[This is part I or a 2 part series on the underlying statistics of the Great Leap Forward. Part II can be found here]
Recently, Ray wrote a great post – and readers added valuable comments – that provided some contexts surrounding the Great Leap Forward. When people discuss the Great Leap Forward (1958-1961), the starting point is almost always the millions killed. I want ask: how fair is that starting point? 1
In this post, I want to briefly focus on specific issue of the underlying statistics – and the often-made claim that millions and millions (I have heard upward of 70+ million!) died in the Great Leap Forward.
According to official Chinese data released in 1983, some 16 million died during the years of the Great Leap Forward (1958 to 1961). But how good is this number (or any other number)?
The most direct way to come up with such a number is if the government could actually count up all those who died during the years of the Great Leap Forward. To do so, however, the government would have needed to keep a good record of those who died before, during, and after the Great Leap Forward. We need data before and after to obtain a baseline against which deaths during the Great Leap Forward can be compared. The government would also have needed the capability to preserve those records and then later to count them up.
Unfortunately, such direct counts is next to impossible, for at least two reasons.
First, China was a poor country, made up predominately of poor rural population. It was not unusual at all to find a person alive who did not have a birth certificate and who died without a death certificate. Second, the birth and certificates – even when issued – were issued at a local level, often informally. The records were not standardized, were not systematically well kept or maintained, and were not regularly transferred to a central location. It is next to impossible now to demand a reconstruction and aggregation of all such records throughout the nation to determine a national death count.
The next best way to come up with a death count of the Great Leap Forward is by a good statistical estimate of death rates before, during, and after the Great Leap Forward – and from that, calculate based on an estimate of the population size at the time, the death attributed to the Great Leap Forward.
Unfortunately, no reliable estimates of Chinese national death rates exist for that time either. One might not see discussion of this simple fact in the standard Western narrative on the Great Leap Forward. However, outside the context of the Great Leap Forward, this fact is readily appreciated.
For example, while it is widely accepted that China’s development over the last 60 years has lifted millions out of poverty, scholars in the West have also argued that it is next to impossible to quantify exactly how many. As Banister and Preston noted 2:
It is generally believed that the People’s Republic of China has achieved significant mortality decline during the last 30 years. However, until very recently, the only data available from China showing this mortality decline were reported crude death rates for occasional years. The sources of these reported crude death rates have almost never been revealed, and the data could have been based on the total number of registered deaths compiled from localities throughout the whole country or only from selected areas of the country. Because of the possibility of serious underregistration of deaths or reliance on an unrepresentative sample of localities, analysis of China’s population have been reluctant to take these reported death rates at face value. Infant mortality rates, as occasionally reported for cities or rural areas, are often implausibly low, so that serious underregistration of infant deaths in particular appears likely.
Reported data on China’s level of morality have been scattered and of questionable validity, but data on the age patterns of mortality for the country as a whole have not been reported at all. … It is likely … therefore … that the government had no more idea of China’s pattern of mortality since 1949 than foreign analysts.
In 1980, for the first time, the People’s Republic of China informally released age-specific mortality data collected in a massive survey of registered deaths and causes of death. We report and analysis these data in an effort to describe China’s recent mortality level and pattern.
Without good national statistical estimates of the the death rate for the Great Leap Forward, the next best thing is to indirectly estimate the death rate by looking to Chinese census data, and from that estimate derive a death count.
The PRC carried out a first census in 1953, and a second one in 1964. 3 The results are shown below.
If one could estimate (guess) a reasonable normal baseline growth rate (growth rate = birth rate – death rate) during “normal years” in between these census years (1954-1957, 1962-1964), then by comparing the 1964 census figures and the expected 1964 figures, it is plausible to estimate how many died abnormally during the Great Leap Forward.
Indeed, many of China’s official government population statistics on population, birth rate, death rates during that period 4were derived thus.
|year||population||birth rate||death rate||growth rate*|
*growth rate is calculated via birth rate – death rate.
Table 1: Government population statics of China, 1953 – 1964
A graph of the birth rates and death rates in the table is provided below.
