Is Demography Just a Numerical Exercise? Numbers, Politics, and Legacies of China’s One-Child Policy

UNPD has continued to revise and refine its population projections since then, as it did in the past, with updated information of demographic parameters as input and new assumptions. The latest edition, published in 2017, projects a peaking year of 2029 with a peaking population number for China at 1.442 billion (UNPD 2017).

We use the term “spreadsheet demography” metaphorically. This is not a label for works carried out with spreadsheet programs, but a term to characterize demographic exercises (including population projections) undertaken without considering their empirical or sociological meaning, that are devoid of critical self-reflection, and that take primacy of population over people.

Although not all Chinese were equally restricted in their births over the policy’s three and half decades, the one-child restriction applied to a majority of the population (Gu et al. 2007). The Chinese government formally announced the ending of the one-child policy in October 2015 (Wang et al. 2016). China still has a birth control policy in place today, with restrictions on births at parity three and above. This remaining practice is likely a transitional step before totally abolishing all forms of mandatory birth control.

In Goodkind’s (2017a) article, our research is used as a straw man. Yet, we and almost all those whose works are criticized in that article were not invited to review and to respond prior to the paper’s publication.

As Goodkind (2017a) acknowledges, his article took several iterations before being published in Demography, including two versions presented at the 2016 and 2017 annual meetings of the Population Association of America (Goodkind 2016, 2017b), and an earlier working paper (Goodkind 1992). We refer them collectively as “the confirmation exercise.”

The real change of crude birth rate in China between 1950 and 1970 was anything but a clear linear trend. During that period, China experienced huge swings in birth rates—from 34.0 per thousand in 1957 to 18.2 per thousand in 1961, and then to 43.4 per thousand in 1963—a result of the Great Leap Forward movement and the following famine.

Goodkind (2017a) provides no explanation or justification why 2060 was chosen as the end point of his examination. One likely reason is that because his fertility convergence assumption (to a TFR of 1.5) determines the total population of his 16-countries counterfactual to have negative growth around 2060, thus he would produce a smaller number of so-called averted births had he further extended his projections to a future time—say, 2080.

Our use of the 16 countries as comparators, as we discussed in our published papers (Wang and Cai 2010; Whyte et al. 2015), was not to claim China was similar to these countries, but to expose how simplistic the calculations sponsored by the Chinese birth control agency were, and to illustrate how these countries with similar birth rate level back in 1970s all experienced fertility decline in the absence of an extreme birth control policy like the one in China.

The 4.3 billion scenario was based on a constant TFR of 3.0 for 100 years; the TFR was from a 1975 survey. In a later edition of their calculation, when more reliable fertility data had become available showing that the 1975 TFR was about 3.6, Song and Yu (1985:246) claimed in a footnote that “such a difference in fertility has no effect for population projection.”

This statement is not consistent with the fertility assumption he used in his projection, which assumes that fertility for all the counterfactuals converge to a TFR of 1.5 by 2060. If that assumption is true, China’s very low fertility is not “two or three decades too soon,” but more than a half-century too soon.

Although the debate about the role and the effectiveness of government policies in fertility transition is ongoing (Bongaarts 2011; Bongaarts and Sinding 2011; Bulatao 1993; Casterline and Sinding 2000; Jain and Ross 2012; Tsui 2001; Shen et al. 2017), general consensus is that socioeconomic development is the fundamental force driving fertility change in the long run but that government policy could play an instrumental role in the relatively short term under the right social and cultural contexts.

Goodkind also suggests that in our earlier publications, we were neglectful in not including Vietnam as one of the 16 comparison countries. He further states that “Given all these similarities, the near total neglect of Vietnam as a comparator by China population experts is conspicuous” (Goodkind 2017a:1380). Had he followed more closely the data sources on Vietnam (given his job responsibility at the U.S. Census Bureau’s International Program) or checked with us, he would have learned the reason. In 2010, when our initial article was written, we used data provided by the World Bank (2010) to find countries that had crude birth rate (CBR) of between 30 and 38 per thousand (enveloping China’s CBR of 33.4 in 1970) and a population of 1 million or more in 1970. Vietnam’s crude birth rate in the WDI 2010 was 40.6 (with a TFR of 7.0), outside the range that we chose. The UNPD subsequently revised Vietnam’s population data between 2010 and 2014. By 2014, the UNPD had changed Vietnam’s CBR for 1970 to 36.5 (and its TFR to 6.4). This revision could have been partly due to a major adjustment in Vietnam’s mortality, which would lead to adjustment in fertility as well: between 2010 and 2014, UNPD revised Vietnam’s life expectancy for females in 1970 from 51.5 to 65.0, a jump of 13.5 years! This adjustment, as the examples we give in the beginning of this article for China, illustrates again that estimates are subject to revision and demonstrates how serious demographers do their work.

