in the UK has been 2¼ per cent, slightly above the middle of the inflation target. Similarly, in Australia the average inflation rate has been 2.4 per cent, again right around the middle of the inflation target.

How much these outcomes can be attributed to monetary policy settings, both before and after the onset of the financial crisis, is a hotly contested debate. I will not add to that debate here, but simply note that the unsatisfactory post-crisis growth experience is a significant contributor to the current scrutiny of central banks and their independence.

In this regard, I am concerned that along with insulating the setting of monetary policy from the political business cycle through central bank independence came a belief that monetary policy was the only game in town for demand management. This belief was reinforced by the success of the macroeconomic management after the 1990s recessions up until 2007.

After the onset of the crisis, with a rare exception in 2008-2009, demand management continued to be the sole purview of central banks. Fiscal policy was not much in the mix. This put additional pressure on central banks. Given their mandate, as long as there was something that could be done to achieve the goals given to the central bank, however small the impact, then that something had to be done. Otherwise there was a serious risk of undermining the legitimacy of the central bank's independence.

As we know, that took a number of central banks into the territory of the zero lower bound, and beyond in some cases. In some respects, monetary policy around the zero lower bound, particularly in the form of quantitative easing, starts to morph into elements of fiscal policy. In the framework that Stan and I used to examine the appropriate degree of central bank independence, government spending or fiscal policy more generally had a central role to play that the central bank needed to take account of. There were different outcomes depending on the nature of the interaction between the central bank and the government (in game theoretic terms, it depended on which was the Stackelberg leader or whether it was a Nash solution). But fiscal policy did play an important role in demand management. Monetary policy was not the only game in town.

Central bank independence does not imply the complete delegation of demand management to the central bank. The achievement of the output and inflation goals of the central bank is very much a function of the fiscal response too. There is a risk that the cult of central bank independence has placed excessive pressure on the central bank to always do more.

Stan and I examined the issue of instrument independence in a ‘normal’ macroeconomic world, where changes in interest rates within standard historic ranges were sufficient to achieve the desired macroeconomic outcomes. Again, that is not the environment that a number of economies have experienced over the past decade, where policy settings have moved into ‘unconventional territory’. Unconventional monetary policy has the possibility to undermine the support for instrument independence because of the interaction between the central bank's asset purchases and the government's debt management strategy.

Mervyn King noted that ‘inflation targeting as a practical definition of price stability is separable from our understanding of the economy, which can evolve’.^[13] I would make the same argument about central bank independence as an overarching framework for monetary policymaking. My view is that inflation targeting and central bank independence are being blamed for economic outcomes that are not a consequence of either of them, but rather of our understanding of the way the economy operates. Legitimate criticisms of central banks' understanding of the economy can be made, but that does not delegitimise the overall operating framework. But a great strength of central bank independence and inflation targeting is that one's view of the transmission mechanism can evolve over time while the basic architecture of the policy framework remains unchanged.

Besides that, a new threat to central bank independence comes from the greater focus currently on the distributional consequences of monetary policy. Monetary policy has always had distributional consequences. The primary transmission of monetary policy decisions is through two relative prices: the interest rate and the exchange rate. The interest rate is the intertemporal price of substitution, which has distributional consequences between savers and investors, as well as across generations, as indeed does inflation itself. The exchange rate has distributional consequences between the traded and non-traded sectors of the economy as well as cross-border distributional implications. It is also important to note that unemployment has larger, first order consequences for income distribution than either the interest rate or the exchange rate. But these distributional effects of monetary policy have always been present. Why is this so much more of a threat to central bank independence today?

One plausible explanation is the lower level of nominal growth, and particularly wage growth. The concerns around income distribution are much starker (though not necessarily any less), when nominal wages are stagnant for a significant share of the population. Another possible explanation is that if a primary transmission channel of unconventional monetary policy is through asset prices (including house prices), then the wealth channel may be playing a disproportionately larger role than in the past. The distributional impact of the wealth channel may well be larger than that of other channels of monetary policy transmission. If the distributional impact of monetary policy channels is larger now than it was in the past, this can compromise the legitimacy of instrument independence.

