Recession is as recession does: updating the US NGDP and output gaps
Plus, US July #NFPguesses and defining potential growth down
If your definition of a recession is two consecutive quarters of negative real GDP growth, then the US is in one and that is true by definition. Of course, there are plenty of reasons not to like that definition, but at least be consistent about using it.
The other leading contender for defining a recession is the declarations made by the NBER’s Business Cycle Dating Committee, but the BCDC is not going to settle that argument in anything like real time. I actually like that fact that there is an independent academic arbiter for US business cycle reference dates, even if calling those dates ‘official’ seems like an overstatement. The fact that these dates are considered authoritative in the absence of any formal authority to declare them speaks well to the Committee’s reputation. If you are interested in the ins and out of the Committee’s process, I recommend David Beckworth’s interview with Jeff Frankel from 2017. But as Michael Strain and others have pointed out, there is very little daylight historically between the NBER business cycle reference dates and the two consecutive quarters of negative GDP growth definition when it comes to the question ‘was there or was there not a recession.’
If you want to DIY your own business cycle reference dates, you can always use the Bry-Boschan-Pagan-Harding algo and choose whatever parameters you like. In a previous post, I argued that classical business cycle dating methods had a good conceptual fit with NGDP level targeting, not least in terms of not having to remove a stochastic trend from the data in order to identify the cycle. The whole point of NGDP level targeting is to avoid a demand-induced classical recession. It is also a good conceptual fit with Friedman’s plucking model of the business cycle.
The obvious difficulty with the idea that the US is in recession is the cyclical low in the unemployment rate. But as the Sahm rule reminds us, it is the change in the unemployment rate rather than the level that matters. The unemployment rate is typically trending either higher or lower. Extended sideways moves are uncommon, but that is what we have seen from the US unemployment rate since March.
My model of the US employment rate has been pointing higher on an unrounded basis since March, but still thinks it should be 3.5%. Since the model’s forecast is unchanged on a rounded basis for the monthly of July, we are going to call it at 3.6% again, which is also where the market is at for the month.
For July non-farm payrolls, the model forecast is an above-market 347k (market 250k), which will only add to the cognitive dissonance on the recession call, not least because payrolls employment looms large in the NBER BCDC’s methodology. But I think that’s more of a knock on the idea that contemporaneous labour market outcomes are a good real-time recession signal rather than a rejection of the notion the US is experiencing a downturn. There are plenty of other confirmatory indicators, ranging from an inverted yield curve, to falling commodity prices and lousy sentiment data. The US 3-month/10-year spread seems to be confirming the signal from the 2/10.
A bigger problem for the recession view from my perspective is that the Mercatus NGDP gap increased from 3.4% to 4.1% in Q2, a level not seen since Q2 2000, when it was 4.4%. Moreover, the projected positive gap doesn’t close until Q2 2027.
The Berger, Morley and Wong real output gap was 4.9% in Q2, but is nowcast at 4.3% for Q3 2022.
Their output gap measure was notable for turning positive in Q2 2021 and provided a good lead on emerging inflation pressures, so it may also be a good indicator on the way down and certainly bears watching.
Defining potential down
Last week’s ministerial statement updated the Australian government’s economic forecasts from those contained in the Pre-Election Economic and Fiscal Outlook. Near-term nominal variables were revised up while real variables were revised down, which is consistent with an economy being hit with supply shocks. The expected mean reversion on commodity export prices was kicked-down the road to next financial year, which explains the shift in near-term expectations for nominal GDP. I much preferred when Treasury conditioned on current commodity prices and the terms of trade rather than muddying the waters with an in-house commodity price outlook.
Looking at the back-end of the forecast period, we have what look like equilibrium values for the inflation rate and the unemployment rate, so by implication, real and nominal GDP growth are also consistent with their equilibrium values. Note the associated expectation for real GDP growth is 2.5%, hence the nominal GDP growth assumption of 5%. This is a little lower than my suggested target path of 5.5% for nominal GDP growth. The difference turns on what would appear to be a lower official assumption for trend or potential GDP growth.
This continues a trend on the part of the official sector to define potential growth down. Partly, this is just fitting to recent historical growth outcomes. If your definition of trend growth is some historical average and that history is subject to a weakening trend, then expected potential growth must be lower. My concern is that what we are really fitting over is the recent historical public policy outcomes that have in turn determined trend GDP growth. In defining potential down, we may be encouraging policymakers to lower their sights for future economic growth.
The same issue will arise when the government releases the next Intergenerational Report, for which a downward revision to the long-term productivity assumption has already been flagged in media reports, from 1.5% to 1.2%. The 2021 report assumed productivity growth over the next 40 years will look the same as the last 40. In Treasury’s three “P’s” framework, because population and labour market factors tend to be something of a wash for growth in real GDP per capita, the productivity assumption all but fully determines the projection for future growth in living standards.
At the time of the last IGR, there was debate within Treasury about the productivity growth assumption, with some taking the view it should be trimmed to reflect the expected impact of climate change mitigation/adaptation. If media reports are accurate, the 2022 report will see that assumption trimmed by fitting to more recent historical averages.
For mine, the big question mark over the previous IGR’s productivity growth assumption was that it encompassed the big lift on economic reform in the 1980s and 1990s. The obvious question to ask of future productivity growth is where are we going to see a similar lift on economic reform? That is not to say the opportunity isn’t there, but whether that opportunity is seized is another question. Sam Bowman has a nice discussion about the Booster/Doomster divide in UK public policy and says that ‘the Doomster view seems very powerful at the Treasury.’ Needless to say, we need more boosterism. See also Bryan Caplan on the distributive distraction.
The productivity growth assumption is not unrelated to the concept of a neutral real interest rate. As RBA Governor Lowe argues, productivity trends will ultimately determine any trend in the neutral real rate. It is no accident that real rates have been declining alongside trend productivity growth. But as we suggested last week, in recent decades, tight monetary policy has probably not done either any favours. Glenn Otto published a paper back in 2007 which found that around 30% of the variation in the Solow residual in Australia could be attributed to demand shocks. Zac Gross has a nice discussion at his Substack on different approaches to the neutral interest rate, noting that the RBA is not clear about which concept it is using when discussing the neutral real rate.
For the record, the RBA’s August Board meeting decided to increase the cash rate target by 50 basis points to 1.85%. The RBA said that its central forecast is for CPI inflation to be around 7.75% per cent over 2022, a little above 4% over 2023 and around 3% over 2024. The Bank's central forecast is for GDP growth of 3.25% over 2022 and 1.75% in each of the following two years. The unemployment rate is expected to be around 4% at the end of 2024. The RBA will release a full set of updated forecasts in its August Statement on Monetary Policy this Friday.
One of my undergraduate teachers, Geoff Brennan, has passed away. Tyler Cowen had a nice appreciation of him, noting that ‘he is also one of the few people to have worked with Buchanan and come out of the experience intact.’ I think Tyler is right that he was under-rated by virtue of being Australian. My favourite memories of him: calling into his office at ANU to say hi and finding Loren Lomasky in the room with him and then having an hour-long chat with them both at around the time they were bedding down their book Democracy and Decision; also, watching Geoff and Harold Demsetz debating the Coase theorem at a Mont Pelerin Society meeting in Sydney in 2010. Geoff was playing devil’s advocate to Demsetz’s view of Coase, a role he was very good at. Geoff would always make you re-think your assumptions, but in the most charming way possible.
Warwick McKibbin on the review of the RBA and nominal income targeting.
Projecting the structure of US Treasury debt. A handy tool from Brookings.
China bought none of the extra $200 billion of US exports in Trump's trade deal.