Serious crime in the United States soared to alarming heights beginning in the 1960s, but began leveling off in the 1980s and has declined by one-third during the 1990s. Every category of violent crime has decreased since 1993. Last year, serious crime reported to the police was only 5 percent above the rates for 1970, and in many cities across the country, it matched the crime rates of the 1960s.

A major reason for this reduction in crime is that crime has become more costly to the perpetrators. The likelihood of going to prison for committing any type of major crime has increased substantially. Since 1993:

• The murder rate has dropped 30 percent, as the probability of going to prison for murder has risen 53 percent.

• Rape has decreased 14 percent, as the probability of imprisonment has increased 12 percent.

• Robbery has decreased 29 percent, as the probability of imprisonment has increased 28 percent.

• Aggravated assault has decreased 14 percent, as the probability of imprisonment has increased 27 percent.

• Burglary has decreased 18 percent, as the probability of imprisonment has increased 14 percent.

Moreover, once in prison, criminals are staying there longer. The median prison sentence served has risen for every category of serious crime except aggravated assault.

The best overall measure of the potential cost to a criminal of committing crimes is "expected punishment." Roughly speaking, expected punishment is the number of days in prison a criminal can expect to serve for committing a crime. It is determined by the probabilities of being apprehended, prosecuted, convicted and sentenced, and the median sentence for each crime. Even today, it's amazing how low expected punishment is.

• For every murder committed, someone spends only 37 months in prison.

• Expected punishment for rape is only 119 prison days, for robbery 52 days, for serious assault 13 days and for burglary 8 days.

• For every motor vehicle stolen, someone spends only two days in prison.

Nonetheless, expected punishment is significantly greater than it was in 1980 for every category of serious crime.

• Between 1980 and 1996, expected punishment more than doubled for murder and nearly tripled for rape.
• It doubled for burglary and nearly did so for larceny/theft and auto theft.

Evidence shows that potential criminals respond to incentives. Crime increases when expected punishment declines, and vice versa. Between 1950 and 1980, expected punishment declined more-or-less continuously from an average of seven weeks for every serious crime committed to only 10 days - an 80 percent drop. In response, the serious crime rate more than quadrupled during those years. In the 1980s, expected punishment began to increase, accompanied by the leveling off and then a decline in the serious crime rate. Between 1980 and 1996, expected punishment for serious crimes increased from 10.1 to 21.7 prison days, a 115 percent increase, and serious crime declined. The experience of our two most populous states - California and Texas - confirms the negative association between crime and expected punishment.

• During the 1980s, California increased its prison population at a rate faster than the nation and experienced a decline in serious crime relative to that of the nation.

• Texas, meanwhile, lagged in the growth of its prison population and its rate of serious crime shot up relative to that of the nation.

• The opposite has occurred during the 1990s, as Texas has enjoyed a 33 percent decline in serious crime while sharply increasing its prison population to the highest rate in the nation.

• By contrast, the growth in California's prison population has leveled off and now trails the national average, and California consequently is making only modest progress against serious crime.

If we are to succeed in achieving an even lower crime rate, we must continue to make crime less profitable by further increasing expected punishment. To achieve that goal there are several options. Expected punishment will increase as we:

• increase the proportion of reported crimes cleared by arrest,

• increase the proportion of the accused who are prosecuted,

• increase the proportion of those prosecuted who are convicted,

• increase the fraction of those convicted who are sentenced to prison, and

• increase the average prison time served.

All these options are expensive in the short run. A higher arrest rate requires more money for police staffing, equipment and procedures. Higher conviction and sentencing rates require more resources for prosecution and criminal courts. All three require more prison space. But a tough approach pays, especially over the long run. As the odds worsen for criminals, crimes decline and the same numbers of arrests and convictions begin to reduce the odds favoring criminals.

Although the cost of building and maintaining more prisons is high, the cost of not doing so appears to be higher. One study found that each additional prisoner incarcerated reduces the number of crimes by approximately 15 per year, and yields a social benefit of at least $53,900 annually. Thus, even at $25,000 a year, the cost of keeping the average criminal in prison is worthwhile.