“Don’t believe these phony numbers when you hear 4.9 and 5 percent unemployment. The number’s probably 28, 29 as high as 35. In fact, I even heard recently 42 percent.” Candidate Donald Trump, February 9, 2016.
When governments don’t like the numbers their statisticians report, they have two options. They can modify their policies with the aim of changing the trajectory of the economy. Or, they can push to change the data to conform to what they would like to see. In countries with trustworthy leaders, those who understand the value of objective facts, we see the former. In places where leaders think that facts are a matter of opinion, we all-too-often see the latter. Finding some economic facts inconvenient, President Trump’s inclination appears to be to change the data, not his policies.
Two recent news reports have been particularly troubling, one having to do with trade statistics and the other concerning growth forecasts. On the first, wishing to make it appear that the United States has a larger bilateral trade deficit with Mexico than it does, Administration officials are apparently exploring the possibility of changing the treatment of goods that are imported and then reexported—counting them as imports but not exports. Not only would this break the symmetry in the treatment of exports and imports, but it also means that the fundamental balance of payments identity will not hold in the data. That is, it will no longer be the case that the difference between overall reported imports and exports equals the measured net cross-border flow of capital. Let us simply say that this makes no economic sense. What other purpose could such a change have than to mask an inconvenient fact?
Turning to the growth forecast, we have quite a bit more to say, but our concern is the same: officials must fight the urge to disregard experience and facts in an effort to buttress support for their favored policies. Growth forecasts play a fundamental role in economic policymaking. Perhaps most important, they drive budget projections: stronger growth means both more revenue and less expenditure. When the economy grows more quickly, income and employment are higher. The result is more tax revenues and less need for spending on the social programs like unemployment insurance, Medicaid and the like.
Federal government budgets are planned over a 10-year horizon. Admittedly, this practice creates a perverse incentive for pushing challenging developments into the 11th year; the pre-set expiration of President George W. Bush’s estate tax cut after 10 years is a classic example. Nevertheless, the 10-year horizon is consistent with focusing fiscal policy on medium-term issues. With so many outlays pre-determined by entitlements and demographics, or by other multi-year commitments (such as military outlays), it is important to estimate revenues over a similar time frame.
The 10-year budgeting window gives rise to the need for a 10-year growth forecast. In the past, the Office of Management and Budget (OMB), the accounting arm of the U.S. government, has used numbers that were broadly consistent with (or only mildly more optimistic than) those of professional forecasters both inside and outside of the government. However, the Wall Street Journal recently reported that the Administration is planning to assume that the economy will grow in the 3 percent to 3½ percent range over the next decade, steadying at 3.2 percent in the outyears. (For simplicity, we’ll treat that as a 10-year projection of 3¼% average annual growth.) By comparison, the Congressional Budget Office (CBO) forecasts an average rate of 1.8 percent, the Federal Reserve’s Open Market Committee (FOMC) forecasts long-term growth in the range of 1.6 to 2.2 percent, and the Survey of Professional Forecasters (SPF) puts the number in the range of 2.1 to 2.6 percent. (A press report this morning does suggest that OMB is building the budgets for the two fiscal years starting October 2017 on a more reasonable assumption of roughly 2½ percent growth.)
Presumably, the purpose of optimistic projections is to mask a substantial deterioration in the long-run budget deficit. To see that the 3¼% projection is far too high, it is useful to decompose GDP growth into two basic components: the percent change in the number of workers plus the percent change in the output per worker, or labor productivity. The following chart shows the decadal average growth rate of these two quantities since 1950. The growth in employment is in black; the growth of productivity in red.
Components of GDP growth (average annual rate), 1950 to 2016 plus next-decade simulation
Source: Conference Board Total Economy Database plus authors’ simulation over next decade.
A few things are important to note. First, since 2000, the average annual GDP growth rate has been 2.0 percent. Second, prior to 2000, growth was quite a bit higher. From 1970 to 2000, for example, growth averaged 3.5 percent per year. But critically, this was the period when the baby-boom generation was coming of age and women were joining the work force in increasing numbers. (The working-age population grew by nearly 40 percent, and the female labor force participation rate increased from 42.7 to 57.5 percent.) As result, one-half of the 3.5 percent average GDP growth over those three decades was accounted for by the change in employment. Since 2010, two thirds of the growth has come from the rise in the number of workers, but this large contribution was in primarily a result of cyclical factors. With the current unemployment rate down to a long-run sustainable level, employment growth over the next decade will be largely determined by the growth in the labor force.
