When Should Regulators Bend? Time Consistency and Flexibility in Financial Regulation - NIESR

For some time, and more insistently under the current government, ministers have been pressing the UK’s financial regulators to do their bit for growth. The Financial Services and Markets Act 2023 gave the PRA and FCA a new secondary objective to facilitate the competitiveness and growth of the UK economy; remit letters to the Financial Policy Committee now routinely invoke it; reviews of ring-fencing, the leverage framework and mortgage rules are all in train. And yet we have not seen the kind of wholesale growth-focused recalibration some in government appear to want. The regulators have offered reviews, process changes and some adjustment, while the core post-crisis architecture remains largely intact.

There is a real argument on each side. The government was elected on a promise to lift the UK’s chronically weak growth, and it is not absurd to think an over-cautious regulatory state is part of the problem. The regulators, for their part, can point to a long history of crises that followed bursts of deregulatory enthusiasm. This is what makes the standoff more interesting than a morality tale about technocrats obstructing democracy.

One reading is that the regulators should bend. They are unelected; the government has a mandate; in a democracy the technocrats should not frustrate elected ministers. A second reading is almost the opposite: the whole point of independent regulators is that they should sometimes refuse. If they simply yielded whenever the political pressure rose, independence would be largely decorative.

I want to suggest that neither reading is quite right. The standoff reflects a genuine tension at the heart of financial regulation: between the value of tying policymakers’ hands and the value of letting them learn. A simple model of that tension helps explain both why regulators are often right to resist, and how a government that genuinely wants change should go about getting it.

It helps to step back and see that two quite different forces are in play. The first is a political cycle in regulation, and it is well documented. Politicians have a standing incentive to loosen prudential rules—mortgage limits are the classic case—to let credit and asset prices run, flatter near-term growth, and lift the polls, particularly before an election. The benefits are immediate and visible; the costs, in the form of future financial instability, arrive later and are diffuse.

This is a time-inconsistency problem. From a distance, we know we would like policy to resist the temptation to loosen the rules for short-term expediency. But when the moment arrives, the temptation is there. That is precisely why the FPC and PRA were built as they were after the financial crisis: to take away the punchbowl, as William McChesney Martin put it, just as the party gets going. That design values commitment—tying the policymaker’s hands against a temptation it will predictably face.

The second force pulls the other way. We do not actually know the right level of regulation. The costs and benefits of a given capital requirement or mortgage limit are genuinely uncertain, and they are likely to shift as the financial system evolves. We learn about these costs only slowly, often from experience. A rigid rule set once and never revisited would lock in our ignorance. That argues for flexibility—discretion to adjust as we learn.

So we want commitment to resist the political cycle, and flexibility to act on what we learn. These cut against each other. How should they be balanced? A simple model, borrowed from the economics of self-control, helps.

The framework of Amador, Werning and Angeletos (2006) provides a useful lens through which to understand this problem. They study a consumption-savings problem in which the individual values both commitment and flexibility: commitment because the future self may be tempted to overspend; flexibility, because the future self may also learn something relevant about its circumstances. We can apply the same logic to financial regulation.

Suppose the policy problem is to choose the stringency of regulation, k. Think of k as how tight mortgage or capital rules are. The higher k, the more stringent the regime. Society’s wellbeing depends on k and on a state θ that captures the genuine cost of tight regulation now—how much it restricts credit, entry, innovation or risk-sharing. But tighter regulation also has a benefit: it reduces the risk of a damaging future financial crisis.

A simple way to capture this trade-off is to start from the perspective of a benevolent social planner who wants to minimise society’s total loss. She minimises:

The first term captures the economic cost of stringent regulation; the second term captures the financial-stability benefit. The costs are assumed to be quadratic: the tighter regulation becomes, the faster the economic costs rise.

Now introduce the policymaker of the day, who may be tempted to loosen regulation for political-cycle reasons. We’re going to assume this policymaker minimises:

Where β <1, capturing the idea that the policymaker puts too little weight on future financial-stability benefits relative to the planner.

The optimal regulatory stance chosen by the social planner is:

This tells us that if financial crises are more costly, captured by higher a, the planner chooses more stringent regulation. But if regulation carries larger economic costs, captured by higher θ, the planner chooses less of it.

Contrast this with the regulatory stance chosen by the policymaker:

In other words, if left unconstrained, the policymaker chooses too little regulation in every state. That is the temptation commitment was meant to curb.

If we knew θ, the solution to this problem would be easy: we’d simply write a rule that set k = k* = a /θ for each state θ. But that is not the world we live in. The true costs of financial regulation are difficult to identify, and probably change as the financial system evolves. We cannot simply ask the affected firms; industry feedback contains information, but it is also predictably biased towards the conclusion that regulation is too costly. And there are complex general equilibrium effects that may well dominate the true calculus for society as a whole.

