Recessions and recoveries in New Zealand's post-Second World War business cycles

ABSTRACT

We present preferred classical business cycle turning points for New Zealand's post-Second World War economy, using the Bry-Broshan dating algorithm on a long term quarterly time series of real GDP. From these, we identify nine recessions and their associated recovery paths for a period approaching 70 years and provide evidence on their key characteristics. Two key specific findings are as follows: (1) on average, real GDP and employment cycles have been associated around 90 per cent of the time and (2) the strength of New Zealand's business cycle recoveries has been independent of the depth, duration, or severity of the preceding recession.

KEYWORDS:

• Classical business cycles
• growth cycles
• employment cycles
• recessions
• economic recoveries
• New Zealand

JEL CLASSIFICATION:

• E01
• E24
• E32

Acknowledgments

The views expressed in this paper are those of the authors and do not necessarily reflect the views of the Reserve Bank of New Zealand. We thank two anonymous NZEP referees and an RBNZ Discussion Paper referee for perceptive comments, and acknowledge audience and discussant contributions from a presentation to Motu's Public Policy Seminar Series on 23 October 2013. We are also grateful to Brian Easton for his very helpful insights on key data periods, and for supplying his Haywood and Campbell diffusion index data during our construction of the preferred Hall and McDermott (2011) data series.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. For example, see Kim, Buckle, and Hall (1994, Table 1) for New Zealand and RBNZ 'world' real GDP having percentage standard deviation volatilities of 3.64 and 1.41 for the period 1966q4–1990q1. See also McCaw (2007, Table 1) for the period 1987q2–2006q2, where the standard deviation ranges for HP800, HP1600 and HP3200 real GDP are reported as 1.24–1.67 for New Zealand, 0.95–1.19 for Australia and 0.85–1.08 for the USA.

2. The Hall and McDermott (2011) quarterly real GDP data series for the updated period 1947q2–2014q3, and the linked quarterly Chapple (1994)–RBNZ–SNZ total employment series dating from 1956q1 to 2014q3 are available in Hall and McDermott (2015, Tables A1 and A2). The linked real GDP series reflect the SNEQ series released by Statistics New Zealand on 18 December 2014.

3. This BB algorithm was written in RATS by Dr Kunhong Kim. The initial version of the program was written to replicate successfully Bry and Boschan monthly results, and was then adapted for quarterly data to reflect what King and Plosser (1994, p. 411) have described as BB's handling of quarterly data in a way similar to that of Burns and Mitchell (1946), ‘… by simply setting each month of the quarter equal to the quarterly value and proceeding to set the series as monthly’.

4. For detailed results, see Hall and McDermott (2015, Table 3).

5. It can also be noted in this two exceptions context that (for their data series to 2006q2) Hall and McDermott (2009, Table 1) included in their ‘benchmark’ turning points a peak at 1958q2 and a trough at 1959q1. This reflected the BBQ program not including a smoothing element. The BB program used in this paper includes the smoothing element, and so leads to what was then a very marginal call of a three-quarter 1958–1959 contraction by the BBQ method not being called by the BB method.

6. An alternative approach to producing close to real time business cycle turning points, but which is computationally much more demanding than the technical-recession method, would be to compute a coincident index for the New Zealand business cycle. See Kim, Buckle, and Hall (1994, pp. 147–49) for previous coincident index work for New Zealand.

7. The NBER Business Cycle Dating Committee, which was formally created in 1978, does not make real-time decisions on designating its Peaks, Troughs and recessions. Rather, it waits for all relevant data to become available, for the publication of early data revisions, and for what it considers a sufficiently long period of time for a Peak or Trough not to be in doubt. It therefore also has no fixed timing rule for announcements. For example, its announcement of the December 2007 Peak took place 11 months afterwards (on 28 November 2008), while the June 2009 Trough was not declared for 15 months (on 20 September 2010). See NBER (2010).

8. For detailed results, see Hall and McDermott, 2015, Tables 3, 7.

9. Neither the BB nor BBQ methods picked these 1975 and 1989/1990 technical recessions, though as stated in section 2.2.1 above the BBQ method did call 2010q3 and 2010q4 as a two-quarter recession.

10. The linked series used in Hall and McDermott (2014) included real GDP production based data from 1987q2 to 2013q3, released by SNZ on 19 December 2013 in their SNDQ series; the series used in this paper reflect the updated and substantially revised SNEQ series released 18 December 2014.

11. For detailed results, see Hall and McDermott, 2015, Table 5.

12. These are SNZ's SNCQ, SNDQ and SNEQ releases.

13. For recent work on the robustness of New Zealand's key business cycle facts, see Hall, Thomson, and McKelvie (2014).

14. Chapple's HLFS-consistent series were published as de-seasonalised, but our graphing of his employment series showed that there still remained a very significant seasonal pattern. Accordingly, the results we present reflect our having run the employment series through Eviews’ X13 program. See Figure 2, bottom panel, and Table A2 in Hall and McDermott (2015), for the resulting X13 seasonally adjusted linked total employment series.

15. On the relatively unusual nature of these two cycles, and the cautionary comments on our real GDP series observations prior to 1954, see Hall and McDermott (2011, s. 6)

16. For an assessment of relative timing of 64 time series, including total employment, with respect to a deviation reference chronology over the period 1947–1974, see Haywood and Campbell (1976).

17. Hall (2011, pp. 431–432) defines slumps broadly as ‘extended periods of low resource utilisation’, and identified them specifically as periods when ‘… the employed fraction of the labor aged 25 through 54 … was less than its normal level of 95.5 per cent of the labor force’. Thus, it would last from when employment falls below its normal level during a contraction phase and continue through to when employment regained its normal level during an expansion phase.

18. For recent narrative commentary on factors associated with New Zealand's post-Second World War recessions, see Easton (1997, 2009) and Reddell and Sleeman (2008), brief paragraphs in Hall and McDermott (2009, ss. 3, 4, 6; 2011, s. 5.2), and analysis in Hall and McDermott (2015, ss. 4.3, 4.4) of key real GDP expenditure components for the post-1991q2 and post-2009q2 recovery and expansion paths.

19. For example, see McCaw and McDermott (2000), who establish from Concordance analysis that, for the period 1960–1999, the proportion of time that New Zealand's business cycle was in phase with those of Australia and the USA was approximately 70 per cent.

20. This opposite average two-year recovery pattern for the USA for the updated period has been maintained, despite the recovery from its most recent 2009 trough having been atypically slow (see, for example, Dominguez and Shapiro (2013), DS). DS have attributed this slowness primarily to successive financial shocks from Europe during 2010, 2012, and especially 2011.

21. The significance at the 10 per cent level of the growth-rate-during-contraction variable in the one-year and two-year cumulated growth rate equations is attributable to two outlier observations, and especially so to the observations for the 1948 recession.