WHY AND HOW TO CONSTRAIN THE DISCRETION OF CENTRAL BANKS

The economic theory case for CBI rests on the argument of time inconsistency (Barro and Gordon Reference Barro and Gordon1983; Kydland and Prescott Reference Kydland and Prescott1977). Any government agency that cares both about inflation control and about employment will be tempted to use inflation surprises to stimulate employment. Rational agents anticipate this and thus revise their inflation expectations and wage demands upward. The result is an unchanged level of employment with higher than necessary inflation. By contrast, an independent central bank is able to credibly commit to an inflation target.

Although the details of the time inconsistency argument are subject to debate (e.g., Forder Reference Forder1998; Goodhart Reference Goodhart1994), there is consensus in theory and practice that monetary policy faces commitment problems and that granting independence to the monetary policy authority can help overcome them with the right institutional design (see Keefer and Stasavage Reference Keefer and Stasavage2003).

For an agency such as a central bank to be considered independent, several conditions need to be fulfilled. Those running the agency need to have job security, and the agency needs to have control over policy instruments as well as some autonomy in determining its budget (Tucker Reference Tucker2018, 11). Despite some institutional variation, modern central banks tend to meet these criteria.

From a systemic perspective, removing certain policy choices from elected officials can deliver substantive benefits. However, this strategy also bears risks. As highlighted in the introduction, the creation of independent agencies removes the policy choices in question by one step from democratic control. Whereas citizens can collectively change governments, independent agencies are not directly accountable to voters. As Jon Elster puts it, “very independent courts and banks may be a remedy more dangerous than the disease” (Elster Reference Elster1994, 66–7).

Two issues have to be distinguished here. First, as highlighted by the extensive literature on optimal principal-agent contracts, there is the question of when delegation to bureaucrats is efficient in the sense of utility enhancing for citizens (e.g., Alesina and Tabellini Reference Alesina and Tabellini2008). Models in this category make assumptions about the incentives of politicians (e.g., reelection) and bureaucrats (e.g., career concerns) to then identify circumstances under which delegation tends to be or tends not to be preferable. Legitimacy is either not a concern in this literature or quickly dealt with by requiring that the mandate of IAs remain subject to review and reform by elected officials (ibid., 444).

The second issue, and focus of this letter, is precisely the question of under what circumstances the policy-making of an IA can be considered legitimate. To isolate this issue, I shall assume that we are dealing with a policy area where delegation to an IA is efficient, as is plausibly the case in monetary policy. The challenge at hand is thus to square the circle between the independence of institutions such as central banks on the one hand and the need to ensure their legitimacy on the other.

For a political body, independent or not, to be legitimate, it needs to provide a justification of the coercive political power it exercises over citizens (cf. Peter Reference Peter and Zalta2017). Intuitively, for equal levels of coerciveness, an IA will need to answer to a higher standard of justification compared with elected officials, due to its independence.^{Footnote 2} Applied to the context of monetary policy, a justification that goes beyond the kind of efficiency-considerations discussed above is required for handing independent central banks the power to print money, influence interest rates, and—via inflation—effectively tax economic agents.^{Footnote 3}

In other words, a system of checks and balances is needed to ensure that the IA does not abuse its power and remains sensitive to the political preferences of the population despite not being up for reelection. Traditionally, these checks and balances include both ex ante measures such as limiting central banks’ mandate to “operational” as opposed to “goal” independence and ex post controls such as parliamentary hearings.

The important contribution of Tucker (Reference Tucker2018) lies in providing a much more detailed framework for legitimate delegation to IAs, and to central banks in particular. Tucker’s account has three components (Reference Tucker2018, appendix): delegation criteria identifying a series of necessary conditions for such delegation to be legitimate—for example, that delegation only occurs after wide public debate, that “society’s preferences regarding the policy issue in question are reasonably stable,” and that the “IA will not have to make big choices on distributional trade-offs or society’s values”; design precepts detailing how delegation should be structured to ensure legitimacy over time—for example, provisions concerning the specification of IA objectives, procedures to be followed, as well as the transparency of the IA’s policy decisions; and multiple-mission constraints explaining why IAs should be given multiple objectives only under specific circumstances—for example, that the objectives be “intrinsically connected,” each subject to a commitment problem, and that combining them in one agency “will deliver materially better results.”

Tucker’s strategy is to create the conditions for the legitimacy of IAs despite their independence by imposing institutional constraints on the IAs themselves as well as on the process of delegation. Tucker’s account is rich in detail and represents a welcome contribution to an under-researched question. However, the point of this letter is not to provide an overall assessment of it, but rather to zoom in on one central delegation criterion that strikes me as problematic in the context of central banks and that raises important substantive questions concerning the delegation to IAs and their legitimacy.

According to Tucker, a key precondition for IA legitimacy is that the agency does not take big choices on distributional trade-offs. The rationale for this constraint is that decisions that create winners and losers need to be subject to democratic control and, thus, should not be removed from elected officials. IAs lack legitimacy on such issues.

