The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
BUREAU OF LABOR STATISTICS MEETING June 9, 1998 ..................................................................... I. Introduction A. Thank you. I’m glad to have the opportunity to comment on the many BLS products, including those produced jointly with the states, that are of great importance to the Federal Reserve System. B. And its always good to speak to a group that really understands statistics. Unlike this fellow I heard about recently. He read that 80% of accidents happen within 5 miles of home. So he moved. C. Seriously, I am always happy to speak to a labor department audience. As some of you know, I worked at the U.S. Department of Labor back in the 1970s in several different positions, including Undersecretary of Labor. D. Shortly after leaving DOL, I was fortunate to serve as a member of the National Commission on Employment and Unemployment Statistics, known as the Levitan Commission, which in 1979 made a number of recommendations on the future direction of U.S. labor statistics. 1. The Commission was chaired by Sar Levitan, a well-known labor economist who died several years ago. 2. Sar was a good friend for many years, and he made a number of contributions to the labor field over the years. E. My experience on the Levitan Commission has taught me the virtue of patience — many of the recommendations contained in our report have been implemented by the BLS. Michael Moskow Speeches 1998 343 1. But in some cases, it has taken nearly 20 years. Of course, that was usually because of difficulties getting the necessary appropriations. 2. In any case, I’m pleased to have the opportunity to offer a few more thoughts on how the BLS could better serve the Federal Reserve System, thoughts which I should note partially reflect a very informal survey we did of Fed economists II. CPI A. The Federal Reserve System is committed to the goal of raising national living standards through the promotion of maximum sustainable economic growth. 1. The qualifier that the growth be sustainable is very important. 2. A brief spurt of growth that is so rapid that it quickly burns itself out and is followed by recession does not raise living standards. B. Some people overestimate what monetary policy can do. 1. They think the Fed could make the economy grow more quickly over the long haul simply by loosening policy. However, the Fed doesn’t have such power. 2. When the Fed eases policy by providing more liquidity to the financial system, it doesn’t directly create any wealth. Wealth consists of physical and human capital. 3. But, the Fed doesn’t build any new factories or upgrade any workers’ skills. An easing of monetary policy can speed up economic growth in the short run, but any prolonged period of ease beyond the capacity of our physical and human capital would only raise inflation. C. So what can the Fed do to promote maximum sustainable growth? The answer is to provide price stability – that is, low and stable inflation. 1. Without low and stable inflation, consumers and producers will not be able to pick up the important price signals that are the central nervous system of our market economy. 2. When inflation is high and variable, it’s hard to know whether the relative price of a product has gone up or whether a price change is just part of a general inflation. 3. Such confusion lessens the effectiveness of the price system. High and variable inflation also makes it very difficult to plan for the future and discourages investment. 4. Conversely, an environment of price stability facilitates planning, encourages investment and generally creates the conditions necessary for sustainable growth. D. That, of course, brings me to the first BLS product I want to discuss. If we can’t measure inflation accurately, we can’t very well guarantee that it stays low and stable. 344 Michael Moskow Speeches 1998 1. So, of all the important products the BLS provides, the most important to the Federal Reserve System must be the Consumer Price Index. 2. For the Fed, the CPI is an important monthly report card. It tells us how well we are doing in our efforts to create a sound economy. E. Moreover, the BLS’s price measures are ingredients in many of the other statistics we routinely use to measure and understand the economy, such as those on real growth, productivity, and real interest rates. F. Thus it is vital that the price indices produced by the BLS be of the highest possible quality. 1. Lately there has been a good deal of discussion of how much “bias” there might be in the CPI. This issue is of great importance to the Federal Reserve System. 2. If the findings of such groups as the Boskin Commission are correct, we are closer to price stability than the monthly readings on the CPI indicate. 3. Of course as Commissioner Katherine Abraham has pointed out on many occasions, a good deal more research needs to be done before we can be confident of the magnitude of bias in the CPI. 4. We hope that such research will be a high priority for the BLS. G. Fortunately, the BLS has been at the forefront of research on cost-of-living measurement. 1. BLS researchers have identified many of the technical problems with the CPI that are at the heart of current discussions. 