A first glance at Fivethirtyeight.com

Like many other internet blog-nerds, I eagerly awaited the launch FiveThirtyEight, a data-based website founded by celebrity statistician Nate Silver. (Happy to finally put those two words together).

I’m just taking a browse through the site and I’m pretty pleased so far. 

The first thing that caught my eye was a follow-up post referencing an earlier analysis of the statistical merits of steals as a measure of a player’s value. Yeah. Long title.

I also enjoyed an analysis on the development process behind Paul Ryan’s proposed budget, including a potential error he might have made. Ryan used something called “dynamic scoring.” The term is unfamiliar to me, but from reading between the lines, it seems to be involve including “the second-order fiscal effects” of whatever happens in the budget. Author Andrew Flowers (@andrewflowers) noted this as unusual, which surprised me. 

Say, for example, a government invested $3 billion in preventative health care. $1 billion was invested towards measures with a 1-year expected return (although I can’t think of any health measures with a 1-year return), $1 billion towards measures with an expected 5-year return and $1 billion towards measures with an expected 20-year return. Initially, costs would be $3 billion higher. At the end of the first year, costs would be ($3 billion – 1st year of expected return on $1 billion investment) higher. The reduction in costs would continue to compound, increasing in rate as the 5-year returns kicked in and so on.

Now, the rate of return is just an expectation. Perhaps some of the measures weren’t carried out efficiently. But, unless the predictions on the efficacy of a certain health measure were entirely inaccurate, SOMETHING would have taken place. So, it would be sensible to use a conservative estimate of those compounding impacts, but it would also be sensible to include it – which is what Ryan’s budget does.

From my understanding, this is what Flowers is commenting on as unusual. I would be happy if anyone could read the article and clear that up for me. For that matter, any comments here would be welcome. I’m sure there are some formulas that could have summarized that process much simpler than my written explanation. ha!

Flowers goes on to note that the Congressional Budget Office (CBO) makes similar compounding predictions, but only on the effects of overall deficit reduction on Gross National Product (increasing it). Ryan, meanwhile, assumes his deficit reduction invokes economic growth, increasing tax revenue, further reducing the deficit.

Flowers does make an insightful critique: if Ryan is proposing cuts to the Department of Education, that will likely have an impact on future economic growth, versus cuts to other departments. This might offset his predictions of future growth causing tax revenue increases.

At the end of the day, I am left confused by Flowers’ distinction between the CBO predictions and Ryan’s. Ryan’s uses specific policies, yes, which will have specific (yet-to-be-determined) implications. But the CBO’s guess – are they just predicting what will happen if deficit abruptly decreases without a change in specific policies? That seems like a relatively toothless prediction.

I should also add: I don’t support a MORE DEFENSE, LESS SPENDING, NO TAXES budget. I just thought that Ryan’s use of dynamic scoring seemed sensible in the context of a federal budget.

Interesting post from commenter Trey Castles:”It would be nice to see congress posit a sophisticated budget with spending increases for a higher than expected growth year and spending concessions for lower than expected growth years.”

Anyhow – the comments section has a whole series of references that I don’t understand, but hopefully an understanding of what I don’t know is one of the things I’ll gain from the website!



  1. andrewflowers538 · · Reply


    Hey, this is Andrew Flowers. Thanks for reading my article and writing such a thoughtful, smart blog post. I think you think through the issues very well, and actually answer your own question.

    Where you write:

    “But the CBO’s guess – are they just predicting what will happen if deficit abruptly decreases without a change in specific policies? That seems like a relatively toothless prediction.”

    The answers is yes.

    The CBO paper cited by Ryan is a weird, abstract exercise. Of course reducing the deficit–ALL THINGS EQUAL–encourages growth by spurring greater investment, as interest rates are lower, due to less “crowing out,” etc, etc. In theory.

    So, yes, you’re right. I agree it is a toothless prediction by the CBO; but they are hamstrung by a Congressional mandate to just analyze proposals “as is.” For example, they don’t analyze–or, more precisely, they *can’t* analyze–issues like whether Ryan’s projections of revenues and spending could actually hold at their levels over a 10-year horizon. That seems crucial, no? Yet this is the weird world the CBO operates in. I give them a ton of respect, but they are constrained.

    The dynamic scoring approach as used by Ryan was “unusual” because it broke the customary “rules” often agreed to for budget proposals. The numbers stemming from the dynamic scoring were insignificant; it was the approach itself that is noteworthy. (Though, to be fair, it was used in the justify the Taxpayer Relief Act of 1997.)

    Furthermore, you write: “I just thought that Ryan’s use of dynamic scoring seemed sensible in the context of a federal budget.”

    I think you’re getting at an important issue I maybe didn’t make clear enough: dynamic scoring seems smart. I say “seems” because it really does sound great in theory–why wouldn’t you want to know the second-order effects of a change in policy?

    But problems arise because dynamic scoring is SO susceptible to manipulation. If history is any guide, budget drafters can use dynamic scoring to contort their projections and make whatever they want seem reasonable. Make sense? Hence why we just abide by strict static scoring. Here’s a good critique: http://www.cbpp.org/cms/?fa=view&id=3598

    I think you wrote a very fair, smart post. I hope to be more clear and elaborate on these complex issues in the future. Thanks for reading!


    1. Hi Andrew,

      Firstly – thank you for taking the time to offer such detailed and positive feedback! I sincerely appreciate it.

      You have helped clear up a few specific items of confusion… but also left me with many more questions to puzzle out!

      I think abstract really is the word I was looking for to describe the CBO’s policy. I suppose I can understand the intent behind the Congressional mandate (limiting partisanship within the CBO), but ultimately, it seems like they’ll be left with less realistic predictions. We know that certain kinds of investment spending will have second-order effects, with either positive or negative effects on economic growth. The discrepancy lies in the extent. If the CBO were to offer ranges in their predictions, that might help. Then again, that would also give us budgets with less specificity, which would be harder to use as a political tool (irrelevant) and harder for the general public to comprehend (relevant).

      I can also much more clearly understand exactly why Ryan’s behaviour was considered “unusual.” The Center on Budget and Policy Priorities article by Paul Van de Water had some very reasonable arguments for why dynamic scoring should not be used. If there can be no agreement on the range of effects, a budget prediction would be similarly useless. Additionally, the section on budget credibility makes sense: I can see how dynamic scoring could be manipulated.

      One question that remains for me:

      Van de Water acknowledges the benefits of investment spending: “Government investments in infrastructure, education, and basic research can boost long-term economic growth, just like private investment.” If dynamic scoring is eliminated from the budgeting process, how are benefits from investment in education, preventative health, and employment properly accounted for?

      It would be interested to see, as Trey Castles mentioned, a high-estimate and low-estimate budget, with the median serving as the baseline of expectations. This would undoubtedly involve quite a bit of additional work.

      Thanks again for your response. It’s given me some interesting things to think about and read into.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: