Regional President Speech

2014 INTERNATIONAL RESEARCH FORUM ON MONETARY POLICY MARCH 21, 2014 Disclaimer • The views expressed in this talk are my own. • They may not be shared by others in the Federal Reserve System ... • Especially my colleagues on the Federal Open Market Committee. Acknowledgements I thank Ron Feldman, Terry Fitzgerald, Samuel Schulhofer-Wohl and Kei-Mu Yi for comments. Monetary Policy and Financial Stability • Major element of monetary policy conversation: Easy monetary policy could create risk of financial instability. • My view: It is preferable to mitigate such risks using supervisory tools. • But in reality: Supervision may leave residual systemic risk. How should this residual risk affect monetary policy? This Talk • A framework to incorporate systemic risk mitigation into monetary policymaking. – Theme: Systemic risk creates a mean-variance trade-off for policy. • A suggestive calculation based on the framework. Outline 1. A Mean-Variance Framework 2. Suggestive Calculation 3. Conclusion A MEAN-VARIANCE FRAMEWORK Simple Model • Monetary policymaker (MP)’s goal is to set a gap X equal to zero. – X could equal inflation minus target . – X could equal natural unemployment rate (UR) minus actual UR . • Note well: X is based on macroeconomic outcomes. • MP can increase X by raising accommodation A. • After MP chooses A, X is also affected by a number of shocks, cluding shocks to the financial system. in The Central Banker’s Problem 2 • MP’s loss is given by the square of the gap (that is, X ). – Standard: MP wants gap to equal zero. – Equally bad to have positive or negative gaps. • Recall: X depends on shocks realized after A is chosen. • MP chooses A so as to minimize the mean loss associated with A: 2 Mean(X |A) Usual Approach • Mean loss equals squared mean gap + variance of gap: 2 [Mean(X|A)] + V ar(X|A) • Typical assumption: MP can’t influence variance of shocks. • Then, minimizing expected loss is same as minimizing squared mean gap: 2 [Mean(X|A)] ∗ • Solution is to choose accommodation A that eliminates mean gap: ∗ Mean(X|A ) = 0 Incorporating Financial Stability Risks • Suppose higher A increases the risk of financial instability that lowers X. • Then, higher A increases V ar(X|A). • MP’s problem is to choose A so as to minimize: 2 [Mean(X|A)] + V ar(X|A) • Now: MP’s choice of A trades off mean versus variance. Mean-Variance Trade-Off ∗∗ • Trade-off means that MP’s appropriate choice A will result in: ∗∗ Mean(X|A ) < 0 • That is, on average, the gap is negative under appropriate policy. • MP gives up some mean X in order to get less risk in X. • But exactly how much mean X should MP give up? Comparing Two Monetary Policy Alternatives ∗ • It is appropriate for MP to choose A over A if A reduces risk suffi- ∗ ciently relative to A : ∗ 2 V ar(X|A ) − V ar(X|A) > Mean(X|A) • Central banks know a lot about assessing the RHS – that is, the mean of X given choice A. – In my view: The RHS remains large for current choice of A. • Key question is about the LHS: How do we assess the difference in the risk implied by policy choices? A Possibly Helpful Simplification • Suppose that a crisis causes the gap X to fall by ∆. • Suppose that monetary accommodation A implies that the probability of a crisis is p(A). • Then (assuming statistical independence of the crisis from other shocks): ∗ ∗ 2 V ar(X|A ) − V ar(X|A) ≈ [p(A ) − p(A)]∆ ∗ • Then: Given any policy choice A or A , we need to assess: The implied probability of a crisis and its impact ∆ on X . SUGGESTIVE CALCULATION Crisis Impact • Assume: the natural UR is approx. 5% in 2017. ∗ • Assume too that, under current policy A , projected 2017 UR is 5%. ∗ – That is, E(X|A ) = 0 in 2017. • Suppose too that a financial crisis would generate 2017 UR of 9%. • In other words: The impact ∆ of a crisis is 4%. According to the Survey of Professional Forecasters ... • How likely is a crisis? As of 2014:Q1, the average SPF prediction is that: Pr(UR ≥ 9% in 2017) = 0.29% ∗ • So, if A is current monetary policy: ∗ p(A ) ≤ 0.0029 – t’s an inequality because there are noncrisis sources of high UR. I (Implausibly) Highly Effective Monetary Policy (cid:48) • Suppose monetary policy A eliminates any chance of a crisis. (cid:48) (cid:48) • That is, A is a policy such that p(A ) = 0. • Then: ∗ (cid:48) 2 [p(A ) − p(A )]∆ = (0.0029)(0.0016) 2 ≈ (0.0022) (cid:48) ∗ ∗ • Should the FOMC be willing to adopt A over A (when E(X|A ) = 0)? (cid:48) • Only if the (implausibly effective) policy A doesn’t increase projected gaps too much. • Simple calculation: nly adopt tighter monetary policy A (cid:48) f: O i (cid:48) A raises UR to less than 5.22%(!!). • Main take-away: Current SPF forecasts imply that Little benefit to reducing or eliminating the probability of a crisis. CONCLUSIONS Financial Stability Framework: What We Need To Know • Mean-variance framework implies that policymakers need to assess: (cid:48) V ar(X|A) − V ar(X|A ) • Possibly could simplify this problem to gauging: (cid:48) 2 [p(A) − p(A )]∆ Assessing Crisis Probabilities • Key measurement questions: what is the probability of a crisis? • Current SPF forecasts suggest that it is very low under current policy. • Some might argue that professional forecasters tend to underestimate probabilities of tail events. • It would be useful to develop other approaches: – Model-based probability assessments of tail events – And market-based probability assessments of tail events

Narayana Kocherlakota (2014, March 20). Regional President Speech. Speeches, Federal Reserve. https://whenthefedspeaks.com/doc/regional_speeche_20140321_narayana_kocherlakota

@misc{wtfs_regional_speeche_20140321_narayana_kocherlakota,
  author = {Narayana Kocherlakota},
  title = {Regional President Speech},
  year = {2014},
  month = {Mar},
  howpublished = {Speeches, Federal Reserve},
  url = {https://whenthefedspeaks.com/doc/regional_speeche_20140321_narayana_kocherlakota},
  note = {Retrieved via When the Fed Speaks corpus}
}