1In many socialist countries, the transition to a market economy in the late twentieth century led to major political and economic upheaval. With several decades of hindsight, the short- and medium-term demographic consequences of these changes can now be analysed. More specifically, how did they affect the health and mortality of the populations concerned ? In this article, Jiaying Zhao, Edward Jow-Ching Tu, Guixiang Song and Adrian Sleigh examine the case of Shanghai in China, which underwent major economic reform from the 1990s. Using data from the death records of the permanent Shanghai population between 1974 and 2007, the authors show that the economic instability and insecurity of the period produced a temporary increase in mortality for certain population groups, followed by a resumption of progress and a faster rise in life expectancy.
2While the complex mortality patterns associated with the reforms of China’s socialist economy have not been adequately analysed, the mortality surge accompanying the transition to market economies in the former socialist economies of central and eastern Europe has been widely studied (Nolte et al., 2000a, 2000b ; Stuckler et al., 2009). This European research constitutes the main body of knowledge on the link between mortality and the economic transition from long-standing socialism to one or other variant of capitalism. Studies of the Chinese experience will expand our knowledge of how mortality trends are affected by reforms and institutional changes (and the speed and depth of these changes), by compensatory factors (e.g. GDP growth, health and social welfare policies), by the political environment and levels of social cohesion.
3Several pioneering studies have used available (but limited) national data to study economic reform and health effects in China (Liu et al., 1998 ; Banister and Zhang, 2005). However, the demographic and epidemiological characteristics of Chinese mortality patterns associated with economic reform are yet to be documented. This article addresses that knowledge gap by focusing analyses on Shanghai, China, a city with reliable death registration, a large population, and substantial economic activity.
4Interpretation of mortality trends in China is not straightforward. However, it is generally agreed that, except during the famine period, mortality in China fell continuously during the second half of the twentieth century (Banister and Hill, 2004). Mortality decline continued during the economic reform period beginning in 1978, but was not always observed at regional or county levels (Banister and Zhang, 2005). The initial living conditions, the timing of the reforms and the pace of change were not geographically uniform (Jefferson and Singh, 1999). Interpreting mortality trends in China is even more difficult when cause-of-death statistics are incomplete or inaccurate (Rao et al., 2005). In addition, European experience suggests that the adverse effects of socioeconomic reform on mortality may be more pronounced in certain urban areas (Walberg et al., 1998).
5For these reasons, analysis of the links between mortality trends and economic reforms should be restricted to Chinese sub-populations for which reliable information is available. This is the case for Shanghai, which has high-quality death statistics for a large share of its population. Major reforms of its socialist economic structure began in 1992 with large-scale privatization, along with the introduction of contract labour and the abolition of life-time job security.
6We analyse the effects of the economic reforms on mortality patterns among permanent residents who are identified as having a Shanghai hukou (household registration) in order to exclude migrants who might introduce selection bias, given the huge increase in the floating population (i.e. without a Shanghai hukou) since the 1990s (Census Office of Shanghai, 2002). The experience in Shanghai is analysed with reference to eastern Germany (East Germany until October 1990), Poland, the Czech Republic (called the Czech Socialist Republic until March 1990), and Russia,  which show similarities and differences in the links between mortality patterns and economic reforms, and reveal diverse mortality patterns and policy approaches (e.g. in health and social welfare) during the economic reforms (Cornia and Paniccià, 2000).
7This article begins by describing China’s economic reforms from planned socialism to state capitalism, spanning the period from 1978 to 2007. We develop a theoretical perspective on reform and mortality, drawing on experiences in the former socialist economies of central and eastern Europe. The focus then moves to Shanghai, and cause-of-death patterns in the city are analysed for three economic periods : the early reform period from 1978 to 1991 ; an initial period of major reforms from 1992 to 1996 ; a further period of major reforms from 1997 to 2007. The mortality experiences in eastern Germany, Poland, the Czech Republic and Russia during the corresponding periods serve as references for discussion purposes. The article concludes with a discussion that links the observations in Shanghai to theoretical perspectives on economic reform and mortality.
I – China’s transition to state capitalism
1 – Scope of the Chinese economic reforms
8The Chinese economic reforms occurred in two major stages. The first stage, that began in 1978 and continued for at least a decade, introduced rural “household responsibility” for marketing of farm products and opened coastal areas to international investment. The second stage began in 1992 when China’s 14th National Congress endorsed a “socialist market economy” as a model for economic reform and began the privatization of state-owned enterprises, largely affecting those living in urban areas.
9Since 1978, China has implemented major institutional change to promote its economic growth in a steady, systematic and staged manner. Its economic system has been radically transformed as a consequence (Li et al., 2000), although its political system has barely changed. The reforms involved several policy changes, including an opening-up policy for economic growth, enterprise and labour market reform, and reorganization of health care.
2 – Opening-up policy and GDP growth
10From 1980 to 1992, China gradually succeeded in opening itself up to the world. Shanghai was granted “open coastal city” status in 1984, which allowed the city to offer preferential conditions for foreign capital investment, including corporate income tax relief. However, from 1984 to the early 1990s, Shanghai remained tightly controlled by central government. Little was done to facilitate the entry of non-state firms, encourage foreign direct investment, deregulate trade and prices, liberalize the labour and capital market, or expand enterprise autonomy (Perkins, 1999). Shanghai’s economic growth was below the national average from 1980 to 1991.
11In the early spring of 1992, the senior leader Deng Xiaoping visited Shanghai, Guangzhou, Zhuhai and Shenzhen, sending a signal of commitment to greater reform and accelerated growth in China (Ash and Kueh 1996). Shanghai quickly emerged as the hub of a key region for open economic development in China. The city has seen double-digit annual growth in GDP since 1992, with a surge in foreign direct investments and an increase in income that has raised population living standards (Figure 1).
