Automated Forms Processing and Paper User Interfaces
for Data Collection from Village Microfinance Groups
Tapan Parikh
Dept. of Computer Science
University of Washington
Seattle, WA 98195 USA
tapan@cs.washington.edu
Vijay Pratap Singh Aditya
ekgaon Technologies Pvt. Ltd.
Kennet Cross Road
Madurai, Tamil Nadu, India
vijay@ekgaon.com
Muthu Velayutham
Covenant Centre for
Development
Kennet Cross Road
Madurai, Tamil Nadu, India
muthu@ekgaon.com
ABSTRACT
Microfinance is defined as the provision of financial services
to poor, disadvantaged and under-privileged members of so-
ciety, particularly in developing countries. Since the success
of the Grameen Bank in the late 1970s, microfinance has
emerged as a sustainable and effective method of poverty
alleviation and local financial development. Recent years
have witnessed an unprecedented growth in the scale and
reach of microfinance services around the world.
For that trend to continue, there is a need for increased data
collection and management capacity within microfinance in-
stitutions (MFIs). Accurate documentation of local trans-
actions is required for MFIs to track the performance of
their clients and to be in a position to make sound finan-
cial decisions. We have conceptualized a scalable, flexible
and accessible way for MFIs to collect data from village lo-
cations using automated forms processing and paper user
interfaces. In this paper we describe the technologies re-
quired in the proposed system and an operational protocol
for deploying it in the field.
1. BACKGROUND
Microfinance is defined as the provision of financial services,
including credit, loans, savings and insurance, to poor, dis-
advantaged and otherwise under-privileged members of so-
ciety, particularly in developing countries, who would other-
wise not have access to such services. Over the last twenty-
five years, since the pioneering work of Muhammad Yunus
and the Grameen Bank in Bangladesh [3], microfinance has
emerged as one of the most effective methods of financial
development and poverty alleviation in these communities.
In recent years microfinance activities have grown dramati-
cally and extended their reach across the globe. Although no
definitive statistics exist, it is widely believed that microfi-
nance institutions (MFIs) cumulatively process hundreds of
millions of dollars each year. Rates of return on microfinance
loans have in some cases equaled or surpassed those in the
mainstream financial sector. This has led banks and formal
financial institutions to begin lending to these groups, open-
ing up previously inaccessible sources of capital for local use.
To continue this rapid development in scale and perfor-
mance, further formalization and accountability is required
within microfinance institutions. Documenting each client’s
Figure 1: Working with microfinance group mem-
bers to elicit system needs and evaluate design al-
ternatives.
transaction and repayment history is important in order
to track the financial performance of individual members
and groups. This allows institutions to better manage their
funds, lending to individuals and groups for purposes that
will maximize repayment and overall benefit to the commu-
nity. This kind of detailed reporting is also required when
accessing capital from the formal financial sector.
This task proves to often be a large burden for several rea-
sons. Most microfinance transactions typically occur in far-
flung rural areas, making interaction with the field cumber-
some, costly and potentially error-prone. Also, the raw num-
ber of small transactions that occur in a typical operation is
so high that data collection, distribution and management
become a serious issue. This is one of the major overhead
costs involved with conducting microfinance activities and a
major reason for its perceived un-sustainability.
All of these factors lead to the conclusion that having an
integrated top-to-bottom management information system
(MIS) for managing grassroots operations is an attractive
possibility, both from an operational and financial view-
point. If there was some way to capture local transactions

and have this data transparently and efficiently integrated
into the institution’s central accounting system, it would
make data collection more accurate, less cumbersome, and
potentially cheaper as well. This would allow the entire in-
stitution to become more efficient and allow the end users
to obtain more useful information and services.
This problem has been the subject of several recent at-
tempts to introduce digital devices into village microfinance
groups to record transactions [6]. However, these efforts
have largely met with mixed results due to lack of scalability
and poor integration with the central institution’s systems.
Another issue is the user demographic. Many members of
microfinance groups, especially in agrarian societies, are un-
educated agricultural workers or laborers. The low literacy
and educational levels of these users, as well as their unfa-
miliarity with technology, makes the design of an accessible
client interface a challenging task and a crucial factor to the
system’s success. No one has yet addressed this issue in a
systematic way.
We see this as a challenging system design problem, to find
a way in which modern digital technology can be used to
provide these community-based financial organizations the
tools they need to make better decisions and improve their
financial performance.
2. PROJECT OBJECTIVE
After six months of contextual study, systems analysis and
participatory design experiments, we have developed an un-
derstanding of these village end users’ work processes, abili-
ties and interaction preferences. Please review our previous
work for the discussion of this work [7, 1]. One consistent
result is the importance of physical models and tangible ar-
tifacts. Having had years of experience recording data in
ledgers and notebooks, on top of a day-to-day life filled with
numerous physically-oriented tasks, these users are much
more comfortable with artifacts that they can handle and
touch, and have some correlation with the notebook-based
data entry mechanisms they have become used to. More-
over, these users’ current social work practices are built
around the idea of shared physical ledger information - which
can be communally viewed, exchanged and audited among
members of the group.
Even semi-literate group members have learned to interpret
data from paper forms. Through repeated use they have
been able to memorize the tabular formats, so that they
can identify personally relevant transactions. If they are
numerically literate (able to read and write numbers), they
will also understand the values of their transactions. These
are important characteristics to maintain. It is important
that any system put in place does not dis-empower the end
users, as that would defeat the whole purpose of the activity.
