Saturday, September 20, 2008

Qipit?

The biggest convenience of having a phone with a camera is in using it as a quick note taker of sorts. Say you're at a book store and see a book title that looks interesting. You're not ready to buy it yet. You want to go back and think about it a bit. Maybe run a few searches at your favorite online bookseller's. Catchup first with your existing reading stack. Or maybe wait for it to appear at your friendly neighbourhood road-side pirated bookseller.

You could whip out your pocketbook and pen and write down the details, except you don't carry either since getting out of school. Perhaps you could commit the title and author to memory, except your memory is not what it used to be and author names and especially book titles are getting stranger by the day. Why not just take a picture of the book cover on you phone and then look at it later?

Same thing goes when you see a notice at a local grocer offering some used furniture for sale. Or you happen on a hoarding for a new apartment block coming up down your road that you think someone you know might be interested in. The easiest thing to do is to take a snap on your phone rightaway.

Used to be, back in the day when a VGA camera on a phone pushed the price up some 300%, that the picture quality rarely allowed this to be any more than a supplement to, or perhaps encouragement for, one's memory. No longer. Most camera phones today offer resolutions close to what you can get with a budget desktop scannner. Mine does, at any rate.

So when I happened upon Qipit, a promising new camera-phone based document scan service, on a list of sites supported by ShoZu, I was instantly interested.  I went over to the site and got me an accoutn pronto.  I wish I could say the sign up was simple:  it was the most protracted and agonising sign-up process I've been thru (save for some Indian sites) in ages.  It wanted my phone number, and then it made me pick my phone model (which wasn't listed anyways), and as I was doing all this the form changed all over the place (pick a country and the carrier changes automatically; pick a phone and the country changes; go back and pick a country and everything gets reset to default).  

The site seemsed to suggest that I'm allowed only 100 photos to upload.  So there must be a premium account.  Must be, but I couldn't for the life of me figure out what it was and how much it cost.  It is also still not clear to me how the "Publish" feature works.  The site talks about it, but the instructions were not clear to me. I had to use a rather round about way to publish the results of my experiment.

I ran the experiment this Sunday morning. I took a couple of pictures of the day's paper (Hindu's Sunday supplement) and uploaded them to Qipit. Here are the pictures and right below them the PDFs that Qipit generated from them.

Tubingen Connection

 
PDF PDF
First, the bad news: there's no OCR! Maybe it's me, but surely OCR would've been the first thing to have put in?  Surely a lot many more people would wish for that over, say, the ability to fax their docs?

The PDFs do look a lot clearer than the originals, I have to admit, and are a lot smaller to boot.  In the second case ("Connection"), where I took the picture somewhat more carefully than the first ("Tubingen"), Qipit did a much better job in terms of cropping and centering the text.  

I'm disappointed and a little confused (will I have to pay $$ if I really get used to this? do I have to download/email the PDF manually everytime I need to share it?), but I like the idea of this service and am certainly not ready to write it off just yet.

Posted by ShoZu


Wednesday, September 10, 2008

Pricing Long-term Options on the NIFTY (Part II)

In my last post,   I ended up looking at long-term call options on the NIFTY mainly in terms of liquidity.  This one looks at their prices in greater detail.   my motivation is to come up with a decent, intuitively grasp-able, model for pricing these instruments.  That Black-Scholes fails badly here is quite appearent from the base priecs that one sees set by the NSE when these contracts are listed.  These computed base prices show up as the closing price on the F&O Bhav sheets until the first trade takes place.  In all cases that I've looked at the first actual trade happened at anywhere between 20-70% of NSE's base price.  Here's the evidence from the 6 contracts that we're tracking from the previous post:



Contract Last base Price First TradeRatio
24-Jun-10 5000 2001.95 880 44%
30-Jun-11 4000 2353.9 1585 67%
30-Jun-11 4500 2489.55 1470 59%
30-Jun-11 5000 2407.7 950 39%
30-Jun-11 5500 2236.75 600 27%
30-Jun-11 5700 2172.7 519 24%

As a first exercise, I made scatter plots of "Price" vs. "Underlying"  (no time order).  My feeling was that, since the expiries are way out in time, it really donesn't matter when the trade took place over the last 6 months.  All that really counts is that a trade took place on that day.  So I used close price as a proxy for "Price" and NIFTY's closing as a proxy for "Underlying" on all days when any trades took place.  The results appear below.

