Good Math, Bad Math

i-542ad824e08aef78e43e256f986e6f58-nyt-stocks.png

This is just a short gripe at the NYT, and a href="http://www.nytimes.com/interactive/2008/10/14/opinion/20081014_OPCHART.html">feature
that they included in today’s Op-Ed section.

It purports to compare how the economy does under democratic versus
republican administrations. They claim that they’re computing the returns
on a 10,000 dollar stock investment under 40 years of republican
administrations and 40 years of democratic administrations, in the 80 years
since 1929.

The whole comparison is pretty idiotic to begin with. Comparing stock
prices over different time spans is already so noise-ridden that it’s going to be damned hard to get any interesting information out of it. The idea of comparing things this way is really not reasonable, meaningful, or valid, because it’s not comparing like to like. For one example, World War 2 was fought entirely under a democratic administration, and was a period of dramatic growth in the American economy. But the biggest reason for the growth was producing stuff for the war and the reconstruction; the same thing would, arguably, have happened if the republicans had been in
charge. Similarly, it includes the tech-stock bubble as a growth period under Clinton, and includes the tech crash under Bush, despite the fact that the tech bubble would have popped under Gore, had he (correctly) won
the election.

But far worse is the fact that they deliberately used data that
could, at best be described as total bullshit.

How do you make money off of an investment in stock? Ask any
economist that question, and I’ll pretty much guarantee that they’ll
name two things:

  1. Dividends.
  2. Selling the stock.

The latter, selling the stock, is dependent entirely on the share price, which is frequently total disconnected from anything meaningful. (Anyone remember the share price for pets.com?) But that’s the only thing
that they consider. The fact that many large, profitable companies hand
a portion of their profits to their shareholders as dividends – and that those dividends can represent a significant profit for the shareholders – is totally ignored by this.

The NYT considers omitting this to be a trivial thing. It’s so utterly unimportant that it only merits mention in a footnote. In fact, it rates
mention below a footnote concerning the changes in the S&P
market index in 1957.

People who read this blog regularly know that I’m a very vocal liberal
democrat. I think that the policies of democratic administrations are better
for the economy than the policies of republicans. And I think that the constant
conservative harping about the so-called “Liberal Media” is just a convenient tool
that they use for manipulating press coverage.

But this is just
bullshit – this is not a meaningful comparison. It’s a hopelessly
biased, fake, meaningless pile of nonsense. And I find it very hard to believe that they would have published it as a huge graphic in the op-ed section, three weeks before an important election, if it didn’t confirm
their beliefs.

The NYT should be ashamed.

Comments

  1. #1 Walker
    October 14, 2008

    For a more interesting comparison of the economy between administrations, take a look at Cactus’ study at Angry Bear last year where he looked at GDP. I have had some issues with his methodology (e.g. in one study he compares CPI over time, even though that metric has been altered numerous times since then 80s). But he tries hard at least.

  2. #2 Carsten Schmidt
    October 14, 2008

    while this comparison is not 100% i wouldn’t call it bullshit. the NYT uses the S&P which is a cross section of the largest companies in the US. In addition the stock market is an indicator of where the economy is going … if it is looking bleak stocks are going down if it is good stocks go up. Also while the was wwII in the 40s the republicans had the 80s cold war which saw a huge military spend.

    so to say the graph is bullshit is to say that the stock market is not an indicator of the health of the economy … but then what is it?

  3. #3 jb
    October 14, 2008

    I think its very honorable to disagree with a claim that supports your political causes. Well done!

  4. #4 Jon L
    October 14, 2008

    I’m actually willing to agree with conservatives that there is a “liberal media bias” in that, most major tv networks are based in large cities and large cities tend to be more liberal than conservative. It’s not (usually) a conscious (or even unfair) bias but it is likely that a liberal environment (such as a city) will tend to produce more liberal programing.

  5. #5 CBBB
    October 14, 2008

    The Stock Market is not really an indicator of the health of the economy at all. For one thing, the vast majority of people do not derive any meaningful part of their incomes from stocks. And, as Mark notes, the price of stocks is often driven by hype and hysteria not by the real fundamentals of the economy. Finally, the interests of companies does not reflect the interest of the economy.
    Even if you assume the stock price is linked to the firm’s profitability – this does not mean that the firm is reinvesting those profits in the US economy by expanding production facilities or hiring more workers.

    The Stock Market may be an indicator of how well the rich are doing but it is a lousy indicator of the health of the economy and this NYT article is lousy reporting.

