Do central banks with private shareholders differ in their financial behaviour from purely public central banks? Private shareholders might bias central banks toward focusing excessively on profits, dividends, and risks to their balance sheets. The enormous powers assumed by central banks during and after the Global Crisis have highlighted concern with these questions.1 In the recent US presidential campaign, for example, Senator Bernie Sanders and others recommended removing commercial banks as shareholders and directors of the Federal Reserve Banks (Sanders 2015).
These are not distinctively American concerns, however, since the Fed is not the only central bank with private shareholders. Many central banks originated as private banks, and in some cases this legacy persists. In addition to the US, the central banks of Belgium, Greece, Italy, Japan, South Africa, Switzerland, and Turkey all have private shareholders.
The rebuttal to the claim that private shareholders are likely to bias decisions in their own interest – that is, in the direction of policies that produce greater distributed profits – is that such tendencies are restrained by central banks’ governance rules. These rules limit the discretion of central banks to generate, state, utilise, and, specifically, distribute profits. Any tendency toward self-serving behaviour by private shareholders will consequently be restrained.
How much do private shareholders matter? As a start on addressing this question, we study 35 OECD central banks, including eight with private shareholders, using new data on governance rules. We find that central banks with private shareholders do not differ from their purely public counterparts in their profitability, nor are they more financially cautious in the sense of building more loss-absorbing capacity. Surprisingly, their transfers to governments out of current profits tend to be higher, not lower. We find that broader governance rules are more important than shareholding arrangements per se for financial payouts.
Private shareholding today
While dividend policies vary, private shareholders typically receive a very small payout compared to the governments (Figure 1). On average, dividend payments to private shareholders are less than 1% of total profit distributions in Turkey, Japan, Switzerland, and Austria. They are higher in the cases of the Federal Reserve, the Bank of Greece, the Bank of Italy and the Bank of Belgium, where they typically range from 2% to 10%.
Figure 1 Average annual dividend payments as a share of total profit distribution, 1994-2014 (unbalanced)
Institutional arrangements affecting private shareholders also vary across central banks. In general, private shareholders have no direct participation in policy decision making, but they possess voting and nomination rights and have the right to approve the central bank’s financial accounts. Voting and nomination rights range from the election of audit committee members to the selection of regents, counsellors and members of the board of directors. Most central banks with private shareholders do not allow private shareholding to exceed 50 per cent of share capital, although there are exceptions.
Some will argue that central banks with private shareholders are so heterogeneous that they cannot be treated as a group. We take an intermediate position on this issue. The premise of our approach is that the economically meaningful heterogeneity lies in central banks’ different financial-participation and governance arrangements. In our baseline estimates, we test whether the presence of private shareholders, modelled as a 0/1 dummy variable, has an effect on financial behaviour. In further tests we then distinguish between two forms of private financial participation: dividend rules based on current or past profits, and dividend rules limited to a fixed share of paid-in capital.2 Finally, we interact the presence or absence of private shareholders of these two types with different governance arrangements.
Do private shareholders make a difference?
Private shareholders are profit oriented. Our first hypothesis therefore is that central banks with private shareholders make higher profits.
Our sample comprises all OECD central banks plus the ECB and covers the period 1994-2014. When we control for a vector of other potential determinants of profits, central banks with private shareholders are found to be neither more nor less profitable than purely public central banks.3 The dummy for private shareholders remains insignificant when we distinguish large one-time positive shocks to profits (distribution of Swiss gold sale revenues for example).4
We also test whether the influence of private shareholders is constrained and affected by governance rules. Our data on governance are drawn from a specially tailored survey that we administered to our 35 central banks (for details, see Bartels et al. 2016).
We find that governance rules indeed matter for profit distributions. Central banks with discretion over profit distribution (those whose governance rules authorise them to retain some of their profits) distribute less to governments, as one would expect. However, this is not true for the subset of central banks with private shareholders – they tend to transfer more rather than less, though the total effect (the coefficient on the direct effect plus that on the interaction term with private shareholders) is not significant.
Five central banks with private shareholders possess this discretion – those of Belgium, Italy, Japan, Turkey, and Austria until 2010. It is probably not a coincidence that these institutions are also subject to income tax, since this is one way for governments to secure a share of profits. Of our 35 central banks, 25 are exempt from income taxation. But the probability that a central bank with private shareholders is subject to income taxation is higher. In fact, central banks with private owners and subject to income tax transfer a larger share of their profits to the government (the total effect – that is, the coefficient on the direct effect plus that on the interaction term with private shareholders – is positive and significant).
Fifteen of our 35 central banks, including four with private shareholders, have accounting flexibility – the ability to place unrealised gains and losses into a revaluation account, thus sheltering them from distribution. But rather than using this flexibility to reduce their transfers to the government, these central banks actually transfer a larger share of profits to the government. The total effect (again, the coefficient on the direct effect plus that on the interaction term with private shareholders) is positive and significant. The same holds for the ability to build provisions with no pre-defined upper limit.
In seven countries the government is required to recapitalise the central bank immediately if capital goes negative. While central banks with this guarantee might be less concerned about risks to their balance sheets and to make larger transfers to their governments in good times, this does not appear to be the case. In fact, central banks subject to this recapitalisation rule transfer less to the government, possibly because recapitalisation is costly politically and reputationally. Only one central bank with a recapitalisation rule also has private shareholders (Turkey’s), and the total effect is not significant.
Finally, in six cases the government unilaterally determines profit distributions. One would expect higher transfers, and this seems to be the case. The only central bank with private shareholders and this arrangement is that of Greece, and here also this has resulted in higher transfers (the total effect is positive and significant).
Overall, these results suggest that governance arrangements do impact the financial behaviour of central banks. This is in contrast with the insignificant results for private share ownership. Evidently, governance arrangements in general, and not private ownership per se, are what matter for central bank behaviour.
We find that the central banks with private shareholders do not differ markedly from purely public institutions in their financial behaviour. Central banks with private shareholders do not report higher profits. Nor are they more financially cautious in the sense of building more loss-absorbing capacity. If anything, they transfer a larger share of their profits to the treasury compared to purely public central banks. Even this result is not always robust, however. In particular, it is not robust to controlling for large one-off events like the 2004 Swiss gold sales.
In contrast, governance arrangements do impact the financial practices of central banks, although there is again no evidence that governance arrangements differ significantly between central banks with and without private shareholders. Evidently, governance in general, and not private ownership per se, are what matter for central bank behaviour.
Of course, we cannot rule out the possibility that private shareholders influence other aspects of central bank behaviour, for example their supervisory or interest-rate-setting decisions. Future research should analyse whether shareholding arrangements and governance rules also affect these other decisions.
Bartels, B, B Eichengreen and B Weder di Mauro (2016), “No Smoking Gun: Private Shareholders, Governance Rules and Central Bank Financial Behavior,” CEPR Discussion Paper No. DP11625.
Demertzis, M and N Viegi (2016), “Credibility of Central Bank(er)s,” VoxEU.org, 28 June.
Gurkaynak, R and T Davig (2015), “Perils of Central Banks as Policymakers of Last Resort,” VoxEU.org, 25 November.
Sanders, B (2015), “To Rein in Wall Street, Fix the Fed,” New York Times, 23 December.
 For discussion see inter alia Gurkaynak and Davig (2015) and Demertzis and Veigi (2016).
 In three central banks with private shareholders – those of Italy (until 2013), Greece and Belgium – the payouts to private investors are (partly) determined by the size of annual profits.
 Regression tables are available in Bartels et al. (2016).
 Note also that the control variables (not discussed here) mostly have their expected signs.