Financial Stability and Independence

Stan and I also addressed the question about goal dependence coming from the standard macroeconomic perspective of the day. One aspect that has changed post-crisis is the heightened focus on the financial stability goal of central banks. The Bank of England is a prime example of this. One question that arises is how the financial stability goal interacts with the inflation target. Is it a separate goal for the central bank that sets up potential trade-offs or is it aligned with the inflation-targeting goal? If it is separate, should the central bank have independence in how it defines the goal of financial stability? Should it have complete instrument independence as to how it pursues that goal?

The goal of financial stability has generally been left vague. To some extent, it has been defined only in the negative. That is, we can all agree that financial instability was present in 2008 and that was not a good state of affairs. But it is much harder to get agreement on what are the important elements of financial stability and how central banks should go about achieving them. The long-lasting, and still very active, debate about whether the central bank should lean against the wind is the best example of this. It brings to the fore a number of issues: is there a trade-off between the inflation target goal and the financial stability goal? Does the goal of financial stability need to be better defined than it is at the moment? If so, by whom? Should the central bank have complete independence to pursue its goal of financial stability with whatever instruments it so chooses, including when the distributional implications of some of those instruments are much starker? The questions do not have a straightforward answer. I would hope they would be a hot topic for PhD students today in the way that central bank independence was when Adam Posen and I were PhD students.

Accountability

One of the points that Stan and I were at pains to stress was the importance of central bank accountability. It is the very bedrock of central bank independence in a democratic society. It alleviates the criticism that central bank independence is inappropriate because macroeconomic policy decisions are being taken by unelected bureaucrats. I find this criticism misplaced. There are many instances where the goals of public policy have been set by the political process but are then implemented by (unelected) public servants. That is, the bureaucracy has instrument independence in many different areas. This set-up is not unique to central banks. But it does underscore the argument for goal dependence. The central bank is accountable to the political process in achieving the goals given to it.

Over the past two decades, accountability has changed substantially. Again the Bank of England is a good example. There are regular appearances before the Treasury Select Committee. The volume and content of central bank publications is considerably larger and more detailed than twenty years ago. The number of speeches is dramatically higher and the messaging more transparent. No longer is it the case that a Governor would say: if you understood what I said, I must have misspoken.

So with independence has come the accountability. The accountability is to the politicians and the population at large, not a small group of central banking aficionados, nor the financial markets.^[14] That means that, unlike the Delphic Oracle, it is important that the utterances of central banks are understandable.^[15]

Another important distinction, in my view, between the Delphic Oracle and central banks is that the accountability is confined to the goals given to the central bank. The central bank should not be regarded as a font of all wisdom. To do so risks compromising the independence in the areas that matter.

Conclusion

Central bank independence was in the ascendancy twenty years ago when the Bank of England was given its independence to pursue its inflation-targeting goal with complete instrument independence. The framework adopted in the UK was very much along the lines that Stan and I envisioned when we wrote on this topic a few years earlier, namely that a central bank should have goal dependence and instrument independence and should be held accountable for its decisions through the political process. The subsequent economic performance of the UK economy validates that choice of regime. The macroeconomic outcomes through the remainder of the 1990s until 2007 certainly do. Whatever failings led to the financial crisis and its long-lived consequences, I am not convinced that central bank independence was a contributing factor. Indeed, I think the experience of the UK economy through the past decade underlines the value of the independence given to it.

I do not regard central bank independence as fighting yesterday's war. The theoretical and empirical underpinnings of central bank independence remain as valid today as they did twenty years ago. The links between monetary policy, fiscal policy and financial stability policy may be more prominent now than they were two decades ago. One can question the central bank's understanding of the functioning of the macroeconomy and the monetary policy transmission process, but that is a separable issue from the validity of independence. That said, the questioning of the framework itself is a vital quid pro quo for independence. The accountability of central banks to the political process and the public more generally is critical. To be able to take independent decisions about the appropriate stance of monetary policy, a central bank has to appropriately justify them.