  6. #6 Mark C. Chu-Carroll
    October 14, 2008

    Re #2:

    First, I don’t agree that the stock market is a meaningful measure of the health of the economy. I know that it’s currently common to teach that it is, but it really works terribly. (Was Lehman really meaningfully worth billions a couple of weeks ago, when it was days from liquidating?) For better measures, you can use things like GDP, which is at least a relatively meaningful measure of the production of the economy.

    Second, the Times article is very specific about its claims: it’s measuring how an investment of $10,000 would have grown over forty years of either political party. But they deliberately excluded a significant, important factor that is an important part of what they claim to be describing!

    If you invested $10,000 40 years ago, would you really claim that you could get a meaningful measure of the success of your investment by ignoring all of your dividends?

  7. #7 Doug K
    October 14, 2008

    see “Unequal Democracy,” by Larry M. Bartels.

    The fact of the matter is GNP has been consistently higher under Democratic administrations than under Republican. The NYT is wrong in detail but correct in substance.

    See Dani Rodrik for a better explanation than I can provide:
    http://rodrik.typepad.com/dani_rodriks_weblog/2008/03/american-politi.html

  8. #8 evgen
    October 14, 2008

    > (Was Lehman really meaningfully worth billions a couple of weeks ago, when it was days from liquidating?)

    Yes. If Lehman was really worth hundreds of millions several years ago then they were certainly worth billions of dollars a couple of weeks ago when it appeared that the carefully constructed stacks of bets they had made would pay out or at least not blow up. If you made a $10,000 bet on red or black at a roulette table then you could meaningfully say that your bet was worth a little less than $5,000 before the wheel started spinning and in some places in vegas you could have used a bet like this as a cash-like instrument. (Except in this case it was bets on both red and black and the ball ended up in 00.) Now that we know where the ball has landed your incorrect bet is worth the value of the paper it is printed on and nothing more.

    What value do you think Lehman was producing in the first place? They were an oversized bookie who was finally unable to layoff a few big bets.

  9. #9 Steve Sandvik
    October 14, 2008

    #7: still misleading. Take the following semi-hypothetical scenario (which mimics, not all that badly, a good portion of the 80 years in question, to a 0th order approximation):

    Republicans typically gain the white house with the economy faltering. The hard work is done on their watch, and the direction shifts, slowly, grudgingly, haltingly, and finally turns upward late in the term. When times are good, we then elect Democrats, establish expansive social programs, and generally spend the largesse we have accumulated. The eventual result is an economic retrenchment, and things start to fall apart.

    In that event, the Democrats will always enjoy the higher GNPs and GDPs–but the heavy lifting was not done by them.

    Without a real quarter-to-quarter trend line, you can’t judge what was going on.

    And nice post, Mark. I’m a centrist (voting Obama mainly for Supreme Court reasons), but I about spat my drink when I saw that graphic today…but then, I’ve always been too easily offended by abuse of statistics by the media.

  10. #10 mlangner
    October 14, 2008

    And, as Mark notes, the price of stocks is often driven by hype and hysteria not by the real fundamentals of the economy.

    Actually, the market is supposed to be driven by the percieved future value of earnings by companies discounted for risk. Even if hype and hysteria were removed you still have an imperfect forward looking instrument subject to significant analytical errors in the best of times…

  11. #11 Jonathan Vos Post
    October 14, 2008

    I wanted to believe that NYT piece too. I tore it out of the copy that I got at the Starbucks next to my bank, in Pasadena, California, planning on giving the article to a Republican scientist friend of mine. But I decided not to. because it made no sense to me. What kind of person would I be if I tried to win an argument with a claim that I could not support?

    I grew up reading the New York Times. I had a friend who worked for the New York Times. My parents read the New York Times. But “the gray lady” faded over the years. My parents used to point out the rare typographical or grammatical errors in it. By the last year of my father’s life, he was saddened that there were even errors in the headlines.

    It’s the end of the world as I know it.

  12. #12 sulfursurfer
    October 15, 2008

    I agree with many of your arguments, but I believe that indices like the S&P fold the dividends in.

