[Linghui Han]

The market concentration levels in some industries dropped by a larger margin while some others rose by a larger margin and the rest remained roughly the same after this policy treatment. The growth rates in those provinces where industries experiencing larger rise in market concentration level are bigger is lower while the growth rates in their counterparts where industries experiencing larger drop in market concentration level are higher.

[Clyde Schechter]

Yes, it's suitable for that design, and your regression is not a problem, only your use of -margins-.

My personal preference for parallel trends is to simply use -collapse- to calculate mean outcomes in both study groups at each time point and then graph them. The graph ideally will look like two parallel curves. One limitation of this is that it does not account for the additional variables you included in your model. Another approach, more analytic, is to actually fit continuous time trends to the pre-intervention data in both groups and examine whether they have (nearly) equal slopes. You can include the other covariates in that analysis. Now, if the time trends are highly non-linear, this will be a poorly fitting model: so doing the graphical approach first will give you warning of non-linearity, and you can then deal with that by using either some non-linear transformation of the continuous time trend, or using splines.

[Linghui Han]

In some industries the market concentration level rises after 2002 while drops in some other industries and remain roughly the same in the rest. Growth rates in those provinces where industries experience rising market concentration level are bigger are slower than their counterparts where industries experiencing roughly no change in market concentration are bigger and more so than their counterparts where industries experiencing dropping market concentration level. Does this context fit for a DIDID analysis? If yes, how should it be done including checking parallel trend before 2002?

[Linghui Han]

Sorry, I did not see my earlier post and posted it again.

Thank you very much for the answer, Clyde! But I am little confused if the design should be DID or DIDID?

[Clyde Schechter]

Growth rates in those provinces where industries experience rising market concentration level are bigger are slower than their counterparts where industries experiencing roughly no change in market concentration are bigger and more so than their counterparts where industries experiencing dropping market concentration level.

I don't understand this sentence. It has questionable syntax, and to the extent I can parse it, it seems to contradict itself. Can you please clarify?

[Linghui Han]

Growth rates in those provinces where industries experiencing rising market concentration level are bigger are slower than their counterparts where industries roughly experiencing no change in market concentration level. Growth rates in those provinces where industries experiencing roughly no change in market concentration level are bigger are slower than their counterparts where industries experiencing dropping market concentration level are bigger.

[Linghui Han]

Growth rates in those provinces where industries experiencing rising market concentration level are bigger are slower than their counterparts where industries roughly experiencing no change in market concentration level are bigger. Growth rates in those provinces where industries experiencing roughly no change in market concentration level are bigger are slower than their counterparts where industries experiencing dropping market concentration level are bigger.

[Linghui Han]

Is #52 clear?

[Clyde Schechter]

Sorry, no. I still don't understand it. I can't figure out what things "are bigger" and what things "are slower" in these sentences.

[Linghui Han]

Bigger refers to the industry share in total provincial industrial output. Slower refers to the provincial GDP growth.

[Linghui Han]

Bigger refers to the industry share in total provincial industrial output. Slower refers to the provincial GDP growth rate.

[Linghui Han]

Growth rates are slower in those provinces whose biggest industries experience rising market concentration level than in those provinces whose biggest industries roughly experience no change in market concentration level. Growth rates are slower in those provinces whose biggest industries experience roughly no change in market concentration level than in those provinces whose biggest industries experience declining market concentration level.

[Linghui Han]

GDP growth rates are slower in those provinces whose biggest industries experience rising market concentration level than in those provinces whose biggest industries roughly experience no change in market concentration level. GDP growth rates are slower in those provinces whose biggest industries experience roughly no change in market concentration level than in those provinces whose biggest industries experience declining market concentration level.

Sorry for my language ability! Is this time is clearer?

[Clyde Schechter]

Sorry, still confused. Your original sentences don't mention industrial output or GDP. They refer to "growth rates," and "market concentration." I'm not an economist, and while I might guess that "growth rates" refers to "GDP growth" (which is not the same thing as GDP) I don't know where industrial output plays a role in your sentences, nor how it is represented as a variable in your -xtreg- command.

[Linghui Han]

The causal relationship I am looking for is the changes in market concentration levels in industries lead to the change in GDP growth rate at provincial level. However, industrial output will not be represented in the ultimate -xtreg- command, but it is used to classify the provinces into three categories:
Group 0: the provinces whose biggest industries do not experience changes in market concentration level;
Group 1: the provinces whose biggest industries experience declining market concentration level;
Group 2: the provinces whose biggest industries experience rising market concentration level.

Is this clear? I am wondering how I can make this design clearer and more straightforward.

Really appreciate your patience!