Changes in socioeconomic indicators, Shanghai, 1974-2007
Changes in socioeconomic indicators, Shanghai, 1974-2007
12From 1992, the distribution of income in Shanghai became more unequal, and the Gini coefficient increased from 0.16 in 1991 to 0.22 in 1996. Around this time, inflation in Shanghai also climbed sharply, peaking at 23.9% in 1994, before falling back to 2.8% in 1997.
3 – Enterprise and labour market reform
13The privatization and restructuring of Chinese collective and state enterprises were implemented under the slogan of “grasping the large and letting go the small” (zhua da fang xiao) (Nolan and Wang, 1999). Most small enterprises were leased or sold, while large firms were corporatized and merged into large industrial groups under the control of the Chinese state (Hsieh and Song, 2015).
14Initially, the Chinese labour market was modelled on socialist ideology with direct allocation of jobs from “cradle to grave” and minimal job mobility (Walder, 1986). In 1992 and 1993, the policy of state job assignment was largely abolished, with an increase in non-state-controlled businesses (Davis, 1999). State enterprises began to use labour contracts, adopt wage reform, and decentralize labour management. The secure “iron rice bowl” (i.e. a life-long job), enjoyed by urban workers rapidly became a thing of the past. Laid-off workers experienced substantial periods of unemployment with minimal welfare benefits.
15In Shanghai, the iron rice bowl lingered on in the 1980s (Lee and Warner, 2004), but contractual employment started to develop from the early 1990s. Many state-owned or collective firms became bankrupt or were restructured. The resulting labour turnover directly affected more than one-third of the working population (SMSB, 1983-2011 ; SMSB and NBSS, 2009 ; Zhu and Yuan, 2001). From 1992 to 1996, over one million staff in state-owned or collective enterprises – around one-fifth of the total working population – were laid off.
16Unemployment in Shanghai rose from a low of 0.2% in 1985 to reach 1.4% in 1991 and 2.7% in 1996 (Figure 1). Unemployment figures for 1992 actually underestimated job losses in Shanghai because laid-off employees remained registered with their former employer (Lee and Warner, 2004). An ever smaller fraction of the work force was employed in the public sector. The proportion of employees working in state-owned or collective enterprises decreased from 96% in 1991 to 82% in 1996, and had fallen to 22% by 2007.
4 – Health care reform
17The reform of state-owned and collective enterprises also brought changes in health insurance coverage. Before the reform, health services for workers in these enterprises were financed in urban China mainly through the Labour Medical Insurance System (Liu et al., 1999). When the economic reforms accelerated, less profitable enterprises could no longer guarantee full health insurance coverage for their employees and retirees. In urban China, fewer people were covered by health insurance, while the proportion who had to pay for services out-of-pocket rose from 28% in 1993 to 44% in 1998 (Gao et al., 2001).
18From the 1980s, health care became less affordable and less accessible due to lower government support and a new policy authorizing health services to earn profit from new drugs and technologies (Blumenthal and Hisao, 2005). In Shanghai, the mean annual number of hospital visits decreased from 6.9 per person in 1985 to 6.2 in 1991, and to 4.2 in 1996, although medical resources (e.g. beds per person) remained stable (Figure 1). The mean annual number of hospital visits started rising again in 1997.
19In order to solve the problem of loss of health service access, Shanghai began its health insurance reform in 1996. The key features included guaranteed access to basic care, wide coverage, joint premium contributions by employers and employees, and the integration of individual savings accounts and social pooling accounts (Liu et al., 2002).
II – Reform of socialist economies : theoretical analysis of effects on mortality
20The Shanghai experience of economic reform and its effects on mortality will be examined from theoretical perspectives which are framed and tested as three propositions.
1 – Economic reform of socialist economies may increase mortality
21In central and eastern Europe, the transition to a market economy was initially accompanied by a rise in adult mortality (Stuckler et al., 2009). During the transition, major institutional changes followed a hysteresis model of the labour market whereby formerly stable employment became unexpected unemployment (Cornia and Paniccià, 2000). The associated rise in mortality is attributed to disruptions in social organization and economic opportunity, and loss of survival strategies. Institutional changes and the lack of public measures to moderate their impact led to a mismatch between skills and labour market needs, a restructuring of alliances, and the demise of political patrons. All these factors led to psychological stress and, in the case of Russia, alcohol consumption to alleviate this stress (Cornia and Paniccià, 2000 ; Shkolnikov et al., 1998). The situation deteriorated further, with interruptions in social security, lower health insurance coverage, and reduced access to health care and hospitals. Rising inflation may also raise mortality indirectly by increasing poverty, especially among low-income populations (Brainerd, 1998).
22Together these changes led to increases in adult mortality from cardiovascular disease and alcohol-related diseases (e.g. liver disease, accidents, homicide, and suicide) (Leon and Shkolnikov, 1998). High intake of saturated fat was another reason for the high mortality rates from cardiovascular disease in central and eastern Europe (Kesteloot et al., 2006).
23During the economic transformations in central and eastern Europe, changes in mortality were more evident among working-age men than their female counterparts (Leon, 2011). This was attributed to greater psychological stress and increased alcohol consumption due to anxiety about job loss and labour turnover (Shkolnikov et al., 1998). As women tend to cope better with stressful events (Weidner and Cain, 2003), men are considered to be more susceptible to heart disease during periods of economic reform.
