February 2004
The Advantages of High Productivity Growth
U.S. productivity growth accelerated in the mid-1990s and, despite the recession in 2001, has surged ahead at even higher rates over the past two years. Measured as growth in output per hour worked, increases in productivity are essential for sustained economic growth and increases in wages and jobs during the long term, yet higher productivity might also reduce the rate of recovery in the labor market in the short term. This relationship is all the more relevant of late, with periods of high productivity and simultaneous concern over the labor market. A number of economic experts gathered at a February 25 AEI conference to discuss recent developments in U.S. productivity growth and its effects on the U.S. economy.
Is Productivity for Real?
Eric M. Engen
AEI
The economy's current productivity growth pattern has risen at an average annual rate of 3 percent-up from 5 percent in the mid-1970s. In the past two years output per hour has risen 5 percent. Today experts will examine these growth patterns and make predictions for the future.
Kathleen Utgoff
Commissioner of the Department of Labor's Bureau of Labor Statistics
The Bureau of Labor Statistics (BLS) productivity statistics are accurate measures of productivity trends in the United States and are not significantly affected by problems in the measurement of hours. To track productivity, data is compiled from the Bureau of Economic Analysis (BEA) and the BLS that trace the output data from the product side of national income accounts. For the data for total hours, the Current Employment Statistics Survey (CES) or payroll survey is used that consists of polling four hundred thousand establishments based on payroll records monitoring employment, hours, and wages. Another survey used is the Current Population Survey, or CPS, which surveys sixty thousand households. Lastly, data from the Employment Cost Index from the BLS national Compensation Survey is used.
Upon basing productivity projections on this data, a number of questions have arisen regarding the assumptions and calculations that drive the data. The first question revolves around the CPS over-slating hours. To counter this, we plan to use a new series based on a ratio of CPS to take care of the over-reporting that will use a CES base to be introduced in August of 2004 and additional data from the Time Use Survey.
The second question regarding the data is hours worked vs. hours paid, the discrepancy between the two, and how this might affect productivity rating. To get a more accurate reading on a worker's hours, the Cost Index that records benefits such as vacations, holidays, and sick leave is used. Also, the notion of off-the-books hours (hours worked and not recorded such as doing laptop work on a plane) is extremely hard to measure. So while it is clear people are more productive, it is not conclusive that they are working more hours.
The BLS prefers the payroll series as the main component when compiling the data because it has a big sample, more reliable industry identification, a timely annual benchmark to universe data from the unemployment insurance system, and business establishments tend to be more precise in reporting hours than households.
Lastly, there is the question of how offshoring of jobs is affecting productivity. Conceptually, off-shoring does not lead to measurement problems, but it does have second-order economic impacts. Firms that leave the country tend to be less productive, meaning the remaining firms will have a higher productivity. Though this occurs, it is also hard to quantify.
Brent Moulton
Bureau of Economic Analysis
GDP and business sector output measures are imperfect, but any biases are likely to be small and offsetting. Some sources of inaccurate measuring revolve around the topics of off-shoring, price indexing and technology output.
The topic of outsourcing is a hot topic for the way it shows up in U.S. accounts. When monitoring outsourcing, it is extremely important to capture the movement in imports. The issues that arise when measuring imports are reporting for services and reporting for low-value imports. This is not much of an issue for low-value imports in the goods sectors, but it can be quite an issue for measuring services since there is a high cutoff as to the level that companies are required to report. If a company's imports of service are below $1million, they are not required to file a report on the imports. Though most of the big corporations do reach the limit and file, there are definitely some that are below and are getting missed. Another problem arises in tracking new importers of services into the market. Unlike other surveys conducted by the IRS and the Census Bureau, there is no list of all importers of services that is easily accessed, so many new firms entering could be missed from the surveys.
Another problem involves the price indexes used for deflation. There is not a regular set of price indexes that cover imports of services, so even if prices were indexed when channels for purchasing goods and services change, they often get missed from the Price Index and do not capture the full cost saving. While these are all things that would suggest an overstatement of output, these problems mostly have counterbalances on the other side of the equation. For example, there are similar problems in missing new entrants in terms of measuring output and in changing channels of distribution for prices and output.