  13. #13 Christina
    October 15, 2008

    I think it’s a great chart, personally. I think you guys read more into it than it says. It’s just an interesting factiod: If you kept your money under the mattress anytime a Republican was in office, you’d make out like a bandit compared to someone who kept their money out of the market when a Democrat is in office. Of course past performance is not an indicator of future results, but this is damning to the Republican brand. Democrats help the economy grow from the ground up, and Republicans think it all trickles down. Well next time you want to grow a garden, you’ll have to decide whether you want to put fertilizer in the soil, or haul a Redwood in to suck up all the sunshine. As for dividends, the data for the S&P total return index on my Bloomberg terminal does not go back that far. I’ve tried to find it–but cannot. Even using the total return for where the history does exist, it gives both parties a boost, but certainly does not even things out. My guess it was a data isssue? Anyone out there have access to total return data going back to teh 20′s? And, for what it’s worth, Hoover’s presidency followed 2 other Republicans, so it wasn’t like a Democrat set thing up in the term before him.
    Remember, this is published as an Op-Chart, not as a scientific study… If you want to tackle bad facts, methodologies and math, look at the chain e-mails going around surrounding the campaign, saying that your taxes will be higher under McCain which are a total lie. I wouldn’t let my panties get too wrinkled over Mr. McCall’s experiment.

  14. #14 Jim C
    October 15, 2008

    I am surprised that you say you don’t think the media is too liberal with a straight face. The coverage by major media of politics is so biased that it has almost become a parody.

  15. #15 g
    October 15, 2008

    Steve Sandvik, could you be specific about what this hard work is that’s done under Republican administrations and shirked under Democratic ones, and do you have evidence for those claims?

  16. #16 andy
    October 15, 2008

    Interesting that they chose to exclude (well at least indicate the impact of) the great depression whilst the dotcom burst and WWII (as you IMO correctly mention) is not periods that are excluded.

    Seems to be that they’re introducing even more bias into an already pretty biased piece of “art”.

  17. #17 Jonathan Vos Post
    October 15, 2008

    I see many parallels between the current mess and the 1893-1894 mess.

    Appleton, D. (1903). “Appletons’ Annual Cyclopedia and Register of Important Events of the Year”. University of Virginia.

    January 1893(QI) peak, June 1894 (QII) trough, 17 months
    peak-to-trough, 20 months expansion Previous trough
    to this peak, 37 months Trough from Previous Trough, 30 months Peak from Previous Peak;

    So, does anyone want to identify which parties were in power, and what the numbers crunch to on these:

    http://scienceblogs.com/goodmath/2008/09/economic_disasters_and_stupid.php#c1151737

    Comments:

    #65

    Condensation/excerption/modification/personalization of Wikipedia article (which has references) “List of recessions in the United States”:

    Panic of 1797 (1797-1800)
    Side-effect of deflation of the Bank of England from French
    Revolutionary Wars; disrupted commercial and real estate markets in the United States and Caribbean.

    Depression of 1807 (1807-1814) [truncated]

    Panic of 1819 (1819-1824) [truncated]

    Panic of 1837 (1837-1843) [truncated]

    Panic of 1857 (1857-1860) [truncated]

    Panic of 1873 (1873-1879) [truncated]

    Long Depression (1873-1896) [truncated]

    Panic of 1893 (1893-1896) [truncated]

    Panic of 1907 (1907-1908) [truncated]

    Post-World War I recession (1918-1921) [truncated]

    Great Depression (1929-1939) [truncated]

    Recession of 1953 (1953-1954). [truncated]

    Recession of 1957 (1957-1958) [truncated]

    1973 oil crisis (1973-1975) [truncated]

    Early 1990s recession (1990-1991) [truncated]

    Early 2000s recession (2001-2003) [truncated]

    Late 2000s George W. Bush recession (2008-Current) [truncated]

  18. #18 CBBB
    October 15, 2008

    Comment #9, I would say your analysis is totally wrong.
    If anything (and I’m Canadian so I look at this from outside) the US typically likes to vote Republican for President and the exception – the election of Democrats – seems to take place during poor economic times with the Democratic administration ending when the economy becomes prosperous again. For example,

    1932 – Great Depression and first Democratic Presidential Win since 1916

    1992 – There was a Recession this year, Democrat Bill Clinton won

    The 1960 election for Kennedy was very close, and other things that get the Democrats elected are split elections (1912, 1992), and scandal with a former Republican administration (1976). But historically Republicans win far more Presidential elections then the Democrats and so I don’t see how one could say Americans vote in Republicans to fix the economic mismanagement of the Democrats when historically it seems to be the exact opposite – I would say that the election of 1980 is the ONLY example that supports your claim.

    Note, as I posted above, I think this NYT Article is garbage but let’s not build ridiculous historical narratives either.

  19. #19 Eric
    October 15, 2008

    A few thoughts:

    First of all, the whole thing is based on the premise that there is some long-term coherence in the views and approaches of democrats and republicans. Just to pick one obvious example, “fiscal conservative” is a label that has been applied to republicans in the past but is not one that comes to mind WRT the current administration.

