****Accessing the data (the Chinese Household Finance Survey (CHFS))

Interested researchers can register with the Survey and Research Centre for China Household Finance at Southwestern University of Finance and Economics, which manages 
the data. You can register by visiting the website: (https://chfs.swufe.edu.cn/datas/), and directions for registering and accessing the data are provided under the section "Data centre". The contact information 
for the data management team is available on the website under the section "Contact us". 

Original survey data files used for this paper is from CHFS Wave2013: 
-chfs2013_ind_20180119.dta
-chfs2013_hh_20180119.dta
-chfs2013_master_20180119.dta


****The supplemental .7z archive for this publication contains the following two files. Stata files were created using Stata Version 15 and may not be compatible with earlier versions.

1. Stata data file
      Working file.dta 

2. Stata program file to produce data codebook and results of Table 1-5 and Figure 1-2 in this paper 
      dofile.do


****Explanation of the "Working file.dta"

-contain information of 6520 respondents aged 40 to 55, who had at least one surviving parent and one child when the survey was conducted in 2013. 
-derived from chfs2013_ind_20180119.dta, chfs2013_hh_20180119.dta, and chfs2013_master_20180119.dta.
-derived variables includes: 
--outcome variables
----attitudes of old-age care (h3046)
----preferences of old-age care (h3047)
--demographic variables 
----age (ageg)
----gender (a2003)
----number of children (childg)
----whether have at least one son (son) 
--socioeconomic variables 
----education (edu) 
----occupation (occu1) 
----hukou status (hukou1) 
----household assets (continues variable) (Fasset)
----household assets (categorical variable) (Fassetg)
--intergenerational support variable 
----whether co-residence with parents or live nearby or live apart (cores)
--identifier and sampling weight
----unique respondent id (hhid2013) 
----sampling weight (swgt)


****Explanation of the content of the "dofile.do".

1. Data codebook
-the command of " use "C:\Local\Working file.dta" " is to get the analytic data file. Please change the path to a folder where you save this data file. 
-the command of "codebook" offers information about all variables in the data file "Working file.dta", including variable name, label, type, value range, value label, missing data, frequency of each category (categorical variables) or percentiles (continues variables).   
--outcome variables (attitudes of old-age care and preferences of old-age care)
--demographic variables (age, gender, number of children, whether have at least one son) 
--socioeconomic variables (education, occupation, hukou status, household assets)
--intergenerational support variable (whether co-residence with parents or live nearby or live apart)
--sampling weights
--unique id 

2. Profile of analytical sample
-"tab....":unweighted distribution of all respondents' attitudes and preferences of old-age care, demographic and socioeconomic characteristics and intergenerational support. 
-"gen....":generate an integral sampling weight using the original sampling weight
-"tab....[fweight=swgt1]":weighted distribution of all respondents' attitudes and preferences of old-age care, demographic and socioeconomic characteristics and intergenerational support. 

3. Attitudes towards who should be responsible for old-age support
-"tab....,row chi2":cross-tabulations and chi square tests between attitudes of old-age care and demographic and socioeconomic characteristics and intergenerational support, among all respondents.

4. Factors associated with attitudes towards who should be responsible for old-age support 
-"mlogit....":multinomial logistic regression, among all respondents
--outcome variable is attitudes of old-age care. Reference category is "mainly the children".
--predict variables include demographic and socioeconomic characteristics and intergenerational support.  
-"margins....":predictive margins of attitudes of old-age care following the multinomial logistic regression model, among all respondents. 

5. Preferences concerning future living arrangements in later life
-"tab....,row chi2":cross-tabulations and chi square tests between preferences of old-age care and demographic and socioeconomic characteristics and intergenerational support, among all respondents.

6. Factors associated with future living arrangements in later life
-"mlogit....":multinomial logistic regression, among all respondents, among men and among women separately.
--outcome variable is preferences of old-age care. Reference category is "living with children".
--predict variables include demographic and socioeconomic characteristics and intergenerational support. 
-"margins....":predictive margins of preferences of old-age care following the multinomial logistic regression model, among all respondents, among men and among women separately.
-"keep if a2003==1":select men as the analytic sample
-"keep if a2003==2":select women as the analytic sample

7. Average marginal effects of expressing different attitudes towards old-age care by age cohort and Hukou status 
-"mlogit....":multinomial logistic regression with interaction terms, among all respondents 
--outcome variable is attitudes of old-age care. Reference category is "mainly the children".
--predict variables include demographic and socioeconomic characteristics and intergenerational support. 
--interaction term is between age group and hukou status.
-"margins....":predictive margins of the outcome for a particular age cohort and hukou status following the multinomial logistic regression model, among all respondents
-"marginsplot....":plot the predictive margins of each category of the outcome at a particular age cohort and hukou status, among all respondents
-"graph combine....":combine the plot graphs above, among all respondents

8. Average marginal effects of expressing preferred future living arrangements in later life by age cohort and number of children 
-"keep if a2003==2":select women as the analytic sample
-"mlogit....":multinomial logistic regression, among women  
--outcome variable is preferences of old-age care. Reference category is "living with children".
--predict variables include demographic and socioeconomic characteristics and intergenerational support. 
--interaction term is between age group and number of children.
-"margins....":predictive margins of the outcome for a particular age cohort and number of children, among women
-"marginsplot....":plot the predictive margins of each category of the outcome at a particular age cohort and number of children, among women
-"graph combine....":combine the plot graphs above, among women