Modelling HRM / Quantifying a model

["Mooy, R.M."]

Hello all,

For my company division, I am currently trying to construct an HRM (Human
Resource Management) model that should describe the transition of Human
Resources under a new organizational structure.
My first question is: is there any model library that contains a number of
HRM models that I might use? I suspect that this topic has been dealt with
many times already, so Id like to see if I can use (parts of) an existing
model.

The HRM model will probably contain a number of soft (difficult to
measure) variables like "quality of the company" and "attractiveness of the
company as an employer". These variables seem to make up some important
feedback loops.
My second question therefore is: how does one go about quantifying such a
model..? I thought of scaling the variables on a scale of 1 to 10, but maybe
there is a better way.

Thanks,

Rutger Mooy
KPN Telecom
the Netherlands
R.M.Mooy@kpn.com

["geoff coyle"]

Mr Mooy,

I think you are on the wrong track in trying to get an existing model or
part of one to plug into your problem. There are two reasons which are
absolutely fundamental in any modelling, not just SD.

The first is that a model is always developed for a purpose (or it should
be) such as to answer some well-defined questions. Your purpose is highly
unlikely to be the same as that of the previous modeller so that model
segment may be utterly irrelevant to your own work.

The second is that you would be taking it on trust that the model did not
contain any errors, such as typos or dimensional errors.

There is a third reason, which is that, if you know what you are doing, SD
models are easy to build. Youd probably spend more time fiddling with
someone elses model than building your own.

There has been a lot of debate on this chat line about soft variables.
CERTAINLY they exist in the real world but putting numbers and equations to
such ill-defined factors as quality of company and attractiveness to
employees is very problematical. To write a reliable quantified simulation
model of the effect that one has on the other is verging on the impossible.
For example, what is quality? There are MANY ways of describing that:
likelihood that it will not go bust while I work there, free lunches,
pension rights, friendliness of colleagues, tyranny of management, product
line (some people would not like to work for an armament manufacturer while
others might relish the technological challenge) and so on and on and on.
Try making a list of another 10 aspects of quality - you wont find it
difficult - and then decide which ones affect the attractiveness to
employees.

A scale for 1 to 10 is easy to make up but means nothing. In any case,
whatever scale you use should start at 0.

The end result of all this is that I dont think that you can quantify such
a model in any reliable way. Of course, I could write a set of equations for
it and it would be fun playing with the software but Id hate think that I
could draw any meaningful conclusions from the model outputs. See my
forthcoming paper in SD Review.

I hope this is helpful, but give me a call if you want to talk.

Regards,

Geoff

Professor R G Coyle,
Consultant in System Dynamics and Strategic Modelling,
Telephone +44 (0) 1793 782817, Fax ... 783188
email geoff.coyle@btinternet.com

["Mooy, R.M."]

Hello all,

Thanks for replying on my request on how to handle soft variables when
quantifying a model, and whether there exist a library with useful HRM
models.

To start with the latter: it was =not= my intention to use an existing
model in the sense of incorporating this into mine. My intention is to see
what are the typical variables that are taken into consideration in HRM
models, and to see how others have quantified soft variables, which are
likely to be present in HRM models. Id just like to see what a typical HRM
model (if you can speak of such a thing) would look like. I already received
some resources on this.
I agree that I should build my own model - not use any existing ones.

Many of you seem to disagree on the subject if I should try to quantify soft
variables, and how to do this. It seems to me that I cannot avoid
quantifying soft variables, but the commentary on exploring the meaning of
soft variables and the effects they have is very useful. This raises new
questions, and will help the model stakeholders in further exploring the
problem at hand.
Anyway, I will be very cautious to draw any conclusions from the soft part
of the model.

Thank you all,

Rutger Mooy
From: "Mooy, R.M." <R.M.Mooy@kpn.com>
the Netherlands

[Yaman Barlas]

> One word of advice: Try not to cascade two "soft" lookup functions. So for

Although this rule of thumb may be useful in avoiding unnecessary cascading,
it must be used with some caution. Simplicity may mask the "real" causality.
For instance, in Jims example if the "quality of the company" had an effect
on the "retention" in real life AND if there was some other variable (like
"training") affecting the "quality", then of course salary may go up, yet the
quality and hence retention may go down. Perhaps this is what Jim means by "If
what you are trying to model permits..."
regards
Yaman Barlas
From: yaman barlas <ybarlas@boun.edu.tr>


---------------------------------------------------------------------------
Yaman Barlas, Ph.D.
Prof. of Industrial Engineering
Bogazici University,
80815 Bebek, Istanbul
Turkey. Fax. +90-212-265 1800. Tel. 90-212-263 1540; ext.2073
http://www.ie.boun.edu.tr/faculty/barlas
SESDYN Group: http://www.ie.boun.edu.tr/sesdyn
-----------------------------------------------------------------------------

[Bill Harris]

I started to suggest that HPS has some aging chain models in some of
their books which could be described as HR models and the Vensim
molecules provide others, but I think Id want to hear more about the
problem youre trying to solve first. Maybe you cant divulge the
problem in a public forum, but I think making it crystal clear at least
to yourself and focusing on it, not the HR system, will help you
succeed.