2. We trust that the BLS will continue to do the kind of careful, credible research that will put estimates of biases on a firmer foundation. H. Of course, where possible, it is better to remove a bias than to simply measure it. Indeed, the BLS has made several important improvements to the CPI over the last several years. Such improvements typify the kind of excellent work that we always expect from the BLS. I. However, I want to point out that these improvements to the CPI, while absolutely the right thing to do, create a significant communication problem for us within the Fed and more importantly between policymakers and the general public. J. Our reading of the research is that the improvements made by BLS since 1994 have had the effect of lowering the measured rate of CPI inflation by around four tenths of a percentage point this year relative to what would have been recorded using the 1994 methodology. • Moreover, more improvements are coming for next year that would have the effect of lowering the measured inflation rate by another two or three tenths of a percentage point even if there were no change in the true rate of inflation. Michael Moskow Speeches 1998 345 K. One of the reasons it is a challenge to create an environment of low and stable inflation is that inflation has a tendency to creep up very gradually. • Moreover, once inflation has increased, it becomes built into people’s expectations and can be very difficult to reduce. Thus we need to be alert to even relatively small changes in the rate of inflation. L. Changes over time in the methodology used to measure inflation complicate the task of identifying the trend in inflation. A change of four or seven tenths of a percentage point in the rate of inflation, the magnitude of the methodological changes instituted the last several years, is a significant change by our standards. M. Within the Fed, the difficulties in interpreting the inflation data are a relatively minor nuisance. But, they present us with a larger problem when it comes to communicating with the general public. 1. We often hear that we are overly concerned with inflation, that the battle against inflation has already been won. 2. While it is true that real progress has been made, some of the apparent progress is BLS methodological progress, not true progress on inflation. 3. My concern is that it will be difficult to maintain public support for sound monetary policy when the problems we are trying to address are partially obscured by changes in measurement methodology. N. In this regard, the BLS could do the Fed and the public a significant service by releasing a historical data series on inflation computed as much as possible using current methodology. • With such a series we could easily see what inflation would have been in prior years if today’s methodology had been in place, thus giving a truer picture of the changes over time in actual inflation. O. There is precedent for such a series. When the method of estimating housing cost inflation was substantially revised in the early 1980s, the BLS produced a series on inflation applying the new methodology to earlier years. Competent researchers typically work with this series, known I believe by the highly memorable name of CPI-U-X1, rather than the standard CPI-U series. P. 346 Having a series, maybe called CPI-U-X2?, that applied current methodology to a long time series would help us to focus on the economics of inflation rather than the measurement. 1. If the BLS produced such a series, it would surely become the standard used by researchers and over time could become a tool we at the Fed could use in our communication with the public. 2. Such a tool would ultimately help in building support for sound monetary policy. Michael Moskow Speeches 1998 III. Payroll survey A. I’ve talked quite a lot about inflation measurement because providing price stability is our chief concern, but it is not our only concern. 1. We also devote a good deal of attention to employment and the real side of the economy more generally. 2. We do this not only because of our objective of promoting full employment but also because it helps us to forecast inflation. 3. Research shows that monetary policy affects inflation with lags of a year or more. If the Fed were to wait to see inflation rise before changing policy, we would often be too late to prevent a significant increase. 4. Instead, we strive to be forward looking, to base our policy decisions on where we think the economy is likely to be headed a year or more from now. B. The Current Employment Statistics program, the joint effort of the BLS and the state agencies represented here, is one of the key sources of data that help us judge where the economy is now and, more importantly, where it is headed. C. There are a number of reasons the Fed pays close attention to the payroll data. 1. First, they are available with a relatively short lag, providing us with up-to-date information. 