Conditional logit

[Gaurav Bhattacharya]

HI All,

I am facing a problem in running the conditional logit model. My data are on firm location choice which is dependent on firm specific characteristics which do not vary across alternatives and region specific characteristics which are firm invariant. However, the restaurant example in stata manual as well as the fishing mode example in Cameron and Trivedi show that costs and prices vary across alternatives as well as families/individuals respectively.

My problem is that the state specific characteristics do not vary across firms. For example, state gdp for a particular state is the same for all firms which locate to that state. The cross sectional data (hypothetical, meant for explanation only) in wide format are as follows:

Factory id	State chosen	fixed capital	State GDP growth
1	A	1200	6.5
2	B	1900	5.7
3	C	500	7.0
4	D	2900	7.5
5	D	650	7.5
6	A	1090	6.5

Now, state GDP growth is the same corresponding to factory 1 and 6 as well as factory 4 and 5 since each pair locates to state A and D respectively. However,stata manual shows that alternative specific characteristics (state GDP in this example) also varies across individuals (here factory id). In the Cameron and Trivedi example on fishing mode, explanatory variables like price varies across alternatives (beach, pier, etc.) as well as individuals.

I have run the conditional logit model but I could not get results as stata says there is collinearity.

Gaurav.

[Clyde Schechter]

I doubt anybody can help you without seeing the actual command you ran and the actual output that Stata gave you.

Please read the FAQ for good advice on how to post questions in ways that help those who want to help you and that are likely to get a useful and timely response.