Search This Blog

Wednesday, July 15, 2020

Governance in our universities

A few years ago, IIT Kanpur conferred its Distinguished Alumnus Award to Dr. Pramath Raj Sinha, who has been an institution builder, having been associated with ISB and Ashoka University and many other initiatives and institutions. In a public lecture, next day, he was asked for the secret of his success. How come all the institutions he is associated with are doing so well. And his answer was that the most important element in success of an educational institution is its governance structure. Of course, faculty, infrastructure, curriculum, research, and so many other things make up for an excellent institution, but good governance makes all these things happen, and good governance is more likely to happen when there is a good governance structure.

And the problem in most of our institutions is that their governance structure is weak. The number of board members who owe their membership to connections in the government is large. Self perpetuating boards are almost non-existent (perhaps, IIIT Delhi is the only exception in the government sector). The chairperson is usually a political appointee (not usually a politician, but still). Thankfully, it is becoming increasingly common to have some faculty representation in the board.

Even when there are non-government nominees, they are either ex-officio, or decided by the government. So if you are enlightened enough to have an alumni representation in the board, it will be either ex-officio (let us put President of Alumni Association in the board), or let the government decide which alumni. Why can't the board decide who will be the alumni for the next term.

The governance structure within the institution is no better. One either has extremely rigid structures (like every Head has to be through seniority and one can not consider leadership, passion, vision, etc.), or there is no structure at all and a Director/VC can appoint anyone in a dictatorial style.

The selection of Director itself is seriously flawed. Often that process takes a few minutes of interaction. How can the two sides understand each other in a few minutes.

To make matters worse, there are no red lines around any entity. If an employee has a grievance, the email will be sent to board members and they will even oblige by asking the institute questions about that grievance. It is common to interfere in the internal functioning of the institute. Ideally, there should be very clear distribution of responsibilities for various committees and one should not interfere in the functioning of the other. On the other extreme, since the roles are not well defined, some people in leadership positions just refuse to take any decisions, like Heads and Deans will seek approvals from Director/VC for the smallest of things. You can't be nimble in such a setup.

Further, the concept of conflict of interest is not understood at all in our committees. This is such a serious problem that may be one day I will write a full blog article about it.

And we continue to perpetuate the poor governance models. Anyone from IIT system, for example, can tell you that the success of IITs is due to multiple factors, including resources, autonomy, etc., and an extremely important reason is its governance structure. And yet, new institutes keep coming up with other poorer governance models. UGC will even ask institutes as to why they are following IIT model of governance and not the older university model.

Thankfully, some of the newer institutions like Ashoka University are showing the way forward with a much better governance structure and as Dr. Pramath Sinha said that is one of the important factors for its success.

How do we improve governance in educational institutions. It is perhaps possible to change the legal structures, the composition of boards, the selection process of Directors and Vice Chancellors, etc., but it would be very difficult to change the culture. But we must start one day. Remember the corporate governance in Indian companies was considered quite poor just 25 years ago. If we start improving governance in education, in a couple of decades we will be at par with the rest of the world.

Tuesday, July 14, 2020

What do I want in campus placement data

Every year, educational institutions are going to make certain claims about jobs that their graduates have secured, and more often than not, the data will give no real picture to anyone (and surprisingly, even the students about whom this data is will certify on social media that this data is correct, when it is not).

The placement data is the most important parameter in the minds of potential students and parents. It is also a fairly important parameter for NIRF ranking and accreditation. And that gives a huge incentive to institutions to fudge this data, at least in ways in which they can claim that they weren't lying (if data were to leak out). And if some do it, it puts additional pressure on those who wouldn't have done it otherwise. But speaking truth will hurt them disproportionately.

Most institutions will give a couple of numbers: What fraction of students received jobs, and what were the highest and average salary packages.

How are they fudged. The fraction of students receiving jobs is often calculated over their most popular programs only. So in an engineering college, it would be only for BTech program and would not consider MTech students whose placements are typically poorer. Second thing that would happen is that it would be calculated over "eligible students" and not over all graduating students. Different institutions would have different definitions of "eligible." Some would consider those eligible who have a CGPA of higher than a certain threshold (since very poor CGPA students would find it difficult to get a job, so why include them in the announced data). Similarly, some would exclude those with a few fail grades. Some may consider only those who have applied for at least N jobs. (Apparently those who apply for very few jobs and not succeed in them are not really serious in placement. Why should we consider them in our data and reduce our numbers.) But, hey, if you are giving someone a degree, shouldn't you consider him ready to take up a job. Or is it that your degree has no value in the job market.

There is more. Suppose there are 100 eligible students. 70 students get one job offer each. And 10 students get two job offers. So 80 students get 90 offers. Most institutions will call it 90 percent placement, hiding the minor detail that it was impossible for a student to give his/her job away to some other student.

