Provincial Yearbooks' County-level Data Files*
Personal comments and evaluation by Robert F. Dernberger, Professor
Emeritus of Economics and Director, Center for Chinese Studies,
University of Michigan
I have been studying China's economy since 1953 and have finally
reached the ranks of the Professors Emeritae. I guess my generation can
be called the Old "New China" hands. In any event, the first two decades
of my career were spent in learning how to pose as an expert on China's
economy without ever having visited China or even having access to
useful sets of data for that economy. The First Five-Year Plan (1952-
1957), published in 1955, was the first significant collection of
absolute figures for the level of economic activities in China for the
post-1949 period (the hard data published in that volume was for the
year 1952) to be made available in the West and The Ten Great Years,
published to celebrate the first decade of Communist rule in 1959,
included absolute values for a large number of major economic and social
indicators for that decade. [See note 1] The failure of Mao's Great Leap Forward in
1959-61 brought starvation to the Chinese people, but also brought
frustration to any hopes for more published statistics for developments
in China during the 1960's. In fact, the period 1959 through the end of
the 1960's can be called the Dark Ages for quantitative information of
economic developments in China. Propaganda and slogans had replaced
meaningful statistics in published accounts of economic and social
developments in China and we had to b e satisfied with statements such
as "output was higher than last year," "output increased this year," and
"output this year was the highest level since the 1950's."
Our hopes were revived somewhat by developments in the 1970's. The
signing of the Shanghai Communiqu in 1972 created the opportunity for
scholarly exchanges between the United States and China. Thus, for the
first time we could see for ourselves the economy we had been trying to
describe for almost two decades. Furthermore, reports of government
officials and articles in the controlled press became more quantitative
during the 1970s. Nonetheless, the amount and quality of statistics
available for the study of the Chinese economy remained a serious
constraint on sound and detailed, quantitative analyses of economic and
social developments in China. This did not, of course, prevent us from
publishing our assessments of developments in China, often based on
little more than policy statements, scattered Chinese and Western press
reports, and our own and the written reports of others' study trips of
various durations - usually three weeks to a month.
Then came the watershed in the evolution of post-49 China; the
transition of leadership in China from the era of Mao to the era of Deng
Xiaoping, usually dated as the 3rd Plenum of the 11th Central Committee
of the Chinese Communist Party in November/December of 1979. At that
Plenum, the leadership of Hua Guofeng (Mao's chosen successor, Mao
having died in 1976) began to crumble in the face of the reemergence of
Deng Xiaoping (Deng having been purged of his Party positions after the
death of his guardian mentor, Zhou Enlai, in 1976). Hua represented the
forces of the "Whateverist" school, while Deng represented "Truth from
Facts." The Whateverists believed that whatever programs Mao had
advocated on the basis of his voluntaristic interpretation of Marxism-
Leninism were to be implemented in being true to Maoism. Those who
agreed with Deng believed that policies or programs advocated in the
name of Marxism-Leninism had to meet the test of empirical results to
prove they would work before being implemented throughout the economy.
This greater respect for accurate empirical evidence complemented and
facilitated the revival of the collection and publication of statistics
for economic and social developments in China. Indeed, statistical work
became a most serious and high priority enterprise in China after 1978.
Origins of the Provincial Yearbooks' County-level Data Project
at the University of Michigan
During the worst days of the Cultural Revolution, the State
Statistical Bureau (SSB) in Beijing was hit harder than most
institutions in losing professional staff who were being sent to the
countryside to learn from and endure the sufferings of the masses. At a
low point, its professional staff was reduced to only 13 statistical
workers. Obviously, the collection and publication of meaningful
national-level statistics was not possible during those years. As with
other institutions, however, the State Statistical Bureau began to
rebuild and reassert its traditional mission after the end of the
Cultural Revolution. This process was well under way by the time of the
3rd Plenum, but it was not until 1982 that the State Statistical Bureau
had rebuilt its hierarchy of provincial, prefectural, and county-level
bureaus and offices under the Bureau's national-level offices in Beijing
(i.e., the State Statistical Bureau, located within the same buildings
and with close ties to the State Planning Commission). Thus, in August
of 1982, the State Statistical Bureau was able to compile and publish
the first of the annual Statistical Yearbooks of China. [2] Copies of
Statistical Yearbooks for the individual provinces began to appear soon
thereafter. [3] These yearbooks were compiled by the Statistical Bureau of
the relevant province and in the early years were published, in most
cases, by the People's Publishing House of the same province. Over time,
the format followed in these provincial statistical yearbooks became
more uniform and publication has been taken over the China Statistics
Publishers (Beijing), although they still are often printed at a
provincial press.
By 1988, I had been able to acquire copies of provincial yearbooks
for over half of the provinces in China for the year 1987 (statistics
for the year 1986) and these provincial yearbooks were beginning to
include some statistics for the counties and county-level cities within
the provinces. This was occurring rather late in my career, but having
been denied ample detailed statistical data for the Chinese economy for
over three decades, I was willing to grasp at any straw. Thus, I began
what than seemed a plausible project: acquire all the provincial
yearbooks I could and extract the statistics for the counties and
county-level cities they contained. [4] In this manner I had hoped to build
up a data bank, admittedly a biased one, but a sample of economic
statistics for county-level units in China. At first, this project
had seemed plausible because not all the provincial yearbooks contained
statistics for the counties and county-level cities in that particular
province and, for those yearbooks that did include these statistics, the
number of variables reported on was not very large. There was no
question but that the missing variables and missing data problems would
seriously restrict or bias my use of the statistics being collected by
this means. Those desiring complete coverage, were collecting the
statistics for the 30 provincial-level administrative units which were
becoming readily available with the publication of these provincial
statistical yearbooks. [5] However, these provinces are often larger than
European countries, i.e., the statistics collected were at a very high
level of aggregation. With over 2,000 county-level administrative units
in China, quantitative reports on a limited number of important
variables for just half those units would be a data set I could not have
dreamed of only a few years earlier. Of course, it was much easier to
collect and enter data for 30 reporting units than for over 2,000
reporting units and the sources for the former were much more numerous
and available than those for the latter.
Despite my efforts and assumptions made in the mid-1980's, by the end
of the 1980's it was becoming clear that the attempt to extract the
statistics for the country-level units from the provincial statistical
yearbooks was becoming a full time job, with no time to do any thing
else, let alone analyze the data being collected. All the provincial
statistical yearbooks were now readily available, almost all of them
were including statistics for the county-level units, and the number of
variables being reported on for these county-level units was increasing
with each passing year. In fact, by the end of the 1980's, the release
of economic and social statistics from China was becoming a flood and no
one person would be able to manage more than a small portion of the
statistics being released. In one short decade we had gone from too
limited an access to statistics for China to having too many statistics
being made available. Certainly some effort to gain control of these
statistics so we could utilize them in our analyses of developments in
China was called for. But by whom? and how?
Formation of CITAS
I doubt I would have found an answer to these questions myself and my
investment of time and money would have ultimately proved fruitless.
Fortunately, in the Fall of 1991, Prof. William Skinner (University of
California, Davis) sent me a copy of a proposal for a China
Disaggregated Data Bank he was presenting to the ACLS-SSRC Joint
Committee on Chinese Studies, asking if I would care to comment on and
add my support to his proposal. Essentially, the proposal called for an
interdisciplinary and multi-national effort to develop a China GIS
(Geographic Information System) which would collect the available
county-level statistics and make them location specific on a digitized
map of China (also using the county-level administrative level as the
basic unit). This neatly married Prof. Skinner's effort to make location
an important variable in analyses of developments in China with my
efforts to collect the county-level statistics being published by the
Chinese. Thanks to Prof. Skinner's initiative, his efforts resulted in a
group of interested China scholars meeting with CIESIN (Consortium for
International Earth Sciences Information Network) to form CITAS (China
in Time and Space) that would organize and supervise the efforts to
develop the proposed China GIS (Geographic Information System)6 The
management of the CITAS China GIS was to be carried out at the
University of Washington, while acting as subcontractors, Prof. Skinner
would develop and contribute the data he was working with in connection
with his own research projects at the University of California, Davis,
and I would supervise the data entry of the county-level statistics made
available in the provincial statistical yearbooks. CIESIN provided the
funding for our efforts in the first two years with funds they obtained
from NASA. They also will be a server for the GIS on the Internet when
it is made available to the public. Funds made available by the Ford
Foundation enabled us to continue our work for six months beyond the two
years and we are now searching for funding to expand the data coverage
in our CITAS 1990 China GIS and include a China GIS for other years. [7] The
Provincial Yearbook County-level Data Entry Project at the U of M Under
CITAS The discussion in the following section reports on the data
entry work at the University of Michigan after that project became part
of CITAS and a more detailed report on the results of that effort. The
concluding section presents a rather personal evaluation of the county-
level data that was entered into the CITAS China GIS from the provincial
statistical yearbooks. While I may have made the initial efforts to
acquire the various provincial yearbooks and to enter the country-level
data they continued into a spread sheet format (files in Excel), the
real breakthroughs in our work on this project at the University of
Michigan has been due to our ability to find on campus an appropriately
trained and very able person to be the project's Research Assistant, Ms.
Huibin Cai from China. [8] Inasmuch as CITAS was aiming at a GIS for China
in 1990 as their first product, we began to input the county-level
statistics for 1990 found in the 1991 provincial statistical yearbooks.
To do that, of course, we needed to develop the column labels (the
variables) being restricted to eight characters (the limit for DOS
filenames at the time). This was not too difficult in the beginning, but
when the list of variables grew to over 2,000 it became a major task. [9] I
had already developed variable names on the basis of the work I had done
before joining the CITAS project. Ms. Huibin Cai then took over and
added many of her own invention. We also were aided in out efforts when
we acquired an extensive set of variable codes for the annual and
monthly national-level statistics collected and distributed by the State
Information Service of China, which is associated with the State
Statistical Bureau. [10] The problem of the codes for the row variables,
the county-level units, was easier to solve; we simply used the
administrative GB codes used by the Chinese which gives every county-
level unit in every year a GB code number. This is not to say the
managers of the CITAS China GIS have had an easy time in resolving some
the inconsistencies between the GB codes published by the Chinese and
the county-level units designated in the provincial statistical
yearbooks. However, we began by entering the row variable with the name
of the county level unit in Pinyin, adding the GB code when they were
provided to us by the CITAS China GIS data bank managers at the
University of Washington.
Thanks to the efforts of Ms. Huibin Cai, all the statistics for the
county level units to be found in the provincial statistical yearbooks
for the years 1990 and 1991 (i.e., the statistics in the 1991 and 1992
yearbooks, respectively) have now been entered. [11] The results of our
efforts are shown in Table 1. [12] As can be seen, the size of the data
bank for any one province varies greatly, from a very small 18K for
Guizhou (1990) to 850 for Hebei (1990), with the total data bank in both
1990 and 1991 amounting to approximately 10MB. [13] In a preliminary
attempt to estimate comparability in coverage among the provincial
statistical yearbooks, we looked at the availability in each of the
yearbooks for the 333 variables most commonly found in the statistics
reported for provincial level administrative units. [14] The results of
that survey are given below:
Statistics are available for the counties in 29 provinces for the
gross value of agricultural output; 28 provinces for total population
and grain output; 25 provinces for oilseed output, for retail trade, and
for cultivated area; 24 provinces for gross value of industrial output
and local budget revenue; 22 provinces for non-agricultural population,
for meat output, for local budget expenditures, and the gross value of
crop output, of forestry output, of animal husbandry output, of sideline
production, and of fishery output; and 20 provinces for students
enrolled in middle schools. This means that only 17 variables are
available in the statistics for the county-level units reported in two-
thirds or more of the provincial statistical yearbooks in
1990. [15] Important Details on The Sources Used And How They Were Used
Table 1 includes the code reference to the particular sources used,
which are presented in Table 2. This type of documentation will be
familiar to most academics as it follows the usual practice for
footnoting sources used. Only a few comments are needed here, but they
are important to keep in mind by anyone using the CITAS China 1990 GIS.
The normal practice was to use the 1991 (or 1992) provincial statistical
yearbooks for the source of entering the 1990 (or 1991) county-level
data that yearbook contained. If we were lucky, the tables with the
county-level data were to be found in a single section of the yearbook.
However, as Table 2 makes clear, the county-level data often was to be
found in tables scattered throughout the yearbook. All tables in the
provincial statistical yearbooks were searched for any county-level data
they may contain. It is important to note that the yearbooks often
contained text describing developments in the county-level units of the
province, text which may contain the same statistics that are found in
the tables, but may also contain statistics that were not to be found in
the tables. Thus, the provincial statistical yearbooks could yield
additional county-level data to that we took from the tables and this
would be our first choice for searching for additional statistics in an
attempt to achieve complete coverage for each variable.
In most cases, the 1990 county-level statistics came from the 1991
yearbooks, as reported above. However, if in the process of collecting
the 1991 county-level statistics from the 1992 yearbooks it was
discovered that the 1990 county-level statistics in the 1992 yearbooks
differed from that reported in the 1991 yearbooks, the statistics
published at the later date, i.e., the 1992 yearbook, was used to
correct the statistics published earlier, i.e., in the 1991 yearbooks.
