What We Learned From Jason Furman

In July, Jason Furman—Chairman of the President’s Council of Economic Advisers—gave us an invaluable treat. At our Capital Market Initiative 101 event, Inside the Jobs Report, he shared his tips on how he (and now you) interprets economic data from key economic reports like the GDP and Employment Situation survey. You can visit our event page to watch the C-SPAN coverage and read his slides. But here is our cheat sheet from his fascinating presentation.

1. Don’t Fixate on One Number: Never get too excited about one piece of data, because there’s probably more to the story. Most statistics are limited by many factors such as sampling errors, complicated seasonal adjustments or even longer term trends. Be cautious of falling in love with one number.

2. Some Data Is Naturally Noisy: Most economic data is based on surveys and require many revisions. When confronted with noisy data, follow these tips: (1) use data with larger sample sizes, (2) look at the data over a longer period of time and (3) find different measures of the same concept and compare the changes.

3. Where There’s a Fork in the Road, Take It: GDP (Gross Domestic Product: how much everyone bought) and GDI (Gross Domestic Income: how much income everyone earned) essentially measure the same thing. However, both data points are rarely the same. The answer: use both numbers. An average of these two numbers (which the Council of Economic Advisers is calling Gross Domestic Output, or GDO for short) is a more accurate picture of the economy.¹ Additionally, Private Domestic Final Purchases (PDFP), one component of the GDP data, is a more stable reflection of the economy than GDP since PDFP excludes erratic data like inventory and net exports from its calculation.

4. Not All Government Economic Data Are Created Equal: The Establishment Employment Survey and the Household Employment Survey are both used in the monthly jobs report that comes out on the first Friday. But the Establishment survey has a much larger sample representing millions of workers versus only thousands of workers in the Household survey.

5. Be Aware of Real vs. Nominal Dollars: Always make sure that reports are adjusting for inflation in their data (especially for wages). And also be aware that different economists, different agencies, and different reports often use different measures of inflation – sometimes within the same chart! A great example of this came a few years back when a chart compared the gap in wage growth and productivity growth.² Because two different price indexes were used (Implicit Price Deflator for productivity and Consumer Price Index for wages) a large chunk of that gap could be explained away.³

6. Find the Apples-to-Apples in the Fruit Salad: Many commentators focus on the declining Employment-Population Ratio and Labor Force Participation Rate. But much of it has to do with the aging of America and the number of young adults in school. Focus on data points that adjust for demographics—like the unemployment rate—for a better picture of cyclical strength.

7. Adequate Time Horizons Differ Widely for Different Data Points: Productivity—a central factor to the economy when it comes to growing GDP and income—is released quarterly and reported upon breathlessly. It is also revised frequently and best measured over the long-run. In fact, quarterly estimates are so noisy and subject to revision that they provide little information about the underlying economy at the time of release. To get a better sense of productivity, look at the data over a much larger time frame (10- or 15-years). Weekly jobless claims bounce around as well, but a moving average of just four weeks should be more accurate. For wages, about four to six months is an adequate dataset to gauge recent trends.

8. Revisions Aren’t Necessarily Mistakes: Quarterly GDP is released, revised, and then revised again. Why? GDP is based on a series of reports that come in and look back at the previous quarter. Not every one of these underlying reports come out in time to be included in the first estimation, so a survey is used instead. A great example of this is data for manufacturers’ shipments, inventories, and orders (MSM). Comprehensive MSM data isn’t reported until 2 months later, so an MSM survey is used for the first GDP estimate.⁴