How to handle dividends in SPY

Discussion in 'ETFs' started by Eliot Hosewater, Jul 6, 2010.

  1. I've been playing around with analyzing historical data on SPX and SPY. I downloaded the data from Yahoo for ^GSPC and SPY. Since ^GSPC doesn't pay dividends and SPY does the results diverge quite a bit over 10 or 15 years. (BTW, I've searched this site and Googled around, but still haven't found what I'm looking for, but I admit my search skills are usually lacking).

    So how do people usually handle the difference? The SPY data is adjusted for divvies. Should I just go back and add them all back in on the dates they were paid? Is there any other better way to handle it?
  2. ptrjon


    the .spx is only concerned about stock prices. No adjustment is made for dividends paid.

    The SPY tries to track the .spx- but actually does earn the dividends, and it pays them out. So if you bought SPY and you didn't reinvest dividends, your SPY holdings should track the .spx pretty closely.
  3. So you're saying to ignore the divvies? If I run the test with SPX data and simulate buying and shorting "shares" of SPX the short term results track SPY pretty closely, especially during periods between dividends. Over longer periods the results don't track very well.
  4. dave4532


    what resuts?
  5. ptrjon


    SPX ignores dividends.

    SPY, if you're long you get paid dividends.
    SPY, if you're short, you pay margin interest and pay dividends.

    If you look at a price chart, SPY has been just about spot on for that time- I don't know what tests you're doing.
  6. I am playing around with buy and sell signals, and buying and shorting the S&P500. I wanted to go back before 1993 when SPY was first created, so I was simulating buying and selling the SPX (^GSPC on Yahoo). I would take a certain amount of dollars and divide by the current SPX to get the number of "shares" to buy. For example, on 5/1/1990 the SPX opened at 330.80, so $500,000 could buy 1,511 shares of SPX. On 5/1/1991 I could sell 1,511 shares at 375.35 for $567,153.

    Now let's compare with SPY (all prices are at the open):

    1/29/1993 buy SPY 11,371 shares@43.97=$499,982
    1/29/1993 buy SPX 1139 shares@438.67=$499,645

    3/22/1993 sell SPY 11371 shares@44.59=$507,032
    3/22/1993 sell SPX 1139 shares@450.17=$512,743

    3/22/1993 just happens to be the day after a .78 dividend on SPY.

    When I run my little simulation I was expecting the results to be fairly close between SPX and SPY, but they are pretty far off, like SPX will show a healthy profit when SPY can show a loss. They ARE pretty close if I choose a period where there are no divvies, but over a multi-year period they don't match very well.

    BTW, I tried adding all the dividends to the SPY prices over the years and the results still don't match with SPX. They do line up pretty well with SPY with divs and without.
  7. Like any ETF, SPY charges a management fee. SPX is a theoretical ideal.
  8. Thanks, maybe that's it.

    ETA: Actually maybe not. Their expense ratio is listed at 0.09. And you would expect it not to track the S&P very well over the years if the fee were excessive.
  9. ptrjon


    ETF's do not perfectly match the index. You are probably using the last trade of the day for SPY, which probably is not exactly trading at it's NAV. Also, you can't buy or sell the SPX, so why are you doing this?
  10. So I went back to it and decided to try buying SH instead of "shorting" SPX, and it gets even worse.

    On 9-30-2008 the open prices were

    SPX=1,113.78 SPY=113.51 SH=75.15

    On 3-5-2010

    SPX=1,125.12 SPY=113.37 SH=51.37

    SPX and SPY roughly track each other, but SH is 32% lower?

    What gives? These are non-leveraged ETFs that are supposed to track the index long/short?

    Maybe there is an arbitrage opportunity here?

    (BTW, ptrjon I believe I addressed your questions in earlier posts.)
    #10     Jul 20, 2010