As to your quantification question, its often nice to quantify relative
to a nominal value (e.g., <relative attractiveness of your company> =
<attractiveness of your company>/<industry average attractiveness>).
That addresses the likely reality that the variables you mentioned are
likely not absolutes but exist only by comparison to alternatives.
Also, think of how you will use the result. If you will multiply it by
another factor, then a variable that scales from 0 to 1 or that has 1 as
its nominal value has advantages.

Regards,

Bill
From: Bill Harris <bill_harris@facilitatedsystems.com>
--
Bill Harris 3217 102nd Place SE
Facilitated Systems Everett, WA 98208 USA
http://facilitatedsystems.com/ phone: +1 425 337-5541

[Bill Harris]

X-MIMETrack: Itemize by SMTP Server on MAILSRV/IBIMSA(Release 5.0.3 (Intl)|21 March 2000) at
09.11.2000 15:04:14,
Serialize by Router on MAILSRV/IBIMSA(Release 5.0.3 (Intl)|21 March 2000) at
09.11.2000 15:04:15,
Serialize complete at 09.11.2000 15:04:15
X-Notes-Item: 28;
type=300; name=$Hops
X-Notes-Item: Memo;
name=Form
X-Notes-Item: ;
type=501; flags=44; name=$UpdatedBy
Content-Transfer-Encoding: 7bit
Content-Type: text/plain; charset=us-ascii
Status: RO


I started to suggest that HPS has some aging chain models in some of
their books which could be described as HR models and the Vensim
molecules provide others, but I think Id want to hear more about the
problem youre trying to solve first. Maybe you cant divulge the
problem in a public forum, but I think making it crystal clear at least
to yourself and focusing on it, not the HR system, will help you
succeed.

As to your quantification question, its often nice to quantify relative
to a nominal value (e.g., <relative attractiveness of your company> =
<attractiveness of your company>/<industry average attractiveness>).
That addresses the likely reality that the variables you mentioned are
likely not absolutes but exist only by comparison to alternatives.
Also, think of how you will use the result. If you will multiply it by
another factor, then a variable that scales from 0 to 1 or that has 1 as
its nominal value has advantages.

Regards,

Bill
From: Bill Harris <bill_harris@facilitatedsystems.com>
--
Bill Harris 3217 102nd Place SE
Facilitated Systems Everett, WA 98208 USA
http://facilitatedsystems.com/ phone: +1 425 337-5541

["Jim Hines"]

Rutger Mooy asks how to quantify soft variables like quality of company.

One word of advice: Try not to cascade two "soft" lookup functions. So for
example try not to go from "salary" to "effect of salary on quality of
company" and then from "quality of company" to "effect of company quality on
retention". Instead go directly from what the "hard" input to the final
soft variable. For example go from "salary" to "effect of salary on
retention". If what you are trying to model permits you to follow this
little rule of thumb, your model will be easier to parameterize and
understand. Rather than having to judge salarys effect on quality and then
THEN qualitys effect on retention, youll find it much easier to directly
judge salarys effect on retention.

Regards,
Jim
From: "Jim Hines" <jhines@MIT.EDU>

[Jay Forrest]

I heartily support Jims comments but I would suggest that it can be very
educational to cascade "soft" lookup functions for you are likely to learn
something about the consistency of your assumptions. I.e. a cascaded series
of assumptions is unlikely to match your understanding of the macro
behavior. Exploring the cascaded assumptions can provide a basis for
refining your underlying assumptions.

Regards!
Jay Forrest

11606 Highgrove Drive
Houston, Texas 77077
Tel: 713-503-4726
Fax: 281-558-3228
E-mail: jay@jayforrest.com or jforrest@futuresguild.com

["Jim Hines"]

Geoff --

I like your comment about the desirability of building your own model. What
do you (and others) make of the many queries in this list asking for models
on this or that topic?

Jim Hines
jhines@mit.edu