In our view, gross domestic product (GDP) data are deeply flawed.
To form a more accurate assessment of the economy, investors should include other measurements, such as gross domestic income.
Official GDP growth was relatively soft in the first half of this year, but the broader set of activity data suggests the U.S. economy is doing just fine.
Before joining Columbia Management I worked for several years as an economist at a few of the large broker-dealers in New York. One of my primary functions was to maintain an ongoing estimate of growth in the nation’s GDP—a so-called GDP “bean count.” Most investors use GDP as their primary summary measure of overall economic performance, so they are keenly interested in how incoming data are likely to impact the estimates. Our running tally of GDP growth for the current quarter was one of the most sought after pieces of research we produced.
Unfortunately, the saying about laws and sausages also applies to the GDP accounts: as we learn more about how they are made, our appetite declines in equal proportion. The official GDP figures include surprisingly many assumptions, extrapolations and guesstimates, and they fail to incorporate many of the most valuable economic indicators—such as employment, industrial production and business surveys. Partly for these reasons, the GDP data are also heavily revised over time.
Investors got another reminder about the flaws in the GDP statistics with the recent report for the second quarter. With the initial release, the Commerce Department estimated that GDP grew by a below-trend 1.7% (quarter-over-quarter annualized). However, with the release of additional Q2 data—especially a better-than-expected June trade report—most economists now expect that Q2 GDP growth will be revised closer to 2.5%.
Revisions to the GDP accounts affect more than just the latest quarter—they often extend deep into economic history. For example, the following chart shows the estimated annualized growth in real GDP for Q4 1968, as reported at various times by the Commerce Department. At the time of the initial release, the government estimated that the economy grew by a solid 3.9% in that quarter.
However, subsequent revisions lowered this estimate over the next several years, such that by 1986 the Commerce Department was saying that GDP in Q4 1968 actually contracted slightly. Later, in the mid-1990s, growth for this quarter was revised back up again, and it now officially stands at 1.7%. It’s anyone’s guess where this figure will be in another 10 or 20 years.
So if the GDP data are deeply flawed, how can investors more accurately track the state of the economy? We would offer three suggestions:
- Exclude the noisiest parts of GDP. Because of the way they are structured, the GDP accounts include all available expenditure statistics, without regard to their volatility or “noisiness.” This can create problems because the U.S. trade and inventory data are extremely noisy. In fact, these components of GDP contain virtually no information value for forecasting GDP growth in the next quarter. Therefore, one simple fix is to focus on GDP excluding trade and inventories—a line item called “domestic final sales.” These figures are much smoother and less subject to revision than overall GDP. In Q2, domestic final sales increased by 2.0%.
- Keep track of GDI. The Commerce Department actually adds up the size of the economy in two different ways: by expenditure and by income. The expenditure data are used to calculate GDP, and the income data are used to calculate gross domestic income (GDI). Conceptually the two numbers are supposed to be the same, but in reality they often differ. Interestingly, research has shown that GDI, not GDP, is the more accurate measure of the economy’s growth rate in real-time (that is, before the effects of revisions). Investors should put at least as much weight on GDI as GDP when trying to determine how fast the economy is really growing. Recently GDI has been running a little faster than GDP, suggesting GDP may be understating U.S. growth momentum.
- Use alternative summary measures. Investors need simple summary measures that accurately track economic growth in real-time. GDP is ill-suited for this purpose, but there are other tools available. At Columbia Management, we primarily focus on data surprise indexes and a proprietary measure of U.S. growth based on our own statistical analysis. These measures put more value on the indicators with the highest signal-to-noise content—whether or not they feed into GDP. They also tend to overweight survey-based data, which are not subject to major revision. Like GDI, these tools currently show more growth momentum in the U.S. economy than GDP.
GDP contains some value as a cyclical indicator, and serves other analytical purposes as well—for instance, the data are essential for comparing income levels across countries or scaling a government’s budget balance. To the extent that central bankers watch the GDP data they are also important for the monetary policy call.
But in our view, GDP deserves much less attention as a summary measure of economic growth than it usually receives. Official GDP growth was relatively soft in the first half of this year, but the broader set of activity data suggests the U.S. economy is doing just fine—indeed, if anything the pace of growth may have started to pick up.