    The other obvious problem is that they use “stock prices” as a proxy for “economic strength”. That they can do this with a straight face after the events of the last few weeks is laughable – it’s fairly obvious that stock prices don’t have a great correlation with economic strength

    I’m ignoring that fact that the fed policy can have a big impact on stock prices.

  20. #20 Jonathan Vos Post
    October 16, 2008

    Eric is onto something: “it’s fairly obvious that stock prices don’t have a great correlation with economic strength.”

    Is there a good proxy? How about real estate values?

    Someone (don’t recall whom) suggested that the US President be paid a salary proportionate to the value of the average home’s increase in value in his/her administration.

    Also, our economic data goes back to before the founding of these parties…

  21. #21 William Woody
    October 16, 2008

    Something no-one has noticed: if you look at the graph you see that 4 out of 7 Republican Presidents are well above average–in the 10% range. If you look at the Democrat Presidents, all are well below average–in the 7% range: factoring out Bill Clinton’s presidency during the dot-com explosion (yet another economic bubble) consistently Democrats preside over a slower governmental growth rate.

    It’s why there are more statistical tools than the average–the median for Republicans is 10.4% while for Democrats its 7.6%. The median is more suitable for future performance than the average: the average factors in statistical outliers, such as the dot-com collapse, the housing market crash we’re enjoying now, or the Great Depression.

    I’m amazed at the commentary over this chart because the most valid criticism of the chart stands right before your eyes: just look at the number of arrows above the line verses the number below the line, and it’s pretty damning for Democrats.

  22. #22 Prufrock
    October 16, 2008

    The math is, in fact, fairly arbitrary.

    1) When you’re playing with exponentials, starting points are decisive.

    2) When you’re playing with economics, externalities that really motivate market forces often have little or nothing to do with the American executive branch. Congress is a much larger influence (being a policy-making organ) and POTUS mostly takes the blame (or credit).

    Of interest — the only (D) president that performs above their compound growth rate is Clinton — presiding over an enormous boom, then bust, in prosperity (yes, the bust started before W took office). That Bush tailoff is sapped by at least one (and arguably more) pre-existing trainwrecks, just as Clinton inherited some booming externalities.

    This would not (apparently) make sense through the plot as shown — how could the Clinton growth rate possibly have dragged the total so high so fast, with only eight recent years in which to work? How is it that with almost equal time periods, with presidents on the (R) side consistently outperforming the (D) list, could be so influenced by four Nixon (war) years and the Bush (even more recent) years of shrinkage?

    Once again, we look to the starting point of the exponential by which we calculate compound interest. All better. The first (R) that wasn’t Hoover (massive loss and serious handicap as a starting point) was 24 years after the timeline start for the (D) candidates. That may make some difference, I think.

    This is really just a misguided attempt to correlate almost-independent variables.

  23. #23 Jon H
    October 16, 2008

    JvP @20: “Someone (don’t recall whom) suggested that the US President be paid a salary proportionate to the value of the average home’s increase in value in his/her administration.”

    What a horrible idea. That’s a sure way to repeat the real estate bubble that is now deflating. House prices would rise whether or not wages stayed flat, so lending standards would be pushed down and mortgage terms would go bonkers just so people could afford the rapidly-rising price of a home.

    The President and Congress should be paid salaries based on a fixed multiple of minimum wage, such that in order to raise their own pay, they’d have to raise minimum wage. If the economy isn’t doing well enough to handle the increase in minimum wage, then they don’t deserve a raise.

  24. #24 Jon H
    October 16, 2008

    Prufrock @22: ” presiding over an enormous boom, then bust, in prosperity”

    It’s not going to look very enormous when this bust is over.

    Aeron chairs and Sun servers were probably easier to unload than the surplus of homes we’re looking at.

  25. #25 Kathryn Cramer
    October 16, 2008

    Mark: Theo Gray, co-founder of Wolfram Research had a similar reaction to yours and was moved to build himself a model in Mathematica. His conclusion? Results about returns by party vary widely based on one’s choice of assumptions and whether you factor in things like dividends, changes in the Consumer Price Index, etc. Theo’s blog post about it is HERE and his downloadable toy model, which I find quite amusing, is HERE.

  26. #26 CBBB
    October 16, 2008

    The media has really done a bad job reporting about economics to the point where it’s gotten so much of the population to think that the stock market is a really good indicator of economic strength. It is not. Neither are housing prices or the prices of any class of assets.

    Perhaps a better chart would be one looking at GDP growth, worker productivity growth, inflation, and wages. All of these indicators (particularly productivity – which admittedly is not so clearly measured) give a much better picture of how the economy is doing and how it will perform in the future then stock or asset prices.

