E-commerce has significantly increased food consumption in rural China through two channels. First, it reduces the cost of living, increasing disposable income that is partly spent on food. Second, it expands choices of food items available, especially non-perishables. Data shows rural household food expenditure grew more than other items with greater e-commerce. However, the biggest impact was on poor households and food for young children, as online access has reduced breastfeeding and increased formula purchases among the poor. While e-commerce has boosted rural consumption, the nutritional effects on children in poor areas requires further study.
Does e-commerce Increase Food Consumption in Rural Areas? Evidence from China
1. Does e-commerce Increase Food Consumption in
Rural Areas? Evidence from China
Xubei Luo, World Bank
Yue Wang, Cornell University
Xiaobo Zhang, IFPRI and Peking University
Chatrium Hotel, Yangon, Myanmar
October 30-31, 2019 1
2. China has become the largest e-commerce
market in the world
• Annual total e-commerce trade volume in
China increased thirtyfold from RMB 0.93
trillion in 2004 to RMB 300 trillion ($310 per
capita) in 2018.
• China’s worldwide e-commerce transaction
value grew from less than one percent a
decade ago to over 50 percent in 2018.
2
3. What is the implication of e-commerce development on consumption,
in particular food consumption growth in rural China?
3
4. Two potential channels
• E-commerce reduces cost of living, thereby increasing disposable
income. Some of the additional income may be spent for food
consumption, but the proportion falls according to Engle’s Law. In
other words, the impact of e-commerce on food consumption growth
should be smaller than many other types of expenditures.
• E-commerce expands the choices of food items, in particular non-
perishable food. When people consume more of the newly available
food items, total food expenditure may go up.
4
5. Data
• First, we construct a series of indicators to measure the development
of e-commerce at the county level using online sales and purchase
information provided by the Alibaba Group, gross domestic product
(GDP) from the China Statistical Yearbook, and population drawn from
the Population Census 2010.
• Second, we merge the county-level e-commerce development
measures with household consumption data obtained from the
nationally representative China Family Panel Studies (CFPS) survey
administered by Peking University.
5
6. Three features of this paper
• By matching the two different datasets, we can directly examine the
impact of e-commerce development on consumption growth at the
household level rather than at the aggregate city level as in Fan et al.
(2016).
• Our finding is more representative because CFPS covers many more
counties than the sample used in Couture et al. (2017).
• Using the rich consumption information of the CFPS survey, we can
study the heterogenous associations between e-commerce
development and various categories of consumption, including food
consumption, which were not discussed in Fan et al. (2016) and
Couture et al. (2017).
6
7. Indicator constructions: Focusing on the purchase side
• E-commerce penetration indexes
• Share of online buyers in population (%):
number of online buyers
residential population 2013
× 100
• Share of online purchase in GDP (%):
annual online purchase GMV
GDP 2013
× 100
• E-commerce intensity indexes
• Per buyer online purchase (yuan):
𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝐺𝐺𝐺𝐺𝐺𝐺
𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛 𝑜𝑜𝑜𝑜 𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜 𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 2013
• Per capita online purchase (yuan):
𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝐺𝐺𝐺𝐺𝐺𝐺
𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 2013
• Market size indexes
• National share of online purchase (%):
𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝐺𝐺𝐺𝐺𝐺𝐺
𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛 𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝐺𝐺𝐺𝐺𝐺𝐺 2013
× 100
• National share of online buyers (%):
𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛 𝑜𝑜𝑜𝑜 𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜 𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏
𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛 𝑜𝑜𝑜𝑜 𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛 𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜 𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 2013
× 100
GMV: Gross Merchandise Value
7
8. Big challenge to merge the two datasets
• They are saved in different servers and not allowed to share the raw
data.
• Final solution: Normalize the ecommerce index into a z-score
z − score of measure x =
xct − mean of x
standard error of x
8
11. Growth rate of household consumption per capita
and e-commerce intensity level in China
11
12. Model specification
Yic,t
s
= 𝛽𝛽0 + 𝛽𝛽1 𝑧𝑧𝑧𝑧𝑧𝑧𝑧𝑧𝑧𝑧𝑧𝑧𝑧𝑧 𝑝𝑝𝑐𝑐,2013, +𝛽𝛽2 𝑙𝑙𝑙𝑙𝑙𝑙𝐸𝐸𝐸𝐸𝐸𝐸𝑖𝑖𝑖𝑖,𝑡𝑡−2
𝑠𝑠
+ 𝛽𝛽3 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑜𝑜𝑖𝑖𝑖𝑖,𝑡𝑡 +
𝛽𝛽4 𝑎𝑎𝑎𝑎𝑒𝑒𝑖𝑖𝑖𝑖,𝑡𝑡/10 +
𝛽𝛽5 𝑎𝑎𝑎𝑎𝑒𝑒𝑖𝑖𝑖𝑖,𝑡𝑡
2
100
+ 𝛽𝛽6 𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑛𝑛𝑖𝑖𝑖𝑖,𝑡𝑡 + 𝝈𝝈𝒕𝒕 + 𝛅𝛅𝐫𝐫 + 𝜖𝜖𝑖𝑖𝑖𝑖,𝑡𝑡
where
Yic,t
s
= 𝑙𝑙𝑙𝑙𝑙𝑙𝐸𝐸𝐸𝐸𝐸𝐸𝑖𝑖𝑖𝑖,𝑡𝑡
𝑠𝑠
− 𝑙𝑙𝑙𝑙𝑙𝑙𝐸𝐸𝐸𝐸𝐸𝐸𝑖𝑖𝑖𝑖,𝑡𝑡−2
𝑠𝑠
is the growth of log consumption per capita of
category s in household i located in county c between two waves of CFPS survey in
2014 and 2016.
