Capital Ideas

Improvement Goes Unnoticed

by By Jessamine Chan  |  Research by Christian Broda
Published: June 7, 2008

Capital Ideas

Image by Getty Images

Measures of real consumption growth in the United States are likely to be higher and more volatile than suggested by national statistics.


Product innovation is one of the primary forces behind economic prosperity. Computers, airbags, new medicines, and microwaves are among many product improvements that have benefited Americans over the last few decades. At the same time, one of the key ways to gauge a country’s economic growth is through measures of real consumption that are provided by such national statistics as the U.S. Consumer Price Index (CPI). Unfortunately, these measures typically ignore the welfare impact that new and better products have on societies. This important gap between theory and practice led associate professor of economics Christian Broda and David E. Weinstein of Columbia University to quantify the mismatch between standard measures of real consumption in the United States and Americans’ true standard of living.

In their study “Product Creation and Destruction: Evidence and Price Implications,” Broda and Weinstein found that ignoring the process through which new and higher-quality products replace outdated products in the calculation of real consumption understates the growth of Americans’ well-being by almost 1 percentage point per year. This “bias” results from the failure of the U.S. Bureau of Labor Statistics (BLS) to account for the quality upgrades inherent in the process of product replacement in their measurement of consumer prices.

Broda and Weinstein found that more and higher-quality products are introduced in booms as opposed to recessions. This led the authors to argue that Americans’ well-being is also more volatile than what is implied by national statistics. In particular, they calculated that real consumption in the 2001 recession fluctuated by 0.4 percentage points more than suggested by official numbers.

The Tricky Nature of Price Measurement

“Societies care about the variety and quality of the products they can buy,” Broda said. “If these product characteristics are partially ignored by official statistics, then we are understating the growth in Americans’ standard of living.”

This increase in quality has occurred for all types of consumer products. A glimpse at a simple product innovation over the past decade — mass-produced digital cameras — underscores this point.

The digital revolution has introduced cameras that far surpass the quality of those sold 10 years ago. However, when the BLS measures consumer prices, it attributes most of the higher price of new digital cameras to inflation rather than improved quality. This implies that even if consumers prefer the newer and sometimes more expensive cameras to their older counterparts, inflation will be positive and measures of real consumption will fall. This is at odds with the notion that higher-quality products raise overall well-being.

“There are literally millions of goods and services available in the United States,” Broda noted. “New products are introduced all the time, existing ones are improved, and others leave the market. For even the most sophisticated statistical office, it’s hard to keep up with the rapidly changing nature of the economy.”

Using advanced statistical models and a comprehensive data set, Broda and Weinstein estimated the impact that product innovation has on consumers’ well-being.

For the sectors included in the study, the authors used a far more inclusive database — the ACNielsen Homescan Database — than that used by the BLS. ACNielsen gives handheld scanners to approximately 55,000 households, covering a demographically balanced sample of households in 23 U.S. cities. When members of those households return home from shopping, they scan every barcode of the products they have purchased. Overall, the database covers about 40 percent of all spending on goods in the CPI.

“This database provides us with price and quantity data for the universe of barcode products purchased by Americans,” Broda said.

The results suggest that the Bureau of Labor Statistics overlooking quality upgrades leads to the Consumer Price Index overstating the inflation rate by around 0.8 percentage points per year. For instance, if the inflation rate in these sectors were announced at 2.5 percent, the true inflation rate corrected for quality improvement could be around 1.7 percent. Over prolonged periods of time and extrapolating the behavior observed in their sample to the entire CPI, these calculations indicate that the real consumption of Americans has grown substantially more than what is suggested by national statistics.

Such a revelation has the potential to affect public policy debates, including how the standard of living for the poor in the United States has changed over the past decades.

Quality Bias Varies Over Cycles

Broda and Weinstein also investigated the relationship between product innovation and business cycles. The work of legendary Austrian-American economist Joseph Schumpeter in the 1930s not only suggested that economic welfare depends heavily on the creation and destruction of products, but that product creation takes place largely independent of demand booms and busts.

“Schumpeter predicted that we should not see the number of products fluctuate with business cycles,” Broda said. “However, recent findings have confirmed what our intuition suggests. We see more and higher quality products being introduced in periods of expansion than during recessions. That finding is consistent with companies continually innovating, but holding back new products until there’s a demand for them.”

In addition, in periods when the economy is booming, little product destruction takes place. Most product destruction occurs during recessions.

Because product creation occurs more often in economic booms than in downturns, the bias in consumer prices that results from this product behavior also varies over the business cycle.

“It turns out that the bias is high in booms and low in recessions,” Broda said. “That implies our real consumption—whatever we consume divided by the price index—is more volatile than what the national figures suggest.”

Broda and Weinstein show that real consumption in the 2001 recession fluctuated from peak to trough by 0.4 percentage points more than suggested by national statistics. They suggest that considerably more research must be undertaken to measure the overall size of the bias in the CPI. That work will require capturing more and better data about prices and quantities of products in additional CPI sectors not covered in the authors’ sample.

Implications for Real Wages in America

The stagnation of real wages of poor Americans has been documented in a wide range of sources. According to a 2006 Congressional Budget Office publication, the real hourly wages of the lowest 10th percentile of income earners rose only by 0.2 percent since 1979. There has been relatively little attention focused on how nominal wages are converted into real wages.

In particular, the analysis has tended to assume that our measures of inflation correctly capture changes in the cost of living of Americans over time, and hence the inability of wages to keep up with prices has been used as prima facie evidence that the poor in America have not benefited from the last several decades of economic growth.

Broda and Weinstein’s findings, however, suggest that real wages should be calculated using a price index that grows around 1 percentage points slower than the official index. This means that real wages are growing faster than implied by national statistics.

“Using our adjusted measure of the cost of living in America to deflate nominal wages, we find that the wages of the poor in the last 26 years have increased by 30 percent,” Broda said

The reasons for this difference stem from the fact that poor households have benefitted from the introduction of goods such as color televisions and microwaves in the past few decades — goods that would have been unaffordable luxuries in the 1970s.

“Typically, we use BLS price indices to calculate the standard of living,” Broda said. “Our study suggests that the real consumption of the poor is substantially higher than what the BLS suggests, because the poor are able to take advantage of the higher-quality products in the marketplace. That is a fact missed by the BLS.”


Last Updated 5/14/09