Finally, as can be seen from the sources cited in Table 2, an
exception was made to the practice of restricting our search for the
county-level data to that found in the provincial statistical yearbooks:
Jiangsu in 1991. In this case, a separate yearbook devoted solely to
statistics for the county-level units was compiled and published by
Jiangsu Province in 1992 (i.e., with data for 1991). Thus, the
statistics in the data bank for Jiangsu in 1991 were taken from the
later yearbook, as indicated in Table 2. [16] The Nature of the
Statistics in the County-Level Data Bank Having gone through the
trouble of assembling this data bank and including it in the CITAS China
1990 GIS (with the data entered in files for a China 1991 GIS as well),
some comment should be made as to the usefulness and reliability of
these statistics. I believe much unproductive time is spent on
methodological arguments and debates over data reliability and argue
that the proof is in the pudding, i.e., reliability is a relative term,
which depends on what you want to use the statistics for, how you use
the statistics, what results are obtained, and what alternative sources
of statistics are available. Obviously, the use of these statistics
would vary from individual to individual and the options will also vary
with their different uses. Nonetheless, some valid generalizations can
be (and should be made) as a warning to any would be user of the CITAS
China 1990 GIS (and the China 1991 GIS, when that becomes available).
Foremost in our evaluation of these statistics is the recognition they
are collected and published as an integral part of the economic and
political administrative bureaucracy and the primary function of those
who work for the State Statistical Bureau is to provide this information
according to a timetable for the decision-makers in the state's economic
and political system. Historically, these "official" statistics are not
collected for the purpose of academic research or for informing the
public. Rather, these statistics had an operational role to play in the
running of the economy and in making policy decisions. [17] Shortly after
the State Statistical Bureau was established in the mid-1950's, its
Director (Xue Muqiao) became engaged in a debate with Mao; the Director
arguing statistics could play their assigned role best by reflecting
reality accurately, achieving this by keeping the Bureau independent of
the economic and political administrative bureaucracy and utilizing
random sampling and other scientific statistical techniques. Mao did not
lose many debates and, unfortunately, this one was important to him. To
Mao, statistics were not factoids, but were part of the revolution and
class warfare, their main purpose was to mobilize and incite people's
efforts for the development of China. And, as in all other areas, there
was to be no organization that was independent from political engagement
and control. [18] Not only did the Director lose his job in favor of a
more compliant Director, Chinese "official" statistics became hostage to
the economic and political bureaucracy and that bureaucracy had a vested
interest in what those statistics said. One of the most important
hypotheses of information theory tell us that such a system will not
produce a very accurate set of statistics, i.e., the statistics should
be completely independent from the individual collecting them, who
should be neutral as to what those statistics say, and the ultimate
users should be unknown to those who collect the statistics. Just as
with the principle of free markets, no individual should have any
influence over the final outcome. Production units at the lowest level
have departments or assigned workers to collect the statistics for that
unit. These statistics are then packaged and reported to their superiors
at a higher level. Obviously, those doing the collecting are not neutral
to what the statistics report and they know full well who will be using
them and for what purpose. With its revival after the 1970's, the State
Statistical Bureau has tried to establish its own offices at lower
levels to check on the statistics being collected within the units of
production and also has tried to enforce uniform standards and forms for
the reporting of statistics throughout China. These efforts have
achieved some success, but have not changed the basic institutional
organization for collecting and publishing the "official" statistics
found in the statistical yearbooks.
Thus, even today the State Statistical Bureau can claim to have in
its possession only those statistics collected and kept at the county-
level, a rather high level of aggregation even before the State
Statistical Bureau begins to aggregate these statistics even further as
they are passed up through each successive level of the bureaucratic
hierarchy for reporting, i.e., to the Prefecture, the Province, and
then the State level. This process of aggregation for the purpose of
reporting provincial level totals and national level totals, of course,
might result in offsetting differences from the truth being reported by
lower levels. Yet, the deviations from the truth would tend to all run
in the same direction, i.e., the reasons for underreporting or
over reporting affecting all local units alike, causing the totals
reported by higher levels to aggregate these deviations from the truth.
This is one of the major reasons for wanting to collect the statistics
for the county-level units rather than higher level units; this is the
lowest level in the collection and reporting of the "official"
statistics by the State Statistical Bureau. If the researcher wants the
statistics for any lower level, they must go to the township-level units
or the individual units of production and collect those statistics; the
State Statistical Bureau does not possess them, they are kept by the
local units themselves. [19] Finally, if the above were not enough cause
for concern, the State Statistical Bureau does not collect all the
statistics it publishes under its name. For example, the monetary sector
and budget data are provided by the Ministry of Finance, while the
Ministry of Agriculture collects and reports much of the rural and
agricultural data. As for foreign trade statistics, the Customs Bureau
keeps one set of books, while the Ministry of Foreign Economic Relations
and Trade keeps a separate and different (for definitional reasons) set
of statistics. In fact, almost every ministry now publishes its own
statistical yearbook. The State Statistical Bureau has complained about
this need to rely on others for the statistics it publishes in the
statistical yearbooks, as publication of the statistical yearbooks are
often delayed due to the need to wait for other Ministries to release
the statistics scheduled to be included in the yearbook..
From the several arguments as to the institutional organization and
the process which produces the "official" statistics published in the
provincial statistical yearbooks, it would be unreasonable to believe
that these statistics have a terribly high degree of accuracy. But
accuracy in the statistics we use in our analyses is a relative term and
involve the analysist's judgment as to what the acceptable level of
accuracy should be. Unless sophisticated sampling techniques are used,
it is unlikely that any attempt in any country to acquire complete
enumeration by means of statistical reports accumulated up through an
administrative bureaucracy of any kind are likely to be "accurate."
Thus, it becomes a matter of comparative evaluation in making a
judgment as to the relative accuracy of the county-level statistics
found in the official sources.
Before we investigate this question, however, one further
characteristic of the county-level statistics must be noted, i.e., the
difference between the Soviet-type economic system statistics and our
Western economic statistics. Quite simply, due to the needs of the
planners and economic-decision makers in the Soviet-type economies, the
demand was for reports on physical quantities of inputs and outputs.
Because prices and values were less important to the planners (they
could set prices at whatever level they desired), .less attention was
paid to the value of inputs and outputs. In short, there was obviously
a priority in what was important in the collection and publication of
statistics and many of the statistics we are most interested in were of
limited interest to the Chinese and that is a large reason we don't find
them in our county-level data bank of county-level statistics taken from
the provincial yearbooks.