The second major concern is the measuring of technology output. This measure is improving because most countries are switching to use both the hedonic method and the chain-type method. Looking strictly at chain-dollars can be misleading, with people who use this method coming up with much larger numbers than estimated.
Lastly, GDP forecasts that are seemingly always being revised are actually fairly accurate in estimating GDP for the overall picture in terms of whether the economy is growing and whether growth is accelerating or decelerating.
Harry Bosworth
Brookings Institution
The BLS is producing "new" statistics with detailed data that can be used to examine the sources of the recent productivity acceleration seen at the aggregate level. Productivity and its sources can now be examined alongside the BEA in compiling the information. The new info also allows people to see how much of the productivity growth is coming because we have more capital per worker, and how much is coming because that capital is a different quality than it was in the past-i.e., high-tech capital.
Despite the focus on production of the high-tech goods, the data highlights that productivity growth in the United States is more diverse and comes from a variety of sources. With new data that allows us to see what industries are contributing to the productivity, we can examine each sector individually. The most promising aspect of seeing a diverse source of productivity growth is its long-term implications. Productivity growth is going to be durable because of it comes from many sources instead of a surge in a single sector.
The ability to focus on particular sectors with the new data demonstrates that service-producing industries are big applicants of high technology, and productivity realizes the benefits. Though the data for manufacturing may not look too amazing, one has to remember that the definition of "goods" is extremely broad, including construction and mining. These two sectors have been in decline since 1995, dragging down the positive numbers for this broad category. The five industries in the service-producing sector that look exceptionally promising are the wholesale trade/retail trade, finance and health sectors. The wholesale/retail trade in the model of Wal-mart/Big Box-type stores has proven to be successful, while computers have been involved with virtually all finance transactions. The health care sector is being dramatically underestimated in terms of improvements and the output of the medical care industry. Lots of further productivity increases are coming from reallocations among industries at the labor productivity level. Service producing industries are doing so well they accounted for three-fourths of the post-1995 growth in aggregate labor productivity.
Despite the great statistics, there are still some critiques on the data. First and foremost, the timeliness of their release is terrible. The studies we are reporting on are from the most recent data of 2001, nearly two years behind. The second critique is the measurement errors in precise contribution of purchase materials.
Productivity and Wages Panel
Diana Furchtgott-Roth
U.S. Department of Labor
The relationship between productivity and wages is, in light of the most recent economic recovery, a timely issue. Based on the nominal compensation data provided by the Bureau of Labor Statistics' (BLS) non-farm business sector labor productivity data since 1979, there is evidence that increased labor productivity may indeed result in increases both in real wages and in total worker compensation. It is important to note that the BLS measure includes wages, salaries, and benefits, and that it takes into account all workers other than those in the farming industry, government, and nonprofit organizations.
Similarly, it is useful to explore how one converts this nominal data into real compensation figures. The standard method involves deflating nominal compensation by the Consumer Price Index (CPI). Historically, however, the CPI consistently overstates inflation as it fails to account fully for substitution possibilities for new products and price increases. Jerry Hausman suggests that a 1.6 percent average annual bias from 1972-1994 and, therefore, that the standard CPI deflation measure will understate actual real compensation growth.
I propose that we employ the standard CPI deflation measure but recognize that derived measures will consistently have this distorting effect. Using the standard measure, we find that wages, salaries, and total consumption respond to changes in productivity to an even greater degree. They have not yet matched productivity growth rates precisely because of this failure to account for new products and product improvements. In conclusion, growth in labor productivity demonstrably facilitates growth in labor compensation and deflation techniques actually determine which has grown more rapidly.
Jared Bernstein
Economic Policy Institute
The notion that increased productivity equates to higher living standards is valid; however, intervening factors, such as the lagging job recovery of the kind we see now, lead to inequitable distribution of such benefits. In 2003, despite high productivity growth rates, we saw an extremely slow rate of growth (0.5 percent) for those in the twentieth percentile of earnings. This contrasts against the simultaneous growth of productivity and low and medium level wages of the 1990s and can only be attributed to slow job creation.
I disagree with the assertion that real wages are holding up well. Weak labor markets have empirically been shown to result in decelerations in wage growth, and indeed, real working earnings for the tenth median and ninetieth percentile earnings levels are falling for the first time since the mid 1990s. Moreover, it is likely that the boost to income was more the result of declining tax liabilities than wage increases. The claim that job loss during the beginning of a recovery is to be expected is equally unfounded as this recovery has been slower than the average rate of the previous nine.
