I recall reading somewhere a few years ago, that it would actually be less expensive for the state to purchase and replace these targeted, older cars that pollute so highly rather than subject hundreds of thousands of non-polluting cars to expensive yearly tests. But, since I cannot lay my hands upon the study, just take that as an amusing anecdote.
The article I linked to above points out the huge cost savings to the public if we were to implement real-time testing on freeway off ramps or busy intersections, ticketing drivers who were out of compliance (with a grace period to get the car running cleanly before any fines were assessed) in lieu of testing all cars every year when 90%+ of the cars run cleanly.
In my State of Utah, it is the vehicle owner who pays for the testing, not the State. If we accept the claim from article I quoted that 5% of cars cause 55% of the pollution then what is the cost of replacing those 5%?
Assume 1 million vehicles get tested every year (a simplification since we've now moved to testing new cars every other year, and older cars annually, but with a total number of cars at something like 1.2 in 2009..) at something like $25 a pop, that is some $25 million annually spent by auto owners each year. At $15k for a newish, averagish car, that is 1666 cars that could be purchased each year if the $25 million were taken in taxes to buy new cars outright. You could triple that to about 5,000 cars by offering a $5,000 incentive rather than providing 100% of the purchase cost. That works out to about 1/2%. So it would take 10 years worth of testing fees to replace the 5% of dirty cars creating over half the pollution.
Of course, we could reduce the cost to individuals to zero simply not requiring them to test their car every year. Most newer cars self monitor and the "testing" entails having the mechanic's computer as the car's computer if emission levels are correct. When the car says, "Yes", the state is satisfied. (And thus VW gets away with providing customers higher performance by cheating on their emissions results.)
Issue tickets to those whose cars emit too much as they drive down the road past random testing points (no need to stop) and they get to either repair their cars to run cleanly, or to get whatever penalty is deemed appropriate. Drivers could take their chances on getting caught, or have their car tested at whatever interval made sense to them. Which for most modern cars would probably be about the time the "check engine" light came on.
Of course, auto emissions are a bigger problem in some areas than in others. Along the urban areas of Utah (the "Wasatch Front") pollution is a major concern as we sit in a bowl subject to winter temperature inversions in which cold air is trapped in the valley under warmer air above it. Pollution builds until a storm comes along to blow it out. Highly urban areas like Paris and LA have problems just from the volume of traffic. But in areas where air quality is not a problem, spending money to make cars run at their optimal cleanness may not be the best use of limited resources.
Charles