Expense Guidelines

[Jennifer Bell]

Part of what I think this project should accomplish is to show that
most (if not all) government spending is within pre-approved
guidelines. I've just started looking into what those guidelines
actually are, and how they can be quantified, and it turns out that
this idea isn't as straight forward as I'd hoped.

While allowed meal expenses are extremely (!!) well documented,
flights and accomodation guidelines seem basically to be:

- fly economy and try to find a low price
- stay in a hotel from our list

I was thinking that it would be interesting to find a data set of
historical airline and accomodation prices, and use that as a
comparison. While there's an American company that's actually has a
sweet business model based on mining that sort of information:

http://farecast.live.com/about/ourTechnology.do

I haven't yet found a good source for Canada.

Does anyone have any creative ideas on how travel expenses can be
constructively benchmarked?

Jennifer

[David Akin]

Not to be an apologist for the government or anything like that but,
from conversations with various exempt political staff, here's some
things to consider in this issue:

For airfare: Bookings are usually done within a day or two of travel.
As anyone knows who has done this, the fare when you book so late
tends to be a premium one. It's difficult to get around this,
political staff say, as this is just the way government works. (You
can't for example, "plan" for a crisis that you'll be sent to in
three months.) So if you were comparing, you may need to compare
against economy fares purchased within a day or two of travel.
Secondly: Political staff are required to travel economy but a
business class ticket is permitted if the "gate-to-gate" travel time
is going to be eight hours or more. That "gate-to-gate" thing is
important as that includes stopovers between connecting flights. There
are almost no flights of eights hours "gate-to-gate" for travel within
Canada, staffers tell me, but there could be situations where you
could be travelling from an eastern Canadian feeder airport to a
secondary airport in, say, the Western U.S. that could be eight hours
gate-to-gate. Travels to certain European destinations -- to get to
Geneva, where a lot of trade talks happen, from Ottawa, you generally
connect through Frankfurt -- would also likely be eight hours gate to
gate. But here again, bookings are often made within a few days of
travel, hence the premium air fare.

--
David Akin
-------------------
http://www.davidakin.com

[Jennifer Bell]

Thanks. Even so, a good number of the airfares seem strangely high.
I tell you, this data set is getting more interesting by the day.

Someone offline suggested doing benchmarking based on distance-
travelled calculations. I think this could actually be pulled off by
getting the geocodes for the cities mentioned in the travel expense
'destinations' field + doing some calculations -- which seems kind of
fun. Is it worth it? Are there other approaches?

If I gave someone the data, would anyone like to take a crack at doing
the geocode mapping code? Though, testing would have to be done on a
subset of the db as I'd like to keep things quiet for now until the
big unveiling.

Basically, I imagine 1) a db of the geocodes of ~100 cdn cities (can
get this online). 2) a best-effort algorithm to:

- take the destination string, apply some transformations,
- look up geocodes
- associate the geocodes and order via some database structure for
easy recall
- use this to calculate distance travelled.

It would be nice to have destinations geo-coded, as that would enable
all sorts of other visualizations too. Time frame would be in the
next 2-3 weeks, but the sooner the better.

Jennifer

[Jennifer Bell]

FYI: I posted this question to the 'Political Parsers' newsgroup, and
it turns out getting geocodes will be easier than I thought:

http://groups.google.com/group/poliparse/browse_thread/thread/c83660387f482a6c

However, I'm re-thinking the idea that this will produce useful
'distance traveled' metrics, since I'm not sure a departure from
Ottawa can always be assumed. At least, it will be nice to have some
of the data geocoded for maps, etc.

Jennifer