Productivity growth can have a significant long-term impact on living standards, but often this benefit is distributed unequally without concurrent strength in job creation. The policy implication, therefore, is a focus on maintaining full employment in times of recovery so as to ensure equitable distributional effects.
Marvin H. Kosters
AEI
I find several significant points of disagreement with the suggestions Mr. Bernstein makes in his presentation. First, instead of assuming the previous "trough" to be the singular point of origin of the recession, I feel he would be better served to consider the peak of the last cycle as another possible starting point.
I also find no strong historical relationship between changes in productivity and those in inequality, as Bernstein would suggest. In fact, we have seen very little change in wage inequality for almost a decade. We should instead be more interested in looking at changes in average well-being as a whole rather than changes in equality as such.
With regard to productivity rates and job growth, there is obviously an arithmetic relationship between the two, given what happens to overall demand output and a number of other factors. This is not to say, however, that growth in productivity is a strong source of change in labor growth rates. Thus it would not be appropriate to blame slow job growth on strong productivity growth as Bernstein suggests, and I feel the tendency to do so is often a product of political concerns rather than strong economic theory.
As Bernstein's numbers demonstrate, measuring the contribution of productivity to wage growth is a complicated process. It is, however, certainly the case that growth in labor productivity is a source of increases in workers real income over time.
Luncheon Address
Elaine Chao
U.S. Secretary of Labor
We are all aware that productivity growth, despite all its challenges, pays dividends for workers. Specifically, we know that it has a positive impact on real wages because when companies make more, they can naturally afford to pay their employees more. Real compensation has risen 5.7 percent since 2000, according to a nonpartisan BLS report. Higher wages also attract foreign investment, which equates to more jobs for American workers. In the current economic debate, too little time has been paid to these effects.
When wages are high and prices are low, Americans can afford to buy more for their families, and living standards go up. But it is also important to consider the number of hours workers must spend on the job as increased productivity also means that workers must spend less time at work, and more at home with their loved ones.
Productivity growth is vital for living standards and workers compensation, but we must also focus on those displaced, and this is precisely why I feel it is so important that Congress pass President Bush's "Six Point Plan." We must ensure that our jobs training programs are relevant to the roles our workers need to fill given the changes in jobs that technology inevitably brings. The department's primary legislative priority is the reform of the Public Workforce Investment System. This human factor of the debate must not be overlooked.
Since 1995 we have seen productivity growth soar, and how we promote and manage this growth will determine the standard of living for American workers. I cannot think of a single public policy topic that is as timely as productivity, and I thank the American Enterprise Institute for being at the forefront of important economic issue debates such as this one.
Recent Productivity and Potential Output Growth
Robert Arnold
Congressional Budget Office
I work on medium-term (ten-year) budget projections at the CBO that require an economic forecast in which productivity plays a large role. The period being focused on for my talk is 1995 to the present: a period that showed rapid productivity growth. Although necessary for the analysis, it is rather difficult to discern whether or not the present burst being experienced in productivity will resemble those we have experienced in the past. Despite the usual high volatility in the growth of labor productivity, some contest that this has declined in recent times. The distinguishing trait, though, of this productivity growth, is a decline in the tendency for growth rates to dip below the post-1973 averages.
Another perspective on the most recent period-since the 2001 recession-is shown by a cyclocomparison. Productivity has been growing at 13 percent (more than at its cyclical peak in 2001), surpassing its usually expected increase of 6 percent. The model being used does not suffice for a complete explanation of the recent occurrences but can be further adjusted to take into account the discontinuities. They were caused by an introduction of new formulas for computing prices and biases in the estimation of CPI. The model allows labor productivity to be broken down into two components: the growth rate of TFP-a measure of productivity a bit more fundamental than labor productivity-and capital deepening. Capital deepening is the biggest factor in explaining the upswing in productivity growth rates.
