So, the US is on the mend, they say – but what about the invisible cut-backs?

Here's a riddle: If a scientist or engineer is laid off, does it affect gross domestic product? America's third-quarter GDP figures, released last week, showed the economy growing at 3.5 per cent annually, breaking a string of four consecutive negative quarters.

This number was greeted by many economists as confirmation that the recession is over. The rise in real GDP, combined with a sharp fall in employment in the third quarter, implies that productivity also soared during the period. Good news, right?

The trouble is that those GDP and productivity growth figures are probably overestimated by as much as 1 percentage point or more.

That's because the official statistics don't pick up cutbacks in "intangible investments" such as business spending on research and development, product design, and worker training. There's ample evidence to suggest that companies, to reduce costs and boost short-term profits, are slashing this kind of spending, which is essential for innovation. Yet you won't see that plunge reflected in the GDP and productivity statistics, which are focused on traditional sectors, such as motor vehicles and construction.

In effect, government statisticians are trying to track a 21st-century bust with 20th-century tools. Not only is that distorting the critical data that investors, policymakers, and corporate executives use to evaluate the economy, but it might also be creating a false sense of relief as Americans battle a brutal recession.

Here's a sobering sign that firms are robbing the future to pay for short-term profits: Over the past year, US employment of scientists and engineers – the people who create the next generation of products and make the US more competitive in the long term – has fallen by 6.3 per cent. Yet overall employment has fallen just 4.1 per cent.

That's a big problem, because the output of such well-educated workers has become a more important part of the American economy in recent years. New research by the University of Maryland suggests intangible business investment came to roughly $1.6trn in 2007, compared with about $1.2trn spent on tangible assets such as machinery and buildings. In 1995, the two were roughly equal. Going back further, tangible investments in 1985 were about 40 per cent larger than intangibles.

America's Bureau of Economic Analysis is taking steps to deal with the new realities. Software has been treated as investment since 1999, and the BEA plans to include R&D in the official GDP statistics in 2013. But the agency acknowledges that other areas of intangible investment still need to be worked into the numbers. "We think it's important not to ignore the fact that R&D is only part of broader innovative activity," says BEA director Steven Landefeld. For now, though, the US is navigating through the downturn with fragmentary information.

While the statistics don't account for it, there's good reason to suspect intangible investments are falling. Companies are under pressure to cut costs by reducing R&D expenditures and deferring other crucial intangibles.

At the same time, companies, especially those in the pharmaceutical industry, are moving more research to China, India, and elsewhere. They don't want to commit to costly investments if the economy remains weak.

One clear sign that GDP growth is being over-estimated is the sharp drop in venture-capital investment, which goes directly to new businesses. VCs invested about $12bn in the first three quarters of 2009, barely half the $22bn invested during the first three quarters of 2008. Some of this shortfall would have been spent on computers and other physical equipment, which would have been picked up in GDP. But most of the drop in VC money would have gone to pay for scientists, engineers, and new product development.

Similarly, many companies have taken a deep axe to reported R&D spending, which doesn't show up in GDP. Adding to the uncertainty, firms report their R&D only on a global basis. So, even though some are adding to such spending, there's no way to know how much of the increases take place in the US.

The stimulus package passed in February did include extra government funds for R&D. But even with this bump, a just-released analysis by the Democratic Leadership Council suggests total real spending on US R&D is falling for the second straight year. The labour market in particular shows the effects the fall in intangible investment is having, and it's not a pretty sight. In the manufacturing sector, non-production jobs – which include engineers, scientists, and other knowledge workers – declined at a 7.6 per cent annual rate in the third quarter, almost twice as fast as the loss of production workers.

Another big problem not reflected in the GDP statistics is that firms are retreating from development of new products, especially in stressed industries. Richard Shellabarger, 59, was a product development engineer. Before being let go in February he was working on the "next-generation air bag". He says: "I was trying to anticipate what the customer needed." In retrospect, Shellabarger worries that product development wasn't a good place to be in a downturn. "I suppose if you are looking to cut personnel, you don't want to short an area where you are delivering to customers right now."

R&D isn't the only type of intangible investment that seems to be declining. US companies spend $134bn worldwide on worker training – but they have been cutting back. The drop started in 2008, when employers reduced their per-worker "learning expenditures" by 3.8 per cent, according to the American Society for Training & Development. But some alternative sources of training are emerging. Autodesk, maker of AutoCAD design software, has started allowing unemployed architects, artists, designers, and engineers to download free student versions of many of its products to help them hone their skills while they're out of work.

Governments, too, are stepping up funding of training. Stony Brook University, part of the State University of New York, recently tapped federal stimulus funds to begin training displaced finance professionals for certifications in fields such as project management.

Measuring intangible investments such as business R&D and worker training isn't easy – which is one reason why government statisticians haven't yet done it. But including such expenditures could make a big difference in the way companies, investors, and others understand the economy.

Will intangible investments revive soon? Josh Albert of recruitment consultancy Klein Hersh notes that some firms are starting to hire biologists again. "It's the beginning stages of commitment to discovery," he says. But that uptick is likely to be negated by two big mergers in the pharma sector, which will result in huge layoffs and cuts in R&D. The hookup of Pfizer and Wyeth closed on 15 October, and the merger of Merck and Schering-Plough is expected to close in the fourth quarter.

Another major trend that will affect intangible investment in the US is outsourcing. "Companies want us to recruit foreign nationals from the US to work overseas," says Mr Albert. In effect, more and more of the global R&D dollar is being spent abroad, so imports of R&D are increasing.

This shift in intangible investment is also not well tracked by the GDP statistics, although it will depress employment of scientists and engineers in the US. But the economic impacts are murky. Corporate executives have argued that shifting R&D to China and India benefits the US by improving the efficiency of the research process. If you can pay two trained scientists in China the same as one in the US, they say, you can get twice as much research output.

For now, it's enough to note that the difficult environment for knowledge workers means GDP statistics are sending too rosy a message about the economy. And as the old saying goes: "If you can't measure it, you can't manage it." Until that improves, it's going to be difficult to know if the US is really on the road to recovery.

This article was first published in Business Week