Good morning buddy.
You are working so hard, aren't you?

For your Q1, and Q2,
The benefit of a stock dividend is optiion. The shareholder can either keep the shares and hope that the company will be able to use the money not paid out in a cash dividend to earn a better rate of return, or the shareholder could also sell some of the new shares to create his or her own cash dividend. The biggest benefit of a stock dividend is that shareholders do not generally have to pay taxes on the value.

Option here means you have the right to make a selection.

For the tax orientation, here is a default rule: the later the better.
Just like the definition of Present value, the earlier the better. Here for the tax, the outflow, you take opposite position.

For your Q3,
The objective of holding stock is to harvest dividend.
From this standing point of view, permanent non-dividend policy will put your stock on a 0 value position.
No one will actually invest in your entity. You may insisted that your entity will create far more value in the far far future. Yes, you may be right, but no one believe you. I am talking about the practice in real world.

For your Q4,
You couldn't contiue adopting the model simply because it will come up with a negative number, which has no meaning in the real world. What are you going to do?

You may shift to EBITA model to figure out your final evluation.

I hope this help.
Good luck. :)

感谢回答！最后一个问题还是有点不太懂……
我的问题在于，当k<g,这个模型算出来的是负数，所以这个模型就不能再用了，这没错。

可是，一旦红利真的以一个固定比率增长，从这个模型的推导过程来看，这个模型没有任何问题，
看不出模型有任何方面的限制。那么股票的价格来自于未来红利现金流的折现，这是一个公比＞1的等比数列，
那么现在的股票价值应该是无穷大吗？

Thank you buddy for your credits.

In the scenario you refereed, the model is not applicable any more.
You have to adopt multi-stage DDM or the other model I mentioned before.

I don't think the value of stock will go that far both in theory and in practice.
All in all, the model is subject to the axiom "garbage in, garbage out", meaning that a model is only as good as the assumptions it is based upon.

I hope this help. :)

谢谢，弱弱地问一句，当k<g时，股票价格是无穷大吗？

谢谢！！

My pleasure buddy.
You got my attention again.  :)

There are two obstacles on your path.

First, the assumption to keep g beyond k is not applicable in a long run,  normally a temporary maneuver;
second, you need to sustain your model viable for a pretty long period.

Neither of the assumptions is applicable both in theory and practice.

I wish this help. :)