Tim,

Budget concerns are usually a concern among modelers regardless of
scale. What must be done is to try to whittle down (a.k.a. selectively
compress) the number of freight cars which would represent a reasonable
sample of those that operated in your chosen prototype and era.

Other than the prototype, AT&SF, and era, 1944-53, you did not provide
the list of what you are trying. For instance, the Santa Fe had
different freight car demands in California than they did in Kansas. Are
you trying to model a Main Line (where?) or Branch Line (where?).

You seem to be restricting yourself only to AT&SF equipment, but only a
percentage of the freight cars on the AT&SF lines were owned by the
AT&SF - the percent of the Daily Average of Freight Cars on AT&SF which
were owned by the Santa Fe were:

1944 46.7%
1945 43.2%
1946 48.6%
1947 46.1%
1948 53.2%
1949 62.3%
1950 58.7%
1951 55.3%
1952 58.9%
1953 59.3%

This means that if you use the system average, between 37.8% and 56.8%
of your freight cars should be foreign-owned.

What should be the split by car type (box, hopper, gon, tank, etc.)
would be dependent somewhat on the locale being modeled as well as type
of operation (branchline, secondary through line or main line). A main
line will introduce the car demands of many branchlines.

If you model a branch line, the type of cars on that line will be
governed by the lineside businesses served, team tracks and LCL
merchandise service. The ownership of each type of car will vary between
car types - for instance, the percent of AT&SF boxcars on the AT&SF will
be much less than the percent of AT&SF gondolas or hoppers on line.

I am sure that by analyzing what cars you will need, you can reduce the
cost while maintaining the accuracy you want.

Hope this helps, Tim Gilbert