Content uploaded by Shijun Ding
Author content
All content in this area was uploaded by Shijun Ding on Feb 22, 2015
Content may be subject to copyright.
8
Seasonal dimensions of
household wellbeing and labour
migration in rural southern China
Shijun Ding, Haitao Wu, Yuping Chen
Introduction
Seasonal patterns of farm households’ economic activities and wellbeing have
been studied since the 1970s, with most research focusing on tropical agriculture
in countries from Africa and South Asia (Chambers et al., 1981; Devereux, 2008,
2009). It has been generally recognized over the last four decades that seasonal vari-
ations in household agricultural activities and other types of economic activities
explain, to a large extent, rural poverty in low-income countries in general (Harris
and Todaro, 1970; Paxson, 1993; Stark and Fan, 1993; De Haan, 1999; Stark, 2007)
and in tropical countries in particular, and that there are several negative factors
that make the lives of poor households worse during the pre-harvest months every
year. Furthermore, the backgrounds or contexts of agricultural development in
most developing countries have been changing rapidly over recent decades, and
these changes need to be taken into account when designing policy interventions
to reduce the negative eects of seasonality in developing countries’ agricultural
and rural development.
Seasonal variation in agricultural and rural economy is a fundamental phenom-
enon characterizing rural development in China (Zhao, 1999). The country is
located in a wide range of geographic locations, with its southern part in tropical
and subtropical regions and the northern part in a temperate region. While patterns
of seasonal variation of agricultural and rural economic activities in southern
regions may follow what had been studied in African and south Asian countries,
patterns of seasonal variation may have their own features in northern regions.
The seasonality of households’ production and wellbeing in rural China can be
investigated in terms of patterns of income generation from a variety of sources:
sales of agricultural products, wages, family-based non-farm enterprise, remittances
and transfer payments. While households’ income from eld crops will be mark-
edly seasonal, production and sale of cash crops may balance the seasonality of
134 Seasonal dimensions of household wellbeing
agriculture-related activities. In the same way, casual non-farm work may balance
the seasonality of a household’s productive activities. Small farm households in
China have developed dierent combinations of livestock/grain production and
farm/non-farm activities in reducing the negative eects of seasonality. However,
seasonal dimensions in agricultural and rural development in China have been less
empirically investigated, and the strategies employed by small farm households to
cope with seasonal variations are currently poorly understood.
Socioeconomic consequences of seasonal patterns of household income,
expenditure and labour mobility need to be understood in order to design appro-
priate interventions to smooth seasonality and improve livelihoods. For example,
policies encouraging seasonal migration for casual work during the slack season
will certainly help households improve their economic wellbeing; timely arrange-
ments for production loans during the planting season will help households in
purchasing inputs (seeds, fertilizers, etc.) for crop production. This paper aims to
describe patterns of seasonal variation in household income and expenditure, to
document patterns of seasonal migration from rural areas to urban cities in search
of casual labour work, and to investigate the eects of seasonal migration on the
changing patterns of households’ income and expenditure.
Data for this paper come from the Rural Household Survey (RHS) of the State
Bureau of Statistics (SBS) and a household interview conducted by the authors.
The RHS is administered directly by SBS through its provincial and county survey
network, and this ensures that the data collected are free from local interference.
Rather than employing a single-interview approach, each selected household
maintains a daily diary over the entire year. An assistant interviewer is supposed to
visit each household every two weeks to check the diary and assist the household
in completing it and then transfer the information to the county level. A dataset
covering 3,300 households in 33 randomly selected counties (from a total of 75
counties) during 2004 to 2007 in Hubei, containing variables on household demo-
graphics, monthly migration, monthly income and expenditure and others, is used
for this paper. A case study from Guangxi on seasonal variations of household cash
income and expenditure is also provided. County-representative rainfall data from
ten selected counties come from the Meteorological Bureau in Hubei, containing
monthly rainfall data from 1982 to 2001.
