Stata syntax to create some variables related to work hour loss using UKHLS Covid-19 survey data
This Stata syntax file uses data from the Covid Survey Waves 1-3 to produce variables to measure (i) experience of loss of work hours (ii) reasons for experiencing loss of work hours.
/*-------------------------------------------------------------------------------------------------------------------*
| This file is for those who use the info of reasons for work hour loss during Covid-19 |
| Users should be aware of how the information is collected. |
| |
| In the UKHLS covid survey, there are four situations in which individuals are counted as “working hours loss” |
| and are asked for reasons. The criteria are |
| - 1. Those who did not answer previous waves. They are employed/self-employed in Jan/Feb and |
| their work hours now are fewer than Jan/Feb. Note that their working hours in Jan/Feb and |
| now are not missing. |
| - 2. Those who did not answer previous waves. They are employed/self-employed in Jan/Feb and not working now. |
| - 3. Those who answered previous waves. They are employed/self-employed now. Their work hours are fewer |
| than previous waves. Of course, work hours in both waves are not missing. |
| - 4. Those who answered previous waves. They are employed in previous wave and not working now. |
| |
| The reason why we have to be causious when using this variable is that |
| for individuals who do previous waves, their info compares with previous wave, and |
| for individuals who do not do previous waves, their info compare with Jan/Feb. |
| |
| That means you cannot take one single wave to know if a person experienced work hours loss and for what reasons |
| during the pandemic. For example, people who experience work hour loss in wave 1 but do not have a further |
| decline in work hours in wave 2&3, are not captured when only using wave 3. |
|--------------------------------------------------------------------------------------------------------------------*/
* This file just takes Covid w1-w3 as an exmaple for users
*== Work Hour Loss ==*
*-1. Merge Covid w1, w2, and w3 data (make sure every data is sorted by pidp)
* use "~\ca_indresp_w.dta", clear
* sort pidp
* merge pidp using "~\cb_indresp_w.dta"
* sort pidp
* merge pidp using "~\cc_indresp_w.dta"
*-2. Generate accumulative info for experiencing work hour loss
* Experience work hours loss between Jan/Feb - Apr
gen workhrloss1=0
replace workhrloss1=1 if (ca_ff_prevsurv==0 & (ca_blwork>=1&ca_blwork<=3) & (ca_hours<ca_blhours&ca_hours>=0&ca_blhours>=0))| ///
(ca_ff_prevsurv==0 & (ca_blwork>=1&ca_blwork<=3) & ca_sempderived==4)
replace workhrloss1=. if !inlist(ca_outcome,11)
label var workhrloss1 "Experience work hours loss between Jan/Feb - Apr"
label define workhrloss 0"No" 1"Yes", modify
label val workhrloss1 workhrloss
* Expereince work houes loss between Jan/Feb - May
gen workhrloss12=0
replace workhrloss12=1 if (cb_ff_prevsurv==0 & (cb_blwork>=1&cb_blwork<=3) & (cb_hours < cb_blhours&cb_hours>=0&cb_blhours>=0))| ///
(cb_ff_prevsurv==0 & (cb_blwork>=1&cb_blwork<=3) & cb_sempderived==4)| ///
(cb_ff_prevsurv==1 & (cb_sempderived>=1&cb_sempderived<=3) & (cb_hours < cb_ff_hours&cb_hours>=0&cb_ff_hours>0))| ///
(cb_ff_prevsurv==1 & (cb_ff_semp>=1&cb_ff_semp<=3) & cb_sempderived==4)
replace workhrloss12=1 if workhrloss12==0 & workhrloss1==1
replace workhrloss12=. if !inlist(cb_outcome,11)
label var workhrloss12 "Expereince work houes loss between Jan/Feb - May"
label val workhrloss12 workhrloss
* Expereince work houes loss between Jan/Feb - June
gen workhrloss13=0
replace workhrloss13=1 if (cc_ff_prevsurv==0 & (cc_blwork>=1&cc_blwork<=3) & (cc_hours < cc_blhours&cc_hours>=0&cc_blhours>=0))| ///
(cc_ff_prevsurv==0 & (cc_blwork>=1&cc_blwork<=3) & cc_sempderived==4)| ///
(cc_ff_prevsurv==1 & (cc_sempderived>=1&cc_sempderived<=3) & (cc_hours < cc_ff_hours&cc_hours>=0&cc_ff_hours>=0))| ///
(cc_ff_prevsurv==1 & (cc_ff_semp>=1&cc_ff_semp<=3) & cc_sempderived==4)
replace workhrloss13=1 if workhrloss13==0&(workhrloss1==1|workhrloss12==1)
replace workhrloss12=. if !inlist(cc_outcome,11)
label var workhrloss13 "Expereince work houes loss between Jan/Feb - June"
label val workhrloss13 workhrloss
*-3. Reasons for experiencing work hour loss
* Note 1: people can choose more than one reasons
* Note 2: codings below is just an example for users who would like to use the variables
* Note 3: some new categories shown since wave 2, such as variation or quit
* Note 4: UKHLS collect data separately for those employed, those self-employed, and those both employed and self-employed
* Experiencing work hour loss during w1-w3, due to
//a. furlough
gen furlough13=.
foreach w in cc cb ca{
replace furlough13=1 if (`w'_hrschange15==1|`w'_hrschange35==1)
}
label var furlough13 "ever furloghed between April-June"
//b. being laid off
gen laidoff13=.
foreach w in cc cb ca{
replace laidoff13=3 if laidoff13==.&(`w'_hrschange11==1|`w'_hrschange12==1|`w'_hrschange13==1)|(`w'_hrschange31==1|`w'_hrschange32==1|`w'_hrschange33==1)
}
label var laidoff13 "ever laid off between April-June"
//c. (Users can do more depending on their research interests)
* Shih-Yi Chao
* S-Y.Chao@soton.ac.uk