Arnold Kling  

Wages Move Toward Equilibrium

Outsourcing = Economic Activit... Ideas and Growth...

Bruce Nussbaum of Business Week writes,

the surge in companies going to India, China, and Eastern Europe in search of very cheap brainpower may soon be coming to an end -- far sooner than anyone has anticipated.

Why? Simply put, the wage gap between the U.S. and Asia is shrinking. Pay scales are rising fast in India and China for college-educated, English-speaking professionals. As U.S. and European companies send more of their call-center, design, accounting, medical-service, and legal-service business overseas, demand for folks to work at these centers has soared.

And since these Indians and Chinese aren't anyone's fools, they've been demanding -- and getting -- increasingly higher compensation...

Global labor arbitrage is hard at work narrowing the international wage gap among educated workers. That may not be terrific for companies hoping to save costs by sending service operations overseas. But it's a good thing for Indian, Chinese -- and American -- workers.

The anti-outsourcing hysteria assumed an infinite supply of highly-educated foreign workers willing to work for coolie wages. The reality is that the supply is far from perfectly elastic, and markets work.

TRACKBACKS (1 to date)
TrackBack URL:
The author at Division of Labour in a related article titled Round-up of Interesting Stuff writes:
    A Clint Eastwood character once said, "A man's got to know his limitations." At the moment, one of this blogger's many limitations is that I have nothing more interesting on my mind than some of the posts I've found elsewhere.... [Tracked on September 27, 2004 12:53 PM]
COMMENTS (7 to date)
Lawrance George Lux writes:

Markets do work, but not quite as fast as suggested. Nussbaum has a habit of assumptions which require considerably more time than desired. One must remember Business Week proclaimed during the Nixon visit to China that the Chinese market would finally be open to American companies. lgl

William Woodruff writes:

Lawrence makes a great observation. I might add as well that, taken in isolation, the Global Labour Arbitrage has not fully worked its way through.

The advantage of a reduction in these (labour) inputs vanishes once all sellers have fully implemented (them).


Shaq writes:

Again its so easy for Economists to stand behind their formulas, and there rationalization that the markets eventually do reach equilibrium. There is a real and not necessarily measurable social loss and impact to families and individuals who are in the latter half of their working careers, and are certainly not as capable as younger workers of changing careers, and usually tend to have greater financial responsibilities.

the other bob writes:

Is the Offshore Labor Pool MUCH smaller Than Advertised?

Many of my colleagues tell me that hiring software programmers and IT administrators in India and other popular 'offshore' locations takes 3 to 5 times as long and requires 10 or 20 times the number of interviews as in the US and UK. I discovered firsthand in Malaysia and India, that the available, qualified labor pool seems to be much smaller than expected.

I propose below a few straightforward demographic reasons that the available and qualified hiring pool may be far smaller in offshore locations. These constraints apply even before considering the more often discussed cultural, communication and education differences. As a consequence, the hiring process can take far more hours and extend to a greater duration than might be planned.

Average Age

Indian cities have vast populations, but the population is very young. In both India and Malaysia, nearly 50% of the total population is under the age of 20, compared to 25% for the US and the UK. (

Of course, the US and Europe have a larger percent of population beyond what might be considered a typical retirement age and developing countries will have shorter life span. Still, assuming that 20 is likely the minimum age for the vast majority of people to have gained the education needed to enter the software programming labor pool, then the available labor pool is likely to be 20% to 25% smaller simply on the basis of age distribution.


When hiring for positions based in the US or the UK, your hiring pool includes most of the best-and-brightest from all of India and several other English-speaking countries, notwithstanding the stricter H1 and L1 visa rules in force today. In contrast, when hiring for positions based in India, the hiring pool is more restricted to the vicinity of your chosen work site for several reasons.

Indian cities, even the most modern, such as Bangalore, are not as attractive to highly qualified candidates as cities in North America and Europe. Also, at least in the in the opinion of some job applicants, it is more difficult to relocate from Delhi to Bangalore, for example, than to Boston. Apartment rents, relative to pay seem to be higher in Bangalore than Boston and accommodations in desirable neighborhoods are more difficult to come by.

Commuting to and from work in most large cities in India is quite time consuming. Trains are unreliable and traffic is far more congested than the worst situations in the US. This can further constrain the available hiring population to the area accessible by reliable transportation. In many cities in India transportation constraints can eliminate 50% of the entire population from your available labor pool.


Smart people are farther between in low-cost labor pools than in the US and UK. (See the table below: Average IQ by Country from Lynn and Vanhanen’s book IQ and the Wealth of Nations)

Now, I don’t think that IQ is a measure of a person’s worth as a human being. IQ is an entirely manmade construct, as is programming. But, IQ is a good predictor of success and productivity as a programmer.

What IQ level is required for software programming and IT administration jobs? There is a wealth of information, created largely for the Human Resources profession, attempting to estimate minimum IQ requirements for a plethora of skills and job functions. Minimum IQ requirements for enterprise-level software programming are generally estimated at about 110 to 115. Minimum IQ measurements for computer operators and administrators are generally estimated at 100 to 115.