24China’s reform of its socialist economy, which started in 1978 and accelerated in 1992, tended to increase mortality. The dynamics of interacting processes in China are summarized in Figure 2. First, available published reports indicate that institutional reforms of enterprises and labour destabilized adult Chinese workers, causing a deterioration of working conditions and an increase in stress, especially for men (Li et al., 1999 ; Wang, 2002 ; Wang and Chen, 2004). Second, enterprise-related health insurance lapsed and the health care system was marketized (Blumenthal and Hisao, 2005). The resulting loss of access to health care would be expected to contribute to an increase in mortality (Gao et al., 2001). Third, the opening-up policy and consequent economic take-off, with rapid growth of GDP, had both negative and positive effects on Chinese mortality (Liu et al., 1998). Negative effects included a rise in road traffic accidents and associated mortality due to an increase in car and motorbike use without adequate safety or road infrastructure (Zhao et al., 2012). Higher incomes also led to changes in diet (e.g. increase in the proportion consuming a high-fat diet) which may have raised the risk of mortality from cardiovascular disease and related chronic diseases (such as diabetes) during the reform period (Popkin, 1994).
Conceptual framework for links between reforms and mortality in China
Conceptual framework for links between reforms and mortality in China
2 – Reform-associated mortality can be reduced under certain conditions
25The adverse effects of economic reforms on mortality in the former socialist economies of central and eastern Europe were not uniform. In Russia, for example, during the economic transformation from 1990 to 1994, male life expectancy fell by six years (Leon, 2011). But several other transition economies, including Poland, eastern Germany, and the Czech Republic experienced much smaller falls in male life expectancy – about one year or less (Cornia and Paniccià, 2000).
26Stuckler and colleagues explored the effects of reform on mortality in 25 European countries (Stuckler et al., 2009). They reported a link between the extent of privatization and a short-term increase in adult male mortality due to increased unemployment. Unfavourable mortality trends linked to economic reforms of socialist economies can be attenuated by economically successful reform, high social capital and strong institutions (Brainerd, 1998 ; Kennedy et al., 1998 ; Popov, 2012). New foreign direct investment encouraged by these reforms provides employment and lowers mortality (Stuckler et al., 2009).
27The variation in mortality associated with economic reform in former socialist economies of Europe suggested that the mortality effect in China should be moderate. This is because China had a strategic approach to privatization. The strategy was to retain strong institutional capacity and social stability while raising living standards through rapid GDP growth, a surge of foreign direct investment and relatively low inflation rates (Figure 2).
3 – The adverse mortality effects of Chinese economic reform might also be transient
28Within a few short years, a downturn in mortality was observed in some former socialist economies of Central Europe (e.g. former East Germany, Poland, the former Czech Socialist Republic), in parallel with economic stabilization, the adaptation of populations to the new social and economic environment, economic growth, better diets, and improved medical care (Grigoriev et al., 2014 ; Leon, 2011 ; Nolte et al., 2000b). However, in Russia, after an initial rise and fall in the early 1990s, adult mortality increased again from 1998 to 2003, coinciding with a second economic crisis (Zaridze et al., 2009).
29Policies aimed at supporting employment help to contain the rise in unemployment, limit the fall in living standards and maintain income levels, thereby reducing stress-related mortality (e.g. cardiovascular disease mortality, suicide ; Cornia and Paniccià, 2000). For example, in the Czech Republic, active policies were applied in a timely fashion over the period 1991-1993 when the structure of the labour market underwent radical change. These policies effectively supported high flexibility of the labour force, significantly moderated structural unemployment, and reduced mortality among the working-age populations. Since 1991, life expectancy in the Czech Republic has increased.
30China also adopted policies to mitigate the adverse effects of economic reforms. In 1996, a first provincial Re-employment Service Centre was established in Shanghai (Lee and Warner, 2004), the first city to implement a combination of welfare provision, employment services and re-training programmes. Hence, considering the substantial GDP growth, the improvement in health insurance programmes after the mid-1990s and the implementation of effective policies to fight unemployment and mitigate the effects of institutional change, we would expect mortality to fall rapidly in Shanghai after an initial increase.
III – Methods
1 – Data on mortality in Shanghai
31We used mortality data for Shanghai, obtained from the official registration system, to address the above three propositions. Before 1991, annual mortality data were aggregated by sex, age, and cause ; from 1991 individual data were available. Causes of death from 1974 to 2007 were based on various codes : the Chinese classification of diseases for 1974-1988, the Ninth Revision of the International Classification of Diseases (ICD-9) for 1989-2001, and the Tenth Revision (ICD-10) for 2002-2007. We classified causes of death into seven categories over the period 1974-2007 : infectious diseases, neoplasms, cardiovascular diseases, respiratory disease, digestive diseases, external causes, and other diseases. From 1991, we used more detailed cause-of-death analyses because monthly records of individual death certificates became available.
32Death certificates are completed by community doctors for natural deaths at home, by hospital doctors for natural deaths in hospital, and by coroners for unnatural deaths (e.g. suicide, homicide). Deaths are recorded by the local police based on the deceased person’s hukou (household registration) and reported to the health bureau for coding. Quality measures in place throughout the 1990s for Shanghai mortality data include physician training, registration guidelines, computer checks, case reviews for non-specific causes, and special hospital procedures to detect child deaths among children known to have chronic life-threatening illnesses.
33In China, newborns are registered at the permanent household residence (hukou) of their father or mother rather than their usual residence (Banister, 1984). The conversion of a hukou from one locale to another requires official approval, a process subject to conditions set out in numerous regulations (Zhang, 2012). Burials do not take place until the deceased person’s hukou records are confirmed, ensuring complete and accurate mortality data for permanent residents of Shanghai. “Floating” persons (without Shanghai hukou) are not included in these death records.
34To assess the completeness of registered deaths in Shanghai households, we compared the number of deaths based on the official registration system with those from the 1982 and 1990 census. These two censuses are generally of high quality, with reliable mortality indicators (Li and Sun, 2003). In Shanghai, the censuses recorded floating populations of less than 0.5% in 1982 and 4% in 1990 (Census Office of Shanghai, 2002). Deaths recorded in the 1982 and 1990 censuses included these floating populations and exceeded the registered deaths by approximately 0.5% and 4%, respectively. We conclude that the difference between registered deaths based on hukou data and and deaths based on census data increased from 1982 to 1990 because of a rise in the floating population rather than any deterioration of register data quality.