Furthermore, aggregates of economic activity were simply sums of the
output in the five material producing sectors (agriculture, industry,
construction, transportation, and commerce), while the service sectors
and services within the five material producing sectors were ignored or
explicitly excluded. This means that the totals for economic activity
excluded some activities included in Western statistics for Gross
Domestic Product and for National Income, while other activities were
double counted, i.e., the iron mined was included in output of the
mining sector (industry), and the value of the iron was included in the
steel produced that is included in the value of industrial output, while
the value of the steel is included in the value of the output of the
truck - also included in the value of industrial output. In addition,
depreciation was more an accounting entry with little relevance to the
"using up" of the physical capital stock. Financial flows and profit and
loss estimates were manipulated by the state authorities by merely
adjusting the administered price system and planned credit allocations.
Many categories of fixed capital investment statistics are available as
these were related to construction projects which were of more
importance to the planners than current levels of output. All this is
currently changing, the Chinese scrapping their old system of social and
economic accounting and adopting standard Western standards and
definitions. Unfortunately, the year 1990 is a bit early to take full
advantage of this change, but it is hard to fault the CITAS China 1990
GIS because it fails to include many variables of most interest to
Western analysts.
A last word of warning concerns the definition of the variables,
i.e., what do the Chinese include when the say "industrial output,"
"capital construction investment," "grain output," etc. Probably not
what we mean when we say industrial output, fixed investment, or grain
output in our statistics. Thus, any user of the CITAS China 1990 GIS
would be well advised to carefully check the definition of the
variables, contained elsewhere in the documentation. Unfortunately, the
provision of this documentation for all of the more than 2,000 variables
is a most ambitious effort. While, the variable documentation provided
with the initial release of the CITAS China 1990 GIS will include those
definitions provided in the statistical yearbooks, many variables
reported on in these yearbooks are not defined and a complete
documentation for all the definitions must await a search in other
publications of the State Statistical Bureau. Even though the definition
may appear understandable from the variable itself, i.e., production of
sewing machines, does this include only production in state-owned
enterprises, only in enterprises under the Ministry of Light Industry,
only in state-owned and cooperative enterprises at the county-level or
above, only in independent accounting enterprises, and does it include
sewing machines produced in workshops attached to garment enterprises?
It should be clear that these definitions have much to do with the
question of accuracy.
Despite all these qualifications and warnings about the statistics
included in the CITAS China 1990 GIS, the statistics that are included
would seem to allow for a significant amount of meaningful social and
economic analyses to be made of contemporary developments in China. Yet,
we should be more forthcoming in our judgment concerning whether or not
those variables included in the CITAS China 1990 GIS are accurate
enough.
The Question of Accuracy
While the proof may be in the pudding (i.e., the judgment of the
recipe and the cook depends upon the judgment of those who eat the
pudding), those who would use the data in the CITAS China 1990 GIS
deserve some prior assurances (from someone) that their willingness to
devote the time and effort to acquire the knowledge and resources so
they can use these statistics in their analyses of developments in China
has some promise of paying off. Perhaps no Western economist has put
the "official" statistics collected and published by the State
Statistical Bureau to as many different tests as has Prof. Gregory Chow
of Princeton University. Prof. Chow is an acknowledged expert in
econometrics, is well-trained in economic theories, and is very
knowledgeable about China's economy. He has been most active in visiting
China and meeting with representatives of the State Statistical Bureau,
giving lectures and seminars in China, and in publishing statistical
analyses of the Chinese economy - analyses which rely on statistics
provided by the State Statistical Bureau. [20] Although somewhat dated
(1986), his article on "Chinese Statistics" remains a useful evaluation
of the quality of the statistics published by the State Statistical
Bureau. [21] According to Prof. Chow, compared with the developed countries,
China's statistics are limited in scope because they lack the trained
personnel that is available in the developed countries and because some
of the statistics collected are considered as state secrets and not
published. As for the quality of the statistics that are published,
their quality may be judged as poor due to such reasons as the lack of
qualified statistical personnel and the fact that the statistics
collected are biased due their use by the authorities in determining
rewards (both positive and negative), promotions (both up and down),
assignments (both targets to be achieved and resources to be received)
to be given to the local unit that begin the process of collecting the
statistics, and the monopoly of the State Statistical Bureau over the
collection and reporting of statistics in China. On the positive side,
Prof. Chow felt that the degree of central control down to the lowest
level was an argument in favor of the Chinese ability to collect better
statistics than was true for the developing countries. Finally, Prof.
Chow concludes those statistics which involve mere counting and are not
influenced by political pressures from above are "less likely to be in
error," but statistics that require technical sophistication to collect
"may be subject to question."(p106)
Obviously many changes took place in regard to the collection and
publishing of statistics in China between 1986 and 1991 (and 1992), when
the statistics included in our CITAS China 1990 (and 1991) GIS were
published. Along with the other indicators of China's modernization, the
supply of qualified statisticians has increased, the reasons for reporting
false statistics has been decreased, the scope of the statistics
reported has increased, and they are more readily accessed in numerous
publications. On the negative side, however, the State Statistical
Bureau still holds a monopoly over the collection and reporting of
statistics in China and a law was passed to make it a crime to violate
this monopoly control. [22] More important than these considerations,
however, is a point that was made by Prof. Chow in his 1986 article; the
Chinese provide very weak documentation as to the method used to collect
and report the statistics published. What is included, what is excluded,
what prices were used and how were they computed, etc. Obviously, as
with other aspects of China's modernization, they are getting better on
the problem of documentation and an extensive research effort,
especially a visit to the State Statistical Bureau and an interview
with a representative of the Bureau, now could possibly provide the
necessary documentation. On the other hand, the Chinese have a long way
to go to achieve the type of documentation that is readily provided by
most developed countries with their official statistical publications.
Finally, a contemporary development as a result of the economic reforms
has greatly weakened the State Statistical Bureau's ability to collect
and publish good statistics with complete coverage for the Chinese
economy - the significant increase in the decentralized and marketized
sectors of the economy. The strict control of the state over local units
and its ability to require accurate statistical reports from those lower
levels is seriously weakened. Thus, the State Statistical Bureau has to
rely on samples, spot checks, tax reports, transit point reports, etc.
to estimates activities at the local level and some of these means are
as biased as the former reports collected up the administrative
hierarchy.
