There seem to be two schools of thought regarding dividend adjustments. One school of thought argues that dividends are part of the return and price data should be adjusted to include dividend payouts. This allows chartists to see the “total return” picture. Others argue that the dividends were paid out and should not be factored back into the price data. Furthermore, dividend adjustments result in artificial price levels that do not reflect true price levels at the time.
So who’s right? Today I will weigh in on this issue with charts for comparison. In particular, I will look at some long-term issues and then compare price data the day before the dividend adjustment with price data the day after the dividend adjustment. It is important to understand that historical data actually changes after adjustment. This means data you were looking at one day will be different after the dividend is factored into to historical data the next day.
Spoiler alert: When analyzing price charts, I prefer data that is adjusted for splits, reverse splits and special dividends, but NOT adjusted for ordinary dividends. The adjustment issue is part of the reason I am moving to Optuma for charting.
Industry Standards
The industry standard for charting sites is to adjust for capital reconstructions, such as splits and reverse splits, and special distributions, but NOT ordinary dividends. This is the default data setting for BarChart.com, TradingView.com, TC2000.com, MarketWatch.com (WSJ), Investors.com (IBD), Bloomberg, Reuters, Nasdaq.com and NYSE.com.
StockCharts is different because their default is to adjust for capital reconstructions, special distributions and ordinary dividends. Everything. Unadjusted data at StockCharts is not adjusted at all (not for splits, reverse splits, special dividends or ordinary dividends). Chartists can enter the normal symbol to chart adjusted data at StockCharts (SPY) and precede the symbol with an underscore to chart unadjusted data (_SPY).
Data must be adjusted for splits and reverse splits. This is a no brainer. Data should also be adjusted for special distributions because they are usually quite large and are one-time events. For example, a $60 stock may have a $15 distribution because of an asset sale. Prices should be adjusted for this distribution to prevent a massive hole in the price data.
Ordinary dividends are different because they are regular recurring payouts. This means data adjusted for dividends is subject to change with each dividend. Moreover, each dividend adjustment affects the entire historical price series. These historical adjustments create artificial highs and lows that are not based on an actual trade. Moreover, these price discrepancies grow over time.
Now let’s hit the charts so you can decide for yourself.
Adjusted Prices do not Reflect Reality at the Time
To get an idea of the long-term effect of dividends, let’s start with two examples using the S&P 500 index, the unadjusted S&P 500 SPDR (_SPY) and the adjusted S&P 500 SPDR (SPY). The first chart focuses on the period from 1999 to 2013 and shows at least three discrepancies. First, adjusted SPY (with dividends) hit a new high in November 2006, almost a year earlier than the S&P 500 and unadjusted SPY (without dividends). Second, the S&P 500 and unadjusted SPY closed below the 2002 low in February 2009, while adjusted SPY did not. And third, adjusted SPY recorded an all time high in September 2012, around six months ahead of the S&P 500 and unadjusted SPY. Personally, I would rather take my cues from the S&P 500 and unadjusted SPY.
Also notice that the price scale for adjusted SPY is different than the price scale for unadjusted SPY. The last price date on the chart is on 11-April-2013. There was an actual trade at 159.19 on 11-April-2013 and this is where SPY closed that day (unadjusted). Adjusted SPY, in contrast, shows the closing price to be 138.50. Sorry, but SPY did NOT trade at 138.50 on 11-April-2013 and did not close at this level on that day. This value equals the unadjusted price plus all of the dividend adjustments. It is an artificial price.
According to adjusted SPY, the ETF traded below 60 in 2009. In reality, this was not the case because there were NO trades below 60 in March 2009. It is an artificial low created from the dividend adjustments. In contrast, notice that the S&P 500 did not break 600 and unadjusted SPY did not break 60. These two numbers are not even on the price scale. Here are some recent lows for comparison. You can see that the S&P 500 lows equate to the unadjusted SPY lows.
Data for 10-October-2002:
- SPY adjusted artificial low = 54.19
- _SPY unadjusted true Low = 77.07
- $SPX true low = 768.63
Data for 6-March-2009:
- SPY adjusted artificial low = 53.38
- _SPY unadjusted true low = 67.10
- SPX true low = 666.79
Data for 9-February-2018:
- SPY (adjusted) Artificial low = 242.17
- _SPY (unadjusted) true low = 252.92
- $SPX true low = 2532.69
The next chart shows more recent data and at least four discrepancies. First, the S&P 500 and unadjusted SPY barely recorded new highs in late April 2019, but adjusted SPY clearly made a new high (red text). Second, SPX and unadjusted SPY broke the 52-week SMA in late May 2019, but adjusted SPY held this moving average (green text). Third, SPY broke above its July high and recorded a 52-week high a week before SPX and unadjusted SPY (blue text). And finally, unadjusted SPY and SPX closed below the December 2018 low in mid March, but adjusted SPY did not (gray text).
Data Consistency and Changes after the Adjustment
Data consistency is also an issue when using adjusted charts. Even if you are using dividend adjusted data, note that price bars since the last adjustment have yet to be adjusted. Thus, if you are charting with adjusted data, the data before the last dividend has been adjusted and the data after the last dividend has yet to be adjusted. You are looking at two different types of data.
To test the effect of the dividend adjustment on recent data, I am showing two charts for Caterpillar with the same indicators. The last dividend adjustment was on January 17th. Both charts cover the exact same range (17-Dec-19 to 16-Apr-20). The “before” chart was created before the most recent dividend adjustment, which was on April 17th. The “after” chart was created the dividend adjustment. The “before” chart shows what we saw the day before the adjustment and the “after” chart shows what we saw after the adjustment.
Price data literally changes after the dividend adjustment and this affects peak, trough and moving average values. First, the 200-day SMA was at 128.43 before the adjustment and at 129.60 after the adjustment. Second, the April high marked resistance at 128.42 before the adjustment and at 129.60 after the adjustment. This is especially crazy because there were NOT any trades above 129 in April. The unadjusted high reflects the highest value based on an actual trade (128.42). 129.6 reflects the true high plus the dividend adjustment.
Third, the March low was at 86.70 before the adjustment and at 87.50 after the adjustment. Again, this is crazy because the unadjusted chart shows us that there was an actual trade below 87 on March 12. The low at 87.50 is an artificial price based on the true low and the dividend adjustments. As these two charts show, chartists focused on price and moving average levels would have seen a big change from one day to the next.
And now for the good news. The values for most indicators did not change after the adjustment. The MACD-Histogram, which is the only indicator shown on the charts, did change. I did not show the other indicators to save space, but note that following indicators showed the same values before and after the adjustment: RSI(14), PPO(1,50,0), PPO(1,200,0), Fast Sto (65,3), Fast Sto (125,3), ROC(65) and ROC(125). This is good news for scans and indicator-based signals. I suspect that MACD and MACD-Histogram are subject to change because they are based on absolute numbers.
Conclusions
When analyzing price charts, I am of the view that price data should be adjusted for splits, reverse splits and special distributions, but NOT adjusted for ordinary dividends. Dividend adjustments change the historical data series on a regular basis. This means support and resistance levels will change after each adjustment. Moreover, dividend adjustments result in artificial historical prices that do not reflect the reality at the time. As we saw in the adjusted SPY and CAT charts, price lows and highs do not reflect an actual trade.
Most indicator values are stable because the entire price series changes after the dividend adjustments. This shift affects moving averages, price lows and price highs equally so their relationships stay the same. This is good news for scans, indicator signals and backtesting. However, keep in mind that this dividend adjustment shifts can result in new highs and lows that were not present prior to the adjustment.