My attempt to do a cost/benefit ranking of the T50 Forum

[Nagster5]

I wanted to try and rank the USN T50 schools according to a cost/benefit analysis. Feel free to critique my methodology. Schools are awarded separate scores for attending at sticker and at average discount, and IS or OOS for applicable schools. I am by no means an excel or math wizard so if you see a better way to do something, or have ideas for useful additions, please let me know.

Link

Assumptions:

- This is Biglaw oriented, since there's really no hard numbers to compare other competitive opportunities (any ideas for a good metric to make a PI or Gov list please let me know).
- All numbers are from the most recent LST figures unless otherwise noted.
- All financing will be done via loans. Stafford loans will be used up to the annual limit, and Grad PLUS loans afterwards. Rates can be manipulated in the table on the right for future years.
- For schools at discounted rate, I multiplied the median award by the percent of the class receiving aid.

Method:

- A school's score is given positive points for placement in elite employment (full-time employment in firms with 250+ lawyers or federal clerkships), low tuition, and a low rate of unemployed graduates. Obviously this fails to capture unique opportunities at schools like HYS, but there's really no way to quantify that without mnually adjusting the numbers, at which point it's not really data, just my pinion, so it's gone quantified. This wasn't a project to prove how much better Penn is than Yale.
- Total CoA was calculated by using 2014 tuition accounting for the schools' average increase over the last 6 years, plus a 2% increase in CoL/year. Also loans and stuff.
- The "score" formula can be weighted using the three boxes on the right, depending on where an individual's personal preferences lie. The default setting is 50% cost, 30% employment, and 20% underemployment.

Interesting Superlatives:

Highest average annual tuition increase: Alabama (7.21%) for IS students
Lowest average annual tuition increase: Arizona (-5.95%) for OOS students

Largest in-state discount: Utah ($23,156)
Smallest in-state discount: UVA/Michigan ($3,000)

Largest average award: Indiana ($22,667)
Smallest average award: Colorado ($792)

Largest CoA: Stanford (S) ($283,913)
Smallest CoA: Iowa (IS) ($69,407)

[Specter1389]

Did you include clerkships in your analysis? I know you said this is big law focused, but the majority of clerks end up at big law firms and I imagine that would change the numbers quite a bit.

[tsujimoto74]

Did you include clerkships in your analysis? I know you said this is big law focused, but the majority of clerks end up at big law firms and I imagine that would change the numbers quite a bit.

Method:

- A school's "score" is it's Biglaw % (FT jobs at firms w/ 500+ lawyers) plus its federal clerkship %. For HYS, I have boosted these numbers (YLS 56.2 to 80, HLS 60.4 to 75, and SLS 66.7 to 75) due to the unique nature of their employment opportunities that are not reflected in BL/Clerkship numbers. This is clunky I know, but it's a better reflection of reality than not doing it and this table isn't going to help many people with HYS options make decisions anyway. Suggestions for improvements are welcome.

[fliptrip]

I think drawing the line at 500+ attorneys is a little too low. Very prestigious and generously remunerative firms like Williams & Connolly, Boies Schiller Flexner, Cravath, and WLRK all have fewer than 500. I'd imagine firms with 250-499 would be worthwhile to include as well.

Also, why not just use LST's employment score? They've already done some analysis that incorporates good PI work.

The first increment to student borrowing is in most if not all cases the Stafford loan and it's currently 100 basis points cheaper than Grad PLUS in interest and 300 basis points cheaper on origination fee. When just doing sticker calcs, this difference is negligible, but when factoring in discounts, it might be relevant.

Ugh, if the ABA would just release average grant amount, you could compute the effective COA everywhere. We just don't know what the distribution of support looks like.

[Nagster5]

I think drawing the line at 500+ attorneys is a little too low. Very prestigious and generously remunerative firms like Williams & Connolly, Boies Schiller Flexner, Cravath, and WLRK all have fewer than 500. I'd imagine firms with 250-499 would be worthwhile to include as well.

Good point, I can add that easily when I'm back at work tomorrow.

Also, why not just use LST's employment score? They've already done some analysis that incorporates good PI work.

I don't think LST's employment score is a good proxy for the relative placement power of top schools. It's useful for telling just how bad the bad schools are. It's next to useless for differentiating good and great schools.

The first increment to student borrowing is in most if not all cases the Stafford loan and it's currently 100 basis points cheaper than Grad PLUS in interest and 300 basis points cheaper on origination fee. When just doing sticker calcs, this difference is negligible, but when factoring in discounts, it might be relevant.

I can adjust this fairly easily as well

Ugh, if the ABA would just release average grant amount, you could compute the effective COA everywhere. We just don't know what the distribution of support looks like.

No kidding. I know median award x% of class with award is clunky, but it's all we have to go on. I was thinking about doing a different calc for each quartile of support (quarter ride, half ride, etc.), but I think that's bordering on so much information it becomes useless for actually comparing the schools/offers.

I think the main thing to take away from this is that certain schools are great deals for the price, while others are consistently placing much lower than other schools that cost the same or less.