The biggest suprise for me was the fact that the data seems to fit a nice linear regression like some textbook physics experiment.   And the more trades a contract had (see JUN-2011 call @ 5000, for example), the better the fit appeared to be.

I used excel to do the fit and here are the equations it threw up (Price = A*NIFTY + B):



Contract A B
24-Jun-10 5000 0.4074 -1190
30-Jun-11 4000 0.6523 -1768.6
30-Jun-11 4500 0.652 -1965.3
30-Jun-11 5000 0.5677 -1788
30-Jun-11 5500 0.3034 -788.47
30-Jun-11 5700 0.3857 -1196.6


We're now closer we were than at the end of the last post to answer my original question: what is the delta on these things?  But we're nowehere near done yet.










(To be continued...)

Saturday, September 6, 2008

Pricing Long-term Options on the NIFTY (Part I)





NSE notified long term contracts on the NIFTY on Feb 27, 2008. In addition to the near, next and far month expiries, we now had 3 quaterly expiries and 5 half-yearly expiries. In effect, these contracts allow one to take a view on the NIFTY upto 3 years into the future. Trading started on Mar 3, 2008.



When these contracts were introduced, there was much doubt raised to their viability. It was pointed out that the volumes on far (and even next) month options were quite thin for the most part. In a ominous sign a month or two following its launch, ICICIdirect, India's largest brokerage by some measures, suddenly and inexplicably dropped support for these contracts from their trading platform. Press coverage was muted (see here). That NSE's most recent attempt at broadening the derivatives market, the so-called Mini-NIFTY contracts, had been an utter failure didn't help either.



Today, 6 months later, we may safely conclude that these contracts are here to stay. The volumes, while not spectacular, are very respectable. There is a good mix of trading and investment activity from the looks of the order books. Most importantly, investment companies such a KMIL are introducing hedge-fund style products that are built on top of these contracts. The table below gives the open interest position as on Sep 5, 2008 across all expiries.





Series Expiry Open calls (Rs. Cr.) Open puts (Rs. Cr.) Total O.I. (Rs. Cr.)
Near, next and far month25-Sep-2008 8650.319688.9318339.25
29-Oct-2008 441.891121.861563.75
27-Nov-2008 96.02139.05235.07
3 quarterly expiries25-Dec-2008 2037.883200.975238.85
26-Mar-2009 62.30226.47288.77
25-Jun-2009 494.62340.37834.98
5 half-yearly expiries 31-Dec-2009 326.64289.38616.01
24-Jun-2010 98.7366.62165.35
30-Dec-2010 318.56320.85639.41
30-Jun-2011 1272.46893.192165.65
29-Dec-2011 3.050.633.68



I am interested in these contracts as a means of taking leveraged bets on the NIFTY. As a firm believer in the NIFTY's long term growth potential, my options before Mar 2008 were between buying an index linked exchange traded fund, such as the NIFTY BeES, or buying and constantly rolling over NIFTY futures contracts. The former offers no leverage at all, the latter is quite painful to do without a large corpus and full-time staff. Long-term options provide leverage (high-delta) and require no maintenance. But what is the delta on these things? This is the question that I shall attempt to answer in subsequent articles.



For my study, I needed to pick a sub-set of contracts whose pricing could be meaningfully studied. As a long-only investor, I'm naturally only interested in calls. As a long term investor, only the half-yearly expiries interest me. But even after filterting for those two, I'm left with a total of 114 contracts across the 5 expiries. Of these only 44 have ever been traded. A glance down this list of 44, revealed that the action seems to be bunched up around a few contracts. I set the somewhat atrbitrary criteria that I'll picks only those contracts that have seen trades on at least a fourth of the days on which they have been listed and ended up with just six. The table below lists the 44. The ones I picked are in red.




