Is the Taylor rule optimal? Evaluation using a wavelet-based control model

ABSTRACT

This paper extends the wavelet-based control (WBC) model of Crowley and Hudgins (2015) to compare the simulated performance of jointly optimal fiscal and monetary policy, where the policymakers place emphasis on a variety of macroeconomic targets, with the case where monetary policy follows a modified Taylor rule. The results show that the Taylor rule is likely to render higher interest rates, diminished investment, and appreciated real exchange rates compared with an unconstrained baseline simulation. We therefore find that the Taylor rule is suboptimal since it results in higher fiscal deficits due to substitution by policy authorities, thus confirming recent research on the interaction of fiscal and monetary policies. The analysis also compares the baseline and Taylor rule simulated forecasts when the central bank adopts a ‘hawkish’ inflation policy compared to a more ‘dovish’ policy stance.

KEYWORDS:

• Discrete wavelet analysis
• monetary policy
• optimal control
• Taylor rule

JEL CLASSIFICATION:

• C61
• C63
• C88
• E52
• E61
• F47

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ G6 interest rates are sourced from the OECD and US rates are sourced from the Federal Reserve. The G6 rates use real GDP in US$ weights sourced from either the IMF or OECD.

² Further details can be found in Crowley and Hudgins (2019). The shortest cycle is frequency range 1 (6 months to 1 year), with cycle length increasing until the longest cycle at range 5 (8 to 16 years).

³ The FOMC noted that ‘an inflation rate of 2 percent is most consistent over the longer run with the Federal Reserve’s statutory mandate.’ 19 December 2018, Federal Reserve. Bernanke (2015) and Kliesen (2019a, 2019b) discuss modifications of the TR as employed in practice by the Fed, and we have simulated the model under these modifications; however, these do not alter our main findings, and thus are not included in this paper. Our model uses interest rates on short-term US Treasury securities (3-month T-bill rates), which follow the Fed Funds rates closely. See the Fed data for details at https://www.stlouisfed.org/on-the-economy/2017/october/increases-fed-funds-rate-impact-other-interest-rates.

⁴ This balances a real interest rate of 2% with a productivity growth of 2%. For an annual population growth of 0.5%, this is consistent with an annual real GDP target growth of 2.5%.

⁵ The target interest rate is thus growing at a quarterly compounded growth rate of 0.04729. This approximates an interest rate response in the short-term bond market to series of eight semi-annual Fed discount rate increases by 25 basis points over the four-year horizon.

⁶ The literature suggests that younger policymakers that have not experienced high inflation periods, such as the 1970s, are more likely to be policy ‘doves’. Since the primary catalyst for a regime shift is past experience, it would be more likely in the horizon covered in this paper for the regime to shift from ‘hawk’ to ‘dove’ when younger policymakers replace older ones. The regime would also likely shift towards a ‘hawkish’ regime if the current inflation rate began to consistently exceed the Fed’s stated target. We have also simulated a regime shift at different points within the horizon, but this does not qualitatively alter our results.

⁷ To ensure the robustness of our results in a higher inflation environment, we also conducted simulations with inflation starting at 4% with a long run target of 2%, and our results were qualitatively identical in terms of signage and similar in terms of magnitudes of variation from baseline. These additional results are available from the authors upon request.

⁸ As a robustness check, the simulations were also re-run using a higher weight on the output gap in the TR. The results were signed identically as in Table 1, and the magnitude of results were similar.