35Also, when registered death rates in Shanghai were compared to the corresponding rates obtained from regional data in the first (1973-1975) and third (2004-2005) national cause-of death surveys, there were no significant differences in mortality rates (SCDC, 2007 ; Zhou et al., 1986). Furthermore, compared to the third national cause-of-death survey, the death registration was 99% complete and 98% accurate (SCDC, 2007). Moreover, based on our calculations, the proportion of deaths coded as ill-defined was always below 6%.
36The total population (denominator) for computing death rates was obtained from the household registration system, which only includes permanent residents with a Shanghai hukou. Age- and sex-specific populations were obtained for the years 1974, 1979, 1985, 1992, 1996, 1999, 2001, 2006, and 2007. Denominator data for the population in the household registration system were not available in 1982, so we used the 1982 census population age-sex structure as the denominator for that year because the floating population was proportionally very small (<0.5%) (Census Office of Shanghai, 2002). These population data provided denominators for deaths of persons with a Shanghai hukou, avoiding the selection biases introduced by migrants. This restriction is an important issue given the huge increase in the migrant population relative to the total population of Shanghai since the 1990s (4.5% in 1992, 39% in 2011) (SMSB, 1983-2011).
37The population for inter-data years was projected using the cohort-component method (Smith et al., 2001). The monthly population was derived from the inter-census estimation by assuming that population change over a period of two years is linear (Shryock and Siegel, 1973), and mortality rates were standardized by multiplying the World Health Organization (WHO) standard population by the observed Shanghai age-specific death rates and age-cause-specific death rates (Ahmad et al., 2001).
2 – Statistical analysis
38A decomposition method was applied to analyse age-cause-specific mortality patterns and their contribution to the change in life expectancy (Arriaga, 1984). Decomposition revealed the main causes of death for several cohorts.
39Death data were assembled and analysed as monthly rates, creating more statistical power than would have been available if annual totals had been used. To assess the statistical significance of rising and falling mortality over time during the economic reforms, log-linear regression analyses were employed for total mortality and for mortality from certain causes (neoplasms, cardiovascular diseases, external causes, and liver disease). Separate regressions covered the periods 1992-1996 and 1997-2007. To account for seasonal fluctuations apparent when dealing with monthly data, harmonic functions were added to the log-linear regression (Stolwijk et al., 1999). The harmonic functions assume that the seasonal pattern follows a cosine function with variable amplitude and horizontal shifts (Stolwijk et al., 1999). The equation can be represented as follows :
41T indicates month (T=12). The value of β1 shows the annual change in the mortality rate. For example, β1= 0.0296 implies that the mortality rate increased annually by a factor of exp(0.0296) = 1.03, i.e., a 3% increase in one year. By contrast, β1= –0.0296 indicates a decrease in the mortality rate of 3% (exp(–0.0296) = 0.97) annually. The sine and cosine functions (β2, β3) are combined to describe the seasonal pattern. The amplitude of the seasonal changes equals
43When harmonic regression analysis is compared graphically to non-harmonic log linear regression, the lines of best fit are improved by introduction of the harmonic functions.
IV – Results
1 – Trends in mortality and longevity in Shanghai, 1974-2007
44Overall life expectancy in Shanghai increased relatively consistently from 1974 (69.7 years for males ; 74.4 years for females) to 2007 (79.1 years for males ; 83.5 years for females) (Figure 3). The average annual increase in life expectancy differed across periods : 0.22 years for males and 0.17 years for females in 1974-1985, 0.31 years for both males and females in 1986-1991, 0.08 years for males and 0.14 years for females in 1992-1996, and 0.42 years for both sexes in 1997-2007. The average annual increase in life expectancy for both sexes was slower from 1992 to 1996 than before (1974-1991) and after (1997-2007) this period. The period 1992-1996 coincided with the major economic transition in Shanghai.
Life expectancy at birth in Shanghai, 1974–2007
Life expectancy at birth in Shanghai, 1974–2007
45Note that the increase in life expectancy was slow between 1979 and 1983, and this may reflect an apparent increase in infant mortality associated with improved reporting of infant deaths.
46To analyse probability of dying, the population was grouped into four age categories : children and adolescents (0-19 years) ; young working adults (20-44 years), older working adults (45-64 years), and the elderly (65-79 years). During the period 1974-1991, the probability of dying decreased for all age groups (Figure 4). At ages 0-19, it fell by 40% for males and 36% for females, and at ages 20-44 years and 45-64 years it fell by 29-32% for both sexes. The smallest fall was in the 65-79 age group (19% for males, 21% for females).
Probabilities of dying by age group and sex, Shanghai, 1974-2007
Probabilities of dying by age group and sex, Shanghai, 1974-2007
47During the period 1992-1996, the probability of dying rose substantially (by 14%) for young male adults aged 20-44 years (p<0.001) (Figure 4 ; Table 1), but decreased for males in the older age groups and for adult females.
Parameter estimate (β1) for trends in monthly mortality, Shanghai, 1992–2007
Parameter estimate (β1) for trends in monthly mortality, Shanghai, 1992–2007Notes : β1 is a coefficient of time (year) derived from a log-linear regression model of mortality over the modelled period. An 0.03 increase or decrease in β1 signifies a 3% change in annual mortality.
Significance levels : * : p < 0.05, ** : p < 0.01, *** : p < 0.001.
48The probability of dying dropped remarkably for both males and females in almost all the age groups for the period 1997-2007. However, infant mortality showed an obvious increase because, beginning in 2002, deaths of abandoned babies from Shanghai’s Children’s Welfare Institute (SCWI) were included for the first time. Death rates from 2002 did not increase if deaths from SCWI are not taken into account. For males, the most notable change was the falling mortality at ages 20-44 years (Figure 4 ; Table 1).
