Japan April 2023 CPI Distributions and Models Update

Play it Again, Japan

This is our first CPI report for Japan, following the elevated requests. From now on, you should expect an email on Japanese CPI day. Research and some model developments are on-going. Nevertheless, what we show today has been monitored and tested for months.

Figure 1. Estimated MoM (saar) of core inflation measures.

Note: the figure shows the MoM seasonally adjusted at annual rate of three measures of “core” inflation for the Japanese CPI. The level of the series is seasonally adjusted in-house using X-13 (SEATS option).

The YoY of core measures is expected to tick up again in the coming months. The metrics for headline CPI as well as for 3 measures of core inflation are shown in Figure 2. The April report has brought the 3m/3m saar of the index ex fresh food and energy to 5.2%, and the one of the index ex food and energy to 3.4% (bottom panels of Figure 2). In both cases, prices at the margin are running well above the YoY, suggesting that the latter has room to tick up again in the coming months.

Figure 2. Metrics of Japanese CPI indexes.

Headline CPI

Index ex fresh food (BoJ)

Index ex fresh food and energy

Index ex food and energy (US-style core)

Evidence from the distribution

Distribution moved down. In April, the unweighted cross-section distribution of MoM (saar) changes across CPI items moved down, as shown in Figure 3. We are careful in taking signal from this movement for two reasons: first, the distribution shown in Figure 3 is the unweighted distribution; therefore, the number of goods items is proportionally higher than in the weighted counterpart. Second, the distribution reflects movements of 300+ items, none of which is published seasonally adjusted. Therefore, we had to seasonally adjust them one by one; but given the high number of items, we cannot exclude that some residual seasonality remains. For these reasons, we are still careful in interpreting the distribution in Japan.

Figure 3. Cross-section distribution of MoM (%, ar) of CPI items ex food and energy

Note: the Figure shows the median (black thick line) and the percentiles of the cross-sectional distribution of MoM saar items. The distribution excludes food and energy items. The levels of the series (about 300 in total) are seasonally-adjusted in-house using X-13 (SEATS) option. The percentiles shown are the 5th, 10th, 25th, 75th, 90th, and 95th.

Our proxies of the BoJ measures of underlying inflation

Room for BoJ underlying inflation measures to tick up again. Figure 4 shows the three measures of “underlying inflation” published by the BoJ (the blue lines). For each measure, we have calculated a proxy (the yellow lines) starting from the distribution of price changes. This is an on-going research project and we feel confident to improve the fit soon (we were able to get an answer from the BoJ about some technical details only last week). In any case, the takeaway from Figure 4 is that there is room for the three measures to tick up in the next release(s), especially for the trimmed mean and the weighted median.

Figure 4. BoJ measures of underlying inflation and our proxies.

BoJ trimmed mean and our proxy

BoJ weighted median and our proxy

BoJ mode and our proxy

Note: the figure shows the measures of “underlying inflation” of the BoJ and our proxies. All figures are YoY changes, in percentage points.

Medium-term forecast

The medium-term forecast points to upside risks. Figure 5 shows our model-based forecast for the three measures of core inflation. In the case of Japan we employ the model by BoJ Hogen, Kawamoto and Nakahama (BoJ review, 20215). We do not discuss the assumptions behind the results in Figure 5 for brevity but the data are available upon request, as usual. The takeaway is that, to us, Japan looks like a “Play it again” country after what we saw in the US, in the UK and in the EA. Especially the forecast of the index ex fresh food and energy looks like the US in late 2021 / beginning of 2022 when the models started to have troubles in getting the forecast converging towards target. Importantly, the models are (well) above the forecast of the policy board members of the BoJ (page 10 here), already recently revised up.

(Note: For the record, we have also produced forecasts based on our “main” Phillips curve model adapted to the Japanese case. The results are similar. Going forward we will show the results based on the BoJ paper. We do have medium-term model-based forecasts for headline CPI if of any interest)

Figure 5. Medium-term model-based forecasts.

Index ex fresh food (BoJ)

Index ex fresh food and energy

Index ex food and energy (US-style core)

Note: The figure shows the model-based forecast of headline CPI and three measures of core CPI. The model is based on Hogen, Kawamoto and Nakahama (BoJ review, 2015). All figures are YoY percent changes. The yellow shadows are intervals of confidence calculated as quasi-out-of sample exercises.

Implications for the BoJ

BoJ to hold… for how long? The BoJ in this moment reminds us of the Fed at the end of 2021. There is no strategy, except for hoping that prices will stop growing on their own. However, if history is of any guidance, with core inflation running as high as 6% (ar) at the margin, hope is really not a strategy because the models say that the music is likely to continue higher in the near-term. As for the BoJ, the focus seems to be on YoY measures and on the “underlying inflation” measures; all of them are well set to keep trending higher in the coming months. The BoJ can hold for a while.. but it should realize pretty soon that its forecast is, once again, too low. How long before it capitulates?

Contact

Want something more tailored?

We provide tailored consulting on ad-hoc projects.