Short run elasticity using ARDL/ECM

[Prabuddha Prakash UTK]

I want to clearly understand how to get the short-run price elasticity of demand using the ARDL setup.


My goal is to estimate the short-run price elasticity of demand. Like SHORT RUN ELASTICITY.



So, I started with ARDL and obviously checked for cointegration using the ec version of the ARDL command. I didn't find any cointegration (the null is accepted), so what do I do now?



Option 1: Do I reestimate my ARDL model with first-differenced variables and no error correction term?



Or, Option 2: I do not re-run the ARDL as in Option 1 above, and just take the coefficient on the ln_p term in the original ARDL command as the short-run price elasticity of demand?

Which one is the correct option when cointegration is not present?

[Sebastian Kripfganz]

If there is no evidence of a long-run (cointegrating) relationship, then both Option 1 and 2 are valid. Option 1 is more efficient - i.e., should tend to produce smaller standard errors - if there is indeed no long-run relationship. Option 2 is more robust, in case there actually is a long-run relationship but it was not detected by the test.