CIUDAD JUAREZ, MEXICO — Bill Randag likes to show visitors to his office the letter he got from a Chinese businessman offering to set up a quick, inexpensive, and accurate data-processing shop in a southern Chinese city.
"I really love the way he signs off," says the senior vice president for Datamark Inc., a pioneer in taking a growing slice of the huge volume of US data processing to Mexico. The last word before the signature "Tanks" - as in "thank you."
Mr. Randag is well aware of the pitfalls that can trip up a data-processing operation in a foreign country - including inadequate language skills. But he is also witness to the factors that are making Mexico the destination of a growing number of data-entry and document-scanning companies looking outside the United States.
Cheap labor costs, a strong work ethic, good telecommunications connections, and proximity to the US are all reasons Mexico is now home to more than 15,000 data-processing workers for US companies - and the number is growing.
"There are other low-cost labor markets," says Leon Satero, president of Newport Beach Data Entry, a California company with 1,500 Mexican employees. "India, the Caribbean, the Philippines, are places you typically hear about, but the work ethic in Mexico is ... a big draw."
Mexico's numbers are still small in comparison to the total industry in the US. Every day more than 700,000 Americans spend their workday keyboarding data into computer files, according to the Bureau of Labor Statistics.
But as companies look for ways to cut costs, "outsourcing" - and "offshoring" of data-entry work continues to grow.
Mexico's attractions for data processors mirror those cited for two decades by US businesses from coupon sorters to parts manufacturers.
But, unlike coupon sorters, data-processing workers often need more developed skills. And companies like Datamark are finding that more Mexican workers now have the skills and basic knowledge of English to do the job.
"Your employees don't have to speak English to key in information ... but they have to have some familiarity," says Randag. "Some worker in India might not recognize 'Connecticut' on a waybill, but in Mexico they do."
For some years data-processing jobs expanded in countries like Ireland and India: One reason was to allow for overnight information entry and processing by taking advantage of very different time zones.
But, as roadblocks developed - Ireland became more expensive, India turned out to be pretty far away for troubleshooting - interest in Mexico and English-speaking Caribbean countries, has grown.
Data processors punch information from express delivery orders into computers, verify checks, process claims, and transfer information from handwritten or typed documents into electronic files. They also verify credit-card accounts.
Ironically, one of the reasons more data processing jobs are transferring to Mexico from the US - a tight labor market - is also a reason US data processors are increasingly expanding away from Mexico's border region and farther into the interior.
For one thing, labor costs are even lower in Mexico's interior than on the border. The rapid growth of manufacturing in Mexican border cities like Ciudad Jurez and Tijuana is leading to a tighter labor market in those cities. And that rapid growth is also fueling problems - with roads, sewage, housing - that are limiting possibilities for expansion on the border, some company officials say.
Low unemployment on the border in Mexico also leads to high turnover, which can exasperate companies that have trained employees for specific tasks.
Some industry analysts claim it still takes about three Mexican data processors to do the work of one in the US. But the lower wages in Mexico - where the average employee might make $4 to $10 a day still make Mexico profitable. And companies with experience in Mexico don't agree with the productivity estimates.
"The language difference alone means you're never going to equal your American workers in Mexico," says Datamark's Randag, "but in our experience the gap doesn't even reach 2 for 1."