The sentence immediately preceding the one noting similarities between China and Vietnam is, “To the constant irritation of the Vietnamese, external observers are struck with the similarities between China and Vietnam” (Womack 2009:10).

What Goodkind misses here was that the 400 million estimate was exactly what the Chinese birth control officials used to defend the one-child policy and to scare the public into believing that another baby boom and population explosion would ensue if the policy were abandoned.

Goodkind (2017a) cites his Fig. 1 in the article and also a paper presented at the 2017 annual meeting of the Population Association of America (PAA) (Goodkind 2017b). The figure in the article simply illustrates the gaps between China and the constructed counterfactuals, saying nothing about the jump itself. The cited 2017 PAA paper is a PowerPoint presentation in which the rebound was illustrated with a hand-drawn line as a gradual process in the early 1980s, peaking at TFR of 4.0 around 1985, and then gradually declining and reaching replacement around 2000, which is certainly not the same as what is presented in his published 2017 article.

The legal minimum marriage age was not lowered in 1980, but rather was increased by two years relative to the 1950 Marriage Law. Compared with the “later” marriage ages enforced during the 1970s, however, the effect was to legally permit earlier marriages.

For a long time, China’s birth control target was phrased in terms of annual number of births, or the (crude) birth rate. With a large cohort born after the Great Leap Forward famine entering their reproductive ages, the birth rate was expected to rise even if TFR were kept at the same level.

One reason for China’s current very low fertility is the continued rise in the age at first marriage. By 2015, the proportion of never-married women at age 30 had increased to over 10 %, from only 1 % in the 1980s. Unlike in the 1970s, when the increase in marriage age was largely due to government policy, this new round of increase is based on individuals’ voluntary choice in a large context of rapid urbanization, rising educational opportunities, and gender equality (Cai and Wang 2014). We expect this trend to continue.

Goodkind dismisses the Yicheng case by quoting statistics from Wei and Zhang (2014): “Yicheng’s crude birth rate was well below the Shanxi Province average from the 1940s through the 1990s, and then rose above the average after 2000” (Goodkind 2017a:1391). The quoted crude birth rate series of Yicheng is highly problematic: it shows no sign of the Great Leap Forward famine and does not correspond with census-based observations.

The full quotation is as follows: “Given China’s colossal footprint as a demographic billionaire, no policy intervention in history has done more to reduce the earth’s human population, and no single statistic better summarizes its impact” (Goodkind 2017a:1376).

Even with the number of only-children families estimated in China to date, it is literally impossible to establish how many only-children are entirely due to the policy versus a result of voluntary family decisions.

It is not entirely clear how Goodkind (2017a) derives his fertility estimates for China. He states, “To construct a fertility series for China, I turn to the two organizations that offer worldwide estimates based on broad demographic analysis: the UNPD and the U.S. Census Bureau. I use the former for estimates from 1970–1989 and the latter for 1990–2015” (p. 1382). However, earlier in that article, he states that “For China, the annual TFR series 1970–1989 is drawn from Banister (1987) and Feeney and Yuan (1994). These estimates are adjusted within each five-year interval to match estimates provided by UNPD (2015)” (p. 1378, in the caption to his Fig. 1). Our examination shows that Goodkind’s TFR series for 1970–1989 is not the same as that of Banister (1987), Feeney and Yuan (1994), or the World Population Prospects (WPP) 2015. No apparent adjustment algorithm could explain such differences.

The assumption that “TFRs in China and each counterfactual will converge to 1.5 by 2060” (Goodkind 2017a:1382) is curious because this convergence level is lower than both the U.S. Census Bureau’s (2017) estimate (TFR = 1.60 in 2050) and the UNPD’s estimate in WPP 2015 (TFR = 1.77 in 2060). Such a choice increases the so-called policy effect in Goodkind’s projection exercise.

Goodkind (Goodkind 2017a:1382) states that “sex ratios at birth come from the same sources.” Reading the sentence in the same paragraph of this statement, one would assume the sources were WPP 2015 for 1970–1989 and the U.S. Census Bureau (2017) for 1990–2015. Our examination suggests that his sex ratios at birth for 1970–2010 are from WPP 2015 and that his sex ratios at birth for years after 2010 are some adjusted version of the U.S. Census Bureau’s estimates.