Some Thoughts after Diving into Prisons Data and Emerging on the Other Side

The National Crime Records Bureau [NCRB] runs the annual Crime in India reports as well as the Prison Statistics India reports. Both get their yearly moment in the sun when, around their publication, news stories with attractive pie charts get published with infographics to show that an unearthly number of cases are pending across the country, that Indian prisons are overcrowded, or that the overwhelming number of prisoners are not convicts but undertrials. The NCRB efforts at logging data came in the news this past week for a different reason — a litigation which resulted in future reports on prisons data carrying details about transgender prisoners as a specific category besides male and female prisoners. 

It so happened that I had spent the past few months working on prisons and prisons data for a project. Reading about the litigation, it struck me that while the project will take some time to come out and is probably yet another lament, it might be useful to talk about the prisons data in some detail, and specifically, about some other areas which the NCRB could perhaps think about including in future editions of its reports besides data about transgender prisoners. I sincerely mean it, because after going through all the Prison Statistics reports over 1995 till 2019, it is a feature of the reports to try and improve upon the clarity with which data is presented (accuracy notwithstanding, of course).

What Kind of Data is Available?

Prison Statistics reports are sources which freeze on a particular frame — 31st December of each year — and then give us information about prisons and prisoners within them. On prisons, we are told about total prisons across the country with explanations behind any change in numbers and specifics about different kinds of prisons; the budgetary expenditure of states on prisons; and the kinds of rehabilitative services, if any, that prisons of different states might have. With respect to prisoners, there is data about the total prisoners at the end of each year, how many of these prisoners are convicts and how many undertrials, and how long is the average prisoner's time in custody. There are also lots of charts, tables, and graphs splicing this data set across various categories: gender, age, caste, religion, educational qualification, etc. All of this data is presented in terms of national figures as well as state-wise figures, which helps to discern the vast differences in the experience of different states. 

Some Notable Macro-Level Trends from 1995 till 2019

Chandra and Medarametla provide some important macro-level trends for data between 2000-2015 in a 2018 study, and much of their findings for this 15 year period are observed even if the data set is extended for the full period from 1995 to 2019. So,

• Average national prison populations as at the end of each year have been on the rise, except for a curious seven year period between 2003 till 2010 where there were some years of a decline in the end-of-year numbers. This coincided with what were two big amendments to the procedural laws — introduction of plea bargaining (2005), and new restrictions on arrest powers (2008). The seven-year period and any linkages between the amendments and prison populations merits close scrutiny, and I would suggest that any such scrutiny will probably show some initial positive effect of the amendments in emptying out prisons. But, the bump brought by these amendments did not last possibly because of how the slow but steady decline in plea bargaining across India, and police simply going back to their old ways and finding work-arounds to new rules.
• The national year-end average for population of undertrial prisoners out of total prisoners has consistently been higher than 65% during 1995-2019, and while the seven-year period of 2003 to 2010 showed a decline of sorts, there has once again been a steady rise bringing us to an average of 69% undertrial prisoners at the end of the year. Of course some states and Union Territories had a horrible ratio for throughout this period — Delhi consistently had over 70% undertrials, with 2019 data reporting 82% prisoners were undertrials. But what worried me is that between 2010 to 2020, the gradual increase in undertrial prisoner populations was not attributable to a few states alone but because most states had gone down this path.
• Reports from 1995 till 2019 suggest that the average length of incarceration for at least 35% of undertrial prisoners is up to three months in jail, and over 60% of undertrial prisoners are in custody for up to six months. This would mean that by the end of the year, which is when the headcount is taken, almost three cycles of prisoners would have been completed. Slowly, but surely, these numbers have changed over the past fifteen years with a reduction in the share of undertrial prisoners in jail up to six months, and an increase in those who are detained for up to a year and above. 
• The pie-chart indicating the different kinds of offences for which persons are incarcerated as undertrials also reflects broad similarities over twenty-five years — offences punishable under the Indian Penal Code constitute the major share, as opposed to offences punishable by various special and local laws. The reports suggest change and stability within these two sets over time. For instance, Offences against the Body (murder, rape, etc.) are still the main Penal Code offences for which people are jailed, but the share of undertrial prisoners arrested for property crimes such as theft, cheating and forgery has increased.     

The Scope for Improvement

There are a lot of important trends that remain difficult to track in part because of the way data is presented in the reports and in part because the data is simply not presented at all. Towards this, some suggestions are flagged below:

• Crime in India reports have tracked arrest data on an annual basis for a number of years which gives an indication about how arrests without warrant are made. But at the same time, there is no clear data for police station or court bails to indicate how many of these were cases where persons had a right of bail — either because the offence was "bailable" or because the police did not complete investigation and statutory bail accrued. This makes it difficult to get an idea about how many cases are those where persons end up incarcerated despite having a pure right of bail.
• For some years from 1995 till 2019, primarily in reports after 2010, there is also about the total number of prisoners admitted to prisons each year with state-wise breakups. But little or nothing has been done with this information, and this is a problem. More details about this figure on total inmates can go a long way in providing a richer picture of incarceration trends. Today we have the National Prisons Information Portal which gives a daily update in broad-level prisons data with a seven-day history. If not the NCRB, then surely the online portal can be improved to not only give more clarity besides telling us how many admissions, releases, and visits take place daily, but also retain that data for longer than seven-day periods allowing researchers to plot trends.
• The statistics for average duration of custody for undertrials are useful, but we still do not have data linking duration of custody to types of offences. So while we do know that over 50% of undertrial prisoners spend up to six months in custody, we have no idea if certain kinds of offences are over-represented in this bracket. Such information is likely to prove critical for making targeted interventions to amend statutes by either decriminalising conduct or at least making it non-cognizable and / or bailable.  
• The NCRB data on duration of incarceration creates brackets that begin with a range of "up to 3 months". This is pathetic, in my opinion, because it is a slap in the face of our rhetoric about how each day in fetters is critical. We need to blow up this detail to figure out (a) the length of police custody detention suffered by persons before they go to jail, (b) if the "up to" 3 months is showing greater concentrations around certain days in custody. 

Like I said at the outset, I genuinely think that the NCRB can work to start looking at these and many other trends if it wants to, because the reports over time suggest a willingness to explore and expand the range of data collection and presentation. If arbitrary arrests, overcrowded prisons, and high rates of undertrial incarceration are amongst the most serious problems that trouble the criminal legal process in India, then one way to try and fix them is by acknowledging that each of these issues comprise of a myriad set of problems that can only be identified if we have the right kind of data. Otherwise, it's little more than aiming our shots in the dark.