2 – Causes of death in Shanghai
49Causes of death can offer another angle to understand mortality trends. The analysis shows annual contributions to changes in life expectancy of deaths at different ages from different causes over three periods (1974-1991 ; 1992-1996 ; 1997-2007) in Shanghai (Figure 5). The bars indicate the total gain (positive value) or loss (negative value) of life expectancy in years by age group. The coloured segments within each bar indicate the relative contributions of seven different causes of death.
Contribution of changes in age-cause-specific death rates to gains and losses in life expectancy (in years) by sex over three periods (1974-1991 ; 1992-1996 ; 1997-2007)
Contribution of changes in age-cause-specific death rates to gains and losses in life expectancy (in years) by sex over three periods (1974-1991 ; 1992-1996 ; 1997-2007)
50Over the period 1974-1991, the annual increase in life expectancy (0.25 years for males ; 0.22 years for females) reflects declining mortality from all major causes of death. For example, the annual standardized death rate (SDR) from cardiovascular diseases decreased from 245 deaths per 100,000 (24.4% of total SDR) in 1974 to 205 (27.6% of total SDR) in 1991 for males, while the corresponding figures for females were 197 (28.5%) and 166 (32.4%), respectively. This falling cardiovascular mortality contributed about 15% for males and 18% for females to Shanghai’s increase in life expectancy. The SDR from neoplasms fell between 1974 (218 per 100,000 for males, 116 per 100,000 for females) and 1991 (202 per 100,000 for males, 106 per 100,000 for females). There was also a decreasing trend in the standardized death rate from external causes for both males (68 per 100,000 in 1974, 52 per 100,000 in 1991) and females (51 per 100,000 in 1974, 41 per 100,000 in 1991).
51By contrast, during the period 1992-1996, while overall mortality continued to fall, but not so dramatically, mortality from certain causes of death increased (Table 1). For example, the annual SDR from cardiovascular diseases for males went up between 1992 (210 per 100,000) and 1996 (217 per 100,000). These changes in SDR from cardiovascular diseases contributed a negative annual component of –0.031 years to overall life expectancy for males, and of –0.005 years for females (Figure 5). For males, the SDR from cardiovascular diseases increased for working-age groups and for the over-65s (Table 1). In addition, male mortality included an annual negative contribution of –0.018 years due to death from external causes, partly as a result of increased traffic accidents (Figure 5). Mortality from external causes (including traffic accidents) increased in Shanghai from 1992 to 1996, especially among young adults, as automobiles became available to inexperienced drivers and road works became extensive (Table 1).
52Conversely, mortality from neoplasms and liver disease fell for both sexes over the period 1992-1996, while suicide mortality remained stable for males aged 20-44 years and fell for older males and for all female age groups. The net effect of these diverse cause-of-death trends was a small decline in overall mortality and a slight increase in longevity over the period.
53From 1997 to 2007, mortality in Shanghai fell rapidly for all major causes of death (cardiovascular diseases, neoplasms, respiratory diseases, digestive diseases, and external causes), leading to a large annual increase in life expectancy of 0.42 years (Figure 5). For males, the standardized death rate from cardiovascular diseases fell from 204 per 100,000 in 1997 to 142 per 100,000 in 2007, and for females from 160 to 110. This mortality reduction contributed 29% of the gain in life expectancy for males and 33% for females. Declines in transportation and suicide mortality were also remarkable.
V – Discussion
54This study has produced three main findings for Shanghai. First, in 1992-1996, a period of major upheaval linked to the economic reforms, the previous steady improvement in life expectancy slowed down. Although mortality from cardiovascular diseases and external causes among working-age men increased, these changes were counterbalanced by improvements in mortality from neoplasms and “other” categories. Transportation deaths increased, while mortality from suicide and liver disease remained stable or fell. Second, the unfavourable mortality outcomes during the initial economic reforms were moderate, and male life expectancy did not decrease overall. Third, the unfavourable mortality patterns were temporary and mortality dropped rapidly after the initial negative impact of the reforms.
1 – Comparison of reform-associated mortality patterns in Shanghai, eastern Europe and Russia
55The patterns of reform-associated mortality in Shanghai display both similarities and dissimilarities with respect to the Czech Republic, eastern Germany, Poland, and Russia. The reforms in these four European countries mainly occurred after the Velvet Revolution (November 1989) in former Czechoslovakia, the fall of the Berlin Wall in East Germany (November 1989), the Round Table negotiations to legalize Solidarity in Poland (the Spring of 1989), and the dissolution of the Soviet Union (December 1991). After these events, these European populations initially experienced unfavourable mortality trends, which lasted for one or more years. Reform-associated longevity changes mainly affected men in the 20-44 age group (The Human Mortality Database, 2012), and this age-sex group therefore acts as a sentinel for the major mortality effects of economic reform. The period of increased mortality for young adult men after the major reforms can thus be characterized, and mortality patterns before and after the reform can be compared.
Period before initial mortality surges
56Mortality trends in Shanghai from the 1970s until 1991 differed from corresponding trends observed during the pre-reform period in the four selected populations. In 1974, life expectancy for females in Shanghai was similar to that in certain former socialist economies of Eastern Europe and the Soviet Union (around 74 years). For males, life expectancy was highest in Shanghai (70 years), followed by East Germany (69 years), Poland (68 years), the Czech Socialist Republic (67 years), and Soviet Russia (63 years). From 1974 to 1991, life expectancy increased steadily for both males and females in Shanghai, but male life expectancy improved very slowly in East Germany and the Czech Socialist Republic and even declined in Poland and Russia (Cornia and Paniccià, 2000 ; Meslé and Vallin, 2011 ; Nolte et al., 2000a). Female life expectancy in East Germany and the Czech Socialist Republic improved, but more slowly than in Shanghai (Nolte et al., 2000a ; The Human Mortality Database, 2012), and remained stable in Poland and in Russia (Nolte, Shkolnikov and McKee, 2000a ; The Human Mortality Database, 2012 ; Vallin and Meslé 2004).