While all of the above arguments may be factors that must be born in
mind when trying to reach a judgment about the quality of the
statistics in the CITAS China 1990 GIS, they do not provide us with the
answer we seek - are the statistics good enough (i.e., how does the
pudding taste when we use the statistics provided by the State
Statistical Bureau as ingredients). It is the results of Prof. Chow's
frequent use of these statistics in his econometric analyses that
provides the best encouragement to the would-be user of our CITAS China
1990 GIS. Four of his articles deserve special attention in this regard.
In "Economic Analysis of The People's Republic of China," Prof. Chow
uses the official statistics (both national level and provincial level)
to estimate production functions, consumption behavior, supply reactions
to price changes in agriculture, the quantity theory of money equation,
and a multiplier-accelerator model of growth. [23] In "Capital Formation
and Economic Growth in China," he uses the statistics published in the
Statistical Yearbooks of China to estimates the aggregate production
function for the five major material producing sectors. [24] Then, in "A
Model of Chinese National Income Determination," he again estimates an
accelerator-multiplier model of income determination for China for the
period 1952-1982 with statistics published by the State Statistical
Bureau in 1983.25 Finally, in "Money and Price Level Determination in
China," Prof. Chow uses the statistics published in the Statistical
Yearbooks of China to analyze both a long-run and a short-run model of
price determination. [26] Prof. Chow's major objective in these articles
was to show that modern economic theories were applicable and useful for
analyzing the Chinese economy, despite the differences in institutional
and behavioral characteristics between China and the developed
capitalist economies of the West. In other words, the estimation of
econometric models based on the theories of modern, Western economics
yielded estimates for the parameters of these models that were rational
and consistent with what was known about the Chinese economy and the fit
between the models and the statistics was relatively good, i.e.,
resulted in high coefficients of correlation. This extent and rigor in
testing the data published by the State Statistical Bureau is quite
convincing in arguing that the data is good enough for our purposes,
i.e., they produce estimates that make sense and compare well with the
many tests of similar econometric models for other economies.
Yet, it is important to understand exactly what the above evidence
argues. The statistics used by Prof. Chow were aggregates for all of
China or for the individual provinces (many as large as European
countries) and these aggregates were tested in estimating macro-economic
models. For example, the models developed by Prof. Chow predicted
aggregate behavior for the relationship between the total money supply
and the average price level; the relationship between total inputs of
land, labor, and capital and total output in the agricultural sector,
the industrial sector, and the construction sector; and the
relationship between total consumption and national income. These
statistics met this test rather well and we can feel even greater
comfort in that our CITAS China 1990 GIS has statistics for the county-
level unit, which is at a much lower level of aggregation than the
province.
The Question of Alternatives
Even if the results of Prof. Chow's estimations were less
satisfactory, however, the use of poor quality statistics has been
justified in studies of economies throughout the world, including the
United States, on the grounds that their are no better statistics
available. With its monopoly over the collection and publication of the
official statistics in China, which continues to this day, it is
unlikely that an alternative source of these aggregate, macro-statistics
will be available anytime in the near future. As in any economy
throughout the world, macro-statistics are indeed the province of the
government or government-created institutions and the quantity, quality,
and availability of those statistics also varies widely throughout the
world. We have argued above that these macro-statistics in the case of
China would appear to support reasonable analyses of economic and social
developments in that country.
While smaller than the province, the average county in China is still
quite large, i.e., there are approximately 3,000 counties, county-level
cities, and county-level districts within cities, or approximately
400,000 people per county-level unit. Thus, the user of the CITAS China
1990 GIS is working with aggregates or averages for rather large units
and the argument that these statistics "may be good enough" applies only
to their use in macro-models of the type developed in Prof. Chow's work.
When it comes to analyses of the behavior or performance of individual
units, i.e., the household, the firm, and the individual, we need much
more micro-level data, even the tabular data for the individual units. A
good illustration of this problem is the many studies of the impact of
the economic reforms in China on the efficiency of the individual
enterprise. Initially, the available macro-statistics were used to
estimate this impact; clearly a violation of the appropriate rules of
hypothesis testing in Western scientific analysis. How can one argue
about the changes in the efficiency of THE firm without firm specific
data that allows you to control for location, size, output mix, prices
of inputs and outputs, etc. Using the average level of inputs and
outputs, average prices, average size, etc., of the several thousand
producing units in a given county is clearly not appropriate. Because
macro-level statistics are often more readily available, or possibly
the only statistics available, the use of macro-level statistics to
answer micro-level questions is probably the norm, but this does not
make it correct. [27] Even, if the statistics were to be perfectly
accurate, the estimates they produced could well be in error. [28] It
was in recognition of this problem, i.e., that the official statistics
collected and published by the State Statistical Bureau were not very
useful in analyzing many of the economic problems they faced, that a
quite separate statistical collecting enterprises was founded after
1978: the collection of samples of unit specific statistics. This
activity has now grown into a full-scale industry, with Chinese research
units collecting such samples for their own research projects, other
semi-official institutions offering to collect such samples for foreign
researchers for a fee, and individual domestic and foreign researchers
collecting their own samples of statistics for their research projects.
These samples are sometimes made available to Western academics and it
is possible to find research articles published both within China and
abroad that cite their access and use of these samples. While these
samples suffer most frequently from their small size and their biases,
it is clear to this author that their quality is better than the macro-
statistics collected and published by the State Statistical Bureau and
they are much more appropriate for most of the micro-questions being
raised by the researcher.
If this ever growing body of statistics to be found in the many
samples collected for the individual units in China are felt to be of a
better quality and a more appropriate data base for analyzing micro
questions about developments in contemporary China, why didn't we use
them in our CITAS China 1990 GIS? The answer is really quite simple. Our
GIS is for the entire land surface of China and for a wide variety of
social and economic variables. These samples are not only narrowly
focused on particular variables, more important they are often site
specific and do not represent a sample of local units for the country as
a whole. There are complete censuses and large samples that do cover the
entire country, but the tabular data (the data for the individual units)
are either not available at all (are classified as secret), are available
only for a rather exorbitant price, or are made available in a form
that does not identify the data for a particular unit with its
location. [29] Yet, there is some promise of a better opportunity in the
future. For example, the 10 percent sample of the population census and
some other rather large and national-wide samples are being made
available to some groups and researchers and where those have become
available they are being included in our CITAS China 1990 GIS. In other
words, to achieve its full potential, the CITAS China 1990 GIS of the
future would not rely so heavily as it does now on the county-level data
collected and published by the State Statistical Bureau. Rather, nation-
wide samples where the observations for the various variables can be
identified with a particular type of local unit at a given location,
hopefully, should come to dominate our CITAS China 1990 GIS.
Concluding Remarks
This is our hope for the future, but given what was available and
possible, we firmly believed it is best to begin now: to collect what
official statistics were available, but to focus on the county-level
administrative unit - the lowest level in the official statistical
hierarchy; to make those statistics available in a format that would be
most useful to any scholar engaged in the study of contemporary
developments on China, whether they are China specialists or not; and to
utilize as out format a GIS which makes explicit the relationship
between location and social and economic developments in China, i.e., to
make location an explicit variable in our analyses if these
developments.
