 The earlier description ignored one important relationship between jobs and migration. It takes time for people to learn about jobs and job opportunities and to relocate. It is going to take some time. Further, people probably respond average number of job opening city and not to short term increases or decreases in the job openings. So, that is what you are going to react to. Now, we are going to modify this our existing model by introducing a stock called as perceived job openings and based on a which will be based on a first order information delay with a smoothing constant of say 0.25 per month. So, perceived job openings let us just create that. So, create a stock as soon as create a stock immediately add a flow because stocks can only be changed through flow. Else it will be too tempted to just put out another variable and your system may not work the way you expect it to. So, immediately we just change it ok this is a change in perceived openings and then new information let us model smoothing factor. This perceived job openings reacts to the actual value of job openings right. So, here your reported value or the actual value will be the job opening itself. So, this and your current value will be your perceived job openings and now instead of reacting to the job openings they are going to react migration reacts to perceived job openings. So, I am going to click delete, delete is trash bin then you click this arrow, click arrow, click perceived job openings which feeds into migration. So, now if you select your equation lots are going to be in black let us say smoothing factor is 0.25 1 per month. Change in perceived values is both units are same job openings minus perceived job openings multiplied by smoothing factor units is job unit of job opening is job. So, this has to be job per month oh sorry this should be job per month perceived job openings let us keep it as 800 oh sorry job 800 and migration does not react to job openings it reacts to perceived job openings. So, the equation you have updated into perceived job openings into people to job ratio divided by adjustment time. So, on the equations for this it can be plus minus you can see it. So, now we have two stocks in the system this itself is a delay because adjustment time we have taken as 2 that means after two months it is average time it takes for people to move. So, that is the time it takes to move. Now, we are putting additional thing called as additional information delay where I mean job opening information is received later and then based on the perceived job openings people migrate and people migrate and fill the jobs. Now, let us simulate this that is override now if we let us look at people again just to clarify we can go to jobs job is same constant at 1000 there is no change in jobs job was again at 1000 the perceived job openings are still fluctuating in this case it did not react if it was only a first order delay then from initial value of perceived job opening was 1000 and then it has to was it 800 let me let us make the perceived job openings 0. Let us assume there is no jobs initially now let me simulate it again. So, initially it was 0 and then when additional 200 jobs came about there is a time 0 came about additional 200 jobs. So, initial people was 800 jobs was 1000 when it changed similarly started to increase, but it did not saturate at to fill the gap it started fluctuating let us see why because if it is only a first order system it cannot fluctuate we know that it has to smoothly reach the target we just simulated it for demand right the same behavior we need to expect. So, the only variable which also can keep changing can cause this behavior is not the jobs, but we are taking the difference between jobs and the people there is second stock in the system. So, let us see the behavior of that let me get all the output variables jobs people this is fine. So, this is a perceived job openings in the people. So, as soon as there are some jobs that are perceived then people started to slowly come in and as they started coming in you know because of the delays involved it overshot the number of people in system overshot that means I have more people than the jobs. So, then it try to compensate because of delays it over compensated again it fell down too much then it can take and then again it kind of fluctuated and then finally, damped finally saturated at the or reaches steady state around 0 here and around 1000. So, steady state values continue to be 1000. So, these kind of oscillation we call it as damped oscillations where oscillations are not consistent kind of damps. So, to get a system to oscillate we need minimum 2 stocks and just this information asked us to allow the system to first exceed then what it has to be the goal and then it over compensates down and again it increases etcetera and slowly only it is going to converge. So, you will end up choosing say mechanical and safe disciplines I do not know or may choose unsafe new disciplines I do not know. So, this allows this so that means you can you just saw what happened right because of the small delay in information more people were there then jobs available. So, more people are there jobs less jobs are there then instead of migration you can imagine migrating to that particular field or number of people are going to join that right. So, then again people overcompensate then suddenly there is more demand system is going to fluctuate. But in this case the delays were short we are adjustment time of 2 months and smoothing factor of 0.25 when reality is much smaller the smoothing factors