3 While all of these explanations seem to have merit, there is one dominant reason why the percentage of GDP and employment dedicated to services has continued to increase: low productivity. According to Baumol's cost disease hypothesis (Baumol, Blackman, and Wolff 1991), the growth in services is actually an illusion. The fact is that service-sector productivity is improving slower than that of manufacturing and thus, it seems as if we are consuming more services in nominal terms. However, in real terms, we are consuming slightly less services. That is, the increase in the service sector is caused by low productivity relative to manufacturing. The implication of Baumol's cost disease is the following. Assuming historical productivity increases for manufacturing, agriCUlture, education and health care, Baumol (1992) shows that the U. S. can triple its output in all sectors within 50 years. However, due to the higher productivity level for manufacturing and agriculture, it will take substantially more employment in services to achieve this increase in output. To put this argument in perspective, simply roll back the clock 100 years or so and replace the words manufacturing with agriculture, and services with manufacturing. The phenomenal growth in agricultural productivity versus manufacturing caused the employment levels in agriculture in the U. S. to decrease rapidly while producing a truly unbelievable amount of food. It is the low productivity of services that is the real culprit in its growth of GDP and employment share.