In these Comments for the CITAS 1990 Data Files we have tried to
explain the origins and reasons for our collecting the county-level data
bank, both before and after the organization of CITAS; the sources of
the statistics in the data bank and their limitations and weaknesses;
and our judgment as to their accuracy and the availability of
alternative sources of data. Obviously, this first step in our efforts
to provide a CITAS China 1990 GIS does serve two important purposes very
well. It has collected the available county-level statistics from the
Provincial Yearbooks and placed them in a machine-readable format that
is accessible and usable by any interested individual scholar. To that
extent, at least, individuals working in isolation do not have to
replicate out efforts. We have already extracted the county-level
statistics for 1991 found in the 1992 Provincial Yearbooks and, with
additional funds, will proceed to do the same for 1992, 1993, etc. [30]
This will provide a considerable sample of official county-level
statistics that will allow for significant projects in both cross-
section and longitudinal analyses of social and economic developments in
contemporary China. This was my original objective when I began to
extract the county-level statistics from the Provincial Yearbooks
several years ago. But our CITAS China 1990 GIS makes a much more
significant contribution that just the provision of the sample of
county-level statistics in the Provincial Yearbooks. The major
contribution is to place these statistics in a GIS which relates all the
statistics to a particular location on a map of China. This, combined
with the layers for topography, hydrology, transportation, settlements,
cultural characteristics, the hierarchy of urban places, and the
boundaries for the core and periphery of the economic regions of China
place these county-level statistics in their very rich and important
geographic environment. This is the core purpose of a GIS and was the
major purpose behind Prof. Skinner's efforts in the CITAS project. There
will, of course, be those who desire to use the CITAS China 1990 GIS
merely for the sample of county-level statistics to use as gist in their
mills for estimating multiple regression equations and there will be
those who use the GIS merely for the purpose of generating very
impressive and colorful maps, but those uses of our GIS will sadly deny
the tremendous opportunity provided by the marriage of advanced computer
software technology and our painstaking efforts, i.e., to move our
analyses of developments in China to a much higher plane - to bring
location and environment into our analysis. Few who are knowledgeable
about developments in contemporary China would deny the importance of
location and environment: it is an explanatory variable which belongs in
our analyses. We hope we have made doing so easier.
FOOTNOTES
* This document contains Chinese Characters, generated by TwinBridge
Chinese Partner for Windows 4.0. If your computer cannot read and
display Chinese characters, don't worry; the Pin-yin romanization and
English translation for all Chinese citations are also included.
[1] First Five Year Plan for Development of the National Economy of the
People's Republic of China (Beijing: Foreign Languages Press, 1956).
State Statistical Bureau, Ten Great Years, Statistics of the Economic
and Cultural Achievements of the People's Republic of China (Beijing:
Foreign Languages Press, 1960).
[2]
1981 1982 (Guojia Tongjiju
Bian, Zhongguo Tongji Nian Jian: 1991 {Beijing: Guojia Tongji Chubanshe,
1982}) [Edited by the State Statistical Bureau, China Statistical
Yearbook: 1981 (Beijing: China Statistics Publishers, 1982)] Although
published in 1982, the first Statistical Yearbook of China carried 1981
in its title as the statistics in the yearbook referred to the year
1981. All later yearbooks in the Statistical Yearbook series had the
year they were published in their title, the statistics they contained
being for the previous year. The first yearbook in the series contained
over 500 pages and measured 5 1/2 by 8 inches. The 1994 edition contains
just under 1,000 pages and measures 7 1/2 by 10 1/2 inches. The latest
edition also is the first edition to carry table headings and labels in
both Chinese and English and the explanatory notes are in both
languages. See 1994
1994 (Guojia Tongjiju Bian, Zhongguo
Tongji Nianjian: 1994 {Beijing: Zhongguo Tongji Chubanshe, 1994})
[Edited by the State Statistical Bureau, China Statistical Yearbook:
1994 (Beijing: China Statistics Publishers, 1994)].
[3] The earliest provincial statistical yearbook I have in my collection
is the Hunan Province Statistical Yearbook for 1982, published, July,
1984. 1982
1984 (Hunansheng Tongjiju Bian, Hunansheng Tongji
Nianjian: 1982 {Changsha: Hunan Renmin Chubanshe, 1984}) [Edited by the
Hunan Province Statistical Bureau, Hunan Province Statistical Yearbook:
1982 (Changsha: Hunan People's Publishers, 1984)]
[4] My effort to acquire all the provincial-level statistical yearbooks
continues and, although I do not have complete coverage, if my current
orders for provincial yearbooks are filled, I should have over 90 per
cent of the 210 volumes for the years 1987-1994. I also hope to xerox
copies of the provincial yearbooks that are missing from my collection,
but held by libraries and others in the US, in the near future in order
to acquire a complete colletion. I wish to thank Ms. He Xiaobin,
Direcctor, Export Department, China National Publishers Import-Export
Corporation, for her most generous help, not only in supplying current
issues of the yearbooks, but also in obtaining several volumes I did not
have for the earlier years. I would also like to thank CITAS, U.of
Washington, for a small grant which provided partial support for
obtaining these yearbooks. When my project is finished, I plan to house
these yearbooks in the Asian Library of the University of Michigan,
where they will be available for research by others (there is a wealth
of data and information in these volumes besides the statistics for the
county-level units in each province).
[5] In 1990, the State Statistical Bureau and its publishing house brought
out a volume which published the economic and social statistics for each
province in a uniform format of over 330 variables for the years 1949-
1989.
1949-1989
1990 (Guojia Tongjiju Zonghe Si, Quanguo Ge Sheng, Zizhiqu, Zhixiashi
Lishi Tongji Ziliao Huibian: 1949-1989 {Beijing: Zhongguo Tongji
Chubanshe, 1990}) [Compilation Department of the State Statistical
Bureau, Compilation of Historical Statistical Materials for Each
Province, Autonomous Region, and Minicipality for The Whole Country:
1949-1989 (Beijing: China Statistics Publishers, 1990)]
[6] The history and details of the organization of CITAS is to be found in
another part of the documentation.
[7] The CITAS CHINA 1990 GIS consists
of the base map. with county boundaries, layers of geographic
information, and the social and economic data for county-level units
for 1990. In addition, we also have all county boundary changes from
1982 to 1992 and the Provincial Yearbook County-level data bank for
1991.
[8] Ms. Huibin Cai received her BA in Electrical Engineering from
Fuzhou University and was a researcher at the Research Center for Eco-
Environmental Sciences, the Chinese Academy of Sciences, before coming
to the United States. Her contribution to our project, including help in
setting up the GIS and making it operational on our PCs here at the
University of Michigan cannot be overestated.
