Idle ramblings of a wandering mind...

I have written before about growing wind power capacity in India and also leadership roles of Indian wind power companies like Suzlon. Since that post in April 2006, the growth has only accelerated. Here are the numbers updated up to June 2006 (source).


Growth of wind power in India is quite remarkable. In fact, preliminary sources indicates that wind power capacity has reached 6018 MW in India as of Sept 2006. The wind power story is now getting attention of popular press as well. Following is an excerpt from the article in New York Times:

Suzlon has expanded rapidly as global demand for wind energy has taken off. Its sales and earnings tripled in the quarter ended June 30, as the company earned the equivalent of $41.6 million on sales of $202.4 million.

The demand for wind turbines has particularly accelerated in India, where installations rose nearly 48 percent last year, and in China, where they rose 65 percent, although from a lower base. Wind farms are starting to dot the coastline of east-central China and the southern tip of India, as well as scattered mesas and hills across central India and even Inner Mongolia.

As oil prices stay high, renewable energy sources are becoming more competitive. For this reason, I am not too concerned about oil reaches $100 or beyond - that will only be helpful in the long run as it will focus our attention on developing renewable energy sources like Wind.

Labels: Alternative Energy, India, Wind Power

   

In my last post on borrower selection at Prosper.com, I emphasized drilling into the delinquencies data that Prosper.com provides. It was in response to StingyJoe's question in his excellent blog - StingyFinance. StingyJoe was kind enough to put a link to my post on his blog post. I am continuing the chain of thought and detailing various kinds of data exploration you can do and helpful insights that you can gather.

If you are a member if Prosper, then you can access all the data on the Lending -> Performance page. You can change performance criteria and see the impact on default rates. Here are the results of my dabbling with this page (all these results are for loans originating between July 1 to Oct 20).

Rules for Investing in Prosper - II

There is a sweet spot: Most people believe that one should focus only on really high credit borrowers (AA, A and B rated) on Prosper because risks are too high. However, when you go with AA and A borrowers, you are likely to get an interest rate of 8-10% - not too different from what you could get from a CD with minimal risk. I believe that there is a sweet spot in the middle where interest rates are high and risk is lower than the reward. For me - this sweet spot is credit rating D. Look at default rates by different credit ratings:

Credit Rating---AA---A----B----C----D----E----HR---Overall
Default Rate(%) 0.00 0.59 1.05 2.72 0.67 4.38 6.28 2.36

Look at the amazing performance of D credit - its almost reaching levels of A rated borrowers. You can get an interest rates of 18-22% for a D borrower and for the performance shown above - it is a bargain. That's why if you look at my portfolio in Prosper - you will find that D borrowers have the biggest claim on my money. Look at the screenshot of my portfolio at Prosper:


Past predicts the future: If a borrower is carrying multiple current delinquencies on their credit right now then their cost of adding one more to the list is quite low. However, if a borrower has zero current delinquencies, then they would be extra careful to cross over to one current DQs. So - if you want to play safe - lend only to borrowers that have zero current DQs. Lets see what is the performance of different credit ratings if we limit ourselves only to borrowers with zero current DQs.

Credit Rating---AA---A----B----C----D----E----HR---Overall
Default Rate(%) 0.00 0.38 0.57 2.28 0.00 0.53 4.73 0.79

Wow! Look at that. Compare this table to the table in the beginning and see the difference it makes when we limit ourselves to lending to borrowers with zero current DQs. Performance improves for all credit grades. Improvement is dramatic for D credit: Zero Defaults!! So if you are lending to D credit with no current DQs - you can sleep soundly at night - your money is not going anywhere while earning you a solid 15-18% interest.

