Don’t shoot until you see the whites of their, Ah, screw it! cut! Cut! CUT!

The Reserve Bank of New Zealand cut interest rates by 0.5% on Wednesday 7th August 2019. The equal biggest cut since the Great Recession in 2009! Clearly the New Zealand economy is going badly and things are expected to get much worse, right? Not according to the latest data on the Reserve Bank’s goals of employment and inflation. Instead the Reserve Bank is acting to head off potential bad future outcomes that it’s own forecasts do not think will happen. Setting monetary policy in direct conflict with your own evidence and expectations does not seem like a solid basis for good monetary policy.

Let us begin with the present. The Reserve Bank of New Zealand, henceforth RBNZ, has two official goals when setting the interest rate: maximise sustainable employment and inflation targeting. On employment, the employment rate at 67.7% is up over the last two years, currently reaching multi-decade highs. The unemployment rate at 3.9% is down over the last two years, currently the lowest since the Financial Crisis over a decade ago in 2007. Neither appears to have recently changed direction, with the latest labour market survey released just days before the RBNZs decision showing both continuing to improve. So clearly the decision to cut interest rates was not a reaction to current employment conditions. On inflation targeting, the RBNZ has a stated goal of “future annual inflation between 1 and 3 percent over the medium term, with a focus on keeping future inflation near the 2 percent mid-point.” The reading for headline inflation in the latest quarter is 1.7%, following 1.9, 1.9, and 1.5 in the previous three quarters. Underlying inflation was between 1.9 and 2.1%, according to Statistics NZ. With inflation well within the target range and about as near as feasible to the 2 percent mid-point the RBNZ was also clearly not reacting to current inflation conditions.

But efforts to measure the effects of monetary policy on the economy all suggest that the impact of a change in interest rates today on the economy will be roughly zero for the next six months with the main effects on inflation, employment, and GDP occurring between one and two years from now. So current economic conditions are anyway outside the control of the RBNZ and not what they should react to, except in so far as the present helps guide our understanding of where the economy will be in the near future. Understanding what is going on in the economy right now is also difficult, with noisy data making it undesirable to overreact to the latest measures; this is why the official wording of the inflation target is around “medium term” inflation. Any countercyclical monetary policy must therefore be based around expectations of where the economy will be a year or two in the future. So while current conditions are fine the RBNZ cut interest rates by 50 basis points because it expects things to get much worse, right? Wrong! The RBNZ forecasts inflation essentially equal to the 2% mid-point of it’s target range from mid-2020 on. The unemployment is rate is predicted to fall over 2020-21 (although since the forecast was not updated to reflect the current 3.9% it actually falls to higher than the current estimates). While the latest data on New Zealand GDP show a slight decrease in GDP growth, the forecasts show growth increasing again during 2020-21; in part because the latest Fiscal Budget states that the New Zealand Government will be spending more by then, in part as some current headwinds are expected to pass by then. It is important to note that the RBNZ forecasts are all based on a future that involved a smaller interest rate cut in the present than the 0.5% cut that was decided upon, as seen in their future forecasts for the policy interest rate (the OCR). To reiterate: since monetary policy is estimated to act with a lag the forecasts for one to two years from now are what matter most when setting monetary policy today.

The RBNZ press conference emphasized ‘asymmetric risks’ and ‘regret analysis’ as a key part of their decision. By regret analysis, they mean that they view the cost of making the mistake of setting rates too low if the future economy turns out to be booming as less than the benefit of setting low rates if the future economy turns out to be recessionary. These partly reflect a value judgement about the relative costs of high inflation versus high unemployment, and partly a view that societal welfare is a concave function of output (so increasing output in a recession leads to larger benefits than the costs of an equal sized decrease in output during a boom). That this value judgement is an opinion is worth recognizing, however it is one I largely share —adding a caveat about the important distinction between transitory and permanent shifts in inflation— so I will not be criticising it. What I will heavily criticise is that their ‘regret analysis’ appears to have been thought through only in a static one-shot context which is of little to no relevance to monetary policy. It completely ignores the fact that the RBNZ will be back again less than two months from now making another decision about monetary policy and interest rates. Any intelligent analysis requires recognizing that they regularly meet and decide on interest rates, and of the advantages and disadvantages of waiting versus acting now. Economists have long understood that even for a central bank that has only the best interests of the country as a whole in mind, and that is omniscient about the present and future, reacting as is best in the moment to current developments may have the longer-run effect of worse outcomes, increasing economic volatility (Kydland & Prescott, 1977); analysis based on deviations of actual policy from monetary policy rules (“Taylor rules”) suggests this does occur in practice (Belongia & Ireland, 2019). Adding in realistic issues of informational challenges around understanding what exactly is going on in the economy adds to the argument against monetary policy overreacting to current developments (Orphanides, 2001; Orphanides & Williams, 2007; Justiniano & Preston, 2010).