Researchers have identified a technological change in the IT sector- characterized by a TFP growth in the production of computers rather than increases in their usage- as another source of productivity growth. Findings by Bosworth of EPI show that the productivity gains from IT are starting to spread outside the narrow IT sector, and we have been adjusting our term projections to take this into account. Potential output is projected to grow at 2.8 percent, and productivity will grow more slowly than it did in the late 1990s but faster than the post-1973 average.
Adam Posen
Institute for International Economics
Potential output is a good summary data point to look at when considering the economy's speed limit. There are problems with overestimating output, and underestimating it prevents us from living up to full potential as in the case of Japan in the 1990s. There is partial validity in the argument that the overestimation of output-presumed by the downward shift in productivity growth in 1973-caused the inflation of the 1970s. Though difficult to discern, there are three ways to determine potential output: from the top down (the statistical method), from the bottom up, and by the rule of thumb. There are problems with all three: in using the statistical method we tend not to pick up things that change, the bottom-up way tends to carry political pressures in practice and forces you to make very strong assumptions about the economy which may be inaccurate, and the rule of thumb has ambiguity as to at what point the trend will change.
It is not technology or trade that creates job loss, but rather privileged interest. In Japan, with its massive financial deregulation, telecom and energy price declines, restructuring in few industries and large IT investment, they experienced high productivity growth which should improve potential output. Despite these significant contributions, Japan still does not have job growth. The problem is they are not prepared to adjust: they have failed to facilitate freer mobility. Countries that are unwilling to provide essential government assurances, fail to provide job training, and produce environments in which people are incapable of switching jobs if necessary generally have difficulty converting productivity into growth.
An example of when underestimating output can make you forgo potential investment is the case of the 1990s when the FOMC decided to hold off raising interest rates. Many feel the decision was made to gamble and get inflation earlier since the growth losses from small amounts of inflation, when you are near zero inflation, are low. It seems we are in a similar kind of situation now, and I worry about the other major economies, such as Japan and the Euro zone, which are not as willing to take such gambles with potential output. All these things are structural and certainly not the central bank's responsibility.
What makes the U.S. productivity growth higher than in much other countries is not only capital deepening, but also how these new technologies and skills are utilized. One of the main drivers for the great productivity gains experienced by the United States has been labor changes. As in the examples of GM, Kodak, and IBM, much of the productivity gains can be attributed to the redistribution from overpaying low skilled workers into corporate investment.
Bruce P. Mehlman
Computer Systems Policy Project
In America, over the past century, there have been extraordinary improvements in healthcare, nutrition, the environment, living and working conditions, and life expectancy. The fact is that America enters the twenty-first century with the strongest and most vital economy in the world. America's economic engine is based on three elements: innovation, entrepreneurial business climate, and productivity. One benefit to innovative capability is the insourcing of jobs. As Secretary Chao mentioned, about 6.5 million jobs from foreign companies insourced in the United States occur because of productivity and innovative capability. Innovation creates jobs and wealth. Productivity explains how Americans can spend far less of their time working while producing far more and enjoying a much greater standard of living. Information technology is crucial to advances in productivity. We believe that the resurgence in American productivity over the past nine years is a result of IT and networking. Real world examples such as Dell, UPS, and the IRS attest to these huge benefits from technological advancements.
I would note that America does not always lead in productivity. Many of our west coast docks have been pretty slow to computerize due to opposition from longshoremen's unions. Docks in Singapore and Hong Kong, by contrast, can handle five times as much cargo, and their computerized scheduling system allows their trucks an average wait of less than one minute as compared with the U.S. average of 2.5 hours to carry cargo away. Thirty-one percent of health care costs are administrative overheads. IT technology promises to make health care more affordable, more accessible, and less error-prone if adopted and embraced.
Investment and policy decision-making need to encourage productivity through fiscal and regulatory policies. We must ensure that our business climate awards innovation and entrepreneurship, and encourages productivity growth by making policy choices that best invest in our future. Productivity is most needed where it is currently least felt, for example, in education, health care and government-critical sectors primarily challenged by rising costs. Third, the rising tide of productivity should lift all boats. Education, job training, and portable health care are needed to facilitate better job transition and compensate the resulting job loss that accompanies progress. Provided these key things, along with innovation, entrepreneurship and productivity growth, America should never have to compete in the battle to see who can pay the least wages.
AEI interns Adam King, Kimberley Lue Lim, and Brian Long prepared this summary.