adjustment time is much larger. So, these fluctuations are very large and there are more delays involved we just had 2 forms of delay one is as migration delay other is the perceived openings delay we have so many other delays which is going to sustain the fluctuations for a really long time in the 50 years 60 years 100 years that is really really long time before it can actually reach whatever its carrying capacity is going to be like here we assume the smoothing constant is constant smoothing constant variable is constant but that can also change I can update it based on new information available I am going to I am going to decide to react quickly to the new information or not then that can be another factor which will again further cause more dynamics in the system. But now we are going to see real non-linear dynamics when you introduce 2 stocks. So in Moodle only the first model is there this one is there. So, you have to make a dimensional consistent then add the next stock and flow only then you will be able to see the dammed oscillations. So, you have to do it if not now later now higher order information delays what does it mean and how do you model it in first order delay the beliefs change immediately on changes to input when higher order delays it takes time for belief to change it responds only after some delay a simplest form of it is a pipeline information delay where if it is you are only going to report a measurement delay or reporting delay where values are not changed I am just going to communicate it then it is you can model it very similar to a pipeline material delay. So, that information is conveyed there is no change in the information. Let me just give example of the third order information delay the third order information delay in in material delay what do you do we had a first order material delay and we connected three in series to call it a third order material delay. In a similar concept we will do a third order information delay where we will cascade three first order information delay. Let us do third order information delay that is define a stock S1 change in S1 let us take that as a input and let us take average delay. So, this is the easiest right whatever input changes this is a first order delay right just written in a drawn in a slightly different fashion, but it is nothing but a simple first order delay where your equation for dS1 by dt is nothing but minus S1 divided by d by 3. Third order delay I am going to divide the total average delay into three equal compartments. So, whatever input minus S1 is what I am changing now when happens when I want to start drawing the second order and third order what I am going to do is define another stock S2 change in S2 change in S2 is get affected by S2 and the information from S1 so dS2 by dt is nothing but S1 minus S2 divided by d by 3. So, now if I increase my input initially assume input is equal to stock 1 equal to stock 2 let me finish it and then I will explain let us put a stock 3 then I have it as change in S3 then I have final output and your dS3 by dt is S2 minus S3 divided by d by 3 and their output is nothing but this stock value S3. So, what happens when I change my input assume input equal to stock S1 equal to S2 equal to S3 equal to output the input is equal to output the system is steady state no change suppose input changes by one unit then based on the difference first this stock changes the next time period only this stock is going to change and reach the value change the value of new value of input and then this value is going to change. So, this is how you model the information delay since again as you can see information is not conserved suppose input increased from 0 to 10 pulse input and then it became again 0 or time 1 then this will increase so, based on average delay it will increase say one-third of it this will increase one-third of this this will increase one-third of even that right. So, it is like almost like Chinese whispers you start with 1 and this may can maximum grow up to one-third of it assume delay is you know one-third of based on the delay value and again it will go one-third of that and then one-third of that. So, each time we are going to keep reducing a final output will be a third order delay of this input and just like we had what is the function delay and function in Vensim to model these in shortcut Vensim has something called as a smooth function in Vensim in this smooth function you can directly connect the input to the output and do the average delay it will smooth the input based on the order of the delay and do the value of the output you can play it with the examples already available in Vensim for the smooth function to directly do that. Because as you can see if you are going to model as a fifth order delay just making all these stocks itself is going to make your model so complicated so that is when this smooth function helps most of the systems we deal with will be a first order delay like demand forecasting people use exponential smoothing very few cases we explicitly go for second order delays and other higher orders just because it is so difficult to comprehend once you start looking at higher order information delay so it may be easier for us to explicitly model them like we did in the previous example we explicitly model change in the perceived value and change in the migration separately may be easier for us to model that so there is some physical relation so there is physical context to the model that we are building but in case if we want to higher order delay this is what is going to happen.