This letter accepts the plausibility of this argument but suggests that at least in the context of central banking but potentially more generally, it presents us with a dilemma. Given that contemporary monetary policy has significant distributive consequences,^{Footnote 4} we either have to refrain from delegating monetary policy to an IA, thus risking higher than necessary inflation, or we have to delegate anyway, violate Tucker’s delegation criteria, and risk a monetary policy authority that lacks legitimacy.^{Footnote 5}

The next section will underpin the empirical premise of this dilemma, namely, the idea that monetary policy decisions do entail big choices on distributional trade-offs.

THE DISTRIBUTIVE DIMENSION OF MONETARY POLICY

Although monetary policy inevitably creates winners and losers, the nature and magnitude of these effects changed with the financial crisis of 2007. The distributive effects of conventional monetary policy pre-2007, at least outside of Japan, were limited to differentially affecting savers and debtors via their effect on the interest rate. The post-2007 unconventional monetary policies, by contrast, have more significant distributive consequences.

Both so-called quantitative easing, that is, the outright purchase of financial assets by central banks from the financial sector, and low-interest loan programs such as the European Central Bank’s (ECB) Long-Term Refinancing Operations (LTROs) have injected massive amounts of liquidity into the economy. Their intended effect, through the portfolio balance effect, is for this injection to stimulate investment and consumption. In practice, a substantive portion has been used to buy existing assets instead, thus leading to booms on real estate and stock markets and to unintended distributive consequences (e.g., De Haan and Eijffinger Reference De Haan and Eijffinger2016; White Reference White2012). Many central bankers, including Tucker, acknowledge this. The Governor of the Bank of England, Mark Carney, for instance, stated that “the distributional consequences of the response to the financial crisis have been significant” (Reference Carney2014).^{Footnote 6}

The reactions of central bankers to this state of affairs are instructive and show that they are well aware of the threat the distributive consequences of unconventional monetary policy pose to their legitimacy. Let me distinguish three types of response here.^{Footnote 7}

First, central bankers are quick to emphasize that distributive concerns do not figure in their mandate. This point is well taken, and some, perhaps most, of the blame for neglecting these unintended consequences should thus be directed at the politicians formulating the mandate instead. However, note that this issue is orthogonal to what concerns us here: Even if distribution is not part of central banks’ job description, distributive side effects of monetary policy can still undermine central bank legitimacy.

Second, central bankers say they had no choice (cf. Fontan, Claveau, and Dietsch Reference Fontan, Claveau and Dietsch2016, 336–7). The distributive consequences of unconventional monetary policy are the price we had to pay for saving the financial system from collapse. Note that this argument entails the claim that no other policy instruments were available that could have avoided financial meltdown without incurring inegalitarian consequences.

Third, several central banks have recently published articles that suggest the overall distributive effect of their crisis response has not exacerbated inequalities and has potentially even had an equalizing effect (Ampudia et al. Reference Ampudia, Georgarakos, Slacalek, Tristani, Vermeulen and Violante2018; Bunn, Pugh, and Yeates Reference Bunn, Pugh and Yeates2018). To understand this claim, we have to take a more detailed look at various channels of redistribution of monetary policy. Although the literature identifies up to six such channels (e.g., Colciago, Samarina, and de Haan Reference Colciago, Samarina and de Haan2019, 5; Coibion et al., Reference Coibion, Gorodnichenko, Kueng and Silvia2017, 81–5), what matters for our purposes is the distinction between two kinds of effects monetary policy has on distribution. “Direct effects include the impact of the different paths for nominal and real interest rates on households’ savings incentives […] and on households’ net financial income. […] The indirect effect operates through the general equilibrium responses of prices and wages, hence of labor income and employment” (Ampudia et al. Reference Ampudia, Georgarakos, Slacalek, Tristani, Vermeulen and Violante2018, 5).

Based on this distinction, central bankers argue that the inegalitarian direct effects of unconventional monetary policy on asset prices (see above) are compensated, or even outweighed by the egalitarian indirect effects on employment and growth. Note, however, that this line of reasoning accepts that monetary policy does have distributive effects. This turns the spotlight back on the second response above, claiming that monetary policy space was limited and that the precise configuration of distributive side effects was inevitable. Yet, might there be monetary policies that produce the desirable effects on employment and price stability without the undesirable distributive consequences? If so, then the central bankers’ argumentative strategy fails. The third response by central bankers ignores this relevant counterfactual. Pointing out that things could have been worse in terms of inequality had central banks done nothing does not establish that things could not have been better in terms of inequality without prejudice to price stability, or achieved a better compromise between the two objectives, had they acted differently.

Three observations are in order at this point. First, it can indeed be argued (e.g., Fontan, Claveau, and Dietsch Reference Fontan, Claveau and Dietsch2016, 336–7) that central bankers did not give serious consideration to ostensibly less inegalitarian policy alternatives, such as for instance helicopter money (e.g., Blyth and Lonergan Reference Blyth and Lonergan2014). Second, it is one thing to claim that monetary has no significant distributive impact, but quite another to claim that it has significant distributive consequences through various channels of redistribution, but that these different effects more or less cancel each other out. The latter strikes me as a rather ambitious empirical claim. Third, even if the empirical claim held for a certain monetary policy mix, choosing this policy mix over another still involves making choices with distributive consequences.

