Opinion February 2019

Broad, detailed and imprecise

The truth about Japanese statistics

Broad, detailed and precise. These were supposed to be the highly acclaimed virtues of Japanese economic statistics. Broad and detailed I can confirm from personal experience. In my previous job as a member of the economic forecasting team at the Mitsubishi Research Institute, Inc., analysing Japanese statistics was a vital and persistent part of daily professional life.

I also had plenty of opportunities to compare Japanese data with those of other countries and regions, as well as those compiled by interna­tional organisations such as the International Monetary Fund, the Organisation for Economic Co-operation and Development and the United Nations. Japanese figures excel all the way in terms of breath and detail. You ask for it, they can provide it. Even things that you do not dream of asking for, they are prepared to provide.

I can vouch for all of the above. Where I went wrong was to assume that broad and detailed automatically begets precision. It has now been revealed that, for well over a decade, some of Japan’s key economic statistics had been compiled using unauthorised methodology.

Who knew?

The problem first came to light with regard to the Monthly Labour Survey statistics put together by the Ministry of Health, Labour and Welfare. Those figures play a critical role in determining the level of unemployment and other welfare benefits. It now turns out that vast sums of underpayments have occurred over the years because of the faulty data provided by the survey in question.

Numerous other key statistical series have sub­sequently been found to contain compi­la­tion issues. Where all relevant firms were supposed to be surveyed, only a handful of limited samples were taken. Where onsite interviews were supposed to be conducted, only postal surveys were sent out.

It is not just that the methods used were faulty or illegal. The question now being asked in the Diet—as well as in media—is whether all this is a case of cock-up or conspiracy. Who knew what when? Who chose to ignore what where? Who decided to keep mum about what? These are the questions swirling around what people are yet to start calling Datagate, but may well begin to do so in the days to come.

In the trenches

It is not as though Japanese official statisticians do not deserve our sympathy. They are said to be vastly understaffed, which is probably very true. Broad and detailed weigh on them heavily. Sending out surveys to each and every large business sounds like mission impossible. Securing a relevant number of responses is no doubt even more an impossibility. Going out to ask questions up and down the country on the doorsteps of small businesses beleaguered by accounting issues and painful decision-making must be a stressful chore at best—and damaging to mental health at worst. Those long-suffering number crunchers are not Hercules. They can­not be expected to achieve the feats of that mytho­logical hero.

Moreover, they enjoy neither the acclaim nor the status of a demi-god. They are very far removed from centre stage in the theatre of bureaucratic performance. There is little incentive for them to make the impossible possible, and every incentive for them to opt for the possible in place of the impossible.

Needed reform

There is a good case to be made for revisions and simplifications to the way Japanese economic statistics are put together. There is absolutely nothing wrong with taking samples rather than checking up on each and every firm. When it comes to finding out the price of things, doorstep interviews are not an absolute must.

The problem with this whole saga is that the people concerned have not taken the trouble to point out the difficulties with existing methodology and to propose changes. They just took the easier way out. People in charge seem to have endorsed that option without actually acknowledging the fact. They turned a blind eye, in other words. The result is that precision became a sacrificial offering at the altar of broad and detailed without leading to any rewards, let alone heavenly blessings.

Personal experience

In light of this news, not-so-fond memories come back to me. I remember instances when I have struggled to make sense of strange move­ments in those broad and detailed statistics which now turn out to have been not so precise. I would try so hard to spin a tale about data which seemed to be telling me no coherent economic story.

I think I had better stop this walk down memory lane before I either become totally mad or decide to place a curse on all statistical departments in all government institutions. Or both, of course. It now occurs to me that the far less broad and far less detailed data that other countries and global organisations produce, compared with Japan, may have been actually more precise than Japanese data after all. The very breadth and detail of Japanese statistics should have warned me about their precision. Oh dear.