Search This Blog

Monday, January 08, 2018

One For Erp

Teachers unions may not be the root of all evil, but they may well be counterproductive for society in aggregate. Here's an excerpt from a recent paper from Cornell University:
We find robust evidence that exposure to teacher collective bargaining laws worsens the future labor market outcomes of men: living in a state that has a duty-to-bargain law for all 12 grade-school years reduces male earnings by $1,493 (or 2.75%) per year and decreases hours worked by 0.52 hours per week.

13 comments:

erp said...

I don't care what the pay scale difference is between male and female teachers. In fact, I'd be happy with robot teachers as long as the unions were ousted and school districts could re-institute the pre-cultural revolution curriculum.

Clovis e Adri said...

Well, considering all the evil Erp atributes to them, looks a pretty small difference...

Bret said...

Small difference?

Yeah, a hundred billion here, a hundred billion there, pretty soon your talking real money!

Clovis e Adri said...

Only if you entirely believe their results, Bret. And I don't.

Their findings are very peculiar. The fact they find those negative results for only the population of men, and not for women, is a strong hint their methodoloy may be measuring something else.

They know that's a serious problem, and try to get away with this:
--
"Why men would be so adversely affected but not women is an open question that our empirical approach admittedly cannot address, but these findings are consistent with emerging evidence that boys’ long-run outcomes are more susceptible than are those of girls to negative shocks that occur in childhood (Autor et al. 2016; Fan et al. 2015; Autor and Wasserman 2013; Bertrand and Pan 2013)."
--
Oh, poor boys, so fragile.


Furthermore:
--
"However, collective bargaining laws have only a modest effect on educational attainment. Our estimates therefore are consistent with the lack of strong effects on high school graduation rates found in earlier work (Lovenheim 2009) and suggest that union effects on labor market outcomes affect human capital in ways that do not show up in years of educational attainment."
--
Oh-oh. The results didn't fit with the conclusion they wanted, so they needed to invent another metric where they show what they do want:

--
"This finding motivates our analysis using the 1979 National Longitudinal Survey of Youth (NLSY79) that shows declines in cognitive and non-cognitive skills due to collective bargaining exposure."
--
Hmm... how that works?
[to be continued]

Clovis e Adri said...


--
"We now turn to direct evidence on how collective bargaining influences student
cognitive and non-cognitive outcomes using data from the NLSY79. This is a nationally-representative dataset of students aged 14-22 in 1979, covering the 1957-1965 birth cohorts.
These cohorts thus overlap with much of the variation in the passage of teacher collective
bargaining laws shown in Figure 1.

Respondents in the NLSY79 data take the Armed Forced Qualifying Test (AFQT), which
is our measure of cognitive skill. These scores are reported in age-specific percentiles. Non-cognitive skills come from three measures: the Rotter Locus of Control, the Rosenberg Self-esteem Scale and the Pearlin Mastery Scale. The Rotter Locus of Control measures the extent towhich students believe they have control over their own lives. Thus, it is a measure of perceived self-determination, with higher scores indicating less internal control. Higher scores on this measure therefore translate into lower non-cognitive skills. The Rosenberg Self-esteem Scale is designed to measure a student’s self-worth. Higher scores indicate higher reported self-esteem. Finally, the Pearlin Mastery Scale is a measure of the extent to which individuals perceive themselves in control of forces that significantly impact their lives. Respondents with higher measures report increased ability to determine the course of their own life."
--

Oh, I see, studying under a bunch of communists controlled by unions make the boys feel less in control of their lifes. Makes sense, right? It is the exact result one would wait if we believe unions work to make people less less individualistic, and more collectivist.

There is only a tiny problem, still. The girls, again. They apparently refuse to bound:

--
"The estimates for women tend to be smaller in absolute value though in similar direction to those of men. In particular, the effect on AFQT scores is less than half the size of the male estimate."
--
You go, girls!

To worry, all those results are taken from a single year only - 1979 - less than fifteen years after the first states passed duty-to-bargain laws, and less than 7 when the majority did. That's pretty fast! One could wonder what would that analysis give if taken a few years before 1979 (preferentially, 15 years before, when such laws were virtually null), and afterwards. I wonder why the authors would have restricted themselves to that single year. No word about that in the paper. A serious omission, I think.