In sum, it seems fair to say that unconventional monetary policy of the kind we have seen since the financial crisis inevitably involves choices that have significant distributive consequences. Even if the precise direction and magnitude of these consequences is subject to an ongoing debate, it is in the very nature of the massive liquidity injections that central banks have resorted to at the zero lower bound of interest rates that they have an important distributive dimension. Incidentally, this holds for alternative policy instruments such as helicopter money, too, though their distributive impact would likely be very different.

Given that the present economic recovery has not been accompanied by a rise in interest rates, and given the difficulty in normalizing central banks’ balance sheets, it is also plausible to anticipate that unconventional monetary policy and its significant distributive implications are here to stay.

Therefore, both in the present and for the foreseeable future, monetary policy is set to violate Tucker’s delegation criterion, stating that independent agencies should not make big choices on distributional trade-offs. How should one respond to this tension from an institutional design perspective?

INSTITUTIONAL DEVICES FOR CENTRAL BANK LEGITIMACY IN THE FACE OF DISTRIBUTIVE TRADE-OFFS

Commentators on the role of IAs in democracies have often assumed “the possibility of separating efficiency and redistributive concerns because such a separation is crucial to the substantive legitimacy of regulatory policies” (Majone Reference Majone1996, 296). The previous section has demonstrated this assumption to be problematic when it comes to contemporary monetary policy. Even beyond monetary policy, the distributive dimensions of regulation are often neglected (see Robinson, Hammitt, and Zeckhauser Reference Robinson, Hammitt and Zeckhauser2016), though considering other policy fields in any depth lies beyond the scope of this letter.

What are the implications of this observation for delegation to IAs, and to central banks in particular? Can we overcome the dilemma identified at the end of Section 1 between politicized and thus suboptimal monetary policy on the one hand and delegated but illegitimate monetary policy on the other and, if so, how?

Given the trend towards delegation to IAs, this question deserves considerably more attention in the literature than it has received thus far. The goal of this last section will be to sketch a menu of available options of how our societies might manage the tension between delegation and legitimacy in light of the distributive impact of monetary policy.

The first option is to bite the bullet and simply accept the distributive impact of monetary policy and that it undermines the legitimacy of independent central banks. This strikes me as just as unattractive as the other horn of the dilemma, namely suboptimal monetary policy formulated by politicians.

Second, and more constructively, one might envisage a more fine-grained division of labour and coordination between central bankers and elected officials when it comes to monetary policies that will foreseeably have significant distributive consequences. One example in this category is the choice of assets included in quantitative-easing-style purchasing programs. Since being included in such a programme brings significant advantages to the emitters of the assets in question, the choice of assets is a deeply political question (Dietsch, Claveau, and Fontan Reference Dietsch, Claveau and Fontan2018, 31–2). An asset purchase programme that is both optimal and legitimate might be one where the central bank decides the size of the stimulus, but elected officials decide its direction, that is, which assets are included. Another example for better policy coordination is the establishment of joint committees composed of central bankers and representatives of the finance ministry, as were common in the postwar era.

Third, the mandate of central banks could be widened to incorporate the distributive side effects of monetary policy into the decision-making of central bankers. For instance, one might require that whenever unconventional monetary policies with a foreseeably significant distributive impact are adopted, this impact should receive a to be specified weight in the central bank’s policy-making (see Fontan, Claveau, and Dietsch Reference Fontan, Claveau and Dietsch2016, 342–3). Advocates of narrow mandates focused exclusively on price stability are likely to object that such an arrangement would undermine the effectiveness of monetary policy, but the literature does not seem to support this objection (see Dietsch, Claveau, and Fontan Reference Dietsch, Claveau and Fontan2018, chap. 2).^{Footnote 8}

The fourth option shifts our focus from central banks as agents to governments as their principals. Once governments are aware of the distributive side effects of monetary policy, one might argue, it is their responsibility to respond to mitigate these effects. One way to ensure they live up to this responsibility would be to enact a distributive neutrality commitment on the part of the government. This commitment would require automatic fiscal corrections for any distributive consequences above a certain threshold that monetary policy might have. One argument against this option is that redistribution is always costly, and that it might be preferable to prevent the distributive side effects of monetary policy from occurring in the first place.

This list of potential ways to reconcile independence and legitimacy in the face of distributive trade-offs is by no means exhaustive.^{Footnote 9} Furthermore, what I have presented here are mere ideas that will need to be fleshed out in theory and tested in practice to assess their merits. This is precisely the point of this letter: To highlight a serious gap in our theories about delegation to IAs when it comes to the distributive side effects of their policies. These side effects undermine the legitimacy of IAs, including central banks. To reduce this tension, we can either reinforce the democratic control over IAs, albeit without undermining the initial rationale for delegation (see option 2), or control and mitigate the distributive consequences of IA policies (see options 3 and 4).