>OECD Economic Outlook reports that, depending on which measurments you use,
>labor productivity growth in the U.S. could be as low as 1.0% or as high as
>1.9%. They say you can measure output for these calculations by both an
>income and and expenditure measure. How do they account for the differences
>in the two measurments? Likewise, there's an establishment and household
>account of employment, and they are different. Do they have a way to account
>for statistical discrepancies in these figures?
The text Christian is talking about is below. This is the OECD's own productivity measure, not the one the U.S. BLS produces. They're playing with dividing two measures of GDP by two measures of employment. They're not using hours of work as the denominator, which is what the BLS and most other people do. GDP is measured in two ways, in terms of product (which for the private sector means sales) and income (the sum of wages, interest, profits, etc.) Since income is charged against production, the two estimates should be equal, but since they're derived from different sources, they're not, and the difference between the two (product less income) is called the statistical discreptancy.. The BEA says that product is the more accurate measure. Usually the number is pretty small (and largely disappears after final revisions), but for the last few years it's been very large, with income running well ahead of production. I'm going to make some calls later in the week to find out what the BEA thinks the reason is - and the figure may shrink with the October revisions - but I wonder if foreign capital inflows that end up in what Keynes called the financial circulation without ever entering what he called the industrial circulation have anything to do with it.
The employment measure is less of a problem. The payroll figures are the ones to use. The payroll figures are based on a monthly survey of employers - a representative sample, not a full count. But those are revised every year with official records that cover 98-99% of the employing establishments. Household measures are good at turning points, because the BLS builds some momentum into its payroll estimates (in expansions, it imputes a certain rate of job gain based on recent trends to cover new businesses that elude its survey). But in recessions, lots of people become "self-employed" - they're counted as employed in the household survey, but their jobs disappear from the establishment survey. Someone holding multiple jobs (and there are more of them then there are unemployed) would count as just one job in the household survey, but as multiple jobs in the establishment figures. I don't know why they brought this up.
Doug
----
OECD Economic Outlook, June 1999
Box 1.5 What is happening in the US economy: measurement issues
Structural changes in the US economy may have affected the statistical information base on which economic policy formulation depends. Partly this is because financially stretched statistical agencies may not have fully adapted to these changes (in particular, agriculture and manufacturing remain covered considerably more comprehensively than services) and partly it relates to conceptual problems, notably those relating to quality changes and price-volume splits, whose importance increases as the share of output which cannot be measured in physical terms rises. Where regulatory arrangements provide collection points for data, liberalisation may also have had an effect.
The recent behaviour of labour productivity is a case in point, since there have been important discrepancies between the various measures of both output and employment growth. The income measure of GDP has been showing a significantly stronger rise in Output than has the most widely accepted measure based on expenditures. At the same time. the establishment-based payroll measure of employment has risen more than employment gains recorded in the household survey. As a consequence, the average productivity growth rate over the 1993-1998 period is almost one percentage point higher if derived from the income measure of output and the household measure of employment than if it is derived from the expenditure measure of output and the payroll measure of employment (see table). Cumulatively over the period, this differential amounts to nearly 5 per cent, a figure that is large relative to most plausible estimates of slack in the economy.
Unfortunately, it is not possible to state categorically what productivity measure is the most reliable until, and unless, the next census provides more accurate information about employment and the next comprehensive benchmark revision of the national accounts better reconciles the different output measures.
------------------------------------------------------------------------ average growth in GDP per employed person, 1993-98 percent
household survey payroll data GDP expenditure measure 1.6 1.0 GDP income measure 1.9 1.4
Note: the expenditure-based GDP price felator is used to convert the income-based GDP series into a constant price series. ------------------------------------------------------------------------
Questions about the quality of data or how to interpret them are not confined to output and employment data.[1] The publication of the Boskin Report[2] called attention to the complexities of measuring inflation, national accounts and flow-of-funds data show different pictures for household saving behaviour, and external financial accounts probably have large errors and omissions. Questions also extend beyond the United States, whose economic and financial statistics are generally thought to compare favourably in international terms. The elimination of customs points within the European Union as part of the single market programme, for example, has affected the accuracy of European trade statistics and the growth of electronic commerce will pose challenges to collection of statistics not only for external transactions but more widely. The importance of economic and financial statistics has been emphasised by both G-7 countries[3] and the IMF, which has established a Special Data Dissemination Standard to encourage more comprehensive provision and timely reporting of information. Concern to improve economic and financial information internationally has largely been motivated by a desire to encourage better reporting by emerging market countries. But, as the size of revisions being made to national accounts in EU countries suggests (see Box 1.2), adapting information systems to a changing world will be a challenge even for those countries where statistical systems are most advanced.
1. A further elaboration on these and other statistical issues is contained in OECD Economic Surveys, United States, 1999.
2. Report of the United States Senate Advisory Commission to Study the Consumer Price Index, Washington, 1977.
3. See, for example, the G-7 leaders' Statement on the World Economy and the accompanying Declaration of G-7 Finance Ministers an Central Bank Governors, 30 October 1998, in which G-7 countries committed themselves "to deliver greater transparency ... in disclosure of economic statistics and key indicators".