RDP 2024-04: Nowcasting Quarterly GDP Growth during the COVID-19 Crisis Using a Monthly Activity Indicator Appendix A: Additional Monthly Activity Indicator Details

Luke Hartigan and Tom Rosewall

July 2024

A.1 Additional MAI dataset details

Table A1: Monthly Activity Extended Dataset
1978:M2–2022:M7
No Series Source Category Start date End date Transformation code
1 Total employment ABS Hard 1978:M2 2022:M9 LD
2 Full-time employment ABS Hard 1978:M2 2022:M9 LD
3 Part-time employment ABS Hard 1978:M2 2022:M9 LD
4 Unemployment rate ABS Hard 1978:M2 2022:M9 FD
5 Underemployment rate ABS Hard 1978:M2 2022:M9 FD
6 Hours worked ABS Hard 1978:M2 2022:M9 LD
7 Job advertisements DoE Soft 2006: M1 2022:M9 LD
8 ANZ Job vacancies ANZ Soft 1999:M7 2022:M9 LD
9 NAB Business conditions NAB Soft 1997:M3 2022:M9 LV
10 NAB Profitability NAB Soft 1997:M3 2022:M9 LV
11 NAB Trading conditions NAB Soft 1997:M3 2022:M9 LV
12 NAB Employment NAB Soft 1997:M3 2022:M9 LV
13 NAB Forward orders NAB Soft 1997:M3 2022:M9 LV
14 NAB Stocks NAB Soft 1997:M3 2022:M9 LV
15 NAB Business confidence NAB Soft 1997:M3 2022:M9 LV
16 NAB Capacity utilisation NAB Soft 1997:M3 2022:M9 LV
17 AiG Performance of manufacturing index Ai Group Soft 2001:M5 2022:M9 LV
18 AiG Performance of services index Ai Group Soft 2005:M9 2022:M9 LV
19 AiG Performance of construction index Ai Group Soft 2003:M2 2022:M9 LV
20 New company registration rate ASIC Soft 1978:M2 2022:M9 LV
21 ANZ-Roy Morgan Consumer financial situation next year ANZ-Roy Morgan Soft 2008:M10 2022:M9 LV
22 ANZ-Roy Morgan Consumer confidence index ANZ-Roy Morgan Soft 2008:M10 2022:M9 LV
23 WMI Consumer family finances next 12 months WBC-MI Soft 1978:M2 2022:M9 LV
24 WMI Consumer sentiment index WBC-MI Soft 1978:M2 2022:M9 LV
25 Retail trade ABS Hard 1982:M4 2022:M8 LD
26 Sales of new motor vehicles VFACTS Hard 1978:M2 2022:M7 LD
27 Revenue passengers, international, inbound DoT Hard 1985:M1 2022:M7 LD
28 Goods and services credits ABS Hard 1978:M2 2022:M8 LD
29 Goods and services debits ABS Hard 1978:M2 2022:M8 LD
30 Building approvals – residential, private ABS Hard 1978:M2 2022:M8 LD
31 Building approvals – housing, total ABS Hard 1978:M2 2022:M8 LD
32 Building approvals – other dwellings, total ABS Hard 1978:M2 2022:M8 LD
33 Building approvals – alterations and additions, total ABS Hard 1978:M2 2022:M8 LD
34 Building approvals – non-residential, total ABS Hard 1978:M2 2022:M8 LD
35 Auction clearance rate CoreLogic Soft 2008:M5 2022:M9 LV
36 Credit – total RBA Financial 1978:M2 2022:M8 LD
37 Credit – housing RBA Financial 1978:M2 2022:M8 LD
38 Credit – other personal RBA Financial 1978:M2 2022:M8 LD
39 Credit – business RBA Financial 1978:M2 2022:M8 LD
40 3-month bank accepted bills/negotiable certificates of deposit RBA Financial 1978:M2 2022:M9 FD
41 Yields on Australian government bonds – 3-years maturity RBA Financial 1992:M6 2022:M9 FD
42 Yields on Australian government bonds – 5-years maturity RBA Financial 1978:M2 2022:M9 FD
43 Yields on Australian government bonds – 10-years maturity RBA Financial 1978:M2 2022:M9 FD
44 Yield spread, AGS 3-years less 3-month bank bill RBA Financial 1992:M6 2022:M9 LV
45 Yield spread, AGS 5-years less 3-month bank bill RBA Financial 1978:M2 2022:M9 LV
46 Yield spread, AGS 10-years less 3-month bank bill RBA Financial 1978:M2 2022:M9 LV
47 AUD trade-weighted index RBA Financial 1978:M2 2022:M9 LD
48 S&P/ASX 200 Bloomberg Financial 1978:M2 2022:M9 LD
49 Index of commodity prices RBA Financial 1978:M2 2022:M9 LD
50 Home value index CoreLogic Financial 1980:M1 2022:M9 LD
51 SWIFT customer-to-customer RTGS RBA Financial 1998:M10 2022:M9 LD
52 Credit card payments RBA Financial 1985:M1 2022:M8 LD
53 Debit card payments RBA Financial 1994:M5 2022:M8 LD
Notes: ‘ABS’ is Australian Bureau of Statistics, ‘DoE’ is the Department of Employment, ‘DoT’ is the Department of Transport, ‘WBC-MI’ is Westpac and Melbourne Institute. ‘Transformation code’ indicates the method used to transform the data to be stationary if necessary, ‘FD’ indicates first difference, ‘LD’ indicates log difference and ‘LV’ indicates level.
Figure A1: MAI Dataset – Targeted Predictors
Top 30 series by Wald statistic
Figure A1 shows a single panel bar chart. Each bar is a series from the dataset. The series are ranked by the Wald statistic. The ranking is from largest to smallest. A higher value means that the series is more able to explain movements in quarterly GDP growth. Consumer sentiment has the highest rank while Retail sales has the lowest rank. There is a mix of Hard, Soft and Financials data categories.

