Microsimulation: Barriers to entry

1
Microsimulation Barriers to entry

• Building and maintaining client relations

Robert Tanton Research Director June 2009
2
Building and Maintaining client relationships

• Building and Maintaining client relationships
  essential for microsimulators if working with
  clients
• Barrier not normally considered, but no point
  doing modelling if dont have clients who are
  going to use the modelling

3
Why you need to build a strong client relationship

• If dont have strong client relationship, you
  either wont get a model they want to them, or
  the model will probably be of little use to them.
• If dont have a strong client relationship, model
  wont be valued
• Need someone on client side who understands value
  of microsimulation models
• Need salesman on client side someone who
  shares your vision
• Need to get this shared people move on

4
How to build a client relationships Clients
understanding you

• Sell the output, not the process
• Dont get technical straight away For most
  clients, dont need to get technical at all
• Explain model with examples
• Need some flexibility for client who is
  understanding your model

5
Technical explanation of spatial microsimulation
- for a statistician

• The reweighting process in SpatialMSM uses an
  iterative constrained optimisation technique
  which minimises a Chi squared distance function
  to calculate weights. When applied to the survey
  data, this technique provides weights that will
  give the best estimates of the Census Benchmarks.
  The technique uses a calibration estimator
  initially outlined by Singh and Mohl (Singh and
  Mohl 1996) and described and implemented by the
  ABS (Bell 2000) in a SAS macro called GREGWT.

6
Explaining spatial microsimulation to a
non-statistical client

• When the ABS issues the microdata files from its
  national sample surveys, it attaches a weight
  to the record of every household within the
  sample. For example, the weight attached to the
  first household within the sample file represents
  the number of households within Australia that
  the ABS believes are the same as that particular
  household. These weights are the mechanism used
  to gross up from the sample survey results to
  estimates for the whole of Australia. In a series
  of recent research projects, NATSEM has been
  refining the technology to weight the ABS sample
  survey files to small area targets derived from
  the census.

7
Building client relationships you understanding
the client

• Need to be able to read client technical?
  Only interested in outcomes?
• Clients will usually want output and outcomes,
  not method or process
• Relate to them at the level they want to, not at
  the level you want to
• Need to fully understand what client wants
• Biggest risk doing lots of work and providing a
  client with something they didnt really want
• Ask questions
• Clarify points
• Constant communication check-points, meetings,
  etc

8
Maintaining client relationships

• Most microsimulation models will need updating,
  so need to maintain client relationship so model
  not on the shelf
• Strong relationship between you and client is
  essential
• Means client trusts you with results
• Developed over time dont expect to be instant
• Can be ruined with one bad project be careful
  with all projects, validate all output

9
Maintaining client relationships

• Public trust is important
• If have good reputation, then clients more
  inclined to trust you
• Dont publish anything that isnt validated
• Doesnt mean cannot publish preliminary results
  but validate these, then release as preliminary

10
Summary

• Sell the output, not the process
• Find a salesman on the client side
• Maintain a reputation of quality, with validated
  analysis
• Fully understand your client what they want,
  ensure they understand what you can deliver

11
www.natsem.canberra.edu.au