Given this experience, is it reasonable to expect that the U.S. economy will grow at 3¼ percent over the next decade? The short answer is no. The basic reason is twofold: (1) demographics (combined with immigration restrictions) will lead to modest growth of the labor force; and (2) the potential for policy changes to reverse the sustained slowdown in productivity growth is almost surely modest.
Let’s start with the demographics. With the population aging, and those (like us) born in the 1950s nearing retirement, labor force expansion has been slowing for some time. Even before recent changes in immigration policy, the Bureau of Labor Statistics (BLS) projected that the labor force would grow at only an 0.5 percent average annual rate through 2024. Assuming annual productivity grows by 2 percent, double the pace of the past decade, employment would need to grow 1¼ percent annually, or two and one-half times the BLS projected rate, to achieve OMB’s reported target of 3¼ percent GDP growth. Absent a shocking rebound in labor force participation, that rate of job growth would lower the unemployment rate to 2% around 2020 and to zero a few years later. This is obviously silly. (For the record, in the opening quote, candidate Trump’s 42 percent exceeds the fraction that is not employed of the entire civilian noninstitutional population—including children, the elderly and the disabled.)
Increased immigration would be an effective counter to demographic headwinds, but policy is pointing in the opposite direction. Why does immigration help? Immigrants tend to be young and educated, arriving ready to work and pay taxes. They have been coming at a rate of roughly 1 million per year since 1990; and in areas where there are labor shortages, foreigners should be welcome. The construction industry, with a 6 percent share in total national employment, is a prime example. Of the nearly 7 million people employed in that sector, 60 percent are foreign born. Is this a problem? Surely not. In fact, the average 2016 unemployment rate in construction was the lowest on record, with the average level of job openings at their highest. If, as the evidence suggests, the industry is facing a labor shortage, limits on immigration will make this worse, not better.
What about productivity growth? Is it reasonable to project a doubling over the next decade? We have experienced episodes of higher productivity growth in the past: for example, over the 10 years ended 2005, productivity growth averaged 2.6 percent per year. But the United States employs cutting-edge global technology in many industries. Consequently, the ebb and flow of productivity growth in recent decades has been determined largely by the pace at which firms—especially those engaged in substantial research and development efforts—push out the knowledge and technology frontiers. Compared to this largely random force, there is little evidence that government policies—even those involving large variations in taxes and regulation—alter the long-run productivity growth rate by more than a few tenths of a percentage point up or down. (This does not make us productivity pessimists; it just means that we can find no scientific basis for projecting a large pickup.)
So, if we accept the BLS projection of 0.5 percent labor force growth, assume like many observers that labor resources are today close to full utilization, and extrapolate the decade-long productivity growth trend of 1.0 percent, we would end up with a 10-year growth projection of only 1.5 percent (as shown in the chart above). If you think that is too pessimistic, consider that productivity growth since 2010 has been even lower, averaging 0.7 percent per year. That is, a projection of 1.0 percent already implies a substantial pickup from recent experience. (For a similar conclusion, see here.)
To put this into perspective, a 1.5 percent growth rate implies cumulative growth of 16 percent over the next decade. By contrast, the reported OMB forecast of 3¼ percent would mean that GDP in 2028 would be 38 percent above what it is today. In current dollars, that’s the difference between $21.8 trillion and $26.0 trillion. Other things unchanged, higher GDP is associated with higher tax revenues and a lower debt-to-GDP ratio.
Regardless of anyone’s views of the Trump Administration’s policies, we find the evidence of disregard for inconvenient facts disturbing. This includes the President’s misquoting of unemployment numbers, the reported efforts of his Administration to overrule long-established scientific methods for constructing trade data, and the reported plans to base long-run growth forecasts on implausible assumptions. This pattern of information distortion is one that prevails in places where governments have fewer institutional constraints, and it is destructive.
Without numbers that we can trust—not only labor and trade data, but GDP and its components, prices, financial information, and official budget projections—both private-sector investors and public-sector policymakers will make poorly-informed decisions. Moreover, if investors lose trust in U.S. data, and in the government’s projections, the resulting increase in uncertainty will drive up financial risk premia, including those on U.S. Treasury debt. The result would be an economy that operates less efficiently and delivers a lower standard of living that it can or should.