[9] The labels for the variables in CITAS China 1990 GIS will be those
chosen and specified by those managing the CITAS China 1990 GIS project
at the University of Washington and the definition of these labels are
reported on elsewhere in the documentation. To enter the data in a
spread sheet format (EXCEL) at the University of Michigan, however, we
could not wait until the "official" set of labels had been chosen, but
had to develop our own; a task which took a large part of our efforts in
developing the county-level data bank from statistics in the provincial
statistical yearbooks..
[10] These disks are distributed by the State Information Center of China
and the variable codes are in MEAS Index Code Handbook, State
Information Center of China, November, 1991. MEAS stands for Macro-
Economic Application System of China.
[11] One exception is statistics for the county-level units in Tibet in
1991. We are still trying to acquire the statistical yearbook for the
Tibet Autonomous Region in 1992, but this exception for 1991 does not
affect the complete coverage for the county-level statistics in the
provincial yearbooks for the CITAS China 1990 GIS.
[12] Table 1 refers to the machine-readable data bank created at the
University of Michigan that was turned over to CITAS to be entered into
the CITAS China GIS. Thus, Table 1 refers to EXCEL files (each file was
restricted to a particular province and sector of the economy; the
different files being for the following different sectors - natural
resources, social indicators, population, labor, investment, public
finance, prices, people's livelihood, agriculture, industry, energy,
transport and communications, construction, domestic trade, foreign
trade, banking, city construction and environment projects, education
and culture, and health and welfare. In the CITAS China GIS, these files
are combined into a single national file, i.e., all the provinces, for
each of the above sectors.
[13] It should be borne in mind that our efforts were restricted to the
statistics for county-level administrative units found in the provincial
statistical yearbooks. This proved to be a special problem in the case
of Shanghai. Unfortunately, the county-level administrative units were
combined for the Shanghai Municipality as a whole in most of the
statistics reported in the Shanghai Statistical Yearbooks; when the
totals for Shanghai Municipality were broken down into lower units, the
statistics grouped into three subunits ( the metropolitan area , the
suburban districts, and the counties) and statistics were not reported
for the individual districts and counties. Only in the case of
agriculture were statistics reported for the individual counties and the
two urban districts in which agricultural activities took place within
the Shanghai Municipality. There are, of course, provincial economic
yearbooks, provincial almanacs, and the individual county-level units
themselves are now publishing their own statistical yearbooks. Thus, if
one were to spend more time in searching all these sources, it could be
possible to expand the county-level data bank to several times its
current size. Once the basic CITAS 1990 China GIS is ready for initial
distribution and use, of course, additional county-level data can be
entered to expand its coverage. But as Mao said, a march of a thousand
li begins with a single step. (We feel we have taken more than a single
step, but it is true that we have just begun a never ending process in
the early stages of the new age of information systems).
[14] These were the variables included in the uniform format used to
report the statitstics for each province that were published in the
source cited in footnote 5, above.
[15] It is important to note that districts within large cities also rank
as county-level units, but that the yearbooks that included statistics
for the county-level units in a particular province did not always
include statistics for the districts within a city, although statistics
for the city as a whole are available. This poses a serious problem in
trying to achieve complete coverage, even within those provinces that do
report the statistics for their county-level units, but not the
districts within large cities.
[16] For this source for Jiangsu, see the citation in Table 2. While I
have not made an extensive effort to acquire these provincial
statistical yearbooks which are devoted solely to the reporting of
statistics for the county-level units, those I have acquired include:
1990
1990 (Jiangsusheng Tongjiju Bian, Jiangsusheng Shi
Xian Jingji: 1990 Nian (Beijing: Zhongguo Tongji Chubanshe, 1990})
[Edited by the Jiangsu Provincial Statistical Bureau, 1990 Economic
Yearbook of the Counties and Cities in Jiangsu (Beijing: China
Statistics Publishers, 1990)].
1993 no place:
1993 (Jiangsusheng Tongjiju Bian,
Jiangsu Shi Xian Jingji: 1993 {no place: Jiangsusheng Xin Xinwen
Chubanju Pizhun Chuban, 1993}) [Edited by the Jiangsu Province
Statistical Bureau, 1993 Economic Yearbook of the Counties and Cities in
Jiangsu (no place:Publication sanctioned by the Jiangsu Province Happy
News Publishers, 1993).
1990 1990 (Hebeisheng
Renmin Zhengfu Ban Gongting Gen Hebeisheng Tongjiju Bian, Hebeisheng
Xian Zhen Nianjian: 1990 {Beijing: Zhongguo Tongji Chubanshe, 1990})
[Edited by The General Office of the Hebei Provincial Government and the
Hebei Province Statistical Bureau, 1990 Yearbook of the Counties and
Towns in Hebei (Beijing: China Statistics Publishers, 1990)
1980-1990 1991
(Guangdongsheng Tongjiju Bian, Guangdongsheng Xian (Qu) Guomin Jingji
Tongji Ziliao: 1980-1990 {no place: Guo Guanjiang Jiangxi Yichun Ziliao
Yinshua Chang, 1991}) [Edited by the Guangdong Provincial Statistical
Bureau, National Economic and Statistical Materials : Counties (and
Districts) in Guangdong Province, 1980-1990 (no place: State-run Jiangxi
Yichun Reference Printers, 1991)
If the reader has knowledge of additional volumes of this type, i.e.,
provincial statistical yearbooks devoted solely to the statistics for
the county-level administrative units in that province, I would
appreciate a message to that effect. My email address is rdernber@umich.
edu.
[17] At a meeting at the State Planning Commission, I heard an official of
the State Statistical Bureau plead with his Western critics to
understand his dilemna. He said he knew the various criticisms being
made and the dessirability of carrying our statistical checks and tests,
as well as collecting large unbiased samples, for the purpose of
verifying the "official" statistics released by the State Stataistical
Bureau. On the other hand, his job was to take the statistical reports
being received from lower level units (often late) and merge them into
aggregates for the statistical reports to be submitted to the State
Council in time for their next meeting . In fact, he implied there was
an extensive schedule of these reports and the dates they were due and
additional requests form the decision-makers were being received all the
time. Thus, it is little wonder the annual statistical yearbooks often
appear some eight or nine months after the end of the year.
[18] Lenin had made the same argument in the 20s about Soviet planning.
When some argued for scientific planning that would use forecasting and
mathematics, Lenin denounced the idea and thought that putting
intellectuals in charge of these decisions and placing them outside the
economic and political administrative bureaucracy would pose a threat to
the revolution.
