Household formation

[vkhodygo]

This is a general road map for household generation.

1. Get the total number of households per every constituent country. The most accurate values are provided by UK Census, all intermediate values can be interpolated/estimated using Labour Force Survey/Understanding Society data. This data would also allow to make future projections.
   Q: Is there any official statistics covering this?
   A: See NOMIS. Almost two decades of data, however, it's at least partially rounded and those numbers might not add up.

2. Split this number into 4 different groups of households:

   • one person household;
   • two or more unrelated adults;
   • one family household;
   • multi-family household.

   These numbers must be represented by a series too, some data is provided by ONS, more data is available at NOMIS (see above).

3. Now, we start with the very first year of the simulation, at the moment this is 2011. Since getting access to the official Census data is accompanied by certain difficulties and takes quite some time we employ a slightly different approach to build the initial population.

   The idea here is to start with the simplest household composition that is - surprisingly - the one family one. That also helps with reducing the population pool size. We build couples with no children first, however, certain assumptions must be kept in mind all the times. Whereas the number of same-sex couples has increased fivefold from 2015 to 2020, their share is still about 0.5%. It is clear that this number will continue to rise until it reaches a certain threshold level, but it's too early to make any predictions for them. Thus, all couples in this model are comprised of a single male and a single female.
   Ways to form a partnership vary from couple to couple, still, we make an assumption that the numbers of civil partnership and married couples follow the same probability distribution whatever it is. This part of the simulation also requires age gap tables (see NOMIS for some statistics) only.
   Q: Do we need divorce tables of any kind here?
   A: Yes, definitely. We need to produce a balance equation of some sort and get numbers for civil/cohabiting couples. See also aggregate marriage and divorce numbers for all kinds of couples.

   Next, we exclude lone parents with dependent children. The only restriction here is for any parent to be at least 16 when their kid is born. Other than that we employ ONS tables to get sex distribution of the parents (most of them are females), we also draw children from the population pyramid. Also, women have different fertility rates at different ages, this must be taken into account. Strictly speaking, men are not eternally young either, but that's too much for the model at the moment.
   Q: Does that require population pyramids that go for about two decades into the past?
   A: Apparently, no. To populate year one we sample from the population pyramid for the same year. Any child is already included in the statistics already.
   Q: How do you create households with 2+ children?

   Next step is to form couples with dependent kids. We follow the examples right above to do that.
   There are some corner cases that must be covered here: women can't be too young or too old, whereas men can't be too young only. In addition, families with multiple kids require data about the number of kids per family, it is also highly unlikely for someone who is 16 to have a large number of newborns.
   Q: this has to be investigated further.

   What's left here is one family households with non-dependent children only. This probably works the same way as above, however, age limits should be completely different.
   Multi-family households: the number is low, but the internal structure has a lot of possible combinations.
   Q: this has to be investigated further.

   Two or more unrelated adults. They are most likely to be students, however, there should be a sex-age distribution for them.
   Q: ask ONS.

   If everything is done right the remainder of the population pool should belong to the one-person-household group only.
   Q: request sex-age composition data.

4. Start projection into the future, every year new people are born and some old die. This way, reach year 2021/2022 and compare the results with the census.

[vkhodygo]

@dkopasker @rachelmthomson @andrewbaxter439 I'd like to hear your feedback on this one, feel free to suggest improvements/corrections.

[vkhodygo]

I should've added @vkatikireddi as well.

[mattDevigili]

Dear all, I might have been included in this thread by mistake?

[vkhodygo]

@mattDevigili I'm afraid so, I failed to find your GitHub id mentioned anywhere here. Still, if you have any ideas, let us know.

[rhettlivingston]

I also received this with no clue what it is about. There was definitely a glitch in the matrix. I've had no activity on GitHub for years.

…

On Wed, Jan 11, 2023, 10:57 AM Vladimir ***@***.***> wrote: @mattDevigili <https://github.com/mattDevigili> I'm afraid so, I failed to find your GitHub id mentioned anywhere here. Still, if you have any ideas, let us know. — Reply to this email directly, view it on GitHub <#14 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AB7FGYSUMA47M6AYOUYF7O3WR3J7DANCNFSM6AAAAAASVVMHVE> . You are receiving this because you were mentioned.Message ID: <MRC-CSO-SPHSU/mortality_module/repo-discussions/14/comments/4658022@ github.com>

[swinkz]

Same here. Please remove me from mail loop. WojciechWiadomość napisana przez rhettlivingston ***@***.***> w dniu 11.01.2023, o godz. 17:24: I also received this with no clue what it is about. There was definitely a glitch in the matrix. I've had no activity on GitHub for years.

On Wed, Jan 11, 2023, 10:57 AM Vladimir ***@***.***> wrote: @mattDevigili <https://github.com/mattDevigili> I'm afraid so, I failed to find your GitHub id mentioned anywhere here. Still, if you have any ideas, let us know. — Reply to this email directly, view it on GitHub <#14 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AB7FGYSUMA47M6AYOUYF7O3WR3J7DANCNFSM6AAAAAASVVMHVE> . You are receiving this because you were mentioned.Message ID: <MRC-CSO-SPHSU/mortality_module/repo-discussions/14/comments/4658022@ github.com>

—Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you were mentioned.Message ID: ***@***.***>

[abidikhairi]

Hi, Why am I in this thread ?

…

On Wed, Jan 11, 2023 at 5:42 PM swinkz ***@***.***> wrote: Same here. Please remove me from mail loop. WojciechWiadomość napisana przez rhettlivingston ***@***.***> w dniu 11.01.2023, o godz. 17:24: I also received this with no clue what it is about. There was definitely a glitch in the matrix. I've had no activity on GitHub for years. On Wed, Jan 11, 2023, 10:57 AM Vladimir ***@***.***> wrote: > @mattDevigili <https://github.com/mattDevigili> I'm afraid so, I failed > to find your GitHub id mentioned anywhere here. Still, if you have any > ideas, let us know. > > — > Reply to this email directly, view it on GitHub > < #14 (reply in thread) >, > or unsubscribe > < https://github.com/notifications/unsubscribe-auth/AB7FGYSUMA47M6AYOUYF7O3WR3J7DANCNFSM6AAAAAASVVMHVE > > . > You are receiving this because you were mentioned.Message ID: > <MRC-CSO-SPHSU/mortality_module/repo-discussions/14/comments/4658022@ > github.com> > —Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you were mentioned.Message ID: ***@***.***> — Reply to this email directly, view it on GitHub <#14 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AQP5MN4SDJ77LRJ4Q3DANJ3WR3PFXANCNFSM6AAAAAASVVMHVE> . You are receiving this because you were mentioned.Message ID: <MRC-CSO-SPHSU/mortality_module/repo-discussions/14/comments/4658541@ github.com>

[andrewbaxter439]

Hello all, you're receiving these emails because I think some bots spammed this discussion in December and tagged dozens of random accounts. Github (helpfully \s) thinks you're all still involved in the conversation so is updating you via notification/email. There seems no way of undoing this. I'll lock the conversation for now so that nobody else gets any more notifications.

[vkhodygo]

some bots spammed this discussion in December and tagged dozens of random accounts

Did this happen when I was on leave? Fascinating, since all traces have been removed by GitHub I can't even explain this to people.