ExpiryStrikeListed SinceDays ListedDays Traded Max O.I O.I
24-Jun-2010430010-Mar-2008121310050
24-Jun-2010440005-Mar-200812315050
24-Jun-2010450004-Mar-200812415050
24-Jun-2010490003-Mar-2008125778007800
24-Jun-2010500003-Mar-200812542185750185700
24-Jun-2010510003-Mar-2008125537503750
24-Jun-2010520003-Mar-20081252100100
24-Jun-2010540003-Mar-200812515050
30-Jun-2011350001-Jul-200847211501150
30-Jun-2011400018-Mar-200811550342200342200
30-Jun-2011410014-Mar-200811714075040750
30-Jun-2011420014-Mar-2008117293509350
30-Jun-2011430010-Mar-20081212115050115050
30-Jun-2011450004-Mar-200812456561550561150
30-Jun-2011460004-Mar-200812415875058750
30-Jun-2011470003-Mar-200812515050
30-Jun-2011490003-Mar-2008125152505250
30-Jun-2011500003-Mar-2008125102773400773400
30-Jun-2011510003-Mar-200812522505025050
30-Jun-2011520003-Mar-2008125192509250
30-Jun-2011530003-Mar-20081251100000100000
30-Jun-2011550003-Mar-200812562112000105150
30-Jun-2011570003-Mar-200812558487750469850
31-Dec-2009400018-Mar-200811535000050000
31-Dec-2009410014-Mar-20081172500
31-Dec-2009430010-Mar-200812135050
31-Dec-2009440005-Mar-200812315050
31-Dec-2009450004-Mar-200812410200150200150
31-Dec-2009460004-Mar-200812445050
31-Dec-2009470003-Mar-20081251125000125000
31-Dec-2009480003-Mar-20081253250050250050
31-Dec-2009490003-Mar-200812527500050000
31-Dec-2009500003-Mar-20081258400400
31-Dec-2009520003-Mar-200812512500025000
31-Dec-2009550003-Mar-200812515050
30-Dec-2010430010-Mar-20081213401500401500
30-Dec-2010450004-Mar-200812415000050000
30-Dec-2010460004-Mar-200812425020050200
30-Dec-2010480003-Mar-20081252250250
30-Dec-2010500003-Mar-200812515200400200400
29-Dec-2011360030-Jun-20084815050
29-Dec-2011400027-Jun-2008491150150
29-Dec-2011450027-Jun-2008493600600
29-Dec-2011500024-Jul-200830654005400



My arbitrary critera does leave out some contracts with very large open interest on account their not having been traded on many days. I've indicated some of these (those with O.I. greater than 1,00,000) in green. My problem with these is that since Iwont have price discovery on most days, any attempt to back calculate delta from these is likely to badly misfire. Much better to use my short-list and come back to these to see if the theory fits.



(To be continued...)


Friday, May 30, 2008

Sifting through the Infosys 217(2A) Table

Section 217(2A) of the "Companies Act, 1956" states:

[(2A)(a) The Board's report shall also include a statement showing the name of every employee of the company who-

(i) if employed throughout the financial year, was in receipt of remuneration for that year which, in the aggregate, was not less than 6[such sum as may be prescribed]; or]

(ii) if employed for a part of the financial year, was in receipt of remuneration for any part of that year, at a rate which, in the aggregate, was not less than 7[such sum per month as may be prescribed8; or]

9[(iii) if employed throughout the financial year or part thereof, was in receipt of remuneration in that year which, in the aggregate, or as the case may be, at a rate which, in the aggregate, is in excess of that drawn by the managing director or whole-time director or manager and holds by himself or along with his spouse and dependent children, not less than two per cent, of the equity shares of the company.]

(b) The statement referred to in clause (a) shall also indicate,-

(i) whether any such employee is a relative of any director or manager of the company and if so, the name of such director, and

(ii) such other particulars, as may be prescribed.

Explanation.- "Remuneration" has the meaning assigned to it in the Explanation to section 198.]

With effect from Financial Year 2004, Rs. 2,00,000 per month has been prescribed as the sum mentioned above. For FY07-08, Infosys had 486 employees that were employed for the full year and earned more than Rs. 2 lakhs per month. You can get all the gory details simply by downloading and looking at tehir latest Annual Report. Here's my study of this data.

Before I begin, some disclaimers.

  1. The data is by no means representative of average salaries in the industry. Infosys is not even the largest IT services company in India.
  2. The data is not representative either. It would be incorrect, for example, to conclude that engineers with 10 years of experience drew an average salary of Rs. 28 lakhs in 2007-08 based on this data. This is because we are provided with data only of those engineers that earned more than Rs. 24 lakhs between Mar 2007 and Mar 2008. For all we know, there are legions more of 10 year Infoscions earning a pittance hidden away in the depths within the deep dark underbelly of the Infyplex.
  3. The data does includes only those Infosys employees that are working in India.
  4. Both the aforementioned Section 198 of the Companies Act and the footnotes to the "Particulars of Employees" table in the annual report are silent on the matter of bonuses and stock options. The footnote in annual report I find particularly egregious.
    Remuneration includes basic salary, allowances, taxable value of perquisites, etc.
    Etc.? Excuse me?
  5. After cut-pasting the table out to excel (by no means an easy, error free task in itself), I carelessly sliced and diced it at will. I wasn't being thorough or careful. Remember, I was not, nor do I expect to, get paid for this. I might have made mistakes. I probably made quite a few.