On the asymmetry of risks —what might be called the ‘what-if-there-is-a-trade-war’ scenario— there appears to be no serious consideration given as to how likely this bad scenario is to occur. All the RBNZ forecasts make it clear that it is less than 50-50 odds, but no thought appears to have been given to considerations that the bad scenario needs not only to be bad but also sufficiently likely to occur before it makes sense that it should have a large influence on current policy. This brings us full circle to the current economic data and RBNZ forecasts, all of which argue that the bad scenario is unlikely. This is not to say that the bad scenario won’t occur and it is possible that two years from now it will look like the RBNZ took the right decision by cutting the policy rate by 50 basis points just before the breakout of a major international trade and currency war. However the latest data and the RBNZ’s own forecasts do not expect this to happen. Possible future bad outcomes should form part of a well founded and intelligent monetary policy, but doing so requires a serious consideration of how likely they are to occur and of just how bad such outcomes would be. It should also consider possible asymmetries in the impact of a given monetary policy in terms of its effectiveness in each of those outcomes. This also gets at the core of my current criticism of the RBNZ’s decision to cut rates by 50 basis points: the criticism is not really about the actual 50 basis point cut, it is the massive and jarring disconnect between this cut and what the RBNZ forecasts for the New Zealand economy over the next few years. If the forecasts had taken a turn for the worse then the cut would be justified.

One possible explanation for the RBNZ’s decision is that they are confused about which country they are setting monetary policy for. As I have emphasized the most recent New Zealand data and the RBNZs own forecasts all point to an economy not as weak as the RBNZ had expected, and which the RBNZ expects to be growing solidly by 2020 with both (un)employment and inflation at target over the one-to-two year horizon at which the main effects of any change in current interest rate policy are thought to have their main impact. This stands in contrast to most other economies, with Australia, the Eurozone and China in particular among those with which New Zealand is closely linked economically showing definite weakness in both their latest economic data and –together with the US– showing inflation coming in substantially and persistently below their targets, while inverted bond-yield curves in the US point to an expectation that a future recession is likely, and expected inflation in the US below target according to market measures. But again, none of these —worsening (un)employment, inflation substantially below target and forecast to remain so by market indicators, and predictors of future recessions— is currently present in the latest NZ data nor in the RBNZ forecasts. In fact the RBNZ update from its previous forecast, and the RBNZ Governor during the press conference announcing the cut both pointed out that current growth was stronger than had been expected. The RBNZ forecasts include predictions of a weakening global economy, but do not think it will weigh heavily enough on the New Zealand economy to cause the domestic economy to become derailed in terms of the stated objectives of monetary policy of maximizing sustainable employment and keeping inflation on target. Obviously the New Zealand economy is small enough in a global context that actions by the RBNZ are not about to boost the global economy, but a quixotic attempt to do so, rather than paying attention to the domestic economy, would provide one rationale for the RBNZs decision. Hopefully it is superfluous to add that a good monetary policy should focus on the domestic economy, taking account of others mainly just in terms of how they will impact domestically.

During the announcement of the rate cut the Governor of the RBNZ, Adrian Orr, expressed the view that “monetary policy always needs friends….[including] Government to be spending“. The Treasurer Grant Robinson, discussing the rate cut on the radio stated that “this is what should happen, monetary and fiscal policy working together”. Sometimes it will be true that it is desirable for monetary and fiscal policy to work together, one possible example is providing stimulus during a large recession. But other times it is not desirable for monetary and fiscal policy to work together. As a heavily cooked-up example imagine a major earthquake occurring during an economic boom, it is likely desirable that the Government increases fiscal spending to help people deal with the aftermath of the earthquake, while monetary policy is contractionary to ensure that this increased fiscal spending does not cause the economy to overheat, nor cause inflation to take off. That there are many situations in which it is desirable for monetary and fiscal policy to work in opposition is evident in the design of most modern central banks, the RBNZ included, namely in their independence. If coordination were always desirable the independence of the RBNZ would be counterproductive and the Finance Minister, who is in charge of fiscal policy, would also take charge of monetary policy.