[19] A very informative example of an attempt to collect the statistics
held by the local level, is the attempt by Stephen Butler, and then
Louis Putterman, to obtain the statistics kept by the accountants for
the almost 100 production teams, the basic unit in China's collectivized
agriculture, in a single commune in Hebei Province. This effort resulted
in the collection of the desired statistics for the years 1970-1985. The
information provided by this effort is almost as significant as the
statistics obtained. The individual team's statistics were in the
account books kept by the team's accountant and were the property of the
accountant, i.e., they were the accountant's responsibility. To obtain
the statistics, the accountants had to be tracked down and what Butler
brought back to the University of Michigan was a travel bag filled with
hand-copied pages from the account books. To be complete, it was
necessary to track down one former accountant, then living in Shanghai,
and get a copy of the account book in his possession. The point being
made: at the local level, the statistics being kept were not part of the
state's administration responsibility, but were kept by accountants or
statisticians who were responsibility to the economic unit to which they
belonged, submitting reports on behalf of that unit to higher levels,
where eventually they were included in the aggregate numbers reported at
the county level to the local bureau or office of the State Statistical
Bureau. A second lesson learned from this effort was that it was common
to keep two sets of books, the statistics to be reported to higher
levels and the statistics reflecting reality. By the time Putterman
arrived at the Commune, the reform period was underway and the rural
areas were under a much more liberal and decentralized policy regime.
Thus, the commune authorities saw no problem in giving Putterman both
the official statistics and the unofficial statistics they kept.
Finally, the statistics collected from this effort end with 1985, as by
that time the commune had become a township and a large portion of
activities in the rural areas had been convereted into household or
individual activities for market sale. Thus, by 1985, accountants or
statisticians at the local level had lost control over the statistics
they were supposed to keep and had to relp on sampling, the tax bureaus,
self-reporting, written receipts, etc. See Louis Putterman, Continuity
& Change in China's Rural Development: Collective & Reform Eras in
Perspective (New York: Oxford University Press, 1993), especially
Chapter 3, Dahe Township, pp. 82-121. All of the statistics obtained
from Dahe Commune, both official and unofficial, are available on floppy
disks from the Publications Office, Center for Chinese Studies,
University of Michigan.
[20] A collection of eighteen of his articles on China's economy has been
recently published under the title, Understanding China's Economy. See
Gregory C. Chow, Understanding China's Economy (New Jersey: World
Scientific Publishing Co., 1994).
[21] Gregory Chow, Chinese Statistics," in Gregory Chow, Understanding
China's Economy, op. cit., pp. 101-111. This article was originally
published in The American Statistician, vol. 40, No. 3 (August, 1986),
pp. 191-196.
[22] When the China Center at the University of Michigan was
negotiating with the Institute of Economics, Chinese Academy of Social
Sciences, in 1988 over a questionaire we were going to administer to
industrial enterprises for our collaborative study of the impact of the
economic reforms on the enterprise, our colleagues from China kept
reminding us that this questionaire had to be approved and administered
by the State Statistical Bureau. This did not prove to be a major
problem, however. Even today (1995), any research institute that wants
to collect a sample by means of questionaires must obtain approval from
higher authorities.
[23] Gregory Chow, Economic Analyses of the Poeple's Republic of China, in
Gregory Chow, Understanding China's Economy, op. cit., pp. 124-134. This
article was originally published in The Journal of Economic Education,
vol. 19 (Winter 1988).
[24] Gregory Chow, "Capital Formation and Economic Growth," in Gregory
Chow, Understanding China's Economy, op. cit., pp. 193-222. This article
originally appeared in The Quarterly Journal of Economics, August, 1993,
pp. 809-842.
[25] Gregory Chow, "A Model of Chinese National Income
Determination," in Gregory Chow, Understanding China's Economy, op.
cit., pp. 223-231.
[26] Gregory Chow, "Money and Price Level Determination in China," in
Gregory Chow, Understanding China's Economy, op. cit., pp. 232-246. This
article was published in Journal of Comparative Economics, vol. 11,
(1987), pp. 319-333.
[27] Faced with the lack of firm specific data with
which to test changes in efficience of the firm, in the mid-1950s, Prof.
Robert Solow presented a brief and simple model for using the available
macro data to estimate changes in "total factor productivity. That model
has become the standard for estimating changes in productivity in
Chinese enterprises as a result of the economic reforms, although few
remember that in his original article Prof. Solow admitted there was no
intellectual basis for what he was proposing; he was just trying to use
the statistics available to get an estimate of changes in productivity
and the only defense of his proposed method was that you either accepted
it or you did not.
Although economists had preached the theory of the gains from trade
for over a century, it was not until the World Bank collected a firm
specific data bank for Chile for a period of years that straddled a
change in tariff regimes and allowed a PhD student from the University
of Michigan, Ms. Lili Liu, to use that data set to test the impact of
the reductions in tariffs on the efficiency of the firm that economists
could show conclusively that due to the increased competition from
foreign enterprises, inefficient enterises closed, the enterprises that
remained became more efficient, and the new entrants were more efficient
as well.
[28] To give just a simple example of the problem, lets take the case
where large scale enterprises are much more efficient than small scale
enterprises and enterprises producing motors are much more efficient
than enterprises producing automobiles. If after the reforms, the number
of small-scale enterprises and enterprises producing automobiles were
reduced and the number large-scale enterprises and enterprises producing
motors were increased, even if the efficiency in all enterprises fell,
it is possible that the use of totals for the average enterprise would
show an increase in efficiency.
[29] Take, for instance, the industrial census of 1985, a complete census
of state enterprises throughout the China. Although several more
detailed versions have circulated internally. the published version made
available to the public was in eleven volumes. One volum,e reported on
the identity of each of the enterprises, one volume presented the
statistics for employment, another for capital stock, one fore output
mix, etc. One volume reported on the statistics for the individual
counties, which did include some useful information on the enterprises
in a particular county, but there was no way to relate the enterprises
reported in the volume on location with their statistics on employment,
capital stock, output mix, etc. reported in the other volumes.
[30] The year 1990 was chosen for us by the availability of a county-
boundary map that was digitized by the Chinese and for which we had
permission to distributed it as part of our CITAS China GIS. Our Chinese
collaborators also have providd us with all county boundary changes from
1982 through 1992 and it would be possible for us to develop and
distribute a CITAS China GIS for those years as well. The county-level
statistics available in the Provincial Yearbooks are much less detailed
and plentiful before 1990, howwever, and it was decided to move forward
in entering the county-level statistics from the Provincial Yearbooks
for more recent years. The statistics for 1991 and now available and
work is proceeding on the preparation of a CITAS China 1991 GIS at the
present time.