2. Second, they are based on an enormous sample. The statistical noise, while substantial, is less than in many other series. The large sample also allows us to look at the detailed breakdown by industry to get a better indication of the sources of growth. D. As with the BLS’s price data, the payroll survey data find their way into other important economic statistics, such as the Federal Reserve’s estimates of Industrial Production and Capacity Utilization, and the Bureau of Economic Analysis’s early estimates of gross domestic product. Here also, high quality is vital. E. Unfortunately, though the payroll survey sample covers 40% of total employment, it is not selected using modern probability sampling theory. 1. Moreover, new firms, which tend to grow faster, do not make their way into the sample very quickly. 2. As a result, BLS has had to employ some rather ad hoc adjustment procedures to the employment totals to account for the birth of new firms. 3. Perhaps for these reasons, the revisions to the payroll data at the time of the annual benchmarking process have sometimes been rather large. Michael Moskow Speeches 1998 347 F. 4. While an average revision of three tenths of a percent of employment may not sound large, the effect on estimates of employment growth is much larger. 5. A revision of three tenths of a percent of employment is equivalent to adding or subtracting 30,000 net jobs per month for a year, about 20% of job growth in a typical year. 6. Such a revision can have a significant effect on our view of how the economy is performing. Thus we strongly support BLS plans to improve the Current Employment Statistics survey. 1. Probability sampling and a quicker introduction of new firms into the sampling frame should make an already useful program even more effective. 2. I’ll also note that current plans are generally consistent with the recommendations made by the Levitan Commission nearly 20 years ago. G. The process by which BLS has come to its present plans also deserves praise. The involvement of experts from the American Statistical Association and private survey organizations undoubtedly complemented the substantial expertise of the BLS’s own staff and added even more credibility to the new design. H. If I have any concern with the BLS’s work on the CES it is that almost all of the attention seems to be focussed on the employment totals. 1. But the hours and earnings data derived from the CES are also very important to policymakers. 2. The problems with these data may be even more significant because the annual benchmarking process does not deliver universe estimates of hours and earnings the way it does for employment. 3. In any case, I hope that BLS will give significant attention to the quality of hours and earnings data as well as to employment data. IV. Turnover and Vacancies A. Fed researchers and policymakers are always trying to understand how the structure of the economy is changing. Keeping up with developments in our highly dynamic economy is both difficult and vital to conducting successful monetary policy. B. Change in the labor market may have been especially significant in the last decade. While in many respects the data available for studying labor market trends is excellent, in others it is rather poor. • For instance, we know relatively little about trends in vacancies and turnover, a gap in our knowledge that the proposed new BLS survey of business firms would begin to fill. C. Why should the Fed care about vacancy and turnover data? Because it may help us to understand important economic developments. 348 Michael Moskow Speeches 1998 • For instance, a good deal of attention has been paid to the fact that wage inflation has stayed relatively subdued despite an unemployment rate that has been low by historical standards. D. At the same time that wage inflation has increased somewhat less than expected, another less noticed deviation from past patterns has emerged. 1. Namely, the level of help-wanted advertising, while high, is not as high as might have been expected given the low unemployment rate. 2. Help-wanted statistics are sometimes taken as proxies for job vacancies. It may be that the level of vacancies has not risen as much as in previous periods of low unemployment. 3. If true, this would be an important development. E. Vacancies measure firms’ unmet demand for labor just as the unemployment rate measures workers’ unused supply of labor. F. 1. If the help-wanted data really are a good proxy for vacancies, the fact that they have risen less than expected would mean that the labor market has, in a sense, become more efficient. 2. In the past, when workers were relatively satisfied with economic conditions, firms were relatively unsatisfied. 3. And, conversely, when firms were more satisfied, workers were less satisfied. 4. If the help-wanted data reflect trends in vacancies, then conditions may have improved for workers without deteriorating as much as usual for employers. What could have led to such improvements in labor market efficiency? Perhaps an improvement in the way workers and firms search for suitable employment matches. 1. This might have been brought on by the increased prevalence of fax machines and the internet. 2. Another possibility is the very rapid growth of the temporary services industry, which matches workers and firms in unprecedented numbers. 3. If, for whatever reason, labor market efficiency has increased, it would provide a possible explanation for wage growth remaining moderate. G. However, it is just as possible that developments such as the fax machine and internet could have lessened firms’ use of help-wanted advertising to fill vacancies. 