06 May 2018

Exclusion errors, Inclusion errors and Aadhaar

This post discusses exclusion and inclusion errors in the context of Aadhaar. From a welfare perspective, reduction of exclusion errors need greater attention. The two errors are either inversely related or are positively related. The Aadhaar architecture is designed to address inclusion errors, and hence, it could either add to exclusion errors or require independent efforts to reduce them. Further, identification being independent, Aadhaar cannot address these errors at that stage. Besides, no one can be held accountable for exclusion errors.   

Introduction: Day 36
On day 36 of Aadhaar hearing in the Supreme Court of India, an implicit reference was made to exclusion and inclusion errors in the delivery of entitlement. I have already discussed about exclusion and inclusion errors in the context of Aadhaar in my earlier posts on proportionalityright to physical existence, and nir-aadhaar among others. Nevertheless, the deliberations on day 36 of the hearing urge me to further elaborate on these.

The two errors
  • Exclusion error occurs when a person deserving an entitlement is excluded (we identify entitlement with receiving of benefits, subsidy, and services in Aadhaar Act). 
  • Inclusion error occurs when a person not-deserving an entitlement is included. 

In comparing the two errors, there are two broad perspectives.

First, from a welfare perspective, exclusion errors are considered the greater evil and to ensure that this is reduced the system can be lenient to inclusion errors. Under this, the two errors are inversely related. In this, efforts to reduce the exclusion errors, will lead to an increase in inclusion errors. Similarly, efforts to reduce the inclusion errors will lead to an increase in exclusion errors.

Second, under certain situations when targets are fixed, the two errors can be positively related. The lower the exclusion errors, the lower will be the inclusion errors. As a corollary, the greater the inclusion errors, the greater will be the exclusion errors. If targets remain unaltered, then a reduction in inclusion errors will also imply a reduction in exclusion errors. Otherwise, reduction in inclusion errors cannot automatically transfer to a reduction in exclusion errors. There ought to be independent efforts to address reductions in exclusion errors. 

In fact, as targets ought to be independently assigned from macro aggregates (say, poverty ratio) then any reduction in inclusion errors leading to savings in budget implies that there are greater exclusion errors.  Or, we are no more in the realm of positive relationship between the two errors. We have moved to the realm of inverse relationship. A reduction in inclusion errors imply an increase in exclusion errors. The authentication failures or absence of Aadhaar that add to the exclusion errors will all fall under this. The savings to the state from these would be like a regressive tax.

Additional concerns
The Aadhaar architecture based on biometric authentication to receive entitlement is designed to reduce inclusion errors (reduce ghosts and duplication among others). However, it is possible that it could also add to inclusion errors. There is nothing in the architecture to prevent inclusion errors that happen after authentication. There can be either denial or reduction of entitlement (like less or no ration) after authentication. However, as the amount will be accounted for it automatically changes hands leading to inclusion errors. This is a continuation of denials when authentication was not done biometrically. Such exclusion-cum-inclusion errors, in the past, have been effectively addressed through social audits and that should continue to be be an effective method, as Aadhaar cannot address it.

A recipient of entitlement ought to be identified independently. It is only after identification that the recipient's information can be integrated with Aadhaar for biometric authentication to facilitate delivery of entitlement. This means that that inclusion and exclusion errors at the stage of identification will remain outside the purview of Aadhaar.

Entitlement denial, with or without authentication, exempts accountability from state. Under Aadhaar Act, any future claims on denial of entitlement may not stand the scrutiny of courts in the sense that the implementing entity cannot be penalised. This is so because authentication will be proof for having received the entitlement and authentication failure cannot be held against them. No one can be held accountable for denial of entitlements.

Conclusion
To sum up, between exclusion and inclusion errors the former is a greater evil. The relationship between exclusion and inclusion errors can either be inversely related or positively related. Aadhaar is designed to address inclusion errors, and hence, it could either increase exclusion errors or need additional independent efforts to reduce exclusion errors. It may weed out some ghosts, but not all inclusion errors. The architecture and design of Aadhar not only does not address exclusion errors, but it could also add to exclusion errors. Further, as identification ought to be independent of Aadhaar, the errors at  the identification stage cannot be addressed by Aadhaar. Besides, under an Aadhaar architecture, the implementing entities cannot be penalised for exclusion errors. 


Earlier blog posts on Aadhaar by the author





[The views expressed are that of the author and not that of the institutions/organisations that the author is associated with. Some of the arguments that are not elaborated here may be found in the earlier related blog posts or in tweets by the author. The author comes from a non-legal background and the final outcome of the Aadhaar hearing would depend upon the arguments put forth by the petitioners and the respondents and their considered interpretation by the honourable judges. Comments are welcome.]