Return-on-Investment in Education

During this time of the year, when millions of students are deciding which college to go to and what program to enroll in, one question that is often asked is, "What is the RoI of the program?" RoI means "Return on Investment.

Now, RoI of education is a bit complicated, because it is not just the additional money that you can potentially earn, but also the quality of job, and the quality of life, happiness, and the privileges of connections and so on. But let us assume that everything can be reduced to a number, the financial returns.

What is the financial return of a program. Well, one has to see what would be the estimate of lifelong earnings of the person without doing that program, and compare that with the estimated earnings of the person after doing the program. The difference between the two can be attributed as return due to that program. Now, you can sense that this is not easy. One can perhaps model an average person but it does not work for an individual because your return on investment could be very different from my return on investment simply because you are more prone to doing a corporate job after the same degree while I am interested in becoming a teacher. The financial returns over the next 50 years in an uncertain world under two hypothetical conditions, impossible.

But still we need to decide whether to spend a very significant amount of money or not. Is there a proxy for RoI calculation. Yes, of course. Without this, what will all the consultants do. It is a simple proxy. You take your expense over the program, and look at the first month salary of an average student in the last batch graduated and computed in how many months one would get as much salary as the amount spent in the program. So, if the fees for a BTech program is 15 lakhs over 4 years and the average package in the university for that program is Rs 50,000 per month, then the RoI is 30 months (never mind that RoI in financial world is in terms of percentage, here months will do). One can question the exactness - you haven't taken into account the hostel/mess expense, you haven't taken the cost of time, you haven't taken into account the interest, and all that, but well, it was always supposed to be a proxy, an estimate.

But this has several problems. (And I am not even talking about how do you know these numbers. Everyone seems to present these numbers in the most confusing way.)

First, should you consider the highest salary in RoI computation or the average salary or the median salary. What is your expectation from the program - do you expect to be closer to the highest or average or what. One would say that what is the expectation of an average student. That should determine RoI of the program. But the expectation of an average student is the median placement offer and not the average placement offer. And I have never seen anyone talk about median. Everyone talks about average which could be significantly higher than median. Next, should you only consider jobs in the core engineering or all jobs of students graduating with that degree. If you look at all students, then you will get biased result. The RoI will be biased by the societal bias (good students getting to a program causing higher placement and not something that you learn in the program causing higher placement). On the other hand, if you look at only core jobs that is not right for the undergraduate programs because it is expected that you are learning to be ready for a non-core job.

OK. So we will toss a coin and take a decision of what jobs to consider for RoI. But then this method has other serious concern. If you have a choice between a 4-year BTech and a 5-year BTech-MTech dual degree (or integrated MTech), what would you choose. Your cost has gone up by at least 25% (if you don't include interest and the fees is not going up by inflation) and unless the median/average salary offer is also higher by 25% or more, it does not make sense to do MTech. In certain industries, and certain kinds of jobs, MTech is still considered very valuable and over 50 years of career, you will be compensated for that one year extra education.

Similarly, there are disciplines which have a slower start but you get to the same point in due course. So if you are taking a decision based only on first month salary, you will not choose disciplines with slower start. The reason why some of the engineering disciplines are less popular today is that first people tend to over-estimate earnings based on highest salaries offered in certain popular disciplines, and second, the core engineering have a slower start of the career.

There are other slow-start careers like Design (this is changing though). In disciplines like Design where one works with other designers in small firms, the importance of working with better known designers is so much that you may have to work for a much smaller salary initially till you establish yourself. And, therefore, the number of students interested in design are less and as the opportunities for earning better salaries right in the beginning are improving, the number of students getting interested in design are increasing.

The obvious question then is this. If estimating lifelong earnings is impossible, and just considering the estimated first month salary is deeply flawed, how do we choose between the two programs.

My answer is this. If you are taking education loan, and you don't have family support to pay that loan, then you really have no option but to look at the first month expected salary. (But even in this situation, you should only decide whether you will be able to pay the EMI, and not compare programs on where you will have more savings or less savings after paying EMI.)

