Hi,
I am running a nested logit model as in Huang and Ritter (2009) : Testing theories of capital structure and estimating the speed of adjustment.
I have a few problems, I am using panel data, I have just taken the panel for 1 year as of now.
1. I have the same set of alternatives for both levels.( first level is whether to issue and if yes, whether it should be debt or equity). I use nlogitrum (Florian Heiss,2002) and am giving only the alternatives
for the lowest level. My tree is as below :
. nlogitgen type = subchoice(de : 1|2, no: 0)
new variable type is generated with 2 groups
label list lb_type
lb_type:
1 de
2 no
. nlogittree subchoice type
tree structure specified for the nested logit model
type N subchoice N
----------------------------
de 536 --- 1 402
+- 2 134
no 769 --- 0 769
----------------------------
total 1305
N = number of observations at each level
nlogitrum issue l.x1 l.x2 l.x3, group(cocode) nests(subchoice type)
factor variables and time-series operators not allowed
I need to include lagged operators her , which I am not able to do. Issue decision at time t is decided by firm-specific variables(x1, x2, x3) at time t-1. How to do this?
2. If I run the code below ,removing the lagged operators , I get "unbalanced data" as error.
. nlogitrum issue x1 x2 x3, group(cocode) nests(subchoice type)
tree structure specified for the nested logit model
type N subchoice N
----------------------------
de 536 --- 1 402
+- 2 134
no 769 --- 0 769
----------------------------
total 1305
N = number of observations at each level
unbalanced data
r(459);
Is this because of the degenerate case? if option is not to issue, then the tree has no subsequent branches.. Should I include all the firm variables for all the 3 options and specify a "mode" variable that takes a value 1 for the choice and zero elsewhere? (As in each firm-year variables will be repeated with the three options (no, debt and equity) and the mode variable will be zero for all except the one option the firm chooses that year) The faq here : http://www.stata.com/support/faqs/st...logit-stata-8/ says that unbalanced panels work in versions above Stata 9. Mine is Stata 12. Please help me figure this out...
Thank you for your help,
Suja
I am running a nested logit model as in Huang and Ritter (2009) : Testing theories of capital structure and estimating the speed of adjustment.
I have a few problems, I am using panel data, I have just taken the panel for 1 year as of now.
1. I have the same set of alternatives for both levels.( first level is whether to issue and if yes, whether it should be debt or equity). I use nlogitrum (Florian Heiss,2002) and am giving only the alternatives
for the lowest level. My tree is as below :
. nlogitgen type = subchoice(de : 1|2, no: 0)
new variable type is generated with 2 groups
label list lb_type
lb_type:
1 de
2 no
. nlogittree subchoice type
tree structure specified for the nested logit model
type N subchoice N
----------------------------
de 536 --- 1 402
+- 2 134
no 769 --- 0 769
----------------------------
total 1305
N = number of observations at each level
nlogitrum issue l.x1 l.x2 l.x3, group(cocode) nests(subchoice type)
factor variables and time-series operators not allowed
I need to include lagged operators her , which I am not able to do. Issue decision at time t is decided by firm-specific variables(x1, x2, x3) at time t-1. How to do this?
2. If I run the code below ,removing the lagged operators , I get "unbalanced data" as error.
. nlogitrum issue x1 x2 x3, group(cocode) nests(subchoice type)
tree structure specified for the nested logit model
type N subchoice N
----------------------------
de 536 --- 1 402
+- 2 134
no 769 --- 0 769
----------------------------
total 1305
N = number of observations at each level
unbalanced data
r(459);
Is this because of the degenerate case? if option is not to issue, then the tree has no subsequent branches.. Should I include all the firm variables for all the 3 options and specify a "mode" variable that takes a value 1 for the choice and zero elsewhere? (As in each firm-year variables will be repeated with the three options (no, debt and equity) and the mode variable will be zero for all except the one option the firm chooses that year) The faq here : http://www.stata.com/support/faqs/st...logit-stata-8/ says that unbalanced panels work in versions above Stata 9. Mine is Stata 12. Please help me figure this out...
Thank you for your help,
Suja
