Help: time series data with time lags

Use this forum to post Vensim related questions.
Post Reply
sl590
Junior Member
Posts: 5
Joined: Sat Jun 02, 2018 5:37 pm
Vensim version: PLE

Help: time series data with time lags

Post by sl590 »

Hi all,

I was working on eviews on a macroeconomic model. The model is a system of simultaneous equations, which I think Vensim will be great to work on the simulations. However, the model has the following properties for many variables:
1. k_k(t) = k_k(t-1)*(1 + gr_k(t))
2. gr_k(t) = gamma0 + gammau*u(t-1) - gammar*rr_l(t)
3. u(t) = y_k(t)/k_k(t-1)
....etc.

most variables will have an initial value to start with

I wonder how to make such simulations possible ... Thank you for the help!
Administrator
Super Administrator
Posts: 4573
Joined: Wed Mar 05, 2003 3:10 am

Re: Help: time series data with time lags

Post by Administrator »

This question is far too vague to really provide an answer. What do these equations even mean?
Advice to posters seeking help (it really helps us to help you)
http://www.ventanasystems.co.uk/forum/v ... f=2&t=4391

Units are important!
http://www.bbc.co.uk/news/magazine-27509559
sl590
Junior Member
Posts: 5
Joined: Sat Jun 02, 2018 5:37 pm
Vensim version: PLE

Re: Help: time series data with time lags

Post by sl590 »

Hi there, thank you for the reply. I only took a few equations out of the system, as I don't know how to include lagged variables in a vensim model. However, to explain what the equations are:
1. capital stock grows at a rate of gr_k, hence at time t, k_k(t)= k_k(t-1)*(1+gr_k)
2. the growth rate is defined by a the following equations all gammas are parameters (constants)
e.g. gr_k=1+2*u(t-1)-3*rr_l (t), where u is the capital utilization proxy and rr_l is the real lending rate
3. u = y_k(t)/k_k(t-1) is showing that capital utilization proxy is the ratio between current real GDP flow and the real capital stock from last year
and there are many other simultaneous equations with lagged variables in the whole system.

If you could share an example vensim file for maybe just one of the equations, that'd be very helpful!
sl590
Junior Member
Posts: 5
Joined: Sat Jun 02, 2018 5:37 pm
Vensim version: PLE

Re: Help: time series data with time lags

Post by sl590 »

Administrator wrote: Mon Jun 04, 2018 6:03 pm This question is far too vague to really provide an answer. What do these equations even mean?
Hi, thank you for the reply. I only took a few equations out of the system, as I don't know how to include lagged variables in a vensim model. However, to explain what the equations are:
1. capital stock grows at a rate of gr_k, hence at time t, k_k(t)= k_k(t-1)*(1+gr_k)
2. the growth rate is defined by a the following equations all gammas are parameters (constants)
e.g. gr_k=1+2*u(t-1)-3*rr_l (t), where u is the capital utilization proxy and rr_l is the real lending rate
3. u = y_k(t)/k_k(t-1) is showing that capital utilization proxy is the ratio between current real GDP flow and the real capital stock from last year
and there are many other simultaneous equations with lagged variables in the whole system.

If you could share an example vensim file for maybe just one of the equations, that'd be very helpful!
Administrator
Super Administrator
Posts: 4573
Joined: Wed Mar 05, 2003 3:10 am

Re: Help: time series data with time lags

Post by Administrator »

I think it's probably best if you have a go first and upload the model if you have any problem.

Things like k_k(t-1) is not the right way to think about it when building a model. Vensim steps forwards in "time step" increments, so t-1 can mean today - 1 (yesterday), this month - 1 (yesterday or last month?). What does t-1 actually mean, is it last month, last week, the last time the simlation stepped forward etc.
Advice to posters seeking help (it really helps us to help you)
http://www.ventanasystems.co.uk/forum/v ... f=2&t=4391

Units are important!
http://www.bbc.co.uk/news/magazine-27509559
sl590
Junior Member
Posts: 5
Joined: Sat Jun 02, 2018 5:37 pm
Vensim version: PLE

Re: Help: time series data with time lags

Post by sl590 »

Administrator wrote: Mon Jun 04, 2018 7:05 pm I think it's probably best if you have a go first and upload the model if you have any problem.

Things like k_k(t-1) is not the right way to think about it when building a model. Vensim steps forwards in "time step" increments, so t-1 can mean today - 1 (yesterday), this month - 1 (yesterday or last month?). What does t-1 actually mean, is it last month, last week, the last time the simlation stepped forward etc.
all data are annual data, hence the initial value coincides with year 0 (e.g. 1950) k_k(1951)=k_k(1950)*(1+growth rate). Hope this time is clearer
tomfid
Administrator
Posts: 3804
Joined: Wed May 24, 2006 4:54 am

Re: Help: time series data with time lags

Post by tomfid »

This explains the mapping between discrete and continuous time, and why discrete notation for continuous phenomena is usually problematic:
http://metasd.com/2017/11/discrete-time-stinks/
sl590
Junior Member
Posts: 5
Joined: Sat Jun 02, 2018 5:37 pm
Vensim version: PLE

Re: Help: time series data with time lags

Post by sl590 »

tomfid wrote: Tue Jun 05, 2018 12:35 pm This explains the mapping between discrete and continuous time, and why discrete notation for continuous phenomena is usually problematic:
http://metasd.com/2017/11/discrete-time-stinks/
Thank you, will have a look
Post Reply