Observations on Chinese Statistics: An User Perspective
Observations on Chinese Statistics: An User Perspective
Deepak Bhattasali
Chief, Economics Unit and Lead Economist
World Bank
Commissioner Li Deshui, Chairman Brackstone, Ladies and Gentlemen:
I would like to thank the National Bureau of Statistics and the co-sponsors of this symposium for inviting me to participate today. The World Bank is a major user of Chinese statistics―both the countrywide and micro datasets. Also, we have collaborated closely with the NBS and other government departments on the development of parts of the statistical system. Thus, our interest in the modernization of the system is from the point of view of the user as well as a provider of technology. We appreciate very much the determination and openness with which the NBS and other agencies have approached China’s statistical modernization, and hope to be engaged in this endeavor as closely in the future as in the past.
Today, as an user, let me start by relating some events and issues from my five years of experience in
When I first arrived in
This was all very educational for me. During the course of the negotiations we learned a lot about the organization of China’s statistical system: how little money was allocated to NBS, how overworked the staff were, how NBS’ internal budget allocation system worked, and we began to appreciate the need for some form of cost recovery for the extra effort. However, the experience also revealed issues related to: (a) differential access to data by different users, (b) lack of coordination among the government departments who had requested us to do this work, (c) the private activities of some people within the statistical agencies, who operated on a very different incentive system. We also began to appreciate the problems caused by a lack of adequate public data files, and by a certain lack of understanding that most national statistics are public goods. In fact, the whole experience raised fundamental issues about the integrity and accessibility of the system and the underlying philosophy that guided the statistical effort.
A different set of issues comes up with regard to assessing the rapid structural changes taking place in
For example: Did poverty go down or up during the last year’s of the 9th Plan? Has the services sector’s GDP been frozen at 32 percent of output for a decade, despite it being the only sector generating new jobs at the rate of 7-8 million per year? What is happening to urbanization in recent years―can the hukou-based definition of urbanization capture the right set of structural changes? What was happening to national integration and the flows of goods and services across provinces―the imports and exports data in provincial input-output tables are calculated as a residual, so do not provide much information? Are there meaningful statistics on geographically economic performance or public expenditure that could assist in the formulation of policies for the development of the lagging Western regions―say, for example, at the sub-provincial level? Can we measure the impact of the Five Year Plans in terms of the people-centred development outcomes that really matter? How close are individual provinces towards achieving the MDG? What was happening to real exports? Why is credit rising so fast and who is getting it? Is the size of public expenditure in
In general, even on such central issues, we encounter problems that stem from high levels of statistical secrecy and gaps in coverage, from the uncertain relevance of many statistical collection efforts to important policy concerns at the highest levels, especially the ability of the Government to respond to structural change.
Recently, there has been a lot of concern about whether the economy is overheating. The recognition lag has been very long―it was not until mid-April that, on the basis of the statistical information, analysts were able to see that aggregate demand is running ahead of production capacity. And yet, even now only qualitative analysis is possible on central concerns such as why the money supply has been growing so fast, what is happening to local government investment, what is the size of consolidated government debt? The coherence of the national statistical indicators is the issue here, but also there are issues related to how policy is made and who makes it, and what the effect of such factors is on the incentives or pressures on NBS and other statistical producers in
Ladies and Gentlemen, these descriptions and the questions they raise are merely intended to illustrate the challenges that lie ahead. But, it is vital that the issues are seen in the context of the broader environment in which
The statistical effort cannot but reflect the broader information management climate in a country. Given the transitional nature of the country and existing restrictions on the flow of information generally―through the print and audio-visual media and the internet―national statistics and even special collection efforts and micro surveys are constrained in terms of transparency.
Policymaking in
Another general characteristic, again deriving from the transitional nature of the country, limited transparency of policymaking and, in addition, restrictions on participation, is that the underlying philosophy of modern statistical systems―that most national statistics are a public good (that can be used by multiple users at the same time)―is yet to be established. Generally speaking, statistics are treated as private or club goods, but with highly unequal access due to official or unofficial reasons. Importantly, if this general philosophy continues, progress in modernizing the statistical system and making it useful to both policymakers as well as broader user communities will be slower than desirable.
Given this background, and within the limitations of the system-wide problems I have identified, many of which are beyond the control of NBS and the other major producers of statistics, what are some useful directions for future statistical effort? Those of us who have followed statistical development in
However, I will not comment on the accuracy of the statistics. As we know, there has been extensive discussion of this issue in recent years, and we all know that NBS, mainly, but also the other producers of statistics, have been working very successfully and very hard on improving the accuracy of statistics. Moreover, as needed, they are drawing on the best international expertise in each specialized area, and many of you in the audience are involved in this effort. I have only two suggestions in this regard.
First, to focus on the social and economic indicators that are so vital to the attainment of the MDG goals, as they reach out to the roots of what development is all about and also parallel the effort of the Chinese government to embark on a new, scientific, people-centered and balanced development strategy. Second, the meta-data need to be much more detailed than at present, not just in the Statistical Yearbook and other general publications, but also on the NBS and other websites. Quality declarations should state the statistical bounds of the total errors in estimates. The non-response declarations presented in publications related to the recent population census are a good first step, but more details would be helpful and the reporting on measurement errors is still weak. General users need to go into specialized statistical journal to find a discussion of such important facts and, generally speaking, most statistical publications present almost no information on the model assumptions that underlie particular statistics.
Let me conclude by briefly discussing other areas of quality from the users’ perspectives.
Who is the user? There are two issues here. First, policymakers need good statistics, no doubt, and most of the statistical effort is aimed at providing them with accurate information. The question of timeliness is perhaps equally important.
Why does the system not cater to the needs of the larger constituency? Because there is little dialogue with this segment of users, although already they outnumber the traditional users. We know that the relevance of statistics is determined by each user, it cannot be defined by the producer. But, what can be agreed between user and producer is a basic set of statistics, and this can only be determined through continuous, serious, and scientific engagement with user groups.
And, finally, how to improve dissemination and use? Clearly, differential access to data by different parties is unacceptable, although to be practical we need to recognize that for some statistics this will continue to be true and, perhaps, necessary. Also, cost recovery is acceptable as a principle, but the cost of statistical services for the users should be transparent. The development of a broader public data files platform will also permit the producers to target cost recovery efforts better to special user groups―the majority of users will be quite content with a broader platform of more accurate and timely data available easily through publications and the internet, supported by high quality and reasonably comprehensive meta-data. Release dates need to be publicized and adhered to―NBS, for example, has made a lot of progress on the national accounts and other data series, but what about poverty and other data? And, to encourage learning and prevent erroneous use of data, but also to promote dialogue with user groups, NBS should not shy away from responding openly to cases where the data have been misused or misinterpreted, and should point out what the correct interpretations should be.
Ladies and gentlemen, we need to recognize that statistical agencies in
Thank you.