𝑧𝑧𝑧𝑧𝑧𝑧𝑧𝑧𝑧𝑧𝑧𝑧𝑧𝑧 𝑝𝑝𝑐𝑐,2013 is the z-score of the e-commerce development indicators.
𝛅𝛅𝐫𝐫: the region fixed effect
12
13. E-commerce development vs. growth in total household expenditure per capita
(1) (2) (3) (4) (5) (6)
All Urban Rural East Central West
Z-score of online purchase 0.304*** 0.240*** 0.771*** 0.297*** 0.312*** 0.401***
amount/population
Lagged log household -0.605*** -0.564*** -0.664*** -0.599*** -0.578*** -0.652***
expenditure per capita
Average age of -0.058*** -0.041*** -0.102*** -0.070*** -0.062*** -0.046**
household members/10
Square of average age 0.001*** 0.000*** 0.002*** 0.001*** 0.001*** 0.000**
of household members/100
Household dependent ratio -0.174*** -0.090** -0.285*** -0.093** -0.210*** -0.230***
Urban/rural area dummy 0.270*** 0.245*** 0.212*** 0.378***
(Urban=1)
Regional dummies -0.018 0.029 -0.033
West
Central 0.064 0.036 0.105**
Observations 21,001 9,343 11,658 8,370 6,580 6,051
R-squared 0.301 0.288 0.328 0.301 0.288 0.324
Year fixed effect yes yes yes yes yes yes
Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1
Rural>urban
Central & West>East
13
14. Commonly consumed goods and service: by region
(1) (2) (3) (4) (5) (6)
All Urban Rural East Central West
Cosmetics and beauty 0.359*** 0.304*** 0.874*** 0.414*** 0.248** 1.007**
Food 0.390*** 0.314*** 0.865*** 0.376*** 0.391*** 0.577**
Of which: Food at home 0.389*** 0.326*** 0.801*** 0.360*** 0.438*** 0.438
Of which: Dining out 0.702*** 0.624*** 1.624*** 0.697*** 0.585*** 1.858**
Clothes 0.274*** 0.240*** 0.511*** 0.231** 0.340*** 0.356
Utilities 0.160*** 0.141*** 0.355*** 0.156*** 0.161*** 0.265
Communications 0.270*** 0.232*** 0.553*** 0.249*** 0.293*** 0.371**
Local transport 0.399*** 0.335*** 0.755*** 0.304** 0.462*** 0.724***
The impact on food consumption is greater than other items except for local transportation14
15. Commonly consumed goods and service: by region
(1) (2) (3) (4) (5) (6)
All Urban Rural East Central West
Cosmetics and beauty 0.359*** 0.304*** 0.874*** 0.414*** 0.248** 1.007**
Food 0.390*** 0.314*** 0.865*** 0.376*** 0.391*** 0.577**
Of which: Food at home 0.389*** 0.326*** 0.801*** 0.360*** 0.438*** 0.438
Of which: Dining out 0.702*** 0.624*** 1.624*** 0.697*** 0.585*** 1.858**
Clothes 0.274*** 0.240*** 0.511*** 0.231** 0.340*** 0.356
Utilities 0.160*** 0.141*** 0.355*** 0.156*** 0.161*** 0.265
Communications 0.270*** 0.232*** 0.553*** 0.249*** 0.293*** 0.371**
Local transport 0.399*** 0.335*** 0.755*** 0.304** 0.462*** 0.724***
Rural>Urban
15
16. Food Puzzle
Given Engle’s Law, only a small proportion of the
incremental disposable income thanks to cheaper price
of online items should be spent on food, then why is
there a bigger impact on food consumption growth
than other types of consumption growth?
16
17. Food Puzzle: The impact is stronger for the poor
(1) (2) (3)
Bottom
25%
25%-
75%
Top
25%
Food 0.550*** 0.323*** 0.150***
Of which: Food
at home 0.496*** 0.335*** 0.204***
Of which:
Dining out 1.507*** 0.272 0.353***
17
18. Guessed answer: Baby food and formula
• Baby food and formula are the top two package food sold online in
China.
• Using the CFPS data, we find that breast feeding is negatively
correlated with e-commerce development; the correlation between
food consumption and e-commerce z-score is stronger for households
with young children under three years old.
• It seems that the poor prefer baby formula to breast feeding.
18
19. Conclusion
• In China, e-commerce has boosted food consumption, in particular in
rural areas and among the poorer households.
• E-commerce has reduced regional consumption inequality.
• Question: What is the nutritional impact on children in the poor areas
as people buy more baby food and formula at the expense of breast
feeding?
19
20. The presentation is based on
20
Luo, Xubei, Yue Wang, and Xiaobo Zhang, “E-Commerce Development
and Household Consumption Growth in China,” World Bank Policy
Research Working Paper No. 8810, 2019.
Thank you!
Xiaobo Zhang, x.zhang@cigar.org