Offshoring and Arbitrage
May 18th, 2007 : Austin SmithAbout six weeks ago, ArcStone discontinued its practice of offshoring indefinitely. Since then, I’ve been doing a fair amount of research into the advantages and disadvantages of outsourcing. It’s become clear that a rational examination of this problem cannot occur by examining outsourcing and offshoring without considering the larger economic context.
Arbitrage is the practice of taking advantage of price differences between markets on identical or nearly identical goods. Labor arbitrage is the practice of purchasing labor in one market and reselling it in another. The theory of labor arbitrage is that ArcStone can buy labor in India at, say, $15 per hour, and sell that labor locally for, say $125 per hour (neither of these are accurate, this is just an example). So, in theory, we make a $110 profit. See below for why that’s totally wrong.
In a Special Economic Study on Outsourcing, Morgan Stanley analyzed the present system of labor arbitrage and claimed that this trade likely to continue:
This phenomenon–a byproduct of IT-enabled globalization–is now acting as a powerful structural depressant on traditional sources of job creation in high-wage developed countries such as the United States. That means America’s jobless recovery could well be here to stay.
Morgan Stanley and others appear to make a dangerous assumption about the nature of arbitrage that may work in other areas but clearly breaks down in software–that in fact, the net offshoring a company does will effectively replace the net local work exactly. This assumption is so dangerous because it focuses on price and quantity only, rather than any measure of comparative quality.
Here’s a real world example: We hired a software contractor in India for $15/hour. He takes twice as long to do the work as our local contractor who will work for $21/hour. The loss is dramatic–$9 for every two hours he works. That’s only the beginning. Our Indian contractor takes twice as long, but also creates numerous problems that need to be corrected before we can show our software to our client. Let’s say that for every eight hours he works, we use another hour of our local contractor’s time to fix things and another hour to assign tasks to the offshore coder. Let’s also take into account the opportunity cost of the local coder’s time. In 80 local hours of work (meaning 160 contract hours), here’s what happens:
Local: 80 X 21 = 1680
Offshore: (160 * 15) + (40 * 21) + (40 * 125) = 8240
Quoted and billed to the client: (125 X 80) = 10000
Ouch! We’re losing $6560 of what should be profit every two weeks on this arrangement–far more than we can afford to lose because we’re counting on revenue from this project to pay more than the $1760 remaining in rents, payroll, and the various other bits and pieces of running a business. And this isn’t the only problem we had at the time, either, nor the problem of the worst magnitude. Additionally, in regards to quality, the IIT schools in India are the premier institutions for engineers–and our contractor was a graduate of one of them. English was generally not a problem, and he seemed to work diligently.
I don’t know what this is, but it certainly isn’t the labor arbitrage that Morgan Stanley is claiming. Perhaps other, larger companies can handle this problem by scaling their offshore centers up far enough to see benefits from having local management and good processes for communication. Anyone who thinks this can be done right may be assuming that knowledge workers overseas are as effective on American projects as Americans.
An important factor of arbitrage is eventual price convergence, especially for larger companies. But for us, paying $1/hour to our contractors would have been painful too because the opportunity cost of each hour of our locals’ time is so very high. Besides, happiness and fulfillment is important to us at ArcStone, and we were all constantly tense when we offshored so much of our programming.
To be perfectly clear, I’m not talking about outsourcing, I’m only talking about offshoring (outsourcing to people not in the USA).
I can’t claim that labor arbitrage is ineffective in the auto, textile, call center, or other industries because I don’t work in them. In fact, I can hardly claim that it’s ineffective for large software developers that build and sell discrete solutions and then sell them to multiple clients. But we’re not one of those companies yet, and we got on the train along with lots of other smaller companies, only to get badly burnt because we wrongly assumed that we would benefit from the promise of highly profitable arbitrage.
Thank you, Thomas Friedman.