Greed is bad: The whole idea behind Prosper is that lender will bid down the interest rate to the one that the market can support (an economist will call it price discovery). Why then would a borrower opt for "Automatic Funding" which essentially completes the transaction the moment the loan is fully funded without giving other lenders an opportunity to bid down the interest rate. Its completely irrational for a borrower to have automatic funding, and since my inner economist core strongly dislikes irrational behavior, I stay away from loans with automatic funding. Lets see what that does to performance of loans:

Credit Rating---AA---A----B----C----D----E----HR---Overall
Default Rate(%) 0.00 0.78 0.70 0.52 0.40 2.35 1.81 0.76

Again - using simple logic gives us phenomenal improvement in performance. My favorite D credit borrowers are again providing really good performance for comparatively lower risk.

Okay - so to summarize: if you are lending at Prosper.com - keep in mind the following: D credit borrowers are your best risk-reward bet at the moment, borrowers with zero current DQs are outperforming others by miles and finally - stay away from loans with automatic funding.

Hope all this is of help. As I look deeper into this data I will post updates to how to select the right borrower on Prosper.

Labels: Investing, Prosper

   

This post is in response to a question by StingyJoe in reply to my comment on his article on Prosper.com on his excellent blog StingyFinance. He asked me what are the most important criteria for choosing who to lend to on Prosper. Here are the heuristics I have been using, and they have served me well till now.

Rules for Investing in Prosper

Start small and build incrementally: Prosper is right now a small site with total loans outstanding of around $23 million. While its growing rapidly (new loans of $2 million between Nov 1 and 21), there are so many lenders competing for good loans that it is really difficult to grow your portfolio while keeping a high interest rate. The "gems" (low credit risk with high interest rates) are few and far between. So do not think that you can lend out $20K in one month and remain profitable. Right now I think maybe 10 loans every week fit my risk-reward criteria and if you are bidding $50-$60 on every loan - you can loan at most $500-600 every week. Now, this may be too small for people with little free time who can't do the necessary research for such small investments. However - I recommend that before you put too much money - put a little first, build very gradually and then once you are comfortable, put more money in. I am also in favor of starting off with manual order and using automated standing orders only after 5-6 months of getting to know this market.

Wait for the 4 month shock: The Lender's Forum at Prosper is abuzz with lenders who see a spike in delinquencies about 3-4 months after the loan originates. Basically it takes that much time for the excess liquidity of a new loan to die and some borrowers to fall again in financial difficulties. Some data to prove my point:

When looking at all loans that started between Jul 20 and Aug 20 (its Nov 22 today, so a loan life of 3 months), default rates are 4.63%. If we look for all loans that started between Jun 20 and Jul 20 (loan life of 4 months) 6.19%. The same figure for 5 months loan life is 7.98%, 6 months loan life has the figure 12.29%. Compare these to default rates of 1.29% for 2 months loan life. So clearly, if you have not had defaults on your first two months, do not consider yourself a Prosper genius and start pouring money. Wait for the 4-5 month default deluge - see how you are performing against the market and then, if you are comfortable, increase your exposure.

One should also note that historic default rates in Prosper (12.29%!! for 6 month old loans) is huge! So build that into your expectations. Just because you are lending at 20% rate does not mean that you will have 20% return. Expect something like 8-10% and you should not be disappointed. I personally think that with careful borrower selection a 15% return is possible and thats what I am shooting for but I know the risks and I would be happy if I am making 8% at the end of it.

Know your data, love your data: Good thing about Prosper is that all (most?) its information is online and available for your to see. For example: I can check default rates for all kinds of combinations of borrower characteristics (Credit grade D but no current defaults is my current favorite) and decide my bidding strategy. For example: if you limit yourself to bidding only on borrowers with no current delinquencies, your overall default rate at 4 month loan age is only 0.34% (compared to 6.19% overall). So if you stuck only with borrowers who do not have a current delinquency in their credit file, you would have way way lower defaults. There are several other ways to slice and dice the data and you should spend enough sleepless nights on the computer to *really* know that data. This is the key to succeeding in this market.

