Exponential Decay Estimation

[Christian Rumpelnik]

Dear Statalist community,

I am trying to replicate a study by Da/Huang/Jin (2018) with an alternate dataset and have some problems implementing
a exponential decay specification of a least squares regression in STATA.

Specifically, the regression I am trying to implement is the following regression, where Ri, t-s are lagged past returns

I know that I have to use the nl command to implent that, but i am having trouble coming up with a code that works. Would be very grateful for some
suggestions!!

Thanks in advance
Chris

[Phil Bromiley]

You didn't get a quick answer. You'll increase your chances of a helpful answer by following the FAQ on asking questions - provide Stata code in code delimiters, readable Stata output, and sample data using dataex. Telling us you couldn't get it to work doesn't help much - we don't know what was going wrong. Also, recognize that most of us are not from your area - explain your problem in terms everyone can understand. Finally, we generally do not like screen shots - often they are hard to read and can't be worked with.

This looks like it is quite doable assuming n is relatively small (if n were very large, the nl statement could get very large and hard to handle). This runs, but b2 and b3 are not identified. I'm not sure how the model is identified - can't you can change the lambda2's to compensate for a change in lamba1?.

clear
set obs 100000
g x1=rnormal()
g x2=rnormal()
g y=2 + .1*.2/.6 * x1 + .1*.4/.6 * x1+ 5 * rnormal()
nl (y = {b0} + {b1} * ( ( ({b2}*x1)/({b2}+{b3}) ) +( ({b3}*x1)/({b2}+{b3}) ) ) ) ///
,variables(x1) initial( b0 1 b1 .1 b2 .2 b3 .2)