1. In this case, the relatively moderate levels of help-wanted advertising would not mean as much. 2. Since we do not have good data on vacancies, we cannot really know whether the economy has changed in an important way or not. H. Hopefully, the new BLS proposal to collect vacancy and turnover data for a broad sample of firms, another 1979 Levitan Commission recommendation, will be fully funded. Michael Moskow Speeches 1998 349 I. The new program would also collect data on new hires and separations. Such turnover data would also be very valuable. Indeed, I would like to see the data on separations disaggregated into those due to quits, layoffs and discharges for poor performance. J. A better measure of quits would help in evaluating another possible explanation for the moderate growth of wages in the face of low unemployment – worker anxiety. This theory holds that increased fear of job loss has restrained workers’ wage demands. 1. In order to evaluate this explanation, many have suggested examining quit rates. 2. Workers who are anxious, the argument goes, are less likely to quit their jobs. Unfortunately, our data on quits is not as good as one would like. 3. The quit rate derived from the Current Population Survey covers quits into unemployment. Workers who quit to take another job are missed. K. Perhaps the proposed new survey of vacancies and turnover could be used to collect information on quits. If not, another old Levitan Commission recommendation could be resurrected. 1. Namely adding questions to the Current Population Survey on employment to employment transitions. 2. That is, we could ask employed workers whether they have changed jobs within the last month and, if so, whether the change was a quit or a layoff. L. However collected, better information on quits would complement the new joint state-BLS program on mass layoff statistics. Together they would give policymakers a valuable picture of labor turnover. V. Research A. The Fed benefits significantly from the many long-term research projects conducted by economists within the System as well as by those at universities and other research institutions. 1. Such long-term research provides us with the knowledge base needed to interpret the often conflicting trends that we are constantly examining. 2. Much of the data that underlie that research is provided by the BLS. B. As the Levitan Commission recommended, the BLS has provided important support for longitudinal data collection. The recent launch of the new National Longitudinal Survey of Youth continues the good work of the BLS in this area. C. For some research projects the large samples of the Current Population Survey micro data are ideal. The BLS has made impressive advances in using technology to make this data available to researchers in an extremely timely fashion. 350 Michael Moskow Speeches 1998 D. Unfortunately, the timely release of CPS micro data does not extend to the valuable data collected as special supplements to the CPS. 1. In fact, it still takes quite some time before these data make their way to researchers or even show up in BLS publications and press releases. 2. I’ve already mentioned the hypothesis that worker insecurity may be playing an important role in recent wage setting behavior. 3. Some of the best information on job loss, presumably the source of worker insecurity, comes from the Displaced Worker supplements to the CPS. 4. The last of these was conducted in February, but BLS tabulations of these important data are not anticipated until the Fall. And outside researchers will be lucky to access the data before the end of the year. 5. Even recognizing the limited resources at BLS and Census, I would hope that in the future, the lag between CPS supplement data collection and release could be significantly shortened. E. Much might also be learned by giving researchers access to the micro data underlying such BLS products as the Employment Cost Index, the payroll survey, and the CPI. 1. I understand that the BLS has begun to allow researchers to access such data in an environment designed to insure data confidentiality. I think this is a very positive development. 2. I also understand, however, that this program is still rather restrictive, excluding, for example, one of our Chicago Fed economists from working with the CPI data. 3. I hope that the kinks in the new system can be worked out, increasing research access to more BLS data. It would have real benefits for policymakers. VI. Conclusion A. In conclusion, I’d just like to remind everyone how lucky the nation is to have the highly professional system of state and national data collection that underlies our labor statistics. B. The independent and nonpolitical character of the BLS and the state agencies as well as their high level of professional competence mean that we can truly trust the numbers. C. BLS and the state agencies have been able to attract high-quality, dedicated staff members over the years and our country is fortunate to have a high-caliber, reliable statistics gathering agency to provide important data to policymakers in an objective, unbiased manner. Michael Moskow Speeches 1998 351