The data from Lynn and Vanhanen set the median worldwide IQ at 100 with a standard deviation of 15 points. For those of you who remember a bit of statistics this means that approximately 68% of the entire population is below 115 (mean + one standard deviation or 100 + 15). Consequently, in a country like the UK where the national average IQ is 100, approximately 32% of the entire population has the minimum IQ required to perform the tasks associated with software programming. Since cities tend to attract people with higher IQ, the labor pool of a large city will have an even larger percentage with the minimum required IQ. (Though they certainly are not all in being the software programmers.)

In a nation with an average IQ of 85, this means that the minimum IQ required is now approximately 2 standard deviations above the national average. That means that approximately 95% of the population is below the minimum. Only 5% of the total population has the minimum IQ required to perform typical software programming tasks.

Look in the National Average IQ table from Lynn and Vanhanen (shown below) to see the national average IQ for nations that are popular software and IT job destinations.

(See for interesting perspectives on the affects of average national IQ.)


It is notoriously difficult to measure productivity in software programming. There is simply no good measure of productivity across projects; lines-of-code produced, function point analysis and many other measures each have their own flaws. However, there are many comparative productivity studies that measure productivity of programmers on the same problem. Those comparative studies show very large variation in productivity between the most productive 10% and the average 10% and the lowest 10%. The variation in productivity is on-the-order-of 10-to-1 or even 20-to-1. That is, the most productive 10% is 10 times or 20 times more productive that the average!

It is certainly an observed fact that a task some programmers will spend a few hours to complete will occupy others for a week or more.

If it is true that productivity is related to IQ, then the productivity of the labor pool will vary drastically across labor pools with different IQ distributions.


A salesman from one of the well-known Indian software outsourcing companies once told me that Bangalore was the San Francisco Bay Area of India. His comparison seemed apt to many of the other people in the meeting. Both Bangalore and the Bay Area have about 7 million residents. Both are known as cities where computer software is a local specialty. However, when you account for the difference in average age, geographic constraints, IQ and productivity difference, it is easy to see that the qualified labor pool in Bangalore is likely to be something less than one-fifth the size of the qualified labor pool in the Bay Area.

Average IQ per Nation
Data from Lynn and Vanhanen

Country AvgIQ
Hong Kong 107
South Korea 106
Japan 105
Taiwan 104
Singapore 103
Italy 102
Austria 102
Germany 102
Netherlands 102
Sweden 101
Switzerland 101
Belgium 100
China 100
New Zealand 100
U Kingdom 100
Hungary 99
Poland 99
France 98
Australia 98
Denmark 98
Norway 98
United States 98
Canada 97
Czech Republic 97
Finland 97
Spain 97
Uruguay 96
Argentina 96
Russia 96
Slovakia 96
Portugal 95
Slovenia 95
Israel 94
Romania 94
Bulgaria 93
Ireland 93
Greece 92
Malaysia 92
Thailand 91
Peru 90
Croatia 90
Turkey 90
Colombia 89
Indonesia 89
Suriname 89
Brazil 87
Iraq 87
Mexico 87
Western Samoa 87
Tonga 87
Lebanon 86
Philippines 86
Cuba 85
Morocco 85
Iran 84
Fiji 84
Marshall Islands 84
Puerto Rico 84
Egypt 83
India 81
Ecuador 80
Guatemala 79
Barbados 78
Nepal 78
Qatar 78
Zambia 77
Congo 73
Uganda 73
Sudan 72
Jamaica 72
Kenya 72
South Africa 72
Tanzania 72
Ghana 71
Nigeria 67
Zimbabwe 66
Guinea 66
Congo 65
Sierra Leone 64
Ethiopia 63
Equatorial Guinea 59

Dave Meleney writes:

China is graduating lots more engineers than we are and they are building a highway network that is cutting in half trucking times between cities. Forbes Sept 20 '04 (by Robyn Meredith) pages111-118: ..."Ford Motor, Cummins Engine and Briggs & Stratton have factories in Chongqing. Engineers there earn just $240 a month, versus $665 in Shanghai.

Population of Chongqing: 15 million. I'm not sure I'd expect the effect of low cost competition to moderate too soon.


John Thacker writes:

Shaq-- Yes, older workers do face special problems. Of course, so do younger workers who will have to face the impending doom of Medicare and Social Security. (Unless the older workers get screwed again.) And, of course, the outsourcing is completely in the interest of the poor in various countries who get these jobs, and raise their own income and standard of living in their countries rapidly. Your argument is just as much one in favor of the interests of a few wealthy (especially on a global scale) against that of the population broadly, and some of the poorest in the world especially.

Dave-- And the population of Shanghai is about 12 million officially as well, though with about 6 million migrant workers who lack the proper papers to live there, but live and work there anyway. If Shanghai's wages increased so rapidly, so can Chongqing's.

Sam Jew writes:

It's a fallacy to compare Shanghai and Chongqing on the basis of size. Shanghai is a much more international city, where many multinationals have their regional headquarters. Chongqing, on the other hand, is merely a regional hub.

Comments for this entry have been closed
Return to top