1 / 30

Arguing with Data: Introduction to Descriptive Data Analysis

Arguing with Data: Introduction to Descriptive Data Analysis. Professor Sarah Reber Lecture 7. Today. Demographic Adjustment Understanding trends and levels of spending The economics of tagging and targeting Disability insurance. 5 th Grade ELA Test Scores. LAUSD average is 386

borna
Download Presentation

Arguing with Data: Introduction to Descriptive Data Analysis

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Arguing with Data: Introduction to Descriptive Data Analysis Professor Sarah Reber Lecture 7

  2. Today • Demographic Adjustment • Understanding trends and levels of spending • The economics of tagging and targeting • Disability insurance

  3. 5th Grade ELA Test Scores • LAUSD average is 386 • Santa Monica-Malibu average is 437 • Difference of 51 points = pretty big • How much of this is due to the different demographics of the two districts? • What if… • Both districts had the demographics of the state but the group-specific tests scores we observe? • Both districts had the group-specific tests scores of the state but the demographics we observe? • Focus on economic disadvantage groups

  4. Explaining Differences • Scores can differ because… • Demographics (“endowments”) are different • e.g. LAUSD has more poor kids, who tend to score worse • Group-specific differences in outcomes (“coefficients”) • e.g. Poor kids in LAUSD score better (or worse) than poor kids in SMMUSD • Both can be happening at the same time

  5. Example • Start with fake test score data for 2 schools • Example 1 • Group-specific scores are constant • Share poor varies • Example 2 • Group-specific scores vary • Share poor constant • Example 3 • Both

  6. Example • Use real test score data from LAUSD, SMMUSD, and CA State • First adjust just by economic disadvantage status • What share of the difference between the two districts is due to demographics? • How does that depend on the method • Then by race and economic disadvantage • Overall, how does the demographic adjustment affect the comparison of the two districts?

  7. Interpretation • When would you use this? • What do the diffs in adjusted scores tell us? • “Treatment effect” of district? • Regression adjustment and accountability • Comparing apples and apples? • The soft bigotry of low expectations? • If group-specific outcomes differ AND group proportions differ  may be useful to analyze subgroups directly for more insight • If you have the data

  8. What to do when? • Micro data  Run a regression! • Dummy variables for married • Dummy variables for age groups • Can do direct or indirect with the results • Don’t have age-specific outcome for pop(s) of interest  indirect method • Have age-specific outcomes for pop(s) of interest  direct method

  9. Conclusions • Demographic adjustments are like regression controls • When an outcome has a strong demographic predictor  useful to “take that out” when comparing across groups or over time • The remainder is not a treatment effect, but could be • Examples • Male circumcision • Disability Insurance participation

  10. Budget Changes • Can also decompose changes in spending into its pieces to better understand trends • Which sub-categories have been changing? • Changes in spending in that category • Compared to other categories • Changes in share of total budget • Changes in number (and type) of people served versus cost per (constant) person • Age adjustment • Demographic Adjustment

  11. The answer is staring university leaders in the face, says Maria Maisto, head of New Faculty Majority, which advocates for adjunct professors: Pay college presidents and coaches less, and part-time professors more. "If education is really at the heart of what we do, then there's absolutely no excuse for not putting the bulk of the resources into what happens in the classroom," Maisto says. But that's not what institutions are doing, she says. "In fact, here in Ohio, I have colleagues who have recently had to sell their plasma in order to buy groceries," she says.

  12. Growth vs Share • If you want to save money, need to know where the money is  what share of the budget are we talking about • If something is growing fast and continues  will eventually take a large share of the budget • But if it is a small share to start and grows really fast for a bit  not a good place to look • Examples • Drug costs • Spending on Administrators = # Administrators * average salary

  13. Simple Example • Decompose Medicaid Spending • Hold constant enrollment • Hold constant spending • Interaction

  14. Economics of “Tagging” and Targeting • Consider a simple model • There is one type of work in the economy, if you do that work, you earn a wage • There are two types of people: • those who experience severe pain while working (“disabled”) • those who do not experience severe pain while working but enjoy not working (“not disabled”) • 15 percent are randomly chosen by Nature to be disabled

  15. Continued… • Assume you can easily tell who experiences pain while working (is disabled) • What policy do you propose? • What if you can’t observe who is disabled? • What will happen • What policy do you propose?

  16. Options • Use correlated, immutable, measureable characteristic • E.g. blind • Get people to “reveal their type” • Impose something that is less costly to targeted group than others • Waiting periods • Ordeal mechanisms • Make the program less good • Partial insurance due to moral hazard • Stigma

  17. Questions? • What explains the trends in SSDI? • Demographics, rising retirement age • The economy • De facto program for the outmoded • What should we make of changes in the mix of diagnoses? • Once you’re on, you never leave: good or bad? • Should you be able to work and get DI? • What about the kids? • Can you design a program targeted to these needs that doesn’t entail adverse incentives

  18. The idea that we, as a society, should offer support to disabled children living in poverty, I haven't taken a survey or anything, but I'm guessing a large majority of Americans would be in favor of that in some form. We'd have a hard time agreeing exactly how we want to offer support. But I think there are some basic things we all agree on. Kids should be encouraged to go to school. Kids should want to do well in school. Parents should want their kids to do well in school. Kids should be confident their parents can provide for them, regardless of how they do in school. And the goal should always be for children to become more and more independent as they grow older, and, if possible, support themselves, somewhere around the age of 18. • This program stands in opposition to every one of these aims. • So somewhere around 30 years ago, the economy started changing in some important ways. There are now millions of Americans who do not have the education or the skills for the current economy. And this does not seem temporary. • Politicians certainly pay lip service to this problem during election cycles and such. But American leaders have not sat down and said, look, things have fundamentally changed. What are we going to do for these people? We need a plan. • In the meantime, the disability programs have become the default plan. Programs that, along with the associated health care, cost about $260 billion a year. And I know this is obvious, but a quarter trillion dollars? That's a lot of money. That's eight times more than we spend on welfare. In fact, it is more than we spend on welfare, food stamps, the school lunch program, and subsidized housing combined. A lot more. • Think for a minute of what we could do with a quarter trillion dollars. We could decide what we want is to expand welfare payments or food stamps. We could decide that the best plan is just to hand out cash to every out of work American. Or we could decide not to spend the money. Let everybody pay less in taxes. There are options. We just never made a decision.

More Related