The paper is organized into the following sections. Firstly we describe the
seasonal variations of household income and expenditure; this will be done by
investigating household monthly income and expenditure variations over the year.
In the following section we document the patterns of monthly migration, to look
at the correlation of migration and monthly agricultural activities; this is done by
mapping seasonal migration between peak and slack seasons. We then investigate
the eects of monthly migration on the changing patterns of household income
and expenditure; this is done by comparing monthly income and expenditure
variations between households with and without migrant labour, and modelling
how household income and monthly expenditure are aected by seasonal migra-
tion. Moreover, a unique farm household case, which we interviewed during a
Shijun Ding, Haitao Wu, Yuping Chen 135
previous study, is analysed in the next section to reinforce the conclusions and get
in-depth understanding of the seasonal variation of household wellbeing in rural
China. Finally ndings are presented and implications for designing interventions
to protect farm households’ livelihoods are drawn.
Seasonal variations of household income and expenditure
Rainfall and its seasonal patterns
In subtropical regions, where southern China is located, seasonality seems more
intuitive since summer and winter dier distinctly from each other, with most crops
harvested in late autumn and a slack season in winter and early spring. To investigate
seasonal patterns in agriculture, an attempt is made to dene dry and wet seasons
in the study province in an absolute sense based on monthly rainfall.1 The average
rainfall in each month from ten randomly selected counties (of 75 counties) in
Hubei for the period 1982 to 2001 is shown in Figure 8.1. Rainfall shows a trend
increasing from January and reaching its peak in July, and then a trend decreasing
until December. An arbitrary approach is used to classify the whole year into dry
and wet seasons, in which 100mm of rainfall per month is taken as a threshold.
The wet season can be seen as being from April to September while the remaining
months are seen as the dry season.
A wide range of crops are grown in the area, with the main crops being rice,
maize and cotton in summer and wheat in winter. Rainfall is an important deter-
minant of crop production. Figure 8.1 also depicts a generalized version of crop-
ping calendars in the area. Rice, maize and cotton are generally planted in April and
harvested in September and October. Wheat is sown in October and harvested in
the following May. It can be concluded that the main agricultural activities happen
in the wet season (peak season), and the slack season coincides with the dry season.
Source: pers. comm., Hubei Meteorological Bureau, 2007
FIGURE 8.1 Rainfall and cropping calendar in Hubei
136 Seasonal dimensions of household wellbeing
Seasonal variations of household income and expenditure
Income and expenditure are used to denote household wellbeing in this paper.
Household income consists of wage income, family-based enterprise income and
other sources, while household expenditure consists of family-based enterprise
spending and consumption expenditure.
The seasonal variation of monthly household income, expenditure and balance in
2006 and 2007 shows a similar trend, as can be seen in Figure 8.2.2 Monthly income
is lower than monthly expenditure for most of the year and the only exception occurs
in December. This can be explained by the fact that agricultural and related produc-
tive expenses mainly occurred during spring, summer and autumn when households
invested most of the productive inputs, and that household income is mainly from
selling summer crops at the end of the year and migrants’ wage income which is
mostly received at the end of the year. On the other hand, household income and
expenditure in the peak season are generally lower than that in the slack season.
To investigate the variations of monthly income and expenditure, income
sources and expenditure items are separated out in Figures 8.3 and 8.4, respectively.
The varied shares of dierent income sources and expenditure items show seasonal
patterns of variation over the year:
1 ages are the main income source from January to September, while family-
based agricultural enterprise becomes the main income source from September
as households would sell their agricultural products in late autumn;
2 The income share of family-based enterprise in the peak season is generally
lower than that in the slack season;
3 Consumption is one of the main expenditure items over the year, and is higher
in lunar December when the traditional Chinese New Year comes around, this
is especially true for food consumption;
4 The expenditure share of family-based enterprise is much higher in the peak
season, indicating the higher agricultural inputs use.
