Bad code is not written on a bet. It's written because either competent engineers crunch, or because they are incompetent.

Good code could be costly, yes, but often is not. In fact it's very often economically sound to periodically address tech debt i.e. not allow it to accumulate because it seems to increase exponentially until one day a critical customer-acquisition-stakes feature cannot be shipped on time due to it. Only then do the executives wake up and even then 80% of the time they just blame the engineers and move on. They are never at fault, the angels.

Finally, many companies are already on the market and are relatively OK economically. Let's be honest: most deadlines are entirely artificial and are just power moves by management; they are not mandated by critical business needs.

In my experience a lot of bad was bad even when it was written. And that writing good code is often cheaper (within reason, perfection is bad) over any project bigger than a couple of months. The payoff from having good code shows up very quickly.