I draw no conclusions, make no judgements. Please leave a comment if you do. Now on to the study.

First the big picture. These 486 employees that make the mark draw an average of Rs. 33 lakhs per annum in salary. Only 79 of these 486 are what one may call "Career Infoscions" -- i.e., people whose first job was at Infosys. 71% of them are Engineers, by which I mean they give B.E or B.Tech as thier first degree. Quite a few more clearly are engineers in that they're doing engineering roles; Several have a basic science degree, for example, followed by an MCA. There might have also been an occasional diploma holder or two, who later went on to do his BE that I missed.

The median Infosys engineer making this list is 37 years old having spent 15 years in the industry of which the last 10 were at Infosys. The median salary for this engineer is Rs. 29 lakhs. Pretty modest, eh?

Next, look at the designations. There are 2 directors, 2 CXOs, 1 Co-chairman, 6 SVPs, 25 VPs and 69 AVPs -- A total of 105 senior level executives (22% of all 217ers). Interestingly, while the average VP earns Rs. 12.7 lakhs more than the average AVP, the highest paid AVP actually earns more than the highest paid VP. Only 1 in 5 (20%) of VPs and AVPs are "career Infoscions", the rest are lateral recruits.

One question that I was interested in was in the presence of women in the list. The Infosys
annual report states that women constitute 31% of their total workforce. The 217(2A) table does not mention the sex of each employee, so I was forced to rely on guesswork and Google. By my reckoning, a mere 28 (5.7%) of the employees in the table are women.

Here's a table that lists all of the statistics that I collected by group:



[If the table above is not fully visible, it's not my fault. That would be a Google Docs mess up. Please see here instead]
On the whole, the picture that emerges from the data is less than satisfactory from an analyst's view point. There aren't any clear patterns and trends that jump out pf the page and grab you. Take a look at the histogram and CDF, for example:



My primary purpose as I set out to do this study was to see if there is any sort approximate relationship between number of years of experience and salaries. There is none. See for yourself:

I then tried to see if perhaps there'll be a correspondence of sorts if one looked at the employee's age or the number of years he/she had spent at Infosys. Once again, I drew a blank. The latter is scatter plot is particularly unhelpful. The points came out so scattered, I didn't even add a trend line.

On the verge of giving it all up, I decide I'd give one last try. This time, I'd look only at the
salaries of engineers. I would also track the median salary besides the minimum, maximum and average. (There is something good about medians, as against averages, that escapes me at the moment, but however it be) Here it is:

Now, this is only slightly more satisfactory than the previous two. Besides the guys with more than 24 years of experience were distorting the whole picture. So I narrowed my field of vision to only those with less than 20 years of experience. Here you have it:

I'm sad to report that many hours of effort have been wasted on this quite utterly useless piece of research. I had nursed ambitions of finding relationships and gaining insights that would help me when I next discussed the matter of salaries with my collegues. I would test these new found relationships against 217ers of different companies and gain even more powerful insights. I would plot the year-on-year trends between and across companies. Well. Friggin' waste of time! I vote they remove Section 217(2A) from the bloody act. Thank you for stopping by.

Saturday, March 29, 2008

Interest Rate Differentials

The universal practice appears to be to express interest rate spreads in absolute terms. Now why should that be?

Is it because interest rates are expected (guaranteed?) to lie within a narrow enough bound that the base does not make much difference? Unlikely. Both inflation and real rates are volatile enough, even over shorter periods of time, to give lie to that. Or perhaps the reasons are purely (or mostly) historical.

Or is it because spreads thus expressed are an indicator of bank profitability? If the bank is operating at a fixed spread, then it matters little what the cost of lending and borrowing are. The differential is the return on the asset base (which logically being a fixed multiple of equity) and thus the return on equity.

This explanation is slightly more credible, but fails to take into account changes to the business models of banks over the last decade as well as the involvement of players other than "banks" in the financial system. Not everyone is contrained to leverage on equity. In fact, I'd suspect that the number of players that gear up on cash flows far exceeds the former.