Source: pers. comm., Hubei Statistics Bureau, 2007
FIGURE 8.2 Seasonal variation of income, expenditure and balance, 2006–2007
Shijun Ding, Haitao Wu, Yuping Chen 137
Source: pers.comm., Hubei Statistics Bureau, 2007
FIGURE 8.3 Seasonal variation of income shares from dierent sources
Source: pers.comm., Hubei Statistics Bureau, 2007
FIGURE 8.4 Seasonal variation of expenditure shares for dierent items
Quantifying seasonal variations of household income and
expenditure
To quantify seasonal variations of household income and consumption, a season-
ality index (Walsh, 1980) is used in this paper, and it can be expressed as follows:
138 Seasonal dimensions of household wellbeing
R
R
x
SI
n
n
n
∑
−
12
1
|
12
|
Where SI denotes seasonality index, χn means income/consumption of month n,
and R
– is the mean annual income/consumption. The seasonality index takes into
consideration all months of the year.
Seasonality indices of household income and expenditure from 2004 to 2007 are
calculated and listed in Table 8.1. Firstly, overall seasonality indices of monthly total
income and total expenditure (except for 2006) show a decreasing trend over the
four years, implying monthly income and expenditure are generally becoming more
stable. Secondly, seasonal variations in total income are generally higher than in total
expenditure, and this is because expenditure usually reects households’ long-term
wellbeing, and households are more able to smooth their expenditure than income,
as evidenced elsewhere (for example, Morduch, 1995). Thirdly, looking at dierent
categories of income and expenditure, the table shows dierent patterns. Seasonal
variations for wage income, in the decreasing trends, are relatively higher, ranging
from 1.346 to 1.281, while seasonal variations for consumption expenditure in
the decreasing trends, are relatively lower, ranging from 0.928 to 0.816. Fourthly,
seasonal variations of income from family-based enterprise are greater than those
from wage income, and seasonal variations of expenditure on family-based enter-
prise are greater than on family consumption. This may be because agriculture as
the main kind of family-based enterprise in rural China requires inputs in the peak
season (a few months in March, April and May, for example) but income from
it comes mainly after harvest later in the slack season (for example, in October,
November and December). Seasonal variations of income from, and expenditure
on, family-based enterprise over the four years show increasing trends, and this may
need further explanation.
TABLE 8.1 Seasonality index for income and expenditure, 2004–2007
Year Income Expenditure
Total Wage Family-based
enterprise
Total Family-based
enterprise
Consumption
2007 1.227 1.281 1.682 0.702 0.975 0.816
2006 1.248 1.285 1.667 0.725 0.945 0.827
2005 1.294 1.305 1.658 0.722 0.926 0.842
2004 1.298 1.346 1.650 0.774 0.907 0.928
Source: pers. comm., Hubei Statistics Bureau, 2007
Shijun Ding, Haitao Wu, Yuping Chen 139
Effects of migration on changing patterns of household
income and expenditure
Seasonal patterns of migration and households’ monthly income
distribution
Migration from rural areas to urban cities to earn a living has become an important
phenomenon in China over the last three decades, with a considerable amount of
migration being seasonal. The monthly distributions of the percentage of house-
holds with members migrating during 2004 to 2007 are shown in Figure 8.5. There
are increasing numbers of households with members migrating over the study time
period (from 75 per cent of households in 2004 to 90 per cent in 2007). Patterns
of migration show seasonal variation, with fewer going out in the peak season and
more in the slack season. Most migration occurs in February when the traditional
Spring Festival is just over.