And then we come to packages. Here, the companies will fudge. We will shout at them, but then in our statistics, we will use their fudged numbers. So the maximum possible bonus value which very few employees are likely to get will be added to salary component. The initial joining expenses/reimbursements will be added. The value of shares which you can sell only if you stay on for 5 years will be added (and not even divided by 5). And so on. So the maximum will be higher than what truly will be the income of the graduate. Finally, we will talk about average and not median which is a better indicator of spread in this case. We will do that since average is always higher than median when it comes to placement packages. A few top packages will skew the average.

What would I like to see as a parent of potential student. Ideally nothing. I would only like to check the credentials of faculty before choosing the place to study. But, that is not the answer anyone is satisfied with. So here is the next piece of information I would like to see.

Let us first talk about BTech program. How many of your recent graduates got a job, or admission to a higher degree program, or joined a family business, or started a company of their own. Basically, what fraction of your graduates are doing something meaningful (and I personally don't consider dropping a year to study GATE as meaningful, your mileage may vary). When it comes to a job, I would only consider jobs which are about double of minimum wages plus yearly interest on the cost of degree. So, if the cost of the program was 10 lakhs, and considering that today's minimum wages are about Rs. 12,500 per month, I would only consider jobs which pay about Rs. 35,000 per month or Rs. 4 lakhs a year. (twice of 12.5 K, and 10K per month of interest payment).

When it comes to MTech program, the cost of the program is rather small, since one gets a decent stipend every month, which takes care of a significant part of the cost. But still, if you had a loan from your BTech program, that would have accumulated more interest, and you have 2 additional years. So the expectation should be at least 5 lakhs a year. (Notice, the assumption is that one got a stipend, didn't spend much more than stipend on the education. If these assumptions do not hold for a particular program, then I would only consider jobs of higher remuneration for those programs.)

And similarly for PhD program, I would consider only those jobs which give more than 7-8 lakhs.

Now, will a university tell me what fraction of all their graduates are doing a job with certain minimum compensation, or higher studies, or family business, or startup (include anything else which is a reasonable outcome of education).

The universities won't do that. And their reluctance is not only due to fear of losing competitive advantage, but also internal.

As I wrote in my previous blog, a consistently poor performance of a program would raise questions about its possible closure. No one wants to do that. So it is best that such information is hidden from everyone. In fact, in terms of jobs, we can add one more information. How many of those jobs were in technical areas, and how many of them were in non-technical areas. Now, we should be able to review all programs. Consider a program with 50 students. If only a couple of students go for technical jobs, and another couple of them for technical higher education, while may be another 20-30 get sales/marketing/finance jobs or go for MBA, should we continue with this program, or transfer these seats to a popular program so that more of our students can follow their passion.

Having quality data easily available would help both potential students/parents on one side and the decision makers on the other. But a unilateral declaration by one college would hurt that college in admissions.

How do we go about ensuring that a large enough set of colleges give out quality data in the same year. Or alternately, everyone at least describe their methodology, whom they include/exclude, for example. How do we ensure this.


Monday, July 13, 2020

Why should we (not) run an MTech Program

One of the constant refrains of educational institutions is that they don't get enough funding. And indeed, if you look at the public funding of higher education, it is much lower than other countries at the same level of development. But, are we spending the money efficiently. Is taxpayer getting the bang for the buck.

And, the answer, unfortunately, is No.

One of the problems in our setup is that once a program starts, it is very difficult to close it. Also, there is never a review of programs from the perspective of their growth or closure. Indeed, even when we start a new program, there is often insufficient due diligence.

For example, why do our institutions run MTech programs. Are they cost effective. Can we spend the same amount of money for better social benefit elsewhere.

In my discussion with many educators, I have come up with the following reasons to have MTech programs:

  1. It is a research program. We are not able to attract enough PhD students. So they form the backbone of our research efforts.
  2. We are producing super specialists in areas where there is a significant industry demand. Under-graduate programs are more broad based but industry needs people with more depth.
  3. It is a profit making program, which allows me to subsidize either research or UG tuition.
  4. We cannot admit a large number of students in our 4-year program. So we admit some more in a 2-year program to give them an exposure to quality education that they will otherwise not get.

And, of course, it could be a combination of these reasons. (And, if you think we should run an MTech program for any other reason, feel free to put that in a comment below.)

Now, whatever are the arguments for any particular program, we should have data to support that argument.

Is it a research program. How many research papers of a reasonable quality have come out of thesis work. In places where PhD is the primary research manpower, one sees anything between 0.5 and 1.0 papers per PhD student per year. So a PhD student who takes 5 years to complete the program will publish at least 2-3 papers, and in many cases 5 papers or even more (including some which will get published after the PhD is over). If MTech is the primary research manpower, is it fair to expect 0.1 papers per MTech student per year (which basically means that out of 5 MTech students, there should be one decent paper over their 2 year period). If you are not getting even this much, then your MTech program is not a research program.

If you are producing quality manpower for industry, how is your placement of MTech programs. Are your MTech students getting placed in companies where their technical skills will be useful (or they are going for PhD). Are those companies valuing their skills to the extent that they offer them a higher compensation package than your BTech students. So, if you are able to place 50 percent of your graduates in jobs that will use those technical skills, and which pay at least 10 percent higher than your BTech median salaries, I think it is ok for you to claim that you are running an MTech program to satisfy an industry need. And if industry is really so keen, you should be running your MTech program with a batchsize of 40-60 students, and not 10-15 students. You are not serving the society by graduating a tiny set of students.