Low interest rates in New Zealand and globally potentially provide another justification for the 50 basis point interest rate cut. Namely if the RBNZ decided that the efficient real interest rate was actually lower than it had thought until now, and thus monetary policy until now had been tighter than intended, then it would make sense to cut the policy interest rate. This would reflect less of a change in policy, than an acknowledgement that with the better information now available it had become clear that the policy interest rate had been set higher than intended, and the present cut was simply correcting this. While the RBNZ statements do discuss low nominal interest rates —in particular low interest rates on 10-year government bonds in Australia, Germany, Japan, New Zealand, and the US— they contain no mention of low real interest rates and nothing to suggest that a decision that real interest rates are lower than had been believed, nor that was part of their reasoning in making the present interest rate cut.

The RBNZ thus seems to have made its decision based not on current developments but on a future scenario it considers unlikely to occur. It attempts to rationalize that decision based on a static “regret analysis” that views it as making a one-off decision for life, rather than recognising that it will be back again in two months time to make another decision. Such a policy of major reactions to unlikely futures is likely to lead to regular overreaction, acting to destabilize rather than stabilize the economy. This move to mediocre monetary policy is not a disaster —my own personal reading of the evidence is that the impact of good vs mediocre monetary policy on the economy matters but is not a major driver of longer-term economic and social wellbeing, although bad monetary policy can destroy countries. But the move to mediocre policy is a shift for the worse.

Enough complaints. How could the Reserve Bank of New Zealand do better? If policy is to be made based on asymmetries of risk then forecasts need to incorporate these. Transparent analysis of just how bad and how likely the outcomes of a recession are should form part of the justification for decisions like the latest 50 basis point cut. Among the more obvious and easier changes this means forecasts that incorporate risk and uncertainty.

In summary, the Reserve Bank of New Zealand has adopted a monetary policy of jumping at shadows: making decisions based on risks it does not see. This seems more likely to destabilize than stabilize the economy.

---

Some related links and notes:

A short official summary statement of the RBNZ’s decision.

The other 0.5% rate cut since 2009 was in 2011 and was made in immediate response to the Christchurch earthquake.

The RBNZ’s own forecasts and stated reasons for the decision can be found in the full Monetary Policy Statement for August 2019. The second page of the document contains the formal statement of the RBNZ’s objectives in setting monetary policy. You can also view the press conference announcing the decision.

The comment about monetary policy needing friends expressed by the Governor of the RBNZ could be interpreted as consistent with the independence of the RBNZ, however in context of the press conference as a whole I do not read it that way, and have understood it instead as calling for greater coordination between monetary and fiscal policies.

The interest rate set by the Reserve Bank of New Zealand is formally known as the Official Cash Rate (OCR). The RBNZ performs open market operations to get the interest rate on 90-day New Zealand Treasury bills to essentially equal the OCR.

The underutilization rate, a broader measure than the unemployment rate that aims to capture things like people who work part time but would like full time jobs was also down over the last few years and in the latest numbers.

Back in 2015-16 the headline CPI inflation rate in New Zealand was substantially below target. While inflation has not since gone above the 2% mid-point of the target range of 1-to-3% it is now well within the target range. This can be seen in the above graph of inflation. I mention this to observe that the view that inflation was notably undershooting the target did make sense just a few years back.

Statistics New Zealand chooses to measure underlying inflation as trimmed-mean CPI inflation. This is not my preferred measure, which would be CPI inflation excluding food and fuel. But beggars can’t be choosers.

Statistics New Zealand does not appear to produce confidence intervals for CPI inflation, so whether or not we can be confident in numbers like 1.7% being different from 2% or whether statistical uncertainty may mean we cannot be confident is unclear. Loosely related, Q2 2019 saw an adjustment in the way rental prices are measured in the CPI index. Whether this may have caused the present CPI inflation measure to be higher or lower this quarter than otherwise I am unsure (i.e., I do not know if the new measure is thought to report systematically higher or lower numbers than the previous measure).

To understand the argument about real interest rates note that it is the real interest rate that in principle matters most for economic decisions, not the nominal interest rate. Whether monetary policy is tight or loose thus depends on whether the real interest rate being set by monetary policy (the nominal policy interest rate minus the inflation rate) is higher or lower than what the real interest rate would be if interest rates were not being set by monetary policy (this later, the real interest rate in a world absent monetary policy is often called the efficient real interest rate, although in principle it need not be in any way related to efficient outcomes in the usual economic sense). Since the effective real interest rate cannot be directly observed, people attempt to infer it indirectly. Hence, changes in views on what the real interest is mean changes in our views on how tight or loose current monetary policy is, and would mean that we want to change the current monetary policy so that it is as tight/loose as had been intended/understood until now.