[lacrossebrother]

Nice work. Would be interesting to open this up and allow people to expand as they please.
The real tradeoffs people need to make relate to cost vs. expected income increase (as you account for), but also there's a personal enjoyment factor that is perhaps quantifiable. For instance, it seems odd to me that Minnesota and USC could be peer schools to someone not from those areas. I think we could maybe establish some baseline fun quotient and allow people to tweak. This should probably be as simple as some measure of city quality (non-weather) (0-5) and weather (0-5)

[baal hadad]

500 plus is not a good metric

I prefer LST 100+ but whatever

[Nagster5]

Nice work. Would be interesting to open this up and allow people to expand as they please.
The real tradeoffs people need to make relate to cost vs. expected income increase (as you account for), but also there's a personal enjoyment factor that is perhaps quantifiable. For instance, it seems odd to me that Minnesota and USC could be peer schools to someone not from those areas. I think we could maybe establish some baseline fun quotient and allow people to tweak. This should probably be as simple as some measure of city quality (non-weather) (0-5) and weather (0-5)

I can't use google docs at work, I'll upload it tommorow night. I coul add a "fun" quotient and set it to 0 by default, but honestly I think that's something people can factor in after the hard numbers are accounted for. Or they could just directly add the amount to "score" for sunny schools. If sunshine is worth a 10% reduction in elite employment opportunities, for example, just add 10 to the score of schools in CA.

[packer_22]

Question about the full list incorporating sticker and non-sticker. Does it imply that Chicago at sticker is a better decision than Columbia at median discount?

[packer_22]

Question about the full list incorporating sticker and non-sticker. Does it imply that Chicago at sticker is a better decision than Columbia at median discount?

[Nagster5]

Question about the full list incorporating sticker and non-sticker. Does it imply that Chicago at sticker is a better decision than Columbia at median discount?

"Better" in terms of the methodology, yes. Chicago at sticker and Columbia at average discount (a whopping $6,000/year) are roughly the same price, but UC places 3.5% more students into biglaw/clerkships. Columbia suffers from having both the highest tuition and lowest average scholarship award, along with a high average rate of tuition inflation and very high CoL.

The difference is weighted CoA between the two is pretty small though, ~$5,000.

[run26.2]

Curious to see what it would look like if you didn't boost HYS. Can you post that?

[Nagster5]

Updated the OP with the following based on feedback and new ideas:

- Updated the list to include the USN T50.
- Normalized the formula to a 100pt scale, and added a penalty for high unemployment.
- Added cells to allow users to weight the three factors (Cost, employment, unemployment) to their liking.
- Modified the CoA formula to use the annual Stafford loan limit before using grad PLUS loans.
- Lowered the threshold for Biglaw to 250+ firms.
- Unweighted HYS. People can decide how much they want to weigh unicorn jobs on their own.
- General aesthetics.



Thanks for the input everyone, still looking for more to improve.

[stego]

Why is biglaw 250+ and not 100+ which is how everyone else on this website defines it.

[lacrossebrother]

I don't get why we don't just look up all grads on state bar websites/avvo/LinkedIn and then add a column stating whether we think their job is a suitable outcome?

[Nagster5]

I don't get why we don't just look up all grads on state bar websites/avvo/LinkedIn and then add a column stating whether we think their job is a suitable outcome?

I'll start working on that

[BigZuck]

This thing was better when it had UT as a consensus T10 school

Make that happen again please

Eta: TYIA

[Nagster5]

This thing was better when it had UT as a consensus T10 school

Make that happen again please

Eta: TYIA

Cost 65%
Employment 34%
Unemployment 1%

[PeanutsNJam]

I don't think the weights work right. You'd seriously recommend someone take BYU at 94k over UCLA at 178k, assuming they have to go to law school or they'd die and those are the only two options (and say their goal is as high paying a job as possible)?

[Nagster5]

I don't think the weights work right. You'd seriously recommend someone take BYU at 94k over UCLA at 178k, assuming they have to go to law school or they'd die and those are the only two options (and say their goal is as high paying a job as possible)?

No I wouldn't, and that's not really what the ranking is saying either. This is a cost vs. outcome ranking, so for the given outcome (a weighted metric of desirable job placement at a given school vs. its unemployment rate) the ranking is saying that for each dollar you spend at BYU at 94k you're increasing your chance at a desirable outcome more than if you spent that dollar attending UCLA at 178k. With the weights I used in the example, which can be easily manipulated in the google doc, the penalty for unemployment is low and CoA is weighted more than job score. If you are more concerned with employment than worried about debt, you can reflect that in the weights and BYU falls below UCLA.

Having said that, I am a complete amateur at ranking and methodology type stuff, so if you have any inputs into how to improve this then please let me know. I'm trying to make something useful, not pass this off as the end all be all (or at this point even a good) metric for LS decisions.

[PeanutsNJam]

While it's an interesting take, the main problem is that the average cost of a school should have zero bearing on anybody, unless the average cost is that particular person's actual cost. Since that's the route you're going with, I wouldn't even try to cater this ranking list as something that would assist an applicant in the decision making; rather, it's simply a comparison of schools for the sake of doing so.