Tags:ArcStone Development Management Offshoring Outsourcing productivity
May 21st, 2007 at 11:29 am
I think you make a fairly cogent argument. My take on why it didn’t work:
Labor arbitrage works well only when there is a comparative advantage and that does not only mean price. India does not have a broad comparative advantage in web design and development (which is what I assume your company does. This is the reason its all about ’services’ that are offshored rather than seeing products or a myspace being developed. So you might have better success in doing the creative onshore and then getting the ’service’ of it actually being built offshore ie this is the layout I want. Build it with flash here etc etc etc.
Of course as you correctly pointed out about price convergence. In 2003, the Indian call-center and IT workers were making about one-tenth what their American peers did. Now, just four years later, we appear to be up to one-fourth. So go to Estonia or some other offshoring destinations. Offshoring, ultimately is a short term play for quick savings, and by short term, I mean (depending on the industry) 5-30 years.
Or you just didn’t identify the right person. Just as in America you would interview and identify the ‘right’ person. the same thing holds true offshore. A contractor is a contractor is a contractor is being shortsighted. Skill sets matter and IIT’s are technical institutes - so hiring and IIT guy for web development isn’t probably the right choice. Thats like hiring a Art history major for accounting - which is not to say it can’t work but the odds of it working are higher. If the onshore contractor had the same percentage of ‘defects’ you costs calculations would dramatically change. So your comparison is between (and I’m making an assumption here) someone who you’ve worked with previously and does this for a living by choice and inclination. They know how you work, what you want and the local market vs someone who has n idea how you work, does not know the local market and who hasn’t been interviewed personally.
So the quality argument is kinda specious - like I bought the best pen available why can’t I draw.
Offshore smart and it might work. Going to a ’sofware’ company isn’t probably the best option since you pay the much higher rate due to the company overheads. How about trying the local classified ad’s, get some freelancers, paying them $3 an hour, communicating through Skype. It might actually work better. Sure there might be more defects and time but now you have upto 7 hours of ’savings’ to pay with and as you work longer (and better) together the percentage of ‘defects’ will drop.
May 21st, 2007 at 12:02 pm
And (sorry I had to point this out) the way you’ve used opportunity cost is totally misleading. If you choose to throw in opportunity cost, infact, you should be using the 40 hours saved of opportunity cost as a benefit - for you to use your local contractor on other ‘higher value’ work. Your calculations then look like this
Local: 80 X 21 = 1680
Offshore: (160 * 15) + (40 * 21) = 3240.
This is cost. If you didn’t offshore you’d have to have the local contractor do 80 hours of work. So you actually save 40 hours. Adding in the incremental benefit of those hours
= (40 * (125-21)) = 4040
Quoted and billed to the client: (125 X 80) = 10000
Profit local = 10000 - 1680 = 8320
Profit offshore = 10000 - 3240 + 4040 = 12360.
Ofcourse, technically this isn’t right too, since if you actually realize this profit then you’d be double counting.
As I said skill sets right!!! Leave the opportunity cost calculations to the Finance guys and the Economists :).
May 21st, 2007 at 12:16 pm
Oops that would be
Profit offshore = 10000 - 3240 + 4040 = 10800
May 22nd, 2007 at 4:28 pm
Thank you, Mr. Hoe, very much actually. You’re dead right about the opportunity cost calculations; I didn’t think those numbers all the way through. Rather than correct mine in the original post, I’ll leave them and simply admit my errors here as evidence of the process. I’m no economist, but I certainly could’ve gotten that right had I thought through it a bit more.
We’ve actually tried offshoring a few ways - we have, in fact, gone directly to locals, and in a couple cases it worked great until each, in turn, either disappeared or did something a little wacky. Also, we did turn to other locales–our “last straw” was in Belarus.
“A contractor is a contractor is a contractor is being shortsighted” is also very true. We were pretty targeted, however, and we said “no” to a fair number of contractors. And our IIT guy had 5 years demonstrable experience in web development prior to working for us.
The question of “why didn’t it work for ArcStone?” is still very much open I think, and additionally we may be partly to blame, but I’m still very uncertain that a company of our size, with the clients we have, could pull it off to any useful gains. And while I am still confident that offshoring should only be done cautiously, slowly, and with low expectations (I think you agree here?), you’re right about several things which I will certainly account for in any future argument I undertake with regards to offshoring.