The evidence that the effects of monetary policy changes will have essentially no effect on the economy for the first six months and have the largest impact on GDP, inflation, and employment around 1 to 2 years from now comes from a variety of techniques. All of these techniques involve a combination of some way to measure the size of the monetary policy shock to the economy, together with a way to measure the impact. Early work by Bernanke & Blinder (1992) and Bernanke & Mihov (1998), and a summary of the resulting literature by Christiano, Eichenbaum & Evans (1999) focus on identifying monetary policy shocks by an assumption about timing and contemporaneous interactions (a Cholesky-decomposition), and then estimating the impact of these shocks using VARs (vector auto-regressions). VARs have also been estimated using assumptions about the sign (direction) of the effects of monetary policy shocks, rather than timing (Uhlig, 2005; Arias, Caldara & Rubio-Ramirez, 2019). Another approach focuses on using narrative evidence to identify shocks, and also then estimates their impact using either VARs or linear regression (Romer & Romer, 1989; Romer & Romer, 2004). Once an approach to identifying monetary policy shocks has been decided upon an alternative, and increasingly the standard, approach to estimating their effect is local projection methods (Jorda, 2005). Recent work has focused on other ways to measure the monetary shock, relating to changes in financial markets in the minutes immediately after central bank announcements (Kuttner, 2001; Faust, Swanson & Wright, 2004) and changes in futures markets prices on central bank interest rates (Barakchian & Crowe, 2013). In common among all of these approaches, the impact of changes in monetary policy on the economy is often summarized in the form of Impluse Response Functions (IRFs). One could in principle look at IRFs using an estimated DSGE model, but these would not help answer the current question as the model itself would determine much of the direction, and to a lesser extent, the time-horizon, of the impact of changes in monetary policy. A related issue is how much these estimates pick up the effect of current changes versus the effect of changes in expectations about future monetary policy (Cochrane, 1998; Gurkaynak, Sack & Swanson, 2005). Haug & Smith (2012) provide evidence specifically for the case of New Zealand based on some of these kinds of estimation techniques.

The RBNZ also emphasized in their press conference that they have performed an analysis to check whether the impact of monetary policy has weakened in an environment of low interest rates, and of keeping rates low for long periods of time. This does not actually appear to have been done in a meaningful sense. [Update: Turns out they had done so, but that the analysis was only published a week after the decision, and hence after this post had been written. As such it was unavailable at time of writing and was not explicitly mentioned Monetary Policy Statement.] The Monetary Policy Statement does report that recent interest rates have impacted on mortgage interest rates, and also mentions two RBNZ analytical notes: one finding that changes in interest rates are reflected in bond-yield curves, and the other shows based on a model simulation observes that the flattening of the reduced-form Phillips curve does not necessarily imply a flattening of the structural-form Phillips curve. But none of this analysis talks directly to present concerns in research on monetary policy which suggests that when interest rates a low for extended periods a kind of ‘low rate fatigue’ sets in and the impact of monetary policy changes on the actual outcomes of interest —(un)employment and inflation— are weakened (McKay & Wieland, 2019; Berger, Milbradt, Fabrice Tourre, & Vavra, 2018). It should be noted that this emerging research remains an open but suggestive finding and is far from a settled issue. It is worth observing that these findings are in direct conflict with the RBNZ Governors claim during the press conference that “holding an interest rate back in case you needed to have more in the future just makes no sense whatsoever“. Related to both this question on the effectiveness of monetary policy and the issues of asymmetric risk and regret analysis emphasized by the RBNZ is the question of whether the effectiveness of monetary policy is also asymmetric? Might monetary policy be less effective during recessions? If it is then this both reduces the benefits and increases the costs of reacting now to asymmetries in future risks. While early studies estimating this found the opposite, that monetary policy was more effective in recessions (Garcia, 2002), most work since that has led to an emerging conclusion that monetary policy is less effective in recessions, that a loosening of monetary policy has less effect than a tightening of monetary policy, and further that monetary policy is less effective during periods of volatility or following financial crises (Tenreyro & Thwaites, 2016; Vavra 2014; Angrist, Jorda & Kuersteiner, 2018). It also finds that the alternative options available to central banks when interest rates are near zero such as quantitative easing, negative interest rates, and forward guidance —known collectively as unconventional monetary policy— appear either as (QE) or less (forward guidance) effective than standard monetary policy (Krishnamurthy & Vissing-Jorgensen, 2011; Gertler & Keradi, 2013; Gambacorta, Hoffman, & Peersman, 2014; McKay, Nakamura & Steinsson, 2016; Inoue & Rossi, 2018; Eisenschmidt & Smets, 2019).