In general, if you are being funded through parental savings, then think like this. What are people doing 5-10 years after doing that program. If it is a new program, can you guess based on a "similar" program. If it is a new college, can you guess based on similar program in other colleges where you think the quality of education is likely to be similar. And if that profile is to your liking, both in terms of job content and reasonable financial gains, it is ok to do that course. And if you have a choice between several such courses, choose the discipline you are interested in, or choose the college where you think the quality of education (broadly defined) will be better.

This advice is based on my life principle that while it is important to have financial security and enjoy the good things of life, beyond a certain level of prosperity, one should look for happiness, impact, time for family, and such, and money should become secondary. And if money is primary for you, go back to previous paragraph and see which program leads to best jobs after 5-10 years and one way to check that is to check linkedin profiles of alumni. So check out 5-10 years and not one month.

JEE Mains 2024 Result: Why Advanced Cutoff Highest in 5 years

The result of JEE Mains 2024 is out and every newspaper is reporting that the cutoff for eligibility of general category for JEE Advanced (for admission to IITs) has gone up to 93.2 percentile. In 2023, this cutoff was 90.7 an in 2022, this cutoff was 88.4.

Why has this happened. The simple answer is that the competition has increased. The number of students taking the JEE Mains exam has increased substantially this year and hence the percentile score has gone up.

To explain this with an example. If I want to select 10 students out of 100, I am selecting those who are above 90 percent of the students, or in other words, my cutoff is 90 percentile. But if I want to select 10 students out of 150, I am selecting those who are above 93.33 percent students, or in other words, my cutoff is 93.33.

If we look at the numbers, the total number of students who took JEE Mains (in either session) in 2023 was 11.13 lakhs, while the total number of students who took JEE Mains in 2024 was 14.15 lakhs, an increase of 3 lakh students. So clearly, if the same number of students is being selected for JEE Advanced (roughly 2.5 lakhs), the cutoff will go up. Please note that it is not dependent on the difficulty level of the question papers in different sessions. So one cannot interpret this data to imply that questions were easier or tougher this year.

But if this simple answer was the complete reason for increase in cutoff, I wouldn't be writing this blog. Enough media houses have pointed out the increased number of candidates. In fact, if the reason was only the increased number of students, the cutoff should have been around 92.7 percentile (and not 93.2).

There is another very interesting reason for increase in cutoff. And that is the impact of two sessions of JEE Mains exam.

A student can take JEE Mains once or twice (in January/April). The higher of the two scores is assigned to the student. Now, consider a scenario where 100 students take the exam in the first session, and the same 100 students take the exam in the second session. The top 10 of the first session is the same set as top 10 of the second session. And now, if I am selecting the top 10 students, I will select those who are above 90 percent students in either of the two sessions, or 90 percentile will be my cutoff. But if the two set of 10 toppers are completely disjoint, that is, there is no common person in top 10 (actually top 5), then I end up taking top 5 of first session, and top 5 of the second session, and my cutoff will be 95 percentile. So the interesting consequence of two session exam is that more than 1 percent students can be there within 1 percentile.

So, the cutoff crucially depends on not just the total number of students, but also on how many students took the exam twice and how many of them had inconsistent results in two attempts. Inconsistency here means that their performances in two sessions is significantly different. And that seems to have played a role in increasing this year's cutoff.

Last year, 5.94 lakh students appeared in both sessions out of 11.13 lakh unique students, which means that 53 percent of all unique exam takers appeared in both sessions. This year, 8.23 lakh students appeared in both sessions out of 14.15 lakh unique students, which means that 58 percent of all unique exam takers appeared in both sessions. So a lot more people have worked hard in the intervening period to see a significant increase in their performance compared to session 1, and that has increased the cutoff by another 0.5 percentile.

Please note that the exact cutoff would also be dependent on the distribution of the number of candidates of different categories, how many students from different categories get counted in general category because of their top performance, and so on. But there is no reason to think that across just one year, there would be significant differences in group performances.