57From 1974 to 1991, falling mortality from cardiovascular diseases contributed about 15% to Shanghai’s improved life expectancy for males, while rising cardiovascular disease mortality was a major component of the worsening pattern in Poland and Russia (Meslé, 2004). Deaths from external causes were another important contributor to falling (or rising) male life expectancy during the period. In short, Shanghai had a pattern of changing male mortality that diverged from the patterns manifested by the four selected countries. For females, the pattern of mortality in Shanghai was broadly similar, though slightly more favourable (Nolte et al., 2000a ; Vallin and Meslé 2004). This divergence between the sexes suggests that males in Eastern Europe (but not in Shanghai) were exposed to risk factors (e.g. excessive alcohol consumption) that may have been important for male life expectancy during this period.
Period of initial reform-associated mortality increase
58Male life expectancy fell by 0.37 years annually from 1990 to 1991 in eastern Germany, by 0.58 years in 1990 in the Czech Republic, by 0.40 years from 1989 to 1991 in Poland, and by 2.01 years from 1992 to 1994 in Russia (The Human Mortality Database, 2012). Over the corresponding periods, female life expectancy in eastern Germany increased by 0.11 years annually, stagnated in the Czech Republic, but fell by 0.15 years in Poland and 1.05 years in Russia. During the period 1992-1996, life expectancy in Shanghai rose by 0.08 years annually for males and 0.14 years for females.
59Mortality patterns for the 20-44 age group are compared for the four selected European countries and Shanghai in Table 2. The probability of dying rose by 14% for 20-44 year olds in Shanghai, while the corresponding figure was 10% in the Czech Republic. An even steeper increase in mortality was noted in Poland (18%), eastern Germany (29%), and Russia (68%) during the initial transitions. In all settings, the transition-associated mortality increase was largely due to cardiovascular diseases and external causes.
Selected features of male mortality during economic reforms in Shanghai, Czech Republic, eastern Germany, Poland, and Russia(a)
Selected features of male mortality during economic reforms in Shanghai, Czech Republic, eastern Germany, Poland, and Russia(a)(a) Major causes of death producing loss of life expectancy during the early phase of reform, or contributing to increase in life expectancy during the late phase of reform.
Note : CVD = cardiovascular diseases.
60A notable mortality difference between Russia and Shanghai is linked to the effects of alcohol, suicide, and liver disease. Mortality from suicide and liver disease increased in Russia, as alcohol intake surged (Leon et al., 1997). Alcohol consumption increased moderately in the Czech Republic, fluctuated in Poland, and fell in eastern Germany (Cornia and Paniccià, 2000). In China, especially in Shanghai, alcohol consumption was much lower than in these four countries (WHO, 2004).
61Similar patterns of increased post-reform automobile mortality were observed in Shanghai, the Czech Republic and eastern Germany (Cornia and Paniccià, 2000 ; Winston et al., 1999). In eastern Germany, death rates from road traffic accidents increased fourfold from 1989 to 1991 due to the economic change and to the availability of cars, resulting in a rise in both vehicle ownership and the number of inexperienced drivers on the roads (Winston et al., 1999). In contrast, transportation mortality in Russia went down during 1991-1994, due to a decline in road traffic resulting from increased fuel costs and economic stagnation (Shkolnikov et al., 2001).
Period after the initial mortality surge
62Life expectancy rose remarkably in Shanghai, eastern Germany, the Czech Republic, and Poland after the initial unfavourable mortality patterns (Fihel, 2011 ; Grigoriev et al., 2014 ; Vogt and Kluesener, 2011). In 2007, life expectancy reached 79 years for males in Shanghai (83 years for females), 76 years in eastern Germany (82 years for females), 74 years in the Czech Republic (80 years for females), and 71 years in Poland (80 years for females). The improvement in life expectancy was mainly attributable to a decrease in deaths among the over-50s (Fihel, 2011 ; Meslé, 2004). Mortality from cardiovascular diseases and external causes also fell. The reduction of mortality in eastern Germany, the Czech Republic and Poland from the 1990s was attributed to lower fat intake, increased fruit and vegetable consumption, a better economic environment, rising living standards, and improved medical care (Meslé and Vallin, 2011 ; Nolte et al., 2000b).
63In contrast, Russian life expectancy fell again during a second economic crisis (1998-2003) very similar to that observed in 1992-1994 (Grigoriev et al., 2014). But life expectancy started rising again from 2004, and by 2007, it was 61 years for males and 74 years for females. Cardiovascular diseases and external causes were major contributors to variations in mortality from 1995 (Grigoriev et al., 2014 ; Leon, 2011). There is some evidence that recent increases in Russian life expectancy can be attributed to lower alcohol consumption, healthier diet, improvements in medical care, and rising living standards, while changes in smoking behaviour are of minor importance (Grigoriev et al., 2014).
64In recent years, mortality trends in Shanghai and in the selected east European countries have begun to converge. These populations are joining the cardiovascular revolution, with a shift toward a new epidemiologic profile which is accompanied by an impressive decrease in cardiovascular disease mortality. Some research suggests that a return to mortality stagnation cannot be ruled out in Russia, however (Grigoriev et al., 2014 ; Vallin and Mesle, 2004).
2 – Explanation of reform-associated mortality in Shanghai
65What is the explanation for the initial unfavourable components of reform-associated mortality trends in Shanghai ? We believe that the answer lies in institutional changes in Chinese society (e.g. weakening role of the danwei), consequent psychological stress, and increased exposure to unsafe environments.
