October 17, 2003

Notes: Scratching My Head

I cannot figure out what the usually sound-thinking and highly-intelligent Stephen Roach is getting at:

Morgan Stanley: On the surface, there's no denying the unique character of this productivity-led recovery. In the first six quarters after the US economy officially bottomed in 4Q01, nonfarm business productivity has recorded a 6.7% cumulative increase. That's the fastest six-quarter post-recession rebound since that which occurred after the recession ending in 4Q70. Equally impressive, however, is the extraordinary shortfall in job creation that has occurred since the end of the last recession in November 2001. Private nonfarm payrolls have contracted about 1% (or 1.1 million workers) in the ensuing 22 months since that cyclical trough. That stands in sharp contrast to gains of about 5% recorded, on average, over comparable periods of the preceding six business cycle upturns. In fact, had the current cycle conformed to the prior-cycle norm, today's job count would be fully 4.3 million workers higher....

In other words, for a GDP recovery that has run at half its cyclical norm, it makes sense to consider the productivity implications of an alternative hiring trajectory that would have closed about half the gap between the current cycle and those of the past. On that basis, we can better assess the impacts of this jobless recovery on America's productivity -- results that stem from a shift in the fundamental relationship between aggregate demand and labor input.

If that presumption is correct, then as much as a third of the so-called productivity bonanza of this recovery can be attributed to a shortfall in domestic hiring. Absent that windfall, productivity growth over the first six quarters of this expansion actually would have fallen well short of its typical recovery profile. Obviously, that has not been the case in this jobless recovery. But that doesn't mean aggregate demand is necessarily being sourced by more efficient modes of global production that require reduced labor content. Instead, courtesy of a cross-border labor arbitrage, it may simply mean that there has been a substitution of foreign labor input for domestic labor input. For America, that has the effect of biasing domestic productivity growth to the upside. That's because conventional measures undercount the total labor input -- foreign as well as domestic -- required to generate a given product flow. Conversely, for foreign outsourcers, productivity growth may well be biased to the downside, as low-wage employment encourages more labor-intensive production schemes...

My big problem with this is that GDP is a value-added measure, and so labor productivity is not domestic demand divided by domestic labor hours--it is domestic demand minus the trade deficit, all divided by domestic labor hours. Outsourcing a job reduces the domestic labor-hours denominator, but it also reduces that domestic demand minus the trade deficit numerator. Unless there is a real efficiency gain to the U.S. from the transfer of employment--or unless imports are undercounted, and they cannot be *that* undercounted--the things Roach is pointing to should not affect measured domestic productivity.

Or so I believe...

Posted by DeLong at October 17, 2003 01:04 PM | TrackBack



Brad - Could you explain what Ricardo Cabellero had in mind in the essay you posted yesterday? I am puzzled as you are about Stephen Roach.

Posted by: anne on October 17, 2003 01:13 PM

Could multinational firm tax/transfer pricing schemes be messing up the productivity numbers more than before? But as with your point about imports, the scale of these potential problems doesn't seem right compared to measured productivity growth.

Posted by: P O'Neill on October 17, 2003 02:13 PM

Brad: I think you're right, in a static setting. From an accounting point of view, there's no way that importing services should make labor productivity higher, for exactly the reason you say.

My best guess to explain Roach's comments is that he has a dynamic process in mind. It *could* be the case (and I haven't carefully thought this all the way through yet, so I'm not positive) that if labor-intensive, low-wage employment is increasingly shifted overseas, the implied gradual shift of the US economy toward more high-wage, capital or knowledge-intensive employment will boost productivity.

In other words, perhaps Roach is thinking that by sending jobs overseas, the US is replacing jobs that produce $X of GDP per hour with jobs that produce $Y per hour, Y>X. Over time, that could cause US productivity numbers to look better.

Posted by: Kash on October 17, 2003 02:24 PM

By the way, even my explanation doesn't shed any light on his assertion that "conventional measures undercount the total labor input -- foreign as well as domestic -- required to generate a given product flow." I have no idea what he's referring to there.

I took my theory of what Roach was saying exclusively from the last sentence that you quoted.

Posted by: Kash on October 17, 2003 02:33 PM

I can't explain Stephen Roach either. Usually he writes clearly.

I have a problem that is similar to Roach's with the productivity numbers. I can not understand the high productivity reported.

And the way GDP is calculated makes me suspicious. My understanding is that a computer can be pumped into the GDP at 10 times the end user paid. (5 times faster than a computer 2 years ago and costs half of what it did two years ago is can result in the GDP value 10 times greater than the money paid for the computer).

My computer clock is 5 times faster but, from my viewpoint, the damn thing is only 10% faster maybe. The only saving grace is that it costs half of what I paid last time.

Posted by: David E on October 17, 2003 02:42 PM


Not sure how this gets counted in GDP statistics but let's say a U.S. multinational hires Indian workers to provide a service. Let's also say its invoice from the Indian subsidiary is either above the arm's length price or below it. For example, suppose that the Indian wages come to $100 million and the appropriate return to capital suggests another $5 million in economic costs aka accounting profits. But then for tax plays, the actual intercompany price is set either at cost plus 15% (high) or cost plus nothing (low). Is there anything to the measurement issues from non-arm's length pricing?

Posted by: Hal McClure on October 17, 2003 04:18 PM

Hal: You may have a point. If MNCs are invoicing their imports of intermediate goods into the US at less than the arm's length transfer price (ALTP), then reported output of the US arm of the MNC will be higher than it should be. Yet the same number of workers are employed by the MNC in the US, so it will look like the productivity of the US branch of the MNC has gone up. I think that you've identified a valid theoretical possibility to explain unusually high productivity growth in the US.

Three points regarding the empirical relevance of this issue: 1. To get enhanced productivity in the US, you need a situation where MNCs are trying to shift their profits to the US. That probably happens sometimes, but the presumption is more often that firms try to shift their profits out of the US and to other countries. 2. There is, of course, enforcement by the IRS and other national tax agencies. They generally require that firms use the ALTP. It's difficult to calculate this, and certainly there must be examples every day where a firm can get around the IRS requirements, but nevertheless it's not trivially easy to shift profits to the US. 3. The story you've described yields a one-time gain in productivity in the US. In order to show sustained productivity growth, firms would have to be getting farther and farther away from the ALTP every quarter, or more and more firms would have to be transferring profits to the US every quarter. I'm not sure whether it's plausible to imagine enough new firms initiating this practice every quarter to yield sustained productivity growth, quarter after quarter.

So as a practical matter, putting these three factors together, I'd be surprised if this effect can explain a measurable amount of the productivity puzzle.

Posted by: Kash on October 17, 2003 06:46 PM
Post a comment