Groups don't matter, Group Leaders do: Most groups on Prosper are "zero-value-add" but a few groups have committed group leaders who make a difference. Over past few months, I have recognized some groups with capable group leaders including Texas Prosperers (I am a honorary member of this group even though I have never set foot in Texas) and Sonic Lenders. I am also impressed with DocProsper group and I *hate* this group.

Alright - now the disclaimers. I am a small investor and have only been with Prosper for a few months - so all my "insights" should be taken with a pinch of salt. I am putting my money where my mouth is - but it may not work for you. Also - all my insights may prove to be nothing if my loan portfolio starts backfiring and I am saddled with defaults. I will be posting updates to my Prosper portfolio here - so check back if you are interested in learning how my heuristics are faring.

Have a happy Thanksgiving and hope Notre Dame beats USC this Sat so that Michigan (Go Blue!) goes to the National Championship games and kicks Ohio State's behind! I can't believe I am cheering for Notre Dame!!

Update: I have posted a follow up of this article and its available at this link.

Labels: Investing, Prosper

   

I came across this really nice website that helps catalog personal libraries: LibraryThing. I always wanted something like these to keep a record of all the books I buy or read. I am slowly indexing all my books. You can check my library at this link.

The site also provides this wonderful widget to put in blogs. You can configure this widget to show many interesting things. Right now if you look on the right hand column of this blog - you will see a random selection of books from my library. You can click on them and that takes you to Amazon, where you can read about the book.

Labels: Blog, Books, Technology

   

Face recognition technology has been improving rapidly for past few years. Advances in face recognition, and pattern matching in general, has significant potential for new products in security, personalized service, etc. A proof positive of improvement in this technology comes from this site: MyHeritage. It provides a wonderful free service that matches your face with faces of celebrities and tells you which celebrities your face matches best against. For example: here is what it had to say about me:


Considering that John Cleese is one of my all time favorite actors, I don't mind the above comparison at all. More impressive than the final result is the process though. You can upload any kind of complex image and the site can "identify" a face in that. I tried feeding the site images where my face is in the middle of complex patterns, and it successfully identified a "face" in the middle of all the clutter.

Good Stuff!!

Labels: Technology

   

How does a stock market return of 61% in 8 months sound!! That's my current rate of return, calculated on time weighted average capital employed. I had some free cash in April 2006 - so I started putting some money in my Scottrade brokerage account. I have been consistently adding to my capital and have been following my time tested investment strategy (detailed here) with some modifications. And here I am now!! Even my debt portfolio (at Prosper.com) is doing well. Its a small experimental portfolio with average interest rate of ~20% with zero defaults as yet.

So what am I doing so well that I am getting such unbelievable returns??

• Have faith in emerging markets: When stocks in developing markets crashed a few months back - I loaded up on them. Since I knew India's economic scene very well - I picked up two winners that I have recommended before as well: Tata Motors (TTM) and ICICI Bank (IBN). Both have given me phenomenal returns since then. I also loaded up on TurkCell (TKC) when bad news brought the stock down. TKC is the marker leader and all the management news did nothing to the fundamentals - so buy away.
  
• Patience, Patience, Patience: I bought Seagate (STX) for the first time in April at 26.05. I was convinced that market is beating it too much for Maxtor acquisition - which will in fact be beneficial in medium to long run. However, the stock kept dropping - and I kept picking up additional stakes after each 5-6% drop. My last buy was at 19.95. After that the stock has now climbed up to 25.34, giving me a pretty decent return. A couple of years back I would have panicked and sold off - but not this time. I am taking my losses, sticking with my stock picks and picking additional stakes after each decline.
  