Looking a list of the states that passed those union laws before 1979 (their table 1), one things looks clear: the gross majority of the population is under those union laws in 1979, so he is comparing data sets of very different sizes. To wit, their Figure 5 (Sensitivity of Results to Excluding Each State - Men) present variations far too big, IMHO.



I won't say it is a useless article, but their results look less sound than they let you know in the text.

Clovis e Adri said...

Finally, let's address the hundreds of billions claim.

It is posed this way:
---
Taken together, our results suggest that there are negative effects of public sector
collective bargaining laws for teachers on the long-run labor market outcomes of men. Although the point estimates are relatively modest in magnitude, they are economically significant: increasing male earnings in the 33 states with a duty-to-bargain law by 2.75% amounts to $149.6 billion of additional earnings per year. This is equal to about 137% of annual federal spending on education. Thus, due to the scope of teacher collective bargaining in the US, the treatment effects we identify translate into large impacts on workforce productivity.
---

This is a demonstration of my previous point on how different the data sets are (i.e. the population in duty-to-bargain states versus the rest), so their small difference translated to a big economic impact.

But is it really big? 150 billions is, still, only 0,8% of US GDP. Furthermore, you'd need to take their bold conclusion - that were not such pro-union laws passed, you would have the people in those 33 states earning 2.75% more - at face value, and you need to do some wishful thinking to arrive at such conclusion.

For, if everyone were to work now 0.52 more hours because they got better schools when young, you would have a lot of people offering more of a product (their work) without necessarily a hike in demand. They'd would end up earning how much more in this alternative scenario? Probably not the linear calculation the authors take. Would it be half their predictions? One third? One tenth? I don't know, and I highly doubt anyone does. This is like choosing dots in a screen of white noise.

Clovis e Adri said...

Erp,

---
I don't care what the pay scale difference is between male and female teachers. In fact, I'd be happy with robot teachers as long as the unions were ousted and school districts could re-institute the pre-cultural revolution curriculum.
---

I guess you misundertood Bret's post. The difference is not in salary among teachers, but how much the population who studied in unionized states differ from the non-unionized one.

Throughout the lives of the population they focus on, people today between 35 and 49 years, a lot of things happened. In particular, taking a look at the list of states with the pro-union laws, it looks like the unionized states are also the ones who got most of the immigrants (Latinos) that arrived in between. No wonder, the group where their results is most visible is among... Hispanics. But is it because unions affect hispanic (and hispanic men at that, not women) more, of because you have more hispanics in the states that got unionized?

That's just one example of how they are being too bold and careless, IMHO, by claiming their results point out to causality, not only correlation.


And I am sorry to tell you that, Erp, but if we substitute teachers by robots, the kids will surely NOT get the pre-cultural revolution curriculum. The robots wouldn't go back to that - a curriculum so wrong that you, 80-plus years old American, knows less about Thomas Jefferson than an ignoramus from Brazil. (Yourself claimed to have learned, in school, he set his slaves free).

erp said...

My email is totally verschimmelted. Comments are no longer being received in my inbox????

Clovis, let's not revisit Jefferson please. Why do you think your source is better than the history books in my day when blacks were being lynched on every lamp post and whipped in town squares across the fruited plain by the KKK?

I guess you agree now that academic treatises are very rarely creditable.

Clovis e Adri said...

My source on Jefferson was his Foundation, Erp, you can't get more to the man than that. It is all there, in the Monticello records.




Bret said...

Okay.

A wide range of people question the study.

Do y'all mind if I just delete this whole post (which will also lost the comments)? If any of you prefer I leave it, I don't have a problem with that either, just let me know.

erp said...

Not a problem with me.

Clovis e Adri said...

Bret,

Who else questions it?

It is your post, and your blog, but I don’t see any reason to delete it. Wrong results also teach us something - sometimes, more than ‘right’ ones.

Bret said...

Okay, I'll leave it.

I didn't learn a lot from this wrong one. My biggest mistakes were that I thought it appeared in a peer-reviewed journal (it was just a working paper) and I didn't study it enough before posting. I'm definitely not proud of this one. Oops!