November overseas arrivals data from ABS has flagged a few concerns. Much interest in the Chinese numbers where growth has stalled.
I deliberately post this with the original data as I don’t think any analysis can be valid without reference to the extreme seasonal volatility. ABS applies a specific seasonal adjustment around the lunar Chinese new year. However you can still pick out that diversions in the seasonally adjusted series from the trend will typically be around the spikes. Note also that the secondary peak in Chinese visitation appears in July which is also a peak in education arrivals. This had quite an impact in 2017.
Chinese new year shifts this year from February 16th to the 5th. So we should see increased activity in January but a less volatile data period than the extremes of last year.
While everybody has been looking at the Chinese data almost nobody has been looking at the holiday visitor data which has been flaccid as a not too impolite description for almost two years now.
Perhaps someone should ask the proliferation of tourism marketing bodies we have in Australia “where the bloody hell are ya?” However this is also where we run into the curious problem of why TRA has been unable to provide purpose of visitor data for the IVS since December quarter 2017 attributed to problems with benchmarking this to the ABS data.
The ABS have referenced in commentary issues with data around VFR (visiting friends and relatives) since changes in data collection from July 2017. Not sure what the problem is given that ABS provide a linked page with the relevant arrival form which is just tick the box.
This brings us to Queensland where promising data earlier in the year have crashed in the last two months.
I have gone back to the original data here looking for the source of why there was previously an apparent trend change in Queensland relative entirely to the reverse trend in NSW approximately following a change in data collection methodology from July 2017.
The surprisingly symmetrical head and shoulders type volatility in early 2018 is almost entirely related to lunar shifts around Chinese new year and Easter. This is where you need the ABS seasonal adjustment special sauce. However the shift downwards in NSW is most significant.
The methodology around this changed from July 2017. It was previously departure data based on state of longest stay. The ABS boffins have done their thing to backdate the new series with a warning beyond ten years. To overlap the series back ten years.
Some variation going back but it looks fine? When you go back and look at the arrival form at ABS it’s difficult though to avoid the suspicion that this may be lower quality data with the relevant questions framed as intended address or contact details?
Commentary and trend analysis at Conus: Disappointing Short Term Arrivals data for Queensland