In 2021, the Bank of Japan (BoJ) released a paper, supporting their newly minted measure of inflation – the Price Sentiment Index (PSI). Breaking away from a long-standing tradition, the BoJ turned to alternative data as a way to forecast (or, in some sense, nowcast) inflation.
While we have yet to receive any data on its reliability as it has been originally proposed only in 2018, the Price Sentiment Index seems to be a promising invention in a field where they may come few and far between. Surprisingly, the data collection process for PSI isn’t all that complicated.
It may seem that such an index has use only for the central banks of the world. Price sentiment, however, may have applications in ecommerce. While the smaller businesses in the industry might have a harder time harnessing the benefits, large marketplaces could gain previously unseen advantages.
Bank of Japan’s approach to PSI
Before we get into the ecommerce applications of PSI, there should be some explanation on what the BoJ did and what results followed. Unlike one might expect, the process wasn’t that complicated and unachievable.
There was one outstanding advantage, though. The Cabinet Office, the arm of the Bank of Japan that supports policy planning, conducts an Economy Watchers Survey (EWS), which is a fancy name for the collection of firms’ views on inflation and price predictions.
In short, the EWS is a monthly survey that queries about 2050 people each time and receives about 1800 valid responses. These responses come from people that the BoJ calls “economy watchers”, that is those who are employed in positions which get to deal with prices and changes regularly – from business managers to grocery clerks.
All of the collected answers are separated into four categories by a machine learning model:
- Comments implying inflation.
- Comments implying deflation.
- Comments implying zero inflation/deflation.
- Comments not referring to price developments.
It’s important to note that Japan has been dealing with deflation for nearly two decades and only recently seem to have emerged out of the death-spiral, making comments 2 and 3 much more important than in any Western country.
Calculating the PSI out of the data is relatively simple, as it’s defined as the number of comments implying inflation minus the number of comments implying deflation divided by the number of comments 1, 2, 3.
The BoJ concludes that the PSI is an effective measure, albeit of short-term inflation, which is subject to greater variation than others. For long-term inflation measures, the CPI might be better and subject to less variation. PSI, however, gives a better outlook for nowcasting practices.
While the BoJ has the advantage of having access to lots of personal data with the protection of the government, there’s nothing stopping large ecommerce marketplaces from sending out similar surveys to their customers and partners, as long as it’s done within the rule of law.
Reapplying PSI to ecommerce
Since the Economy Watchers Survey is a relatively simple tool, any business with some capital to invest can do the same in nearly any country. There are even dedicated companies that will take over the survey process, according to the needs of the buyer.
One of the benefits of PSI is that it’s intended to measure the outlook of price expectations of companies. In other words, of those that set the prices at the production level. Therefore, their opinions and forecasts are much more important than, say, those of consumers.
Two important outputs are created during the production of PSI for ecommerce giants. First, if it’s performed regularly, just like the BoJ does, it provides insight into the changing prices at several levels. As raw material costs impact production, the survey questions might unveil any future pricing changes on several levels, allowing companies to prepare better.
Additionally, if any of the Economy Watchers indicate that they are running up costs due to the difficulty of acquiring raw materials, it’s likely that the issue is industry-wide (or even global). If the issues haven’t hit the ecommerce company that has created the survey, they will likely eventually catch up, making room to ease into the price hike.
On the other hand, since PSI decently reflects short-term inflation and price trends, it can be used by ecommerce companies to predict competitor actions and adapt accordingly. If the Economy Watchers imply inflation, it’s likely to reverberate on all ecommerce companies.
As such, competitors are extremely likely to raise prices on affected products in order to not lose out on margins. Someone with the knowledge can take several steps (such as raising prices slightly later after initiating a marketing campaign to acquire customers) to gain an edge against the competition.
Finally, changing price ranges can provide great predictive power and an outline for strategy. Some price hikes, as we have seen in the current days with, for example, oil, are quite large and might force some people to rethink their spending habits. Every user that suddenly shifts towards cheaper products or any alternatives might impact the business health of a company.
Conclusion
Certain inventions by large institutions remain for their own use only. The Price Sentiment Index, however, might have applications outside of the domain of central banks of the world. I’ve tried to outline how PSI might benefit ecommerce companies at large.
One addition I’d like to add before closing down the thoughts is that the BoJ doesn’t seem to apply natural language processing (NLP) for its Price Sentiment Index. Maybe they have discovered that NLP and sentiment analysis is not necessary, but for ecommerce applications, getting a grasp of how heavily Economy Watchers believe their statements might provide additional insight.
In either case, PSI is promising not just in macroeconomics and finance, but in smaller applications. Whether it will survive the coming years is something I will follow closely as it’s one of the most enticing inventions borne out of alternative data.