So if we're not to express spreads (or changes to base rates) as absolute differentials, then how else? Clearly, a purely relative change a la stocks holds less meaning than for the latter. Consider this: "The Fed today decreased its benchmark funds borrowing rate by 28.5%. The market expects further easing of between 33% adn 66% in the coming quarter", instead of: "The Fed today slashed its benchmark funds rate from 1% to 1.5%. The market expects futher easing of between 0.5 % or 1% over the next quarter".

I recently came across a passage in an article by Percy S Mistry in The Financial Express that brought out the absurdity of doing this (emphasis mine):

First, the Fed reduced rates to unprecedented levels. Then, it jacked them up
mechanically to five times the low rate over 21 months. It would be like RBI
raising rates to 30% in the next two years and not expecting trouble to arise!

Clearly Mr. Mistry is equating the tightening from 0.5% to 2.5% to RBI tighening from 6% to 30%. To me, the case under question is a lot closer to a scenario where RBI tightens from 6% to 8%. Not quite the disastrous policy that Mr. Mistry accuses the Fed of. But he does have a point: I'd expect borrowers to have been more adversely impacted by the Fed's action compared to RBI's analogous equivalent action by my definition.

Here's my proposal: express interest rates as (1 + r). This number needs a name. While I'm sure someone has already named it as well as made a similar proposal, it's hard for me to find this out (google search his not amenable to this sort of thing), besides it is an original proposal for me anyhow. So, I propose to call it "pips". A 5% interest rate is 1.05 pips, 0.5% would be 1.005 pips. Now, we can meaningfully express pips differentials as percentages. If the Fed tightens from 0.5% to 2.5%, that's a 1.9999% tightening in pips terms from 1.005 to 1.025. The equivalent RBI action would be to tighten from 5% to 7.1%, i.e. from 1.05 to 1.071 pips.

Wednesday, March 26, 2008

NIFTY Long Term Call Options: Tax Implication

Long term options introduced by NSE starting Mar 2008 allow one to take positions on the benchmark NIFTY 50 index over up to a three year period.

Brisk trading in select contracts. Lots of OI build up. Compared to earlier "innovations" in teh F&O space -- in CNX IT, Junior NIFTY, MINIFTY etc -- this one is clearly a hit.

Two things to notice about premiums: puts are a lot more expensive than calls and calls are often quoted at IV less than next and far month options of similar strike. So what's going on?


Clearly the call writers are writing covered calls and taking a position that NIFTY won't go below a certain level, rather than that the NIFTY will stay below the strike. The downside risk is low on the whole and it is probably quite easy to close out both positions should there be a deep correction and get away with a small loss. On the upside, the call premium goes to fund the margin now and can be taken out at expiry. But even this does not explain the low prices -- FD interest rates of 9.5% over the same period easily beats the strategy.

The real kicker is in the tax implications. The underlying that is purchased for margin, the NIFTY basket, is treated as equity investment , gains on which are tax free beyond a year. Naturally, the call writer expects to pay out the entire resulting gain to the call buyer at expiry. Now that is an income loss, offsettable against other income, and can be carried forward three years.

Here's an example scenario: I sell 1 European call contract at a strike of 5000 expiring in 25DEC2010. I receive a premium of 800 for this. NIFTY is currently at 4800, so I put in another 4000 and buy 1 NIFTY in cash and net this against the margin I need to keep for the open call. No tax implication in FY08-09. No tax implication in FY09-10 either. Now, on Dec 2010 NIFTY happens to be at 7000, say. I'm 2000 out of the money. So, I sell the 1 NIFTY that I hold, get 7000 for it, pocket 5000 and pay out 2000 to the call buyer. I book a loss of 1200 on the call.

I invested 4000. I receive 1000 at the end of 2.5 years. That's approximately 11% tax free return. But wait. That's not all. As far as the tax inspector is concerned, I made a long term capital gain of 1000 on equity which is tax free and a loss of 1200 which gives me a net 399 tax advantage. So its really as if I invested 4000 and got 1399 for it in 2.5 years. That's more than 15% post-tax return pa! The higher the NIFTY ends up, the more I stand to make (assuming I have enough other income to offset the tax loss against, of course).

And the tax free dividend (of around 2-3%) the NIFTY pays out over the 2.5 years. Now thats pure gravy!