Everyone knows that the United States has been experiencing an investment boom, driven by expenditures on information technology equipment. Looking at real GDP (that is, removing the effect of price changes), business investment in structures, equipment, and software amounted to 14.1 percent of GDP in 1999, up from 9.6 percent in 1990 and far above its share throughout the 1970s and 1980s. § However, if one looks at nominal expenditures - that is, not adjusting for price changes - business investment amounted to 12.9 percent of GDP in 1999, up from 10.9 percent in 1990 but no higher than the share of GDP that was devoted to business investment from 1979 through 1982.

More generally, business investment in the 1990s looks significantly more impressive when one adjusts for price changes than when one does not.

What accounts for the difference? The primary explanation is falling computer prices. Computer prices have been declining, and declining rapidly. Computers are also much better than they used to be; you probably could not buy what was available ten years ago even if you wanted to do so. Consequently, price changes for computers are based on estimates of the prices of key attributes, namely speed and memory. As computers get faster, the estimated price comes down, even if the price of the physical machine does not. Thus, converting the growth in nominal expenditures to real, by adjusting for price changes - a process known as "deflation” - results in a much higher rate of growth in real output. While nominal investment in computers and peripherals rose roughly 10 percent per year between 1995 and 1999, real investment rose 45 percent per year.

Which is the better measure, real or nominal? If one is looking for possible explanations for the recent pickup in productivity growth in the United States, real investment in computers is more informative. After all, businesses investing in computers are enjoying huge increases in speed and memory, even if they are not spending that much more.

REAL business investment is spending on equipment, structures, and software measured in constant dollars (usually 1996)

NOMINAL business investment is measured in current dollars

However, if one is concerned about businesses" ability to finance their investments, nominal expenditures is probably the more relevant measure. It is the dollar outlays needed to acquire the rapidly growing capability that determine whether firms can finance their investment programs internally or must look to external funds. Although capability, at least as represented by memory and speed, has grown by leaps and bounds, spending has been more in line with profits and cash flow.

Similarly, computers make up a sizable fraction of U.S. exports and an even larger fraction of U.S. imports. Thus, rapidly declining computer prices boost real growth rates for both exports and imports as compared to nominal growth - and raise real import growth more. This is the right way to look at a number of issues. But if one is concerned about financing the trade deficit, it is nominal expenditures that matter. Between 1995 and 1999, the gap between imports and exports, in real terms, grew from 1 percent of GDP to 3.6 percent; declining prices for computers, peripherals, and parts accounted for 0.5 percentage points of this increase.

Real output figures can also give a distorted sense of the health of the computer industry and the impetus to aggregate demand that investment in computers provides. Based on growth in real spending, one might assume that computer firms have been doing well, with employment growing vigorously. However, employment in the manufacture of computers was only 5 percent higher in 1999 than in 1995. Over the same period, employment in motor vehicles and parts manufacturing also rose 5 percent, while real motor vehicle output grew about 5 percent per year. Not a bad performance, but only about one-tenth of the real growth rate in computers.

Note: Information processing equipment is included in equipment, structures, and software
Source: U.S. Bureau of Economic Analysis

Growth in real output that is associated with rapidly declining prices should be viewed somewhat differently from gains that result from producing more units at a constant or rising price. While rapid declines in computer prices reflect tremendous technological strides, they also are indicative of intense competition that results in these improvements being passed on to customers. Ironically, an easing in competitive pressures could result in a somewhat slower rate of price decline and, thus, somewhat slower growth in estimates of computer output. Yet, workers and shareholders in the industry might be better off, at least in the short term.

As advances in computer technology permit huge increases in speed and memory with much more modest increases in dollar expenditures, it is important that these increases in capability be captured in our measures of output growth. But it is also important that we not look at this "new economy” phenomenon with a mind-set shaped by an old economy in which prices almost always increased.