• Discipline is the key: The trading strategy I follow is simple. Here are the steps:
1. Select a stock. Look here for what constitutes a good stock. This is the most crucial step.
2. Buy the stock (say 200 shares). The price at which you buy (say x) is your anchor point.
3. If stock price reaches 6% higher than the anchor (i.e 1.06x), sell 100 (not the whole 200) shares. Change your anchor to this new trade price.
4. If stock prices reaches 6% lower than the anchor (i.r 0.94x), buy 100 shares. Change your anchor to this new trade price.
5. Unless something fundamental has changed (which is a rare event), go back to step 3. Continue till either you have no cash to buy or no holdings to sell.
• Simplicity pays: The above strategy is simple, almost too simple, but it works as long as you do not put your emotions on line. The 6% limit is written in stone - do not change it. When price drops - buy against all your fears and when price increases, sell against all your greed. This has worked for me - I have put this to many stock price data as simulation and almost always the results are phenomenal. Although everything depends upon selecting the right stock in Step 1 - so you gotta know the industry and know the business - do your research!
• Tradings beats long term investing: I know I am going against conventional wisdom here. Everyone says - buy and hold, hold for long long long - but as someone said - in the long run all of us are dead. The long run holding scheme has two big problems - one, it does not take advantage of stock volatility and second, second, it does not book profits - your gains continue to be in paper form which can get wiped out by one fall. My trading strategy works *really well* with volatile stocks and actually more volatile, better it is, as long as long term trend is positive. And you periodically book profits - thereby reducing your incremental risk. You end up paying more tax on your gains - but hey - if you are a graduate student like I am and hence fit into the low-low-tax-rate category - capital gains tax does not hurt that much. Further, low commissions at Scottrade help too!
  

I am going to continue following my strategy and post updates on how my portfolio is doing.

Labels: Investing, Prosper

   

I am a teaching assistant for BIT 582 - Enterprise Systems Strategy course at Ross. During last class the class discussion went towards a remarkable analysis - which countries has more people looking for information about cutting edge technology . After all, tomorrow's technology leadership will be decided by how many people in which country are knowledgeable about advanced technologies. So here are some startling information from Google Trends, which tracks origins of search requests for popular search terms.

So, to start, lets pick a popular and generic advanced technology: Nanotechnology. There is no doubt that Nanotechnology will be exceedingly important in future. So - which cities have most people looking for information in Nanotechnology? Google trends provides the answer:



Whoa! Look there - of the top ten cities, none are in developed countries, six are in India and all ten are in Asia! May be top cities is not capturing the full information - lets look at the same data by Countries, rather than Cities.


This is definitive now - Developing countries are the one most important in finding out more about cutting edge technologies. US comes 8th in the list - quite troublesome, considering that its current "adversary" Iran is at the 2nd spot. BTW - this is not happening just because developing countries have more people than the developed world. While overall population is surely bigger, the number of people with Internet access is developing world is smaller than that in developed world.

Just for fun, lets take something from manufacturing, say "Lean Manufacturing" - very effective and well known concept. Which countries are looking for information on Lean Manufacturing? Lets also look at something more specific, say "Service Oriented Architecture". Google Trends provides the the following information:



Chart 4: Searching for Service Oriented Architecture: Top Countries (from Google Trends)

The trend is consistent - Developing countries are more inquisitive about technology then developed countries. India seems to be right on top on every technology search. Is this a cause of concern for developed countries? I think so. These kind of things are the leading indicators, the weak signals that portends a structural shift in balance of technology leadership. A few more years of this and we will have a developing world well versed in technology and a developed world ignorant of the same. What makes it even more striking is the fact that *all* advanced education and technology facilities are in developed countries and even then we have this disparity. As developing world improves their education systems and builds technology facilities, this gap will only increase and the trend will get even more momentum.

Some caveats are in order too: Google is mostly used by the English speaking world - that's why China is conspicuously absent from all the charts above. I am quite sure it will be neck-to-neck with India (or even above) if all search engines are taken together.

My take on this: the four billion poor people of the world have finally information at their disposal (Thank You Internet!) and they are voraciously consuming all this information. The knowledge barrier that kept them down has vanished - and they will be down no longer. We are looking at the emergence of a global knowledge base - that will further feed globalization, and the developed countries are looking at strong (and knowledgeable) competition for tomorrow's technology leadership.