To further investigate the eects of seasonal migration on household income
and expenditure, monthly income of dierent types of household are compared in
Figure 8.6: households without migration, with 1–3 months of migration, with 3–6
months of migration and with 6+ months of migration. The log value of income is
adopted for intuitive expression. Households with longer migration times generally
have higher incomes. When considering separate months, it follows from what we
found earlier that household income is lower in the peak season and higher in the
slack season, with December having the highest monthly income. This monthly
pattern of income generation does not seem to t intuitively with the migration
story above. This can be explained as follows. Monthly wage income as remittances
from migrants is sometimes paid by employers at the end of the year (this is espe-
cially true for construction migrant workers) and taken back home by migrants
themselves when they are back home taking the Chinese New Year (the spring
festival) holiday in November and December. Farm households mostly sell their
Source: pers. comm., Hubei Statistics Bureau, 2007
FIGURE 8.5 Percentage distribution of number of households with migration by month
140 Seasonal dimensions of household wellbeing
agricultural products at the end of the year, and traditionally it is common that
households sell their livestock (mostly pigs) shortly before the Chinese New Year
so that they have cash income for the coming festival celebration. As a result, house-
hold income can dramatically increase in December.
Determinants of seasonal variations of income and
expenditure: An ordinary least squares (OLS) analysis
The ordinary least square regression on determinants of – more specically, eects
of migration on – seasonal variations of household income and consumption
expenditure is applied in this section. Considering other factors aecting variations
of income and expenditure, variables such as farm size, working time per labourer
and percentage of household labourers who have received technical training are
included in the analysis. The determinant equation can be expressed as follows:
Y = a0 + a1 farmsize + a2 worktime + a3 training + a4 migration.dummy + u
Where Y represents the log value of household income (consumption expendi-
ture) or seasonality index; farmsize, worktime and training represent farm size (hectares
of cultivated land), working time per labourer (number of months) and percentage
of household labourers receiving technical training, respectively, while migration.
dummy represents numbers of migration labour dummy variables (0 if no migrant,
1 if 1 migrant, 2 if more than 1 migrant); a0 is a constant item, a1 to a4 are the
coecients of corresponding independent variables, and u is the disturbance term.
Using data on 2007, the OLS estimations of four equations are listed in Table 8.2
for income and Table 8.3 for consumption expenditure.
Source: pers. comm., Hubei Statistics Bureau, 2007
FIGURE 8.6 Distribution of monthly income of households with dierent migration
periods
Shijun Ding, Haitao Wu, Yuping Chen 141
The results show that migration has a positive eect on household income
generation, with the eect being statistically signicant (at 1 per cent level) when
the household has more than one migrant labourer. Other variables, including farm
size and percentage of household labourers having received technical training, also
have signicant positive eects on income. Moreover, migration has a signicant
negative eect on the seasonality index of income (at 1 per cent level), meaning
that migration helps reduce seasonal variation of household income.
Looking at the eect of migration on household expenditure, although it is not
statistically signicant, the eect shows a positive sign. Other explanatory variables,
TABLE 8.2 Determinants of income and its seasonality index
Variables Income Seasonality index of income
Coecient T value Coecient T value
Farm size 0.042*** 22.11 0.032*** 20.80
Migration.dummy=1 0.029 1.32 –0.085*** –5.98
Migration.dummy=2 0.113*** 4.88 –0.141*** –9.76
Worktime 0.023*** 3.45 –0.018*** –3.41
Training 0.229*** 8.34 0.016 0.86
Constant 9.251*** 148.82 1.208*** 29.58
Observations 3300 3300
R-squared 0.1566 0.1532
Source: pers. comm., Hubei Statistics Bureau, 2007
Note: *** Statistically signicant at 1% level.
TABLE 8.3 Determinants of consumption expenditure and its seasonality index
Variables Consumption expenditure Seasonality index of
consumption expenditure
Coecient T value Coecient T value
Farm size 0.018*** 9.03 0.005*** 4.46
Migration.dummy=1 0.032 1.38 0.010 1.01
Migration.dummy=2 0.018 0.72 0.042*** 3.76
Worktime 0.025*** 3.59 –0.007* –1.93
Training 0.220*** 7.58 –0.049*** –3.72
Constant 8.809*** 134.56 0.834*** 29.10
Observations 3300 3300
R-squared 0.0582 0.0516
Note: *** Statistically signicant at 1% level. * Statistically signicant at 10% level. Source: pers. comm.,
Hubei Statistics Bureau, 2007
142 Seasonal dimensions of household wellbeing
including farm size, work time per labourer and technical training received, all
have statistically signicant positive eects on household expenditure (at 1 per cent
level). Regarding the eect on the seasonality index of expenditure, while work
time per labourer and technical training received have negative eects, farm size
and having two or more migrants have statistically signicant positive eects (at 1
per cent level), meaning farm size and migration contribute to the seasonal varia-
tion of household consumption expenditure. This may be explained by the fact that
seasonal migrant labourers actually have higher consumption expenditure when
they are away than when they stay at home, as explained earlier.