If it is a profit making program, then of course, no other argument is needed. But the program can be profit making only if you don't have specialized labs, you are recruiting part-time or contractual faculty at low wages. Are you telling all this to students before they seek admission.

If you are doing a favor by admitting students who missed getting admission 4 years ago, and they have now proven that they are worthy of studying in your institution, then again, what they do after graduation should be better than your BTech students. (In terms of jobs, higher education, entrepreneurship, and so on.) Is that happening.

In most of our MTech programs, we really can't argue on any of these lines, and yet we have a large number of MTech programs. Remember, that unless you are running MTech with part-time/guest/low wage faculty, it is a very expensive program, because the class sizes are typically much smaller. It often needs specialized labs which only a few students will use. The thesis work requires much more expensive faculty time and if it is not resulting in research publications, the return on investment is rather low, and we probably should increase the BTech seats instead.

But often there is no evaluation of any program, and once a program starts, it continues for a long time. And so, our very expensive MTech programs continue for pretty much no reason at all. My suggestion is that everyone should look at what MTech graduates are doing: quality of jobs, publications, enrolling for PhD, etc. Each public institution should clearly and publicly articulate when will the program be allowed to continue and when will it be closed. For example, they may say, we will close it if the total number of students admitted is less than 10-15 for two consecutive years. We will close it if less than 50% of graduates in a batch get job with salary higher than BTech median or get admission in a higher ranked college or whatever else. And if the output is not sufficient, close that program.

The same resources can be used for a better program. In fact, not closing current programs is a major reason for the inability to start better programs.

Sunday, July 12, 2020

Grade Inflation

If you are a teacher, whether in a school or a university, ask yourself a simple question: The marks/grades received by your students, do they truly reflect the level of learning of that subject.

I have asked this question to a lot of people in many colleges, and not surprisingly, the answer has been in the negative. If you are an affiliated college, do most students getting close to 100 percent marks a reflection on their level of learning. Nobody would argue in positive there. If you are a university, does everyone getting an "A" grade in final year project reflect the quality of projects done at your university. (Simple test: Would you put up the project report online in a publicly accessible site for all projects who have been awarded an "A" grade?)

And yet, when one raises the issue of grade inflation, the attackers are ready. Anyone talking of grade is giving too much importance to evaluation, and not learning. Evaluation leads to stress and reduced learning. I am going to argue here that having an honest grading would actually support learning.

Another argument one hears is that since everyone does grade inflation, if we don't do it, our students will suffer during placement, which is something that I disagree with.

In an ideal world, we would all have great teachers who have this wonderful command of their subjects, have a great ability to motivate students to learn their respective subjects, and all our students come to college only to learn new things. We probably won't even need evaluations in an ideal world.

In real world, the faculty members mostly know their stuff, but are not able to impress students with their performance in the class. And students have the innate capability to learn those topics, but have been told by their seniors that only a few courses need to be learnt for placement, which can be done in 2 months before the placement season. Yes, CGPA is somewhat important because many companies will do shortlisting based on CGPA. So just worry about a respectable grade, and ignore learning. That can be done, as I said above, in 2 months.

So, in the real world, the only handle a good enough faculty has to "encourage" a good enough student to learn is a grade. Note that some students would learn despite faculty. And some faculty would be able to motivate students despite evaluations. But those are outliers and not the mainstream. And if the only handle you got is a grade, then not using that handle is not just dereliction of duty, but a criminal waste of national resources.

The fear that honest grading will lead to poor placement is far fetched. In fact, opposite will happen. If you insist on hard work, most students will deliver on that and will learn better. You can actually do an experiment. In the final year BTech Project, just give an "Incomplete" grade to a few students and tell them that they are given 15 days to present again. And now notice the amount of hard work they do in those 15 days. It would be absolutely remarkable. After that hard work, in any interview, they will be able to answer any question about their project and will sail through easily.

We should also look at the placement data more carefully. Among the popular jobs, how many students would not have been shortlisted if their CGPA was less by 0.2 or 0.3, and how many of those eventually got those jobs. The number could be non-zero, but will not be large, particularly for technical jobs. And if their technical skills were strong, many more of those shortlisted would have got the job.

A few years ago, I collected data on grades of graduating batch from several institutions. It turned out that the top institutions who are famous for their quality of teaching learning processes had a median graduating CGPA around 7.3. Those which were like just behind had median graduating CGPA around 7.5. And our typical Tier 2 institutions were around 7.9 or 8.0. So it seems that higher the grade inflation, the lower is the placement.

Easy grading is a strong disincentive to learn in the context of Indian institutions, and frankly, it is done not because faculty is concerned about the career of the students, or about comparative grading practices of competing institutions, but mostly because faculty members do not wish to work hard on proper exams and proper grading. As I said above, most faculty members are aware that they are being dishonest in grading. And students are only too happy.