Labels: Development, India, Technology, US

   

One of the reasons I decided to do a PhD was that I wanted to teach. Now, PhD programs are usually a lot more about research than teaching. However, I have managed to do fair bit of teaching and teaching related work. Details of my teaching activities are on my teaching page.

I independently taught the BIT 311 - Decision Support with Excel (for BBAs) course this spring. I was the teaching assistant for the BIT 582 - Enterprise Systems Strategy (for MBAs) course last fall and I am continuing with the same responsibility this fall as well. In my two years with this course, this course has seen more than a doubling of enrollments and a significant increase in evaluations.

I have had the opportunity to write some teaching case studies as part of the Center for Global Resource Leverage: India. These case studies (and the corresponding teaching notes) are available on request for potential teaching or other non-commercial use. Many of these cases are already being used at Ross (especially in BIT 582). Following are the case studies which are substantially finished:

• Evalueserve: Outsourcing Expert Knowledge Services
• 24/7 Customer: Business Process Outsourcing
• IT Enabled Business Processes for Rural Retail Ecosystem: The Case of ITC
• ICICI: Leveraging Technology in Financial Services
• Madras Cements Ltd.: Achieving Operational Efficiency through IT
• Marketics: Providing Actionable Marketing Analytics
• A Different Story on Joint Ventures in Emerging Economies: TVS-Suzuki

I hope all this teaching experience will come handy when I start teaching as a business school faculty.

Labels: CGRL:India, Teaching

   

All these while that I have been rather dormant on this blog, my research has moved into top gear. With my co-authors, I submitted two papers to top journals and had three papers accepted at various research conferences in past three months or so. Here are the details:

The Alignment Between Software Structure and Collaboration Network Structure In Open Source Projects
Submitted to Information Systems Research
Ning Nan, Sanjeev Kumar and Li Wang

Abstract

In this study, we view open source software (OSS) as digital networks of processing elements (e.g., function and classes) and open source development teams as social collaboration networks of developers. By integrating theories and methods from social network, software engineering and organizational behavior studies, we examined the information processing need implied by structure of the digital networks, the information processing capability defined by structure of collaboration networks and the alignment effect between the two on developer productivity. With development data from SourceForge.net, we found that the alignment between the digital and the social networks is not always proper. Particularly, in projects with high software structural interdependency, the current collaboration network structure (larger and less centralized) is associated with lower productivity. Our findings indicate that OSS practitioners still need to deliberately manage collaboration structure of the development teams to enhance productivity. Furthermore, we found that the negative effect of collaboration network size on productivity was diminished in projects with high software structural interdependency but low requirement ambiguity. Therefore, eliciting clear and complete project specifications is a potential remedy to the difficulty of aligning software structure and development team size.

Antecedents and Performance Outcomes of Onshore and Offshore Business Process Outsourcing
Submitted to MIS Quarterly
Jonathan Whitaker, Sunil Mithas, Sanjeev Kumar and M.S. Krishnan

Abstract

Firms are increasingly utilizing onshore and offshore business process outsourcing (BPO) to manage their operations and achieve their strategic objectives. However, there is a lack of understanding of when firms are likely to use BPO, and when firms are likely to see business value from BPO. Building on previous research in transaction cost economics, business process management, strategy and international business, this paper proposes a theoretical framework for adoption and benefits of onshore and offshore BPO, and tests the framework using data on U.S. firms. Our analysis indicates that firms with broader IT application deployment and stronger business process codification are more likely to engage in onshore and offshore BPO. We also find that firms with broader IT application deployment are more likely to experience both cost and quality benefits from BPO.
This study makes two primary contributions to the emerging literature on BPO. First, we draw on previous research in transaction cost economics, business process management, strategy and international business to develop and validate a theoretical model for both the antecedents and performance outcomes of BPO. We extend previous research to understand the relationship between organization characteristics and adoption and benefits of the emerging practice of onshore and offshore BPO. Second, we establish a link between IT application deployment and BPO adoption and performance outcomes. This research highlights and validates the role that IT plays in enabling firms to locate their business processes across vendor and geographic platforms, and in facilitating cost and quality benefits from BPO.