The government’s 16 million estimated death comes from estimating the “normal” baseline death rate to be .0108. Subtracting .0108 from the death rates in Table 1 to get an “abnormal death rate,” and then multiplying that result by the estimated population at the time, one gets an estimate of the numbers dead attributed to the Great Leap Forward for each year of the Great Leap Forward. The process is summarized below.
|year||death rate||abnormal death rate||estimated population (million)||estimated death due to GLF (million)|
|total GLF death (million)||15.30221377|
Table 2A: Estimated number of death in the Great Leap Forward based on data provided in Table 1 and an estimated baseline death rate of .0108.
This appears to be a very rough estimate. For one thing, as noted by jxie in this comment, the .0108 death rate appears to be low for a population in the 1950’s, even by Western standards. If we take an average of the published death rates during the baseline years (1954-1957, 1962-1964), we get a slightly different baseline death rate of .01134, resulting in a slightly lower number of death at 13.87 million. This calculation is summarized Table 2B.
|year||death rate||abnormal death rate||estimated population (million)||estimated death due to GLF (million)|
|total GLF death (million)||13.8661365|
Table 2B: Estimated number of death in the Great Leap Forward based on data provided in Table 1 and an estimated baseline death rate of .01134.
If we do above again using a baseline death rate of .012 (that of West Germany around 1960), then the estimated death becomes 12.13 million.
So given a little uncertainty in the baseline death rate, we get an estimated dead varying between of 12 – 16 million for the Great Leap Forward. But the problem is bigger than that. As it turns out, the census figures cannot per se give a good estimate with any margin of error we need of the death rate. 5
To illustrate this, let’s first calculate an estimated population for 1964 based on the 1953 census figure and the birth and mortality rates shown in Table 1. The process is shown below.
|year||census figure (million)||birthrate||death rate||growth rate||estimated population (million)|
Table 3: Estimated population from 1954 – 1963 based on 1953 census and government published birth and death rates
In this table, the 1954 estimated population is calculated to be 616.86 million, as 601.938 * (1+(.03797-.01318)). The 1955 estimated population is calculated to be 629.39 million, as 616.86 * (1+(.0326-.01228)). The process is iterated until we get the 1964 figure. 6 By this process, one estimates the 1964 population to be 716.36 million, a figure that is only .92% off from the actual 1964 census figure.
At first, the .92% would appear to confirm the death rates shown in Table 1 (a rate that would result in either a 12, 14, or 16 million figure, depending on the presumed baseline rate), with little leeway for the figure to be revised much more upward or downward.
However, it turns out the numbers can nevertheless be revised dramatically either way without violating the census figures. To see, this, one might replace the death rate for 1958-1961 in Table 3 during the Great Leap Forward years by a “normal” baseline death rate of 0.01134, as if no abnormal deaths occurred during the Great Leap Forward.
|year||census figure (million)||birthrate||deathrate||growthrate||estimated population (million)|
Table 4: Estimated population from 1954 – 1963 based on 1953 census and government published birth and death rates, with the death rates during the Great Leap Forward replaced by the “normal” baseline death rate of 0.01134.
Instead of an estimated 1964 population of 716 million, .92% below the 1964 census figure (see Table 3), we now get an estimated population of 731 million, 1.1% above the 1964 census figure. Not bad at all! Without additional information on the margin of errors of the 1953 and 1964 census, 1.1% would appear to be just as acceptable as -.92%.
In other words, to the extent that the census data of 1953 and 1964 support the assertion that 14 (or 12 or 16) million died, depending on the presumed baseline death rate, the data also support with equal likelihood that no one died as a result of the Great Leap Forward!
