Software code for air transport models

Dear Department for Transport,

The DfT document "UK Air Passenger Demand and CO2 Forecasts",
published January 2009, cites and exhibits the results from a
number of economic models.

http://www.dft.gov.uk/pgr/aviation/atf/c...

In particular, the following models and forecasts are mentioned by
name, and calculated results given in a non-computer readable
format:

* Annex A: Unconstrained Passenger Demand Econometric Models
(operated for each of 21 markets).

* Annex B: The National Air Passenger Demand Model, which includes
21 econometric models.

* Annex C: The Passenger Airport Choice Model

* Annex D.3: National Air Passenger Demand Model

* Annex D.5: National Air Passenger Allocation Model

* Annex D.6: CO2 model

* Annex G: Detailed Demand and CO2 Forecasts

* Annex H: Monetised Net Benefits Methodology, comprising several
subsections detailing the benefit types.

* Annex I: Surface Access CO2 Emissions

* Annex J: Estimate Value of UK Aviation's Climate Change Impact

* Annex K: Aviation's Estimated Share of UK Climate Change
Emissions

As you know, there is considerable public skepticism concerning the
claim that air travel can increase by 60% over the next 4 decades
whilst meeting the UK's climate change obligations. It is the
government's claim that these calculated numbers prove otherwise.

Under the Freedom of Information Act 2000, please send me:

* Information about which departmental team or outside consultants
is in charge of developing, revising and estimating the values for
the input parameters in each of the above models, and whether there
is any independent verification of them.

* How frequently the model calculations are rerun with new data and
what the procedures exist for either reviewing public policy as a
result of changes, or varying the input parameters to keep the
output on track.

* What programming languages or software platforms these models run
within.

* An actual copy of the software code for the models, including
inputs, in a form that someone with access to the same software
platforms could rerun them with minimum of effort.

Please advise me of whether you hold this information, and then
disclose as much of it as you can.

There is insufficient information in this document to recreate the
models and it is not realistic to expect that people must do so in
order to independently verify the results given alternative
estimates of the input parameters.

For example, according to the footnote 78 in section C.3:

""The Passenger Airport Choice Model is an application of the
standard multinomial logit formulation commonly used in this
context. The model assumes the proportion of passengers with
journey purpose p travelling to/from UK zone i to foreign
destination j, that use airport A, PijAp, can be represented by the
following very flexible functional form:[78]

...

[78]The form shown is the simplest of those used""

Thank you for your consideration of this request.

Yours faithfully,

Julian Todd

Dear Richard Clarkson,

Thank you for your 24 March response to my 23 February FOI request
for background details on the economic and environmental models
used in the DfT's January 2009 "UK Air Passenger Demand and CO2
Forecasts" study.

You advised me that the time limit for responding to my request had
been extended for 20 days beyond the statutory limit, and that my
request was being handled under the Environmental Information
Regulations.

I am writing to ask for some specific details about the progress of
my request.

Firstly, I am not sure that all of the listed models cited by the
study fall within the definition of "environmental information".
Can you list the parts of my request that fall into the different
categories?

Secondly, section 10(3) of the FOI Act sets out the circumstances
in which the 20 day time limit need not be complied with in respect
to qualified exemptions that include a public interest test.

Can you confirm whether the information I requested is on your desk
now and is awaiting the application of the public interest test
before disclosure, or has the whole process been delayed and the
department has not reached this stage yet.

Please could you also inform me which exemptions are being
considered in respect to each piece of information, as this may
allow me to agree to narrow my request slightly and thus ensure a
speedier disclosure.

Yours sincerely,

Julian Todd

Dear Richard Clarkson,

Thank you for your final reply on 22 April to my request for
information about the eleven social, economic and environmental
computer models that are used as the basis for supporting the
expansion of air-travel policy.

The Department for Transport's opinion is very clear on this issue,
which is why any self-generated data such as from these DfT
generated models needs to be handled with care.

1. I asked for information about who designs these models, and who
maintains them, and whether there is any independent verification.

I have not been given sufficient information to determine whether
there is any independent verification of these models or their
parameters. I do not know if the "independent peer reviewer" or the
"external appraisal expert" are also employed by the DfT, or to
what depth they reviewed the modelling.

Please send me further details that you have, if any, that would
give me confidence that these models are objective. Owing to 4.,
there will have been no external assessment of these models that
has not been authorized by the DfT.

2. Some of the inputs into these models are the GDP growth
forecasts, interest rates, oil price forecasts, and so on, which
the Treasury and other international bodies update on a monthly
basis.

In many institutions that rely on correct assessments from their
economic models it is routine to rerun every time new data becomes
available.

You have informed me that these models are run very infrequently
owing to the enormous amount of effort involved, and that a new run
is done according to a publication schedule that is less than
annual.

Can you confirm that there is no procedure for rerunning these
models and updating the conclusions when input parameters that are
impacted by, say, an unanticipated global recession vary?

3. Your response informed me that the National Air Passenger
Allocation Model (NAPAIM) is a highly complex tool written in the
C++ programming language which requires specialist skills to
operate, understand, and integrate (via another programming
language) with other parts of the suite.

I have received no information about what this other programming
language is, or the programming languages and software platforms
for the other 10 models I listed in my request.

It is manifestly reasonable to expect the DfT to know what are the
languages and software platforms they have developed these
important and very expensive assets on.

For example, if half of them are written on the Java programming
platform, then International Networks Analysis and Support (INAS)
managers must know that they need to retain a minimum core of Java
trained developers on the staff -- particularly in light of the
answer to 4. where it was explained to me that the software is not
in an easy to maintain state and with no written step-by-step
guides.

