The Curious Case of the Missing Productivity
- The United States has consistently experienced very low productivity (less than 0.5%) since 2011
- A survey of households that asks about employment status confirms the recent strong employment growth reflected in the jobs report
- GDP is underestimated, but it is consistently biased, and the gig and free economies are too small to explain the productivity gap
On the first Friday of each month, the Labor Department releases the most widely covered economic indicator in the monthly calendar—the jobs report. Analysts quickly scour the numbers for hints as to how strong the economy is performing.
These efforts to divine meaning from the jobs report are understandable—net new jobs create growth in spendable income and consumer spending makes up about 70% of GDP final expenditures. New workers will spend 95% of their paychecks on goods and services.
Since 2014, job growth has consistently hovered around 2% at an annual rate (Figure 1). The conundrum is that GDP statistics indicate that U.S. economic growth is very erratic, with large growth spikes followed by slow (or no) growth. This period shows a tendency for the first quarter of each year to be the worst, followed by the strongest growth in the second quarter. Many economists believe that the seasonal adjustment of the GDP number is not robust enough to capture the anomalous weather of recent years. On the other hand, employment statistics are also seasonally adjusted and they do not exhibit high volatility.
Over the course of 2015, the economy reportedly grew an average of 2.4%—only 0.3% faster than the 2.1% growth in employment. The very low implied productivity growth is not unique to 2015; the United States has consistently experienced very low productivity (less than 0.5%) since 2011.
Where has the nation’s productivity gone? Many possible explanations for slow productivity growth are circulating among economists, but very little solid evidence exists. For this reason, some analysts question the data. Is productivity growth really that lousy? If not, then statisticians are overestimating employment growth and/or underestimating economic growth.
The problem is less likely to be with employment. Overstating employment growth can occur for a year, but it is not systematically wrong for many years. The Labor Department is able to triangulate to a very accurate number via an annual benchmark and data from the survey of establishments are reconciled to a virtual census of employment from states’ unemployment insurance programs. Furthermore, a separate survey of households that asks about employment status confirms the recent strong employment growth reflected in the jobs report.
That leaves GDP. There are many ways that the nation’s economic growth rate can be understated:
- Unmeasured and nonmarket activity have always been a concern. Household chores and volunteer charity work are not counted; when the same activities are paid for, they are included in GDP.
- The measures of service sector output are imprecise. Government statisticians can sum up the sales and value-added of business consultants, law firms, healthcare providers, etc., but when they have to separate the revenue into changes in prices and changes in quantity, the breakout is very imprecise.
- Inflation-adjusted GDP growth is calculated by dividing final expenditures by a price deflator. If the rate of inflation is overestimated, then GDP is understated. Most products and services are not adjusted for quality improvement, except for computers and electronic products. Critics point to medical care advances and ask if a person would rather have 1980 medical care at 1980 prices or 2016 medical care at 2016 prices. Clearly, today’s technologies are preferred, but the full extent of quality improvement is not captured in GDP. Totally new goods and services have no comparables, so new items and services have no impact on prices. Another issue with deflators is that the price impact of substituting low-cost imports for higher-cost goods and services incorporated into products and services is difficult to capture.
All these limitations have existed for a long time. GDP statistics are not perfect, although statisticians regularly make improvements. The key question is whether any of the statistical limitations have significantly worsened since 2011. Low-cost country sourcing for manufactured components has certainly grown in importance since the 2008-2009 recession; however, Reinsdorf and Yuskavage find the effect of the sourcing substitution biases on the average annual growth rate of inflation-adjusted GDP is less than 0.1 percentage points.
Two recent phenomena that are garnering attention are the “gig economy” formed by digital platforms such as Uber, Airbnb, and TaskRabbit and the “free” internet services that enhance the mobile experience. In the former, transactions occur between individuals and independent contractors. The transactions are difficult for statisticians to track, even if the contractor properly reports the income on his or her taxes (this raises the statistical discrepancy between the expenditure and income side in the national income accounts). In the case of the latter, the argument is that free apps greatly enhance the quality of mobile devices beyond what can be measured by the advertising revenue or business fees earned from the app or site.
Can the gig, mobile, and free economies explain why employment is so strong and economic growth is so weak? While it is true that GDP growth is understated for all the aforementioned reasons, we are talking about $18 trillion in reported GDP. A 1.5% undercount in GDP equates to a huge $270 billion; that miscounting would have to compound year after year to distort the growth rate. Employment data show that internet and telecommunications job growth is small relative to total job growth. In the 12 months ending in April, growth in internet and telecommunications jobs1 was 11,000 per month out of a monthly total of 224,000—only 5%.
Relatively strong post-recession employment growth is not a statistical anomaly and only a small portion of the gain is coming from internet and telecommunications jobs. GDP is underestimated, but it is consistently biased, and the gig and free economies are too small to explain the productivity gap. Rather than blaming statistics, analytical effort is better spent determining the root causes for slow productivity growth.
1. Electronic markets and agents and brokers; nonstore retailers; telecommunications; ISPS, search portals, and data processing; other information services; and computer systems design and related services.
The full Issues in Brief archive can be found here.