Most importantly though IMO, if you can get the % of in-state students, you can just merge in-state/out-of-state numbers.

Another issue is that, as with most things in life, outcome does not scale linearly with cost. The math will get complicated, but you will need to use a differential equation to get an accurate analysis of cost vs employment. If you treat this relationship as linear, low cost schools will have artificially high ranks (like Iowa). Frankly I wouldn't bother since there's no way to get to the proper equation without a disproportionately large amount of effort. Since biglaw allows you to pay off debt, I'd just take a bit of weight off the cost and add it to the employment.

A 250+ firm is not per se more desirable or more selective than a 100+ firm. So long as you're being paid market (which almost all 100+ firms do), you have a "desirable outcome." You still leave out elite boutiques but the numbers there should be more or less diminimis.

In your employment score calculations, I think it would be fair to weight federal clerkship % as slightly higher than biglaw %, so you have a modified employment score number. This would put HYS closer to their proper place. (So 1% of fed clerk counts as 1.5 points, and 1% biglaw is 1 point or whatever, and your employment score wouldn't be a percentage it'd be a raw score.)

[Nagster5]

While it's an interesting take, the main problem is that the average cost of a school should have zero bearing on anybody, unless the average cost is that particular person's actual cost. Since that's the route you're going with, I wouldn't even try to cater this ranking list as something that would assist an applicant in the decision making; rather, it's simply a comparison of schools for the sake of doing so.

The I think the average is useful to people looking for where to apply, since it gives them a good idea of what they can realistically expect from the school. I agree that it is not super useful for evaluating schools independently though, but I did want to recognize in some way that schools with high tuition do not necessarily have a high average CoA due to generous scholarships. I was considering eliminating this metric due to clutter/readability anyway when I expanded to the T50, so I will probably do that anyway.

Most importantly though IMO, if you can get the % of in-state students, you can just merge in-state/out-of-state numbers.

Interesting, I will try to do this, although it's a lot of research compared to copying off LST.

Another issue is that, as with most things in life, outcome does not scale linearly with cost. The math will get complicated, but you will need to use a differential equation to get an accurate analysis of cost vs employment. If you treat this relationship as linear, low cost schools will have artificially high ranks (like Iowa). Frankly I wouldn't bother since there's no way to get to the proper equation without a disproportionately large amount of effort.

I will see what I can do re: a differential equation, it's been a long time since calculus but I'm very bored and I actually want his to be a good product. if any STEM types can help with this please PM me. If not I can figure it out with a lot of work.

A 250+ firm is not per se more desirable or more selective than a 100+ firm. So long as you're being paid market (which almost all 100+ firms do), you have a "desirable outcome."

I didn't think this was correct, but it seems based on feedback I was wrong. The next iteration will have 100+ as the metric.

In your employment score calculations, I think it would be fair to weight federal clerkship % as slightly higher than biglaw %, so you have a modified employment score number. This would put HYS in the proper place. (So 1% of fed clerk counts as 1.5 points, and 1% biglaw is 1 point or whatever, and your employment score wouldn't be a percentage it'd be a raw score.)

I meant to do this but I got sidetracked and forgot, awesome suggestion thank you

[jbagelboy]

the problem with the inclusion of cost in any metric is that cost is individualized; the other factors like BL+clerkship placement are objective, but school cost will differ for each student so the averages dont' mean very much. Penn at sticker, for example, is still a bad choice (even if it ranks #1 on your chart), whereas Penn with a Levy scholarship will almost always be the best choice (unless a Hamilton, Rubenstein or Vanderbilt is in play).

So I appreciate the effort to include cost, but its not particularly valuable in the abstract.

[Tls2016]

the problem with the inclusion of cost in any metric is that cost is individualized; the other factors like BL+clerkship placement are objective, but school cost will differ for each student so the averages dont' mean very much. Penn at sticker, for example, is still a bad choice (even if it ranks #1 on your chart), whereas Penn with a Levy scholarship will almost always be the best choice (unless a Hamilton, Rubenstein or Vanderbilt is in play).

So I appreciate the effort to include cost, but its not particularly valuable in the abstract.

Plus does anyone actually get the average award amount?
(Also, Penn at sticker may be fine if you have family money.) It's a small number of full rides and then partial scholarships averaged with a bunch of people paying sticker
I don't see how this is different than just ranking biglaw and federal clerk rating. You are also including firms that don't pay market, so the outcome in term of salary will be varied if you are looking at money. Not even all 100+ people firms in NYC pay market. I think not even all the NLJ 250 pay market.

Some may feel federal clerk is a better outcome than biglaw in terms of bonus, but not at all for salary. It doesn't benefit corporate people at all.
Also no accounting for the essentially local nature of most law schools.
I'm not sure what this adds?

[RareExports]

While it's an interesting take, the main problem is that the average cost of a school should have zero bearing on anybody, unless the average cost is that particular person's actual cost. Since that's the route you're going with, I wouldn't even try to cater this ranking list as something that would assist an applicant in the decision making; rather, it's simply a comparison of schools for the sake of doing so.

I think your description of this as, "a comparison of schools for the sake of doing so," is precisely what this is intended to be.