Suppose there were a law that forced you to pay a government agency for apples you were supposed to feed your children.

The government didn't care if you grew your own apples or if your neighbor grew apples you liked better than the government's brand -- the law compelled you to pay for the state's product whether you wanted it for your children or not.

Now, suppose many people who actually fed their children public apples discovered something wrong with them. Some apples were bitter, others mushy and others rotten to the core.

When they complained to the public-apple agencies, agency bureaucrats and their union would say: "Excuse me, the bad apples are not our fault. You need to give us more money so we can build better apple storage facilities, and so we can pay better wages to apple handlers."

So the government forced everybody to pay more for its apples.

Now, the public-apple agencies built beautiful new apple storage facilities. They paid their apple handlers handsomely. Still, a disturbing number of apples remained bitter, mushy or rotten to the core.

In the face of new complaints, the bureaucrats and their union declared, "We need a federal Department of Apples."

Conservatives fruitlessly argued that the Constitution does not authorize a federal Department of Apples. Congress created one anyway. The new DOA spent vast sums paying its own bureaucrats and subsidizing local government apple agencies. Still, many public apples remained bitter, mushy or rotten to the core.

A "compassionate conservative" -- N.B. a "big government conservative" -- was elected president. He advocated giving even more federal aid to local public-apple agencies in exchange for a federal "apple accountability" program. Under the program, states were required to test their apples every year, with the goal that after 13 years every public apple would be good enough to eat.

After several years, the tests showed almost no improvement in public apples. Apple agency bureaucrats and their union representatives complained that the apple-accountability standards were unrealistic. So the secretary of apples relaxed the standards, and the compassionate conservative president called on Congress to reauthorize the program.

The public apple in this parable, of course, is public education -- which is indeed rotten in many places.

If there is one thing the Department of Education does well, it is collect statistics about schools. According to its National Center for Education Statistics, Americans in recent decades paid for a massive increase in spending on government schools. Between the 1970 and 2002 school years, average per-pupil spending in public elementary and secondary schools rose 111 percent, from $4,170 (in constant 2001-2002 dollars) to $8,802.

From just 1990 to 2003, average per-pupil spending increased 25 percent, from $7,692 (in constant 2003-2004 dollars) to 9,644.

This big run-up in spending did not cause a big run-up in student performance.

Since the early 1990s, NCES has periodically administered National Assessment of Educational Progress tests to a sampling of elementary school students. The tests are graded on a scale of zero to 500, and students are anonymously assigned an achievement level of "below basic," "basic," "proficient" or "advanced." "Basic" means the student had only a "partial mastery" of the subject appropriate for the grade level.

NAEP reading scores for eighth-grade public school students remained essentially static between 1998 and 2005. In 1998, eighth-graders averaged a score of 261 out of 500 in reading. In 2005, they averaged 260. Only 29 percent were rated grade-level "proficient" or better.

In other words, 71 percent rated less than proficient in reading.

Math results were a little better. Between 1990 and 2005, the average eighth-grade score rose from 262 to 278. Again, only 29 percent were rated grade-level proficient or better.

In other words, 71 percent rated less than proficient in math.

Private schools did better. The 2005 NAEP tests rated students in Catholic and Lutheran schools. Forty-nine percent of eighth-graders in both rated "proficient" or better in reading. Forty-four percent of eighth-graders in Lutheran schools, and 40 percent in Catholic schools, rated "proficient" or better in math.

Increasing per pupil spending by another 111 percent -- whether it is done by compassionate conservatives in Washington, D.C., or plain old liberals in your home state -- will not fix public schools.

It's time to give all American parents vouchers equal to the per-pupil spending in local government schools. Then parents can decide whether the government schools deserve their children -- or whether they will try the apples elsewhere, thank you.