If one insists on a qualitative difference between an error rate of .92% vs. 1.1%, that insistence begins to make no sense when takes into account the birth rates as well. As can be noted by studying Table 1, the birth rates published by the government – even during non-Great Leap Forward years – fluctuated regularly, with a standard of deviation of .00384. Applying that uncertainty to the published published birth rates during the Great Leap Forward by shifting them downward by just one standard of deviation (.00384), one gets an estimated 1964 population of 720.43 million, for an error now of just +.3%, thus putting the error of the null hypothesis under that of the government hypothesis (.92%.) This process is summarized below in Tables 5 and 6.
|published rate||modified rate||difference|
Table 5: Modified birth rates used in Table 6.
|year||census figure||birthrate||deathrate||growthrate||estimated population|
Table 6: Estimated population from 1954 – 1963 based on 1953 census and government published birth and death rates, with the death rates during the Great Leap Forward replaced by the “normal” baseline death rate of 0.01134, and birth rates during the Great Leap Forward lowered by .00384 (the standard deviation of the birth rates of the baseline years).
We cannot therefore make any assertion whether anyone died as a result of the Great Leap Forward because we cannot refute the null hypothesis that no one died due to the Great Leap Forward, as far as the census data of 1953 and 1964 are concerned.
To make things more concrete, let’s now to consider the margin of error in the 1953 and 1964 census figures. The government does not provide an official margin of error for either the 1953 nor 1964 census. Given the stage of China’s development at that time (we are after all talking about the Great Leap Forward, about one of China’s failed attempt to advance and industrialize), it is not unreasonable to assume that those figures would possess high margins of error. The 1953 data – in particular – is especially troublesome.
To provide a bit of perspective, here is a quote from Wild Swans and Mao’s Agrarian Strategy, Australia-China Review, by Wim F. Werthheim, Emeritus Professor, the Univ. of Amsterdam 7
Often it is argued that at the censuses of the 1960s “between 17 and 29 millions of Chinese” appeared to be missing, in comparison with the official census figures from the 1950s. But these calculations are lacking any semblance of reliability.
At my first visit to China, in August 1957, I had asked to get the opportunity to meet two outstanding Chinese social scientists: Fei Xiao-tung, the sociologist, and Chen Ta, the demographer. I could not meet either of them, because they were both seriously criticized at that time as rightists’; but I was allowed a visit by Pang Zenian, a Marxist philosopher who knew about the problems of both scholars. Chen Ta was criticised because he had attacked the pretended 1953 census. In the past he had organised censuses, and he could not believe that suddenly, within a rather short period, the total population of China had risen from 450 to 600 million (by the way: with inclusion of 17 million from Taiwan), as had been officially claimed by the Chinese authorities after the 1953 ‘census’. He would have like to organise a scientifically well-founded census himself, instead of an assessment largely based on regional random samples as had happened in 1953. According to him, the method followed in that year was unscientific. For that matter, a Chinese expert of demography, Dr. Ping-ti Ho, Professor of History at the University of Chicago, in a book titled Studies on the Population of China, 1368-1953, Harvard East Asian Studies No.4, 1959, also mentioned numerous ‘flaws’ in the 1953 census: “All in all, therefore, the nationwide enumeration of 1953 was not a census in the technical definition of the term”; the separate provincial figures show indeed an unbelievable increase of some 30% in the period 1947-1953, a period of heavy revolutionary struggle (PP.93/94)!
My conclusion is that the claim that in the 1960s a number between 17 and 29 million people was ‘missing’ is worthless if there was never any certainty about the 600 millions of Chinese. Most probably these ‘missing people’ did not starve in the calamity years 1960-61, but in fact have never existed.
If we assume a margin of error of a mere 5% for the 1953 census (very generous given the nature of the 1953 census), and allow that margin of error to be propagated, we would get a 5% error for all estimates based on the 1953 on. An error of either .92 or, .3%, or 1.1% error would thus clearly be within the statical error supported by our data. In fact, with a 5% error, presuming no margin for error for the 1964 census, the data supports an absurdly wide range of estimates.
For example, by further assuming the birth rates throughout the Great Leap Forward to be constant, at the 1957 rate, the data support an assertion that as many as 70 million died. The calculation can be found below.
|year||census figure||birth rate||death rate||growthrate||estimated population|
|1964||723.0703||total GLF death (million)||69.3876463|
Table 8: Estimated population from 1954 – 1963 based on 1953 census and government published birth and death rates, with the birth rates during the Great Leap Forward assumed constant to be constant at 1957 levels, and the death rate presumed to be .038, with a baseline death rate of .01134.