Institutional changes : the role of the danwei
66Before economic reform, almost all urban workers in China were organized as part of a work unit (danwei). Danwei had multiple social, political and economic functions and offered employment for life (Naughton, 1997). Workers and families depended upon their danwei for social status, career chances, social and medical insurance (Walder, 1986 ; Wong and Lee, 2000).
67The change from centrally planned socialism to state capitalism was an abrupt and violent shock. Widespread mass layoffs in state-owned and collective enterprises from the early 1990s weakened the traditional life-time dependency of employees on their danwei. Work unit profitability became a key factor in resulting social stratification (Xie and Wu, 2008). Furthermore, downsizing danwei could not reimburse large medical bills for their employees and retirees, who had to pay out-of-pocket for their health care costs (Hu et al., 1999 ; Liu et al., 2002). Income inequity combined with reduced health care access began to narrow China’s mortality advantage over countries with similar levels of development (Zhao, 2006). Although newly expanding private firms became an alternative provider of resources and life chances, they took less responsibility for employees’ social and medical welfare than the danwei system.
68The initially unfavourable mortality pattern also reflects the unaffordable cost of residual social insurance, particularly for health care (Blumenthal and Hsiao, 2005 ; Liu et al., 1999). At that time, health care cost recovery and profit-seeking raised the price of medical care (Blumenthal and Hsiao, 2005). Insurance coverage and health service uptake declined despite growing urgent needs (Gao et al., 2001).
69European evidence suggests that job loss, uncertainty about the future, and loss of welfare entitlements due to institutional changes associated with economic reform produce sharply rising levels of psychological stress (Cornia and Paniccià, 2000 ; Shkolnikov et al., 1998). Such stress affects people who have lost their job or are confronted with unfamiliar labour market conditions, especially when permanent employment was the previous expectation. In Shanghai, many people experienced major changes in their established working and living conditions, such as growing (and previously unknown) uncertainty, and rapid obsolescence of learned behaviour. Job insecurity became the Chinese worker’s top concern during this period (Morris et al., 2001) ; many workers were also stressed due to lost social protection from work units and ongoing hope for a strong state commitment to welfare (Wong and Lee, 2001). Laid-off workers had a higher level of psychological stress and anxiety than others (Li et al., 1999 ; Wang, 2002 ; Wang and Chen, 2004). Stress became unavoidable, particularly among working-age populations.
70The adverse biological effects of psychological stress are complex. Stress impairs cognitive abilities, motivation, confidence and self-respect, and excess alcohol consumption compounds the problem (Cornia and Paniccià, 2000). The effects of stress factors interacting with stress relievers (e.g. alcohol) are complicated further by reduced access to health services (Leon et al., 1997). Psychological stress is also associated with cardiovascular disease morbidity and mortality (Cohen et al., 2007 ; Cornia and Panicci, 2000 ; Rozanski et al., 1999).
71As traditional breadwinners, men often experience more stress than women when facing job loss and uncertainty. In eastern Europe, differential psychological stress contributed to a gender gap in heart disease during the initial phase of economic reforms (Weidner and Cain, 2003). Also, as the economic structure shifted from manufacturing to services, women were more likely than men to adjust to a new service occupation (Sun et al., 1998). Psychological stress is also associated with cardiovascular morbidity and mortality (Cohen et al., 2007 ; Cornia and Panicci, 2000 ; Rozanski et al., 1999).
Increased mortality from external causes
72Increased mortality from external causes during the economic reform in Shanghai was another major component of unfavourable mortality trends. This was an outcome of poor work and transportation safety during fast socioeconomic development. Exposure to injury risks increased, as did labour-intensive manufacturing, intensive construction, and motorization (Zhao et al., 2012). Transportation mortality increased during the reform due to poor preparation for increasing motorization. From 1992 to 1996, the number of vehicles doubled from 170 to 321 per 10,000 population, without the corresponding development of highway or driver safety (SMSB and NBSS, 2009). This outcome was similar to that experienced by the Czech Republic and eastern Germany.
Other possible explanations
73Changes in diet were of minor importance for explaining the short-term reform-associated mortality fluctuation in 1990s Shanghai, especially among males. In contrast to eastern European countries, cardiovascular disease mortality had started to decrease in Shanghai in the mid-1970s. There is some evidence that annual meat consumption increased throughout the 1980s and 1990s, reaching a peak in 2002, and then decreased (SMSB and NBSS, 2009). In other words, changes in diet cannot fully explain the pattern of cardiovascular disease mortality during the reform period.
74Additionally, some scholars have concluded that there is no clear evidence of a rapid change in smoking habits during the transition that might have prompted an immediate mortality increase (Stuckler et al., 2009). Also, suicide and liver disease in Shanghai remained stable or fell, suggesting that alcohol intake had little adverse effect. The rise in pollution due to industrialization is another potential explanation for increased reform-associated mortality. However, older adults and young children are more vulnerable to air pollution than the working-age population, so pollution is not a major factor explaining the reform-associated mortality pattern.
3 – Moderate the impact of economic reforms on mortality in Shanghai
75In Shanghai, economic reform under a stable political environment led to increased economic production without seriously disrupting the social order or increasing crime rates (Shanghai Yearbook Editorial Board, 1996-2000). As in some central European countries, new foreign investments in Shanghai provided economic opportunities that helped combat unemployment (Stuckler et al., 2009).
76It would appear that countries with strong post-reform institutions, either democratic (like the Czech Republic) or authoritarian (like China), tended to cope better with the adverse social effects of the reforms than other socialist economies (Popov, 2012). China managed a strategy of gradual transition that allowed the required restructuring with reallocation of capital and labour over a long period of time. This reduced the degree of stress associated with unemployment and labour turnover. A more gradual transition preserved institutional capacity and avoided or mitigated the collapse of outputs (Popov, 2012).