So what can the planner do? One natural move is to restrict the menu of choices available to the policymaker—the set of k’s it is allowed to choose from. In this application, this takes the form of a floor to regulatory standards. The policymaker is free to tighten regulation above that floor, but cannot weaken standards below a pre-set minimum, k. If the genuine case for stringent regulation is strong, discretion remains: the policymaker can choose a higher k. But if political pressure pushes towards excessive loosening, the floor binds.

Mapped back to financial regulation, the implication is straightforward. Some discretion is valuable because regulators should be able to adapt as we learn more about the costs and benefits of rules. But because the political temptation runs mainly in one direction—to weaken prudential standards in pursuit of near-term growth—that discretion should be asymmetric. The system should make it possible to tighten when risks rise, but should place a hard floor under core prudential standards.

The figure below illustrates the result. The blue hyperbola shows the first-best level of regulatory stringency the planner would choose if she knew θ. The red dashed hyperbola shows the level chosen under pure discretion by the policymaker of the day. The green line, with its hard floor, shows the second-best outcome when the planner cannot condition directly on θ.

Figure: Constrained discretion in financial regulation

The politically tempted policymaker chooses less stringent regulation than the social planner. A regulatory floor prevents standards falling below a minimum, while preserving discretion above the floor.

The first implication is that the current standoff between the government and regulators should not automatically be read as institutional failure. Seen through the model, the independent regulator is the device that enforces the floor. If ministers press for growth and regulators respond with reviews, marginal adjustments and process reforms, but resist a material weakening of core prudential standards, that may be the system working as intended. A commitment device that yielded whenever political pressure intensified would not be a commitment device at all.

That does not make the arrangement undemocratic. The commitment was created democratically, through statute, and can be changed democratically through statute. The proper route is therefore a deliberate and transparent change to the objectives, powers or constraints under which regulators operate, not informal pressure to reach a different judgement. The Financial Services and Markets Act 2023 did exactly this by giving the PRA and FCA a secondary competitiveness and growth objective. But secondary means secondary. Treating that objective as real, while preserving the primacy of the regulators’ primary objectives, is the statutory architecture doing its job.

None of this means the floor is currently in the right place. The post-crisis settlement was built for a banking-centred system, and as risk migrates to non-banks and private credit, parts of it may now be too high, too blunt, or aimed at the wrong margin. The point is only that the case for lowering it has to be made with evidence, in the open—which is exactly what distinguishes genuine learning from political convenience.

The second implication is that none of this is an argument against cutting red tape. Some regulation is simply deadweight cost: duplicative reporting, slow authorisations, unclear supervisory expectations, overlapping data requests, or rules that no longer bind the risk they were designed to address. Removing these costs does not pose the same time-consistency problem. If a rule imposes cost without adding resilience, there is no deep trade-off to manage. The growth agenda should start there: not with lowering the floor under core prudential standards, but with removing avoidable frictions around them.

The third implication is more subtle. The standard monetary-policy literature suggests another possible solution: delegate to someone more conservative than society as a whole. In Barro and Gordon’s classic model, discretionary monetary policy creates an inflation bias. Rogoff’s answer was to appoint a conservative central banker who dislikes inflation more than society does. The parallel in financial regulation would be to appoint regulators who put greater weight on financial stability than elected politicians do.

It turns out that if calibrated well, this conservative financial regulator solution can achieve the first-best in our simple model. To see this, suppose the conservative financial regulator minimises:

where λ>1 captures the degree of conservatism: the greater is lambda, the more weight the regulator places on financial stability benefits. Given this objective, the regulator will choose

If the planner appoints a regulator with λ =1 / β , we get back to the first-best:

So if the regulator is conservative enough to correct the political bias, but no more, the result can approximate the social optimum. But if the regulator is too conservative, regulation will be tighter than society would choose. The risk then is too little entry, too little innovation, too much compliance cost, and too much financial activity pushed outside the regulated perimeter.

But there is a deeper problem than over-shooting. Hitting the first-best this way requires the planner to know the size of the political bias and to set λ exactly right. In practice, that bias is neither observable nor constant. It shifts with the state of the world and the electoral calendar, and the regulator’s independence is under continual pressure. A single dial of regulatory conservatism, set once, cannot track a bias that varies across time. That is precisely why a floor still earns its place: it is the robust backstop for the times when the dial is mis-set, or quietly leaned upon.

This is also why the financial-regulation problem cannot be solved simply by appointing hawks. Nor can it easily be solved by a Walsh-style contract. In monetary policy, inflation is observable and can be written into an incentive scheme. In financial regulation, the object of concern is systemic risk, which is latent, multidimensional and often visible only after the event. Credit growth, competitiveness and GDP are not substitutes, because rewarding them too directly recreates the original political temptation inside the regulator.

So the practical answer is constrained discretion. Keep a hard floor under core prudential standards. Allow regulators to adjust above that floor as evidence changes. Require them to show where rules impose genuine deadweight costs and what they are doing to remove them. And if ministers want the floor itself lowered, they should make that case explicitly, with evidence, through the democratic channels that created it in the first place.