An even higher number of deaths can be supported if we specify a nonzero margin of error for the 1964 census figure.
In light of the fact that the data we have is so spotty that it can support any of several millions upon millions died thesis as well as the no one died thesis, many have turned to anecdotal evidence to make their own estimates of the numbers death. Relying on anecdotal and piecemeal evidence was also the only way a Western researcher could study the Great Leap Forward during the Cold War years. With China closed off in part by an aggressive Western embargo against China, scholars in the West had little information into what China was really like on the ground.
Still, anecdotal evidence inevitably undermines serious research. They are considered bad even in politics. In the case of the Great Leap Forward, anecdotal evidence should be taken with an even more urgent degree of care. According to official government data, the effects of the Great Leap Forward affected different localities very differently. The general extrapolation of data from any one or a few localities is itself an error.
While this piece has focused on the underlying statistics, I also want to point out – as we come to the end – how I think we should also re-evaluate how the millions upon millions died thesis is framed. The Great Leap Forward occurred 1958-1961, but it did not occur in a vacuum. It occurred as part of a determined effort on the part of the Chinese to industrialize even as they strive for political independence. In that vein, perhaps the focus should not be on just the four years, but on the years from 1949 on. It’s been noted, for example, how India and China started on equal footing across many economic and human welfare indices in 1949. However, since that time on, China has outperformed India on almost all indices. On infant mortality alone, the persistent improvements China has made over the years over India means that China had averted hundreds of millions of death that still occurs in India every day today.
In another vein, more light should perhaps be shed on the international embargo directed at China at the time. Had the West relented on her inhumane embargo, much suffering could have been reduced. Whatever your view of the Great Leap Forward, the dominant narrative we hear today is definitely too simplistic.
Back to statistics. I want to conclude by emphasizing that I am not refuting per se that millions upon millions died in the Great Leap. I have no privileged information on which to make such assertions. It may turn out to be the case that millions upon millions did die, or it may turn out to be the case much fewer died than currently believed. However judging how readily and resiliently China’s GDP recovered after the Great Leap Forward (see chart below 8, notice how quickly the after the GLF, the rate of growth rebounded (if a massive number of people died, you might expect the slope to be permanently depressed, until the population could catch up)), my guess is that not that many people actually died – at least not on scale of millions upon millions. Whatever your inclinations, we must acknowledge front and center that no reputable systematic data exists to make broad pronouncements of how many actually died (or did not die) during the Great Leap Forward, and how much piecemeal and anecdotal evidence form the basis of current discussions of the Great Leap Forward in the West.
- For resources that raise doubt, see, e.g., this piece by Joseph Ball, or this interesting piece from “Set the Record Straight” ↩
- http://www.jstor.org/discover/10.2307/1972767?uid=2&uid=4&sid=21101590357931. ↩
- http://news.xinhuanet.com/english2010/china/2011-04/28/c_13850246.htm ↩
- see http://www.stats.gov.cn/english/ ↩
- Much of what I demonstrate below is based on the simple observation that the population estimates at the beginning and end of a period of time is not very sensitive to changes in annual death rates of a small sub-period of that time. The corollary is that if we depend on the population estimates at the beginning and end of a period of time to derive the death rates of a sub-period of that time, relatively small errors in the population estimates would allow for relatively large errors in the death rates of the sub period, so much as to potentially render any estimates of death rates unusable or irrelevant. ↩
- Note, to make these calculation, I have assumed the census population to be the population estimate for the beginning of the year. The estimated population I calculate is for the end of year. Modifications can be made for these assumptions – e.g. census information is mid-year – with some changes in my numbers, but the conclusion drawn would be the same. ↩
- Thanks to www21234 for bringing this to my attention. ↩
- data from http://en.wikipedia.org/wiki/Historical_GDP_of_the_People%27s_Republic_of_China ↩