It is worth noting that higher seasonal variability of household consumption
expenditure cannot be interpreted as problematic. Migrant workers have been seen
as an important engine for China’s economic boom over the last three decades. More
than three quarters of rural households in the sample have at least one member who
has migrated to cities for non-farm work. The migrants have far higher productivity
than they would have had if they had stayed at home involved in agriculture, while
they also have higher consumption expenditure away from home, which contrib-
utes to the higher variability of household consumption expenditure.
Seasonal variations of household income and expenditure:
A case study
A farm household in Guangxi was interviewed in 2002. The household head
voluntarily recorded a detailed cash income and expenditure diary from January
2001 until when we visited in March 2002. Figures 8.7 and 8.8 show dierent
items of cash income and expenditure.
The household depends almost entirely on agriculture for its livelihood. Cash
inows include selling various types of agricultural products such as grains, vegeta-
bles, animals, and so on. The farmer grows single-season rice but does not sell it at
once. Instead, sales are rather scattered across the year, with the largest sales occur-
ring during April to June in the pre-harvest season when the household may be
short of cash income, and when there is also intensive labour input. The constant
cash inow over the year comes from selling vegetables. As for cash expenditure,
constant cash outow occurs in items such as food and other daily necessities. The
sudden increase in expenditure in June is mainly because the farmer is paying to
install a biogas system for agricultural production and for household energy use by
his wife, who cooks for the biogas construction technicians every day.
The constant cash inow from vegetable sales may compensate for the expense
of daily necessities. However, the farmer pays a large sum of money in October for
baby pigs to rear, and this makes him indebted. In November a payment to one of
his relatives as a gift for a ceremony, makes him indebted again, and the situation
is even worse in the subsequent month, when he makes gifts to relatives for other
ceremonies (it is common in rural China for farm households to hold their celebra-
tion/ceremonies later in the year, in the slack season). The balance of monthly cash
inows and outows is shown in Figure 8.9.
Source: pers. comm. with a farm household in China, 2005
FIGURE 8.7 A household’s cash income (RMB Yuan) over 15 months, 2001–2002
Source: pers. comm. with a farm household in China, 2005
FIGURE 8.8 A household’s cash expenditure (RMB Yuan) over 15 months, 2001–2002
Source: pers. comm. with a farm household in China, 2005
FIGURE 8.9 A household’s cash ow over 15 months
144 Seasonal dimensions of household wellbeing
As can be seen, the uctuation of the farmer’s cash ow in income and expendi-
ture over the year shows no rule, with the peaks and lows not matching at all.
He faces a cash decit after October (he actually borrows in this month to pay
for baby pigs), and gets to be indebted by some 300 Yuan by the end of the year.
Furthermore, his indebted status continues at the same level for the next couple of
months. If there were no government support, no social network, and no mecha-
nism for borrowing, he would probably have to sell his productive assets for to repay
his debts. Alternatively, if he didn’t have enough assets to sell, or a family member
suddenly suered an emergency or accident, he would then not be able to sustain
his livelihood, which may collapse.
Concluding remarks
Risks and uncertainties have been seen as fundamental factors leading farm house-
holds in rural China into poverty and desperation. Seasonality in subtropical regions,
such as in southern China, can be an important source of such risks and uncertain-
ties, leading to great uctuations in household income and expenditure. This may
be seen from the fact that seasonal changes in agricultural production explained,
to a large extent, the variations in households’ cash income and expenditure. This
may be especially true for those households that mostly depend on agriculture for
their livelihoods.