77In urban China, the danwei continued to play a very important role for workers in state-owned firms and served as a buffer against employment instability (Gu, 1999). Many laid-off workers maintained ties with their original danwei, which provided a partial living allowance and some other danweirelated welfare (e.g. pensions).
4 – Transience of reform-associated mortality in Shanghai
78Mortality in Shanghai dropped rapidly after 1996, four years after the major economic reforms. The transience of mortality effects was also observed in eastern Germany, the Czech Republic, and Poland. A properly designed and timely employment programme can lower mortality among the working-age population (Cornia and Paniccià, 2000).
79Improvements in the social security system in Shanghai from 1996 offset the negative influence of unemployment. In response to the new circumstances, Shanghai established Re-employment Service Centers in 1996 to combine welfare, employment services and retraining (Lee and Warner, 2004 ; Zhu and Yuan, 2001). Also in 1996, an array of other government measures were introduced to subsidize the cost of hospitals and drugs (Wang, 2008). In 2003, health insurance coverage was further expanded to people in the counties of Shanghai. By then, the worst effects of the reforms had receded.
5 – Limitations of the study
80We do not use statistical modelling to link mortality trends with economic reforms because it would not be possible to harmonize the variables across jurisdictions. The analyses presented in this article were conducted to detect population-level effects, which cannot establish causal relationships that apply at the individual level. However, some of the similarities in terms of mortality changes between Shanghai and entire countries of central and eastern Europe do provide evidence of the impact of economic transition on mortality in Shanghai.
81Of course, Shanghai is not a country, and cannot represent the ordinary Chinese cities. Adverse mortality effects due to the reforms might be stronger in poor cities, such as those of north-east China, than in affluent Shanghai.
82As described in the section on Shanghai mortality data, adult mortality statistics for Shanghai are quite exhaustive and only a small proportion of deaths are coded as ill-defined. Individual mortality data became available from 1991. However, the procedure of collecting and coding data in Shanghai before and after 1990 was similar. The coding rules in Shanghai changed from the Chinese classification of diseases in 1988 to ICD-9 in 1989 and to ICD-10 in 2002. Thus, the coding rules did not change during the reform period 1992-1996.
83There may be coding differences between Shanghai and the countries selected for comparison. However, while such differences may cause comparability problems where more specific categories are used, these problems are relatively minor where mortality rates are estimated by broad groups of causes (Nolte et al., 2000b ; Shkolnikov et al., 2013).
84Our analyses were restricted to deaths of permanent residents (Shanghai hukou) and excluded floating populations (migrants). While floating populations may be initially healthier than permanent residents due to selective migration, migrants do not enjoy the full benefits of local citizenship (e.g. certain reserved occupations for local residents, medical care) (Liang and Ma, 2004). There is some evidence that migrants in Shanghai may have higher risk of dying from injury than permanent residents, and injury is one of the leading causes of death for the working-age population (Zhao et al., 2012). Migrants are disproportionally exposed to hazardous environments, are more likely to engage in high-risk occupations, and lack safety awareness and equipment. Moreover, migrants change diets and lifestyles after they move to cities, and the unfamiliar environment increases psychological stress, raising the risk of cardiovascular disease (Zhao et al., 2014). The mortality rates of working-age populations in Shanghai might therefore be higher if floating populations were taken into account. In the analyses presented here, we focus on the relatively homogenous population of Shanghai permanent residents and we avoid confounding reform-associated death trends with the effects of migration.
85We have examined and verified three theoretical propositions. Analysis of mortality in Shanghai suggests that economic transformation from centrally planned socialism to state capitalism initially increased the risk of death, especially among working-age men during the period 1992–1996 (Proposition 1). However, the negative impacts were offset by gradualist reform strategies and strong institutions combined with rapid economic growth. Orderly economic reform proceeded in a stable social and political environment and was associated with a moderation of negative effects and a net reduction in mortality (Proposition 2). Appropriate social policies were implemented and the unfavourable effects on mortality were transient (Proposition 3).
86Social and institutional changes and their association with health and mortality crises have been documented not only in transitions from socialism to capitalism but also in changes under extraordinary historical conditions (Cornia and Paniccià, 2000). In the USA, for example, the abolition of slavery is a well-documented case of a major social and institutional change which caused a worsening of health conditions among the African-American population (Meeker, 1976). Such changes can create a sudden shock for people before they adapt to the new social environment, which results in increased psychological stress and negative health consequences. However, Shanghai’s experience suggests that these negative health consequences can be managed, mediated, and even reversed by providing gradual transition, strong institutions, socio-political stability, positive economic effects, and proper social policies.
87However, a widening gap in mortality by social status may emerge during economic reform (Murphy et al., 2006). Other research has focused on changes in mortality differentials by educational level, employment status, and marital status.
88It would also be interesting to examine the effect of alcohol on mortality in areas of China with high alcohol consumption, especially northeastern provinces with experience of major state-owned enterprise reform and a high prevalence of hazardous drinking during the transformation period. Alcohol consumption was clearly a major proximate factor of the mortality crisis in Russia during the reform.
89Sociopolitical reform has a long march ahead of it in China, and there will certainly be other episodes of major change in the future. But the lessons learned from the Shanghai experience may be useful for other socialist countries wishing to shift to a market economy.
Australian Demographic and Social Research Institute, and National Centre for Epidemiology and Population Health, Australian National University, Australia.
Correspondence : Jiaying Zhao, Australian Demographic and Social Research Institute, Australian National University, Canberra 2601, Australia, email : Jiaying.email@example.com
Division of Social Science, The Hong Kong University of Science and Technology, Hong Kong.
Shanghai Municipal Center for Disease Control and Prevention, Peoples’ Republic of China.
National Centre for Epidemiology and Population Health, Australian National University, Australia.
The territories of the eastern European countries analysed in this article correspond to their current borders.