Empirical Analysis of Determinants of Perceived Customer Satisfaction with Open Source Software
Accepted for presentation at the OASIS 2006 Workshop of IFIP Working Group 8.2
Sanjeev Kumar and M.S. Krishnan

Abstract

Open Source Software (“OSS”) has attracted significant research interest but research has focused on the development process (“supply side”) of OSS leaving the “demand-side” of OSS relatively unexplored. Further, extant OSS research is lacking in empirical studies. In this study we fill the gap with an empirical analysis of drivers of perceived customer satisfaction with Linux, the most popular OSS product. We used Ordered Logit technique to analyze a dataset collected through a survey of business-technology professionals. Our results suggest that perceived customer satisfaction with Linux is positively influenced by duration of use and the quality of third party support. We also found a strong relationship between perceived satisfaction with Linux and prevalence of other OSS products in the organization. Further, we found that perceived customer satisfaction with Linux is lower for large firms and higher when the largest distribution of Linux is used. Interestingly, internal IT support capability was found to have no significant effect on perceived customer satisfaction with Linux. Our results contribute to research by verifying traditional customer satisfaction models in an OSS context and also by extending these models to include unique aspects of OSS. On practice side, this research provides directions to the OSS community for achieving higher customer satisfaction.


Do CEO Characteristics Influence the Technology Strategy of a Firm?
Presented at INFORMS Conference on Information Systems and Technology (CIST) 2006

Srinivas Kudaravalli, Sunil Mithas and Sanjeev Kumar

Abstract

This study provides evidence for the link between CEO characteristics and technology strategy. We use IT and R&D spending in firms as evidence of strategic orientation to establish this link. The results of this study show that CEO characteristics account for a significant amount of variation in technology spending even after controlling for other firm level variables. In particular, the CEO’s career experiences in marketing and technical (R&D, Engineering or IT) functions and CEO’s stock holding in the firm have a significant impact on technology spending. We use these results as well as the comparison between IT and R&D spending in our sample of firms to illuminate the effect of top management on technology strategy.

Performance Outcomes of Onshore and Offshore Business Process Outsourcing
Presented at INFORMS Conference on Information Systems and Technology (CIST) 2006
Jonathan Whitaker, Sanjeev Kumar and M.S. Krishnan

Abstract

Firms are increasingly utilizing business process outsourcing (BPO) to manage their operations and achieve their strategic objectives. While researchers have noted the increase in BPO since the 2000 timeframe, there is limited work on the implications of BPO for firms. To study the performance outcomes of BPO, we first develop a theoretical model based on literature from operations management, performance measurement, and vendor management. We then validate our theoretical model by performing an empirical study of data from 49 publicly traded firms in the U.S.
Consistent with previous research in operations management and software development, we find that a firm’s performance measurement focus and strategy dictates whether the firm achieves quality, cost and/or time benefits, and that quality benefits from BPO lead to cost and time benefits. These results suggest that firms should place a strong focus on quality when evaluating processes and vendors for BPO, and that firms should incorporate a robust performance measurement system to ensure that their strategic objectives are achieved.
This study makes two primary contributions. First, we integrate literature from operations management, performance measurement and vendor management to develop a theoretical model for quality, cost and time benefits from BPO. Second, this paper extends the emerging literature on BPO. While recent research has investigated BPO benefits from the analytical and case-based perspectives, our empirical research investigates actual performance across a broader range of firms.

Labels: Conference Presentations, Economics, FLOSS, India, Journal Submissions, Research