In the study areas, wage income is the main income source for most of the year
while family-based agricultural enterprise becomes the main income source at the
end of the year. Consumption is one of the main expenditure items over the year,
and is higher in December and this is especially true for food consumption. The
expenditure share of family-based enterprise is much higher in the peak season,
indicating higher agricultural inputs use during that time. As income from agri-
cultural production forms a large proportion of overall household income, public
policies should be carefully designed to help households mitigate the adverse eects
of the seasonal variation in agricultural income. For example, because drought is
one of the major risks in southern China, developing an irrigation system may help
farm households with additional rice harvests (i.e., increasing rice production from
one season to two seasons), and providing drought-tolerant crop varieties may also
help to reduce the variability of agricultural income.
Migration from rural to urban areas for non-farm work has made a signicant
contribution to China’s economic boom over the last three decades. Casual labour
migration is seen as the main non-agricultural income-generating activity for rural
households, and helps a large proportion of rural households climb out of poverty.
On the other hand, seasonal migration has been one of the most important factors
both positively and negatively aecting the uctuation of household income and
expenditure.
Variability in household income and expenditure in the study areas may not
necessarily be problematic. As discussed earlier, although seasonal migration
contributes to variability in household expenditure, it generates higher income
Shijun Ding, Haitao Wu, Yuping Chen 145
and helps smooth household income variability, and consequently, house-
hold expenditure can be increased. That no statistically signicant variations
in seasonal food consumption patterns have been evidenced may indicate that
household food supply in the study area has been secured to some extent, and
there may be other variables, such as variation in household labour supply and
changes in household resource endowment, that explain seasonal variations of
household wellbeing. This nding is somewhat dierent from other studies in
African and south Asian regions which found that seasonal hunger is a serious
concern (Devereux et al., 2008).
One of the variables that very positively aects income and expenditure and
their variability is the technical training provided to the rural labour force. A
labourer who has received technical training (presumably for non-farm produc-
tive skills) on the one hand signicantly increases both income and expenditure
on consumption and on the other hand signicantly reduces their seasonality. This
may have important policy implications. For example, training as a public good
should be government-funded and provided more widely to the rural labour
force, to develop the labour market and generate a higher return to households
who heavily rely on the seasonal migrant labour force. In the same way, house-
hold assets play a very positive role in helping to generate income and smooth
expenditure so further policy intervention should encourage household assets
accumulation.
Seasonal variations in agriculture and their eects on household livelihoods
in rural China have been inadequately investigated, and little evidence has been
collected. Further eorts should be made to identify the nature and extent of
seasonal variations and their impact on households in rural China. Two arguments
can be made:
1 agricultural policies and rural development agendas need to engage more with
agricultural seasonality and look more at household-level evidence, which is
currently rarely considered in policy interventions;
2 more institutional or integrated interventions (e.g., nancial support for agri-
cultural production, subsidies) and technological improvements, rather than ad
hoc arrangements (i.e., social relief programmes), are needed to help smooth
seasonal variations of household income and expenditure in rural areas.
Acknowledgements
The research projects, on which this paper is based, received nancial support from
the National Science Foundation of China (Grant No. 70573122, 70773120). Mr
Shu Zhenbin from the State Bureau of Statistics Hubei Branch helped with data
analysis. We are grateful to Dr Stephen Devereux and colleagues at the Institute
of Development Studies, UK, for comments in revising the chapter and help in
editing.
146 Seasonal dimensions of household wellbeing
Notes
1 More sophisticated studies may use ‘temperature’ as well. To take our study as a starting
point, only rainfall is used. Further investigation of seasonality in a subtropical region
may need to take temperature into account.
2 The Chinese lunar calendar is used in the paper, which is based on the cycles of the
moon. In the Chinese lunar calendar the beginning of the year falls somewhere between
late January and early February. Chinese farmers mainly base their agricultural activities
on the lunar calendar, and it is also used for festive occasions such as the Chinese New
Year.