While the combination of goal dependence and instrument independence in terms of monetary policy has stood the test of time, the equivalence in terms of financial stability is more an open question. Both the goal of financial stability and the instruments are much less clearly defined at present. While central bank independence in terms of monetary policy was all the rage in the first part of the 1990s, a similar focus in terms of financial stability today is warranted.

Endnotes

Thanks to Stan Fischer for his education, example, guidance and support over many years. Thanks to Ed Balls and Bruce Preston for helpful discussions. Ed Balls and his co-authors have recently written a paper covering many of the same issues. [*]

Debelle G and S Fischer (1994), ‘How Independent Should a Central Bank be?’ in Jeffrey C Fuhrer (ed), Goals, Guidelines, and Constraints Facing Monetary Policymakers, Federal Reserve Bank of Boston Conference Series no.38, North Falmouth, Massachusetts, pp 195–221. [1]

Munchau M (2016), ‘The End of the Era of Central Bank Independence’, FT.com site, 14 November. Available at <https://www.ft.com/content/8d52615e-a82a-11e6-8898-79a99e2a4de6>. See also W den Haan, M  Ellison, E Ilzetski, M McMahon and R Reis (2017), ‘The Future of Central Bank Independence: Results of the CFM-CEPR Survey’, 10 January and Fels J (2016), ‘The Downside of Central Bank Independence’, PIMCO Macro Perspectives, May. [2]

Adam Posen reached the same conclusion in his PhD thesis. See Posen A (1995), ‘Central Bank Independence and Disinflationary Credibility: A Missing Link?’ Federal Reserve Bank of New York Staff Reports no 1, May. [3]

Barro R and D Gordon (1983), ‘A Positive Theory of Monetary Policy in a Natural Rate Model’, Journal of Political Economy, 91(4), pp 589–610. [4]

Bernanke B, T Laubach, R Mishkin and A Posen (1999), Inflation Targeting, Princeton University Press. [5]

Rogoff K (1985), ‘The Optimal Degree of Commitment to an Intermediate Target’, Quarterly Journal of Economics, 100(4), pp 1169–89. [6]

Debelle and Fischer (op cit) p 198. [7]

Walsh C (1995), ‘Optimal Contracts for Central Bankers’, American Economic Review, 85(1), March, pp 150–167. [8]

Cukierman A (1992), Central Bank Strategy, Credibility and Independence: Theory and Evidence, MIT Press; Grilli V, D Masciandaro and G Tabellini (1991), ‘Political and Monetary Institutions and Public Financial Policies in Industrial Countries,’ Economic Policy, 13 (October), pp 341–392; Alesina A and L Summers (1993), ‘Central Bank Independence and Macroeconomic Performance: Some Comparative Evidence’, Journal of Money, Credit and Banking, 25(2), pp 151–162. [9]

Haldane A (ed) (1995), Inflation Targeting, Proceedings of a Conference, Bank of England, London. [10]

King M (2016), The End of Alchemy, Little Brown. [11]

Balls E, J Howat and A Stansbury (2016), ‘Central Bank Independence Revisited: After the financial crisis, what should a modern central bank look like? M-RBCG Associate Working Paper Series, No 67, Harvard Kennedy School. [12]

King M (2016), op cit. [13]

See Haldane A (2017), ‘A Little More Conversation, A Little Less Action’, available at <http://www.bankofengland.co.uk/publications/Documents/speeches/2017/speech971.pdf>, Federal Reserve Bank of San Francisco, Macroeconomics and Monetary Policy Conference, 31 March. [14]

That said, the opacity of the Delphic Oracle is a matter of some historical contention. [15]