“Facts are stubborn things, but statistics are pliable.”
– Mark Twain
The majority of U.S. economic data are based on statistical samples and the various figures are typically adjusted for seasonal variation. That means that the numbers are subject to some level of uncertainty. For some reports, the uncertainty is low. For example, the weekly jobless claims data are simply the total of state reports (occasionally, one or two states may need to be estimated due to late reporting). At the other end of the scale, new home sales results are reported with a gigantic level of uncertainty. Sales were reported to have risen 3.7% (±18.5%) in January, meaning that we can bE 90% certain that the true change was between -14.8% and +22.2%. One wonders why they even bother. Data typically take some time to be assembled. Preliminary estimates are subject to revision.
Data uncertainty can create some problems for investors. One should always look skeptically at any single piece of data and consider the broader array of information. Stock market participants often react to surprises in the data (if sufficiently far from the median forecast) even though the surprise may not be meaningful. In setting short-term interest rates, Fed officials look at a wide range of indicators, also paying attention to as much anecdotal information as they can assemble.
The government’s collection of economic data picked up following the Great Recession, with an eye toward understanding and preventing serious downturns. Many of the figures are geared to the old-style economy (plenty of figures on manufacturing, for example). However, despite long-standing budget constraints, the government has adapted to a changing economy over time and its data methodology has evolved. Note that in contrast to Canada, which has one statistical agency, U.S. data are spread all over the place (the Bureau of Census, the Bureau of Labor Statistics, the Federal Reserve, etc.). As an aside, this makes it more difficult to cook the books.
Candidate Trump said that the unemployment rate, officially reported at 4.8% in January, was really more like 42%. That might be true if one counted four-year olds and grandma as being part of the labor force, but it’s unclear why you would want to do that. It’s well known that labor force participation has declined significantly since before recession (from 66% to 63% more recently) and the employment/population ratio has fallen (to 59.8% in January, vs. about 63% just before the recession). Bringing the participation rate back up to 66% would allow the economy to grow faster. However, most of the decline in participation has been related to the aging of the population and some is due to a greater tendency for teenagers and young adults to remain in school.
The employment report is made up of two separate surveys. The establishment survey yields estimates of payrolls, hours, and wages. The household survey generates estimates of labor force participation and the unemployment rate. Note that some undocumented workers are captured in both surveys, but the Bureau of Labor Statistics cannot determine how many. In the household survey, the BLS does ask whether one is foreign born or native born. In January, foreign-born individuals accounted for 17.0% of workers. It’s well known that undocumented workers will often apply for a job using an assumed Social Security number. Hence, they pay into both the Medicare and Social Security trust funds without receiving benefits.
Last week saw two proposals regarding the economic data. One was to not count re-exports (goods that are imported then exported someplace else) as exports. This might make sense of you also don’t count the imported re-exported goods as imported. Otherwise, the change simply expands the reported trade deficit (some say this is the goal), without anything really changing. Needless to say, things get complicated when you consider that U.S. manufacturing relies on supplies, parts, and materials from around the world. Some items are shipped back and forth across borders a number of times throughout the production process. The proposed Border Adjustment Tax has generated a lot of pushback in Washington (in addition to threats of retaliation overseas).
The other proposal was related to budget forecasting. The impact of fiscal policy changes are calculated using static scoring. That is, the economy is assumed to perform the same regardless of any changes in taxing and spending. Depending on how far we are from full employment, tax cuts would normally be expected to boost growth. Dynamic scoring, which takes the expected growth impact into account, should give a more accurate assessment of the loss in tax revenue from cutting tax rates. The reason the government doesn’t do that is because it leads to abuse. One can always claim that growth will be so strong that tax cuts don’t reduce tax receipts. The Wall Street Journal reported last week that the Office of Management and Budget has been ordered to assume that long-term growth in real GDP will be 3.5% per year. That makes it easier to cut tax rates. In contrast, the nonpartisan Congressional Budget Office projects that real GDP growth will be a bit under 2% per year over the next decade, reflecting slower growth in the labor force (it’s also worth noting that about 40% of the labor force growth over the next decade is expected to come from immigration).
Investors need to make informed decisions. While the government’s economic figures are far from perfect, they provide a fairly reliable picture. Undermining the quality of those statistics is something no one should tolerate.
© Raymond James
© Raymond James
Read more commentaries by Raymond James
