You are, of course, quite right that this is an example of people spending money to acheive a specific result - using patent applications as a way of showing results from R&D. The problem is that these patents very often have little value and are simply being made to reach a numerical target set by management - they are not a real measure of the quality of R&D being done by the company. Income from patent licensing is a far better measure, but some companies prefer not to license their technology. Even the patents which might be of value are very often extremely poorly drafted, and companies are unwilling to spend the money to have the quality of their US patents improved through re-issuing.

In fact, the vast volume of patent aplications being made by Chinese companies in the US is much more a measure of how easy it is to gain a patent in the United States nowadays. I know companies with global interests which make upwards of two-three thousand applications per year to the US but less than two hundred in Europe, Korea and Japan put together. It is when you examine the extrememly un-novel and obvious nature of the applications being made, and how very few of the claims survive examination at the USPTO (even given the rather lax standards applied there) that you understand that these applications are much more a PR exercise aimed at potential investors than anything else.

And how many of these "innovations" were lifted from foreign companies in China.

All companies engage in 'writing around' their competitor's patents, this does not mean they automatically have the right to use the invention, as the invention may incorporate technology that has already been patented by a competitor.

Most licensing agreements leave order-export manufacturers free to patent any improvements they might make on the technology which is licensed to them - and so they do, there is nothing wrong with this.