Every January update of the consumer price indices is of particular interest since it sees the publication of meaningful inflation updates for package holidays after eleven months of essentially meaningless updates from February to December. The 12-month inflation rate for CPI package holidays for January 2019 was 5.9%. (This was up from 5.2% in December 2019, not that this means very much. It does tell us, though, that the price change for January 2018 to January 2019 was positive, as it would have been negative if there had been a deceleration.)
The 12-month inflation rate of 5.9% actually tells us that prices of package holidays were up by that percentage for fiscal year 2019, from February 2018 to January 2019, as compared to fiscal year 2018, ending in January 2018. The 12-month inflation rate for January 2018 was 4.9%, which tells us inflation picked up in FY 2018 as compared to FY 2017. There is no breakdown between UK and foreign holidays for the CPI, but there is for the RPI. From it we can glean that UK package holidays had a 5.0% inflation rate in FY 2018, up from 3.6% in FY 2017, while foreign holidays had a 6.4% inflation rate, up from 6.0%. (The fact that the CPI inflation rate lies between the foreign and UK holiday rates in both years is misleading. Foreign holidays are more heavily weighted in the RPI than UK holidays. A properly weighted average of the two would give an RPI inflation rate for all holidays of 6.1% for FY 2018, up from 5.5% in FY 2017. The holidays components of the RPI make considerable use of the Carli formula, so presumably the higher inflation rates are largely due to the formula effect.
The calculation of these fiscal year changes is fine. It appropriately gives monthly expenditure weights to items in this highly seasonal category, and out-of-season holidays just drop out of the index in those out-of-season months as they should do. Otherwise the indices are a mess. Paul Johnson noted vaguely that the indices lagged the rest of the index, which was surely a charitable evaluation under the circumstances. He also said an examination of the method should be made a priority. It obviously hasn't been, since four years have passed and we are no further ahead.
My post implied that the formula effect was the only factor affecting the difference between the inflation rates for holidays based on the CPI and the RPI in FYs 2017 and 2018. Of course this is not true and another important difference would be the shares of UK and foreign trips in the two indices. These are different in 2017 and 2018 for the RPI, and likely for the CPI as well.
The CPI is based on a domestic approach and would include UK holiday trips, as well as the presumably less common foreign holiday trips, booked by foreign residents in the UK. On the other hand it would exclude foreign holiday trips booked by UK residents abroad, which would be part of the RPI expenditure weights, although not part of its pricing. On balance, this would seem to suggest the RPI would give a lower share for domestic holidays than the CPI, and in FY 2017 and 2018 this would also tend to give it a higher inflation rate for total holidays.
Unfortunately, the situation is complicated, needlessly complicated, because the RPI excludes low-income pensioner households and higher-income households from the UK resident population. The low-income pensioner households likely don't spend much on holidays at all, but the higher-income households would likely spend a much greater share of their holiday spending abroad than the population at large. This wouldn't be a problem if the RPI had changed its target population to cover all resident households as was recommended by the 2012 Diewert report. If the RPI is to be reformed as recommended by the House of Lords Economic Affairs Committee, this would be a better place to start than by making changes in the clothing sample.
I would appreciate it if the ONS could provide relative shares for UK and foreign holiday trips in the CPI in 2017 and 2018, since presumably such internal weights are part of their system.
Thank you very much for highlighting the revisions to package holidays in the German CPI and HICP, which is much appreciated. I left a comment on your podcast at your notayesmanseconomics website so I will try just to add to what I wrote there.
It sounds like you know some German. I don't know any but using https://recnik.krstarica.com/ (designed primarily for translation between Serbian and other languages, but which does permit translation from German to English) it appears that the methodology change was, quite depressingly, a move from seasonal weights to fixed weights, not the other way around. It seems that previously "destinations were provided according to their seasonal demand with flexible weights. This meant for example that trips were more heavily weighted to the Balearic Islands in the summer than in winter." With the revision these flexible weights "were converted to fixed annual weights". Month-to-month changes can now be interpreted as "pure price changes", which wasn't the case with the preceding index. It also seems that there was a change in the sample to include package holidays to destinations that have no off-season, so twelve-month pricing is feasible. I asked one of our members, Andrew Lydon, who is fluent in German, if I have got this right, or if he has anything to add.
I have a copy of the 2007 book by the late Peter von der Lippe, "Index Theory and Price Statistics". His views on chain indices (see chapter 7) likely influenced the current German CPI approach, where the index is always a Laspeyres index for the most recent years and the index goes through multi-year revisions. Of course, if you are going to do this, it would probably make better sense to revise the index back to 2013 with a 2015 basket, but so far the Germans haven't done this.
On p.410 of his book, von der Lippe says: "while Israel abandoned the M method [in 1988 in its CPI], it is still in use for example in Germany". Von der Lippe treats the "M method" as synonymous with the Rothwell formula, so perhaps this is what was used to measure holidays in the German CPI prior to this revision. It would be good if this could be verified. This is the formula that I recommended to replace the existing ersatz seasonal weighting approach to package holidays in the RPI.
Eurostat only mandates class-confined seasonal weights in the HICP, a crude and unsatisfactory simplification of the Rothwell formula. Although it has no authority over how member countries calculate their national CPIs, this may nonetheless be having a pernicious impact, with France, and now perhaps Germany, moving away from seasonal weights. Instead of doing this, both countries should have been extending their use of seasonal weights to other categories. France did not have seasonal weights for package holidays, nor Germany for seasonal food groups. Neither had seasonal weights, as does the Netherlands in its CPI, for seasonal clothing.
Thank you again for telling us about this.
Thank you very much for this weighting information. You are the best!
Feb 15, 2019 12:57 AM Feb 16, 2019 2:39 AM
Feb 15, 2019 12:57 AM
Feb 16, 2019 2:39 AM