Macroprudential Policy Under Uncertainty Saleem Bahaj & Angus Foulis 21st November 2016 The views expressed in this presentation are those of the presenter and not necessarily those of the Bank of England or members of the MPC, FPC or PRA Board. Introduction • Post-crisis, macroprudential policymakers have acquired a suite of tools whose impacts are largely untested… • …and a target variable which is inherently unobservable. • This work: how should policymakers react to this? – Is caution warranted? – Or does uncertainty require a more active stance? • What we do in the paper: suite of reduced form models to explore some of the relevant issues. • Key message: uncertainty does not justify inaction. 2 The results of Brainard (1967) Chart 1: Estimates of the GDP impact of a Chart 2: Swathe of Bank Balance Sheet 1ppt increase in Capital Requirements. Indicators (up=more vulnerable) Source: Macroeconomic Assessment Group Source: Bank of England 3 A stylised stabilisation problem Policymaker seeks to stabilise financial stability, • Where is a policy tool. • The impact of the tool, , is uncertain, • The term denotes a shock : . The The policymaker’s optimal choiceloss of function, is therefore:, is: 4 Other stories and mechanisms • But is that the whole story? Are there other ways uncertainty can influence how active policy should be? • In the paper we explore a few other mechanisms: – – – – Asymmetric preferences Robust control Learning Private sector uncertainty • These all could point towards more active policy 5 Is symmetric stabilisation the right way to think about macroprudential policy? • The costs of financial crises may be much larger than overly tight regulation. • Theoretical models that deliver endogenous crisis via occasionally binding constraints (e.g Bianchi and Mendoza (2010)) illustrate how severe asymmetries can emerge. • A stylised way to capture this is to have an asymmetric objective function for the policymaker. 6 An asymmetric loss function: Linex Policymaker seeks to stabilise financial stability, • Where is a policy tool. • The impact of the tool, , is uncertain, • The term denotes a shock The policymaker’s loss function, : . , is: 7 Graph of Linex Function 8 Results with asymmetric preferences 1) for and sufficiently large. • If the loss function is sufficiently asymmetric the policymaker should react to uncertainty about what it’s tool does by using it more to insure against the worse. 2) for and sufficiently large. • The same logic applies. 9 Targeting the worst case outcome (robustly) • The Linex function is an ad-hoc objective and still relies on known distributions. • Financial policymakers may face Knightian uncertainty • An alternative, microfounded (see Gilboa and Schmeidler (1989)) approach is to use robust control (minimax); where policy is set to avoid large losses in the worst case: • Place more weight on models that signal risks. • Already embedded within stress testing. 10 The robust control problem Policymaker seeks to stabilise financial stability, : • Where is a policy tool. • The impact of the tool, , is uncertain, • The term denotes a shock . The optimal policymaker’s choiceproblem of is therefore: is to choose in order to: 11 Graphical Illustration 12 Learning and Experimentation • A benefit of activism: The more a tool is used, the more we can learn about it’s impact. – Natural experiments can help but policy is still a learning by doing process. • Political constraints may be paramount. – We can’t treat the economy as a lab rat… – … but learning is an argument against inaction bias in the face of uncertainty. 13 Learning model Now imagine there are two periods policymaker has objective: And: As before, before choosing Tradeoff: use and the , but the policymaker sees : to stabilise or reduce •14 Optimal Policy Choices with Learning 15 Private Sector Uncertainty • The public are also uncertain about risks to financial stability, the impact of macroprudential policy, and how the nascent macroprudential policymakers will themselves behave. • These uncertainties can reduce the effectiveness of policy. • Specifically, the signaling power of policy is diminished if private sector is uncertain why the policymakers are acting. • A more uncertain public may necessitate a more active policy stance. 16 Setup: Private Sector’s Problem There are two states of the world, high and low. The private sector chooses a level of risk taking, . This increases the probability of the low state and the payoff in the high state, , but lowers the pay off in the low state, , specifically: Where is a variable chosen by the policymaker. •17 The Policymakers Problem The policymaker’s problem is identical with 3 exceptions: 1) The policymaker chooses and not . 2) The marginal payoff from in the high state is lower for the policymaker: . 3) There is an additional cost attached to the low state: •18 Information Sets The private sector has the following priors: The policymaker sector knows same prior over . and The policymaker moves first, hence over for the private sector: , but has the serves as a signal •19 Solution Private sector’s optimal choice is: Method of undetermined coefficients to solve for The final solution for is then: and With the parameters on and positive. Result: Greater private sector uncertainty over policy objectives results in more active policy: •20 Conclusion • Uncertainty comes in many forms and influences optimal policy in different ways. • There is not a general prescription that greater uncertainty requires less active policy. • If anything, in the situation currently facing macroprudential policymakers, uncertainty may call for greater action. •21
© Copyright 2026 Paperzz