Please supply me with the complete answer to my request.

4. If the software code is maintained in such a state that it
cannot be re-installed independently and rerun with a minimum of
effort, please supply me a copy of the software in the actual state
it is.

I have not provided to you my CV, but I can assure you that I
probably have enough of a level of programming expertise to read
and derive truthful conclusions from the raw form of the code. As
you will be backing up the code anyway (to avoid data loss or
computer failures), providing me with the code for each, any or all
of these data models should incur less cost than drafting reasons
for not releasing this data.

Please consider this letter as a request for an internal review.

Yours faithfully,

Julian Todd

Dear Tom Worsley,

Thank you for your comprehensive review of my FOI request for the
software code of the air transport models.

I am a little confused by the consistency of some of the answers.

Firstly, if the Department has commissioned a contractor to develop
and maintain some software then the code is the property of the
Department, unless otherwise specified. There is no reason to
believe it is otherwise specified since Scott Wilson Group Plc does
not appear to be in the software selling business.

Secondly, 9 out of the 12 models run on Excel VBA. I happen to have
a copy of Excel with VBA, so being able to rerun the models would
likely require a minimum of effort were I provided copies of the
files, since this platform doesn't require any configuration. If
they were all in FORTRAN, it would be another matter.

Thirdly, the reasoning about the requirement to provide support for
use of this software has been exposed in Paragraph 45 of decision
notice FS163282 (29 March 2010):

"QUB also emphasised that even if a copy of the raw data was
produced for the complainant, it would be meaningless to the
requestor and could not be put to any meaningful purpose. However
the Commissioner does not consider this to be a valid consideration
when assessing information for possible disclosure under the EIR.
There is no requirement for an applicant to demonstrate how they
would be able to use any information provided."

Thank you for your time.

Yours sincerely,

Julian Todd

Dear Nathan Hill,

I look forward to receiving as much as you can disclose of the
codes for the 12 economic computer models (at least 9 of which are
in Excel VBA, a platform I certainly have access to) that are used
as the basis for national environmental decision making.

I am sure you are aware that software Intellectual Property rights
which you say belong to the contractor are fully protected under
copyright law no matter what form they are published in. They
cannot be copied and reused by third parties without prior written
permission, therefore their economic interests are fully protected.

I hope that for the software models that you do not disclose you
are able to provide satisfactory evidence that the copyright of
them is indeed held by the contractor, rather than the client, as
is normally the case in industry, and that these products are in
fact advertised elsewhere for sale to other clients.

There is little use in me running the models with my own sets of
data, because the outputs will be deemed invalid. The purpose of
disclosure is to enable public participation in this
decision-making, which at the moment can be no deeper than to face
the results of these computer models.

I already know that the models are run too in frequently to take
account of the timing of, for example, the world-wide recession.
There may be other short-comings in these models, only visible
through inspection of the code, which a client that has a
presumption in favour of aviation growth is liable to overlook.

Yours sincerely,

Julian Todd

Dear Nathan Hill,

Thank you for sending me copies of the DfT's air transport software
models on 21 January. In your letter you stated:

"If we adopt a wide interpretation of your request to include, in
addition to the aviation models, the code for the software model
used to extract user surface transport access to airport costs...
[as it is a] proprietary package that has been used numerous times
by Scott Wilson for many different applications and clients."

Referring back to the responses to my initial request of 23
February 2010 I note the existence of the National Air Passenger
Allocation Model (NAAM) which is written in: "SATURN, MINITRAMP or
EMME/2 for skims; then bespoke FORTRAN77 then Excel VBA
interfaces"[1]

Is this the model that relates to the user surface transport access
to airport costs? I can tell I have not been sent it as there is no
FORTRAN77 code in any of the packages that have been emailed to me.

Can you outline what other clients and applications this package
has been used for by Scott Wilson? For example, is it used to plan
the location of, in particular, airports in other countries, or
does it have a use outside this sector?

One final comment: My inquiry was made on the basis of the DfT
publication: "UK Air passenger demand and CO2 forecasts 2009".
Nowhere does it indicate in those pages that these calculations are
not the produced by or owned by the government.

[1]
http://www.whatdotheyknow.com/request/29...

Yours sincerely,

Julian Todd

Dear Peter Parr,

Thank you for explaining the provenance of the User surface access
model -- as being derived from Wootten Jeffreys software bought by
WS Aktins plc in 1994 and provided a secondary interface and
transport model network by Scott Wilson in 1998-1999.

This provides me with the first confirmation of a boundary between
the Department of Transport and the private sector
aviation/construction industries.

Before I take up the offer of purchasing elements of this software
from WS Atkins, can you provide me with a name and description of
this product and what computer equipment I will require to run it?

Also, the name "User surface access model" suggests that this is a
tool for the aviation industry (for all other sectors "surface" is
definitely the default and would not be mentioned in this way).

Can you confirm that Scott Wilson has provided use of this
secondary interface and transport model network to more clients
than it has the DfT's NAPAlM/Spasm software? If the interface is
provided on the same terms as the the rest of the software suite,
then it should be disclosed in the same manner.

Thank you for consideration of these questions in helping me get to
the bottom of this. As Mr Hill explained in his letter on 14
January 2011,[1] there is an extremely large gap in the DfT's
modelling capabilities. By publicising the technical steps and gaps
in what the Department is or is not able to do for itself, it
cannot but help create an environment for alternative suppliers to
come forward.

[1]
http://www.whatdotheyknow.com/request/29...

Yours sincerely,

Julian Todd