Note: Dashed line represents the χ [ 3 ] 2 ( α=0.1 ) critical value = 6.25.

Sources: ABS; Ai Group; ANZ; Authors' calculations; Bloomberg; CoreLogic; NAB; RBA; Westpac and Melbourne Institute.

Figure A2: Pattern of Data Availability by Category
Figure A2 shows a single panel of three stacked columns chart. Each column represents a data category. The sample period is 1978:M4 to 2023:M6. It shows the availability of the series used to estimate the MAI by data category. Generally, there are more Hard data series initially, but a big increase in Soft data occurs in the late 1990s due to the start of publication of many NAB business indicators. The full dataset of 30 series does not eventuate until 2008.
Figure A3: Correlogram – Quarterly GDP Growth
Figure A3 shows a two panel line and column chart. The top panel shows the estimated Autocorrelation function (12 columns) and the 95 per cent confidence interval (two dashed lines). The bottom panel shows the estimated Partial autocorrelation function (12 columns) and the 95 per cent confidence interval (two dashed lines). In both panels, no column is outside the 95 per cent confidence interval. This suggests that quarterly GDP growth in best characterised as being close to a white noise process.

Notes: GDP is first release. Dashed lines represent 95 per cent confidence intervals.

A.2 Additional MAI estimation details

Figure A4: Estimated Number of Dynamic Factors
Log criterion using penalty p3
Figure A4 shows a two line chart which is used to determined the number of dynamic factors. This suggests there is one dynamic factor in the targeted predictor dataset.

The number of dynamic factors is determined by looking for the second ‘region of stability’ in relation to Sc (i.e. a value of 0) and checking which value of qc this corresponds to. See Hallin and Liška (2007) for more details.

Figure A5: MAI Weighting
Top 10 by total
Figure A5 shows a single panel bar chart. Each bar is a factor loading for a series from the dataset. The series are ranked by the total sum of the factor loadings which is also represented by a point. The ranking is from largest to smallest. A higher value means that the series is more correlated to the factor. NAB employment has the highest rank while Total credit has the lowest rank.

Sources: Ai Group; Authors' calculations; NAB; RBA; Westpac and Melbourne Institute.

Figure A6: MAI and Underlying Targeted Predictor Dataset
Figure A6 shows a single panel chart with 31 lines. Each line is a series from the targeted predictor dataset. The MAI is also shown on top. The sample period is 1978:M4 to 2022:M6. The MAI provides a summary of the movements in all the 30 individual series.

Note: Grey lines represent the underlying ‘targeted predictor’ dataset; blue line represents the MAI.