Data: The Understanding Society data used in this paper are available freely to academics without a special licence through the UK Data Service (https://ukdataservice.ac.uk/).  Full, searchable documentation for the Main Survey is available at https://www.understandingsociety.ac.uk/documentation/mainstage. 

Records needed: w_indresp.dta w_hhresp.dta (all waves w) & xwavedat.dta
Variables used:
pidp = cross-wave personal id number
hidp = wave-specific household id number 
hiqual_dv = highest qualification obtained
nnatch = number of own natural children
nch02_dv = number of children aged 0-2 in household
agechy = age of youngest child in household
age_dv = current age
istrtdaty = year of interview
gor_dv = standard government office region

You must set up the Understanding Society data in a panel (multi-level / 'long') format with personal identifier 'pidp' at the top level and 'wave' (within 'pidp') at the bottom level.

The variables above used to construct the birth, parity, education and duration variables shown in the parameter estimation STATA program in File 1 below: 'fertility_parity_DR.do'.

Software: Stata 15.1 & Excel


Files:
1. Parameter estimation and descriptive statistics : use 'fertility_parity_DR.do'

2. TFR calcuation for Cross-validation by individual years: use Excel worksheet 'Cross-validate-years.xlsx'

3. Cross-validation test statistics: use 'modeltfr.do' with 'modeltfr_test.dta'
In the data 'modeltfr_test.dta', 
'actual' refers to actual TFR in England and Wales from registration data
'tfrm0' and 'tfrm1' are model predictions of the TFR from models 0 and 1, respectively
'error0' and 'error1' are model prediction errors of the TFR from models 0 and 1, respectively

4. Bootstrapping results in Table 1
The program applying to the time period 2017-20 is provided: 'PPR_bootstrap_ppr5_17.do'  It is easily adapted to produce the PPRs for the other columns of Table 1. To calculate the mean TFR and its SE you must run the program 'se_tfr.do' after saving the bootstrapped data set from 'PPR_bootstrap_ppr5_17.do'.  

5. Bootstrapping results in Tables 2
The program applying to the time period 2017-20 for degree-educated women is provided: 'PPR_bootstrap_ppr5_17_deg1.do'. It is easily adapted to produce the PPRs for the other columns of Table 2. To calculate the mean TFR and its SE you must run the program 'se_tfr.do' after saving the bootstrapped data set from 'PPR_bootstrap_ppr5_17_deg1.do'.