For a formal statement of conditions under which optimal monetary policy would mean ignoring foreign developments completely, see Galí & Monacelli (2005). While this is likely too strong in reality, it provides an extreme example of the more general point that monetary policy should mostly focus on the domestic economy, and foreign developments are of interest mainly indirectly via their impacts on the domestic economy.

References:
Angrist, Jorda & Kuersteiner (2018) – “Semiparametric Estimates of Monetary Policy Effects: String Theory Revisited“
Arias, Caldara & Rubio-Ramirez (2019) – “The systematic component of monetary policy in SVARs: An agnostic identification procedure“
Barakchian & Crowe (2013) – “Monetary policy matters: Evidence from new shocks data“
Berger, Milbradt, Tourre, & Vavra (2018) – “Mortgage Prepayment and Path-Dependent Effects of Monetary Policy“.
Bernanke & Blinder (1992) – “The Federal Funds Rate and the Channels of Monetary Transmission“
Bernanke & Mihov (1998) – “Measuring Monetary Policy“
Belongia & Ireland (2019) – “Rules versus Discretion: Inference Gleaned from Greenbook Forecasts and FOMC Decisions“
Christiano, Eichenbaum & Evans (1999) – “Monetary policy shocks: What have we learned and to what end?“
Cochrane (1998) – “What do the VARs mean? Measuring the output effects of monetary policy“
Eisenschmidt & Smets (2019) – “Negative Interest Rates: Lessons from the Euro Area“
Faust, Swanson & Wright (2004) – “Identifying VARS based on high frequency futures data“
Galí & Monacelli (2005) – “Monetary Policy and Exchange Rate Volatility in a Small Open Economy“
Gambacorta, Hoffman, & Peersman (2014) – “The Effectiveness of Unconventional Monetary Policy at the Zero Lower Bound: A Cross‐Country Analysis“
Garcia (2002) – “Are the Effects of Monetary Policy Asymmetric?“
Gertler & Keradi (2013) – “QE 1 vs. 2 vs. 3. . . : A Framework for Analyzing Large-Scale Asset Purchases as a Monetary Policy Tool“
Gurkaynak, Sack & Swanson (2005) – “Do Actions Speak Louder Than Words? The Response of Asset Prices to Monetary Policy Actions and Statements“
Haug & Smith (2012) – “Local Linear Impulse Responses for a Small Open Economy“
Inoue & Rossi (2018) – “The effects of conventional and unconventional monetary policy: A new approach“
Jorda (2005) – “Estimation and Inference of Impulse Responses by Local Projections“
Justiniano & Preston (2010) – “Monetary policy and uncertainty in an empirical small open-economy model“
Krishnamurthy & Vissing-Jorgensen (2011) – “The Effects of Quantitative Easing on Interest Rates: Channels and Implications for Policy”
Kuttner (2001) – “Monetary policy surprises and interest rates: Evidence from the Fed funds futures market“
Kydland & Prescott (1977) – “Rules Rather Than Discretion: The Inconsistency of Optimal Plans“
McKay, Nakamura & Steinsson (2016) – “The Power of Forward Guidance Revisited“
McKay & Wieland (2019) – “Lumpy Durable Consumption Demand and the Limited Ammunition of Monetary Policy“
Orphanides (2001) – “Monetary Policy Rules Based on Real-Time Data“
Orphanides & Williams (2007) – “Robust monetary policy with imperfect knowledge“
Romer & Romer (1989) – “Does Monetary Policy Matter? A New Test in the Spirit of Friedman and Schwartz“
Romer & Romer (2004) – “A New Measure of Monetary Shocks: Derivation and Implications“
Tenreyro & Thwaites (2016) – “Pushing on a string: US monetary policy is less powerful in recessions“
Uhlig (2005) – “What are the effects of monetary policy on output? Results from an agnostic identification procedure“
Vavra (2014) – “Inflation Dynamics and Time-Varying Volatility: New Evidence and an Ss Interpretation“