    The stock prices only show how well a very tiny economic elite are doing (those who derive their income from stocks). Since the US Economy is so very dependent on consumer consumption, an economy where the stock market is booming but worker salaries are stagnant and inflation is moderate would imply trouble ahead.

    Overall of course, the particular President in office at the time only has so much impact on the economy.

  27. #27 dc
    October 17, 2008

    The stock prices only show how well a very tiny economic elite are doing (those who derive their income from stocks).
    That’s the second time at least in this string of comments for this falsehood to be mentioned. Millions of middle class individuals invest in the stock market over decades to prepare for retirement. Not everyone is so stupid to think that a corporate pension (if you’re lucky enough to have one), Social Security and Medicare is going to be enough to sustain their standard of living after they quit working. Moreover the mass extortion of the American people at the hands of insurance companies, hospitals and drug makers implies that everyone will be forced to pony up even more as their bodies eventually break down in old age. What’s the alternative? Here’s one: a long term neighborhood resident in my area was recently moved out of his home by family, he was a Navy man who kept a neat house and had never married. The family was literally giving away his belongings on the front lawn in order to make him destitute and therefore able to qualify for government assistance. So, there’s the choice folks, save up the nest egg and hope it’s enough, or give it all away and plead for help.

    Historically the market returns 10%, far better than some piddly money market account paying 2% at a bank. The people hardest hit in the current crisis are people close to retirement who don’t have the time to make up for the Republican lack of fiscal conservatism induced market losses.

  28. #28 JC
    October 17, 2008
  29. #29 Citizen Z
    October 17, 2008

    Once again, we look to the starting point of the exponential by which we calculate compound interest. All better. The first (R) that wasn’t Hoover (massive loss and serious handicap as a starting point) was 24 years after the timeline start for the (D) candidates. That may make some difference, I think.

    Thank you. I thought it was the most common example of the power of compound interest, to compare someone who starts saving at age 20 or so versus someone who starts saving when they’re 30. And we’re just supposed to ignore that long 20 year (D) stretch nearly at the start?

  30. #30 Libertarian
    October 21, 2008

    I just came across Theo Gray’s article and I remembered the similar discussion here, so I thought I would come here and put a link to it, but Kathryn beat me to it.

    I got three points to add:

    1. Theo points out that instead of investing money only when the President of your favourite party is in power, if you kept your money in the market the whole time, you would make ORDERS of magnitude more than otherwise.

    2. As Prufrock has pointed out earlier, you can’t discount the Congress’s influence either.

    3. Personally I think the best combination is Democratic President and GOP Congress. The GOP congress wouldn’t let a Democratic President spend taxpayer money and the Democratic President would dampen the God and Gays fever in the GOP.

  31. #31 Lars
    October 22, 2008

    I agree that using stock prices as an indicator for the competence of the administration is bogus.
    However, the S&P index, other stock indexes and historical stock prices in general factor in dividends payed. (Historical stock prices are furthermore corrected for stock splits). This means it is valid to look at historical stock prices and simply calculate what an amount invested in a stock in the past would have yielded to date assuming that you had re-invested any dividends.

  32. #33 Mike
    October 28, 2008

    Somewhat related is this lovely little piece of code.

    http://www.atr.org/get401kcalc.html

    It purports to calculate for you the “value” of your 401k with respect to the candidate’s “plans” (I suppose tax plans) and those of Congress and ATR itself. It doesn’t bother explaining what is meant by “value” (for example, is it value in situ, or if you take a distribution?), and can’t possibly account for different types of contributions, the user’s age, etc. However, it merrily predicts that you’ll do much worse under Obama than McCain and under the Democratic congress than using ATRs own plan.

    These guys should apply their financial wizardry to the market in general, since they seem to be able to predict outcomes based on so little input.

  33. #34 Joshua K.
    November 12, 2008

    This was printed on the Op-Ed page. That means it’s an opinion piece. At least the NYT still has the decency to put their opinion pieces in the correct place, identifying them for what they are, than other so-called news outlets (Yes, Faux News, I’m looking at you).

    It’s an opinion. It’s on the Op-Ed page. It’s not an NYT-sourced story. What’s the problem?

  34. #35 user unknown
    November 14, 2008

    How would that stock do, if you price it in daily Euro prices, for example? Okay – Euro is to young for a comparision – take Gold.

    I guess this will give a very different impression.

    By what rule should big depressions, or crazy bubbles be excludet? That’s a funny idea! Assume the goverment is only responsible for medial events?