What to do if you didn't meet the cutoff for JEE Advanced but still want to study in an IIT.

Most IITs have a scheme to allow students from other colleges to study as a non-degree student for a semester or two. But pretty much no college has a system of credit transfer despite NEP2020. To the best of my knowledge, the only university in India which sends a large fraction of their students to do coursework in IITs is JK Lakshipat University in Jaipur.

There is also a very liberal scholarship for those who have done well in JEE Mains. If you have 95+ percentile in JEE Mains, you study for free. If you have 90-95 percentile in JEE Mains, you get 75% fee waiver, and so on. And these scholarships can potentially last for all four years based on performance. Check out the details on the scholarship page.

 

Is there a problem if everyone wants to study CS/AI

In every media interaction and now, even social gatherings, I get asked if Indian growth story will get affected if no one studies Civil/Chemical/Mechanical/Electrical and everyone only wants to do CS, AI, and at best ECE. How will the new infrastructure be built. How will we become a hub for manufacturing. Can we be VishwaGuru with everyone studying only AI.

And I tell them to just relax. Lots of students are studying non-CS/IT engineering branches. One can check AICTE data for the same. There is no shortage of pieces of paper declaring someone to be a graduate in Civil Engineering. And it takes only 4 years to print more paper, if there is need.

The problem is that degrees are only on paper and do not reflect learning. Industry is genuinely worried that they are not finding good engineers. But that problem is also true for software industry. They too are not finding good engineers.

Who are joining non-IT engineering programs. There are, of course, rare students who for whatever reasons are interested in these programs. But these are rare. In any given college students prefer CS and related programs over other engineering programs. This is true of IITs and this is true of Tier 3 colleges. But a lot of students think about non-CS in a more competitive college versus CS in less competitive college and there are enough who choose non-CS in a more competitive college. This comparison happens not just between IIT Civil versus NIT CS. It also happens between Civil Engineering at an affiliated college versus CS at a private university. (Affiliated colleges are cheaper and a lot of students/parents believe that getting a degree from a government owned technical university is somehow more valuable and education does not matter.) This comparison is here to stay and we will always find enough students taking up non-IT engineering programs.

There are some smart students who take benefit from this lopsided preference of the society. They can get admission to a college which provides much better overall education than they would have got if all branches had equal interest from students, and they get better placement too. Don't believe me, let me explain.

Most people would look at placement data and see that average of CS batch is higher than average of Civil batch. They also notice that the highest salaries in CS are much higher than the highest salaries in Civil. And they assume that they are the ones who will get those highest salaries. But smart people will think differently. They will understand that it is very difficult to be the topper of CS batch while it is much easier in a less competitive branch. It is aided by the fact that a lot of students who join non-IT branches keep wondering if they took the right decision and whether they should have taken up CS in a less competitive college. Their peers also demotivate them. The smart people will further notice that the top few jobs to Civil engineering graduates are of higher amount than the median of CS. And hence the expected placement is better in Civil for these people. (Of course, I must quickly add that I would never recommend anyone to choose a branch based on placement data or a faulty concept like Return-on-investment. But man people still do that.)

So, overall, I don't see a problem in terms of number of people opting for non-IT branches. The problem is that a very large number of people are not getting the education they deserve in every discipline.

The related question is if it is possible for a top institution to attract better (read, more motivated) students to non-IT programs. Yes, it is possible. The simple method is to project placement data differently (unfortunately, people will look at placement data). Instead of projecting data in terms of discipline, if they project data in terms of CGPA, and point out that the placement is strongly correlated with CGPA (and to a lesser extent with a discipline), the craze for certain disciplines will go down. Also, if they don't already offer minor in CS/AI, do so. The students will have the confidence that if they don't get a job in their chosen discipline, they can go back to IT recruiters in the final year. And, finally, offer dual major. A student admitted to Civil should have the option to pick up his/her second major in CS/AI. This, again, gives confidence to the student in case things don't go right in Civil, and also produces inter-disciplinary engineers which are so important for the industry.