These studies have convinced us that paper cannot be aban-
doned as one of the primary data entry mechanisms. Paper
is cheap, ubiquitous, flexible and comfortable for the end
users. The infrastructure required to support paper based
data-entry is readily available in villages, and it requires lit-
tle additional investment in terms of technology, training
or support. Paper is a flexible, accessible, and sustainable
technology appropriate for use in local villages.
However, paper as an information medium also has its draw-
backs [8]. It cannot easily be searched, indexed or archived.
It is difficult to perform calculations and create reports from
paper data. This is an important limitation, as timely and
accurate reporting is a must for the effective functioning and
accountability of a microfinance institution. Given this, and
the large volume of transaction and account data generated,
digitization of information at some level is required for the
effective management of any middle to large-sized microfi-
nance institution. At the very least, most of the critical data
at the central office should be stored digitally [10].
3. AUTOMATED FORMS PROCESSING AND
PAPER USER INTERFACES
We have recently conceptualized a data collection system
that suits these needs. We envision a system where data is
entered, stored, and updated manually on paper in the vil-
lages. This data is then accurately and efficiently integrated
into the digital data store at the institution’s central office.
We plan to do this using two tightly liked technologies, pa-
per user interfaces [5] and automated forms processing [2,
9].
Automated forms processing is a new technique being used
by large organizations such as governments, insurance com-
panies, census bureaus and tax offices that have significant
data processing requirements. In such a system filled-in pa-
per forms are automatically processed by a scanning and
document recognition system, which can deduce the identity
of the form and extract relevant data semi-automatically.
This process relies heavily on technologies such as image
processing and optical character recognition. Recently this
solution has been implemented by companies like IBM to
automate processes such as census collection and tax form
submission [4].
In our proposed system, we imagine that village end users
will fill out transaction and account data in a format very
similar to that which they use now (see Figure 2). This form
will have been designed to be recognizable by an automated
forms processing system, so that data from the form can
easily be extracted. This paper form (or a digital scan of it)
can be transported back to the main office, processed, and
integrated into the institution’s central data store.
However, automated forms processing is not an error-free
process. A large potential for incorrect data extraction ex-
ists, particularly due to erroneous optical character recogni-
tion. Even now some of the best OCR systems operate only
at about 80-90% accuracy. Therefore there must be a simple
and effective way for end users to review the processed data,
ensure its accuracy and correct any mistakes. For this we
envision a role for paper user interfaces.
Paper user interfaces are paper forms with interface elements
within them, such as checkboxes, buttons and entry fields.
When the form is scanned, the system can parse what has
been entered into these fields and take some action based on
the results. Originally paper user interfaces (or PUIs) were
prototyped at Xerox PARC as an interface to a document
services system, where a user could print, fax or email a
document to a selected list of recipients based on entries on
a paper cover letter, which served as a paper user interface.

Figure 2: A sample spreadsheet format used for data
collection from village groups.
Paper user interfaces would be used in our proposed system
to allow users to correct errors in the automated form pro-
cessing and character recognition phases. After a form has
been processed and the data extracted, a report can be gen-
erated listing what data has been captured. On this report
each data field would be accompanied by a check box, where
a user can mark off whether or not it is correct, by compar-
ing the value to the local physical record. In those cases
where it is not accurate, the user would enter a replacement
value.
After each form (or set of forms) has been automatically
processed, a report can be generated and sent back to the
village for corrections. The users would mark up the form
and send it back to the central office for processing. Once
the data has been updated, another report can be generated
and sent back to the village for verification of the corrected
figures. If the data still is not correct the process can iterate
until no errors remain.
This is a possible protocol using automated forms processing
and paper user interfaces:
1. Villagers fill out transaction entry form.
2. This form is collected by a field representative at the
end of the month and transported to the central office.
3. The form is processed by an automated forms pro-
cessing system, and data integrated into the central
database.
4. A report is sent back to the village at the next field
visit, with check boxes to indicate errors and places to
enter corrections.
5. Local users compare the report to their records, check
boxes with errors and enter corrected values.
6. This report is collected by a field representative at the
next collection time.
7. Iterate until no errors are left in the report.
4. PROJECT BENEFITS
This paper-to-digital data collection system has many po-
tential benefits in this application domain. The system is
flexible, scalable, accessible and self-sufficient, which are all
requirements for success in this setting.
Scalable Since many microfinance institutions operate on
a very large, distributed scale - with many small trans-
actions occurring regularly over a widely spread geo-
graphic area - efficient, cost-effective scalability is a
requirement for any data collection system. Moreover,
since the money amounts of transactions are so small,
it is imperative that the incremental, per-transaction
costs be kept as low as possible to maintain profitable
operating margins.
Paper is a ubiquitous, easily available and familiar
technology. Paper forms can be cheaply created, dupli-
cated, distributed and transported. It is a sustainable
solution for the institution, as the costs and overheads
of paper can easily be borne by institutions in most
circumstances.
Accessible As described earlier, paper is a familiar medium
for the end users. After years of working with note-
books and ledgers, even semi-literate users are able to
use and interpret data on paper forms. Paper also
has social benefits as a communally viewable and ac-
cessible medium, which is important for establishing
mutual trust and cross-verification capabilities.
Self-sufficient It is important that technology not become
a limitation and source of dependence. End users need
to maintain ownership of the local processes. This is
important in order to maintain Mahatma Gandhi’s vi-
sion of gram swarajaya, or village independence. Since
paper is such a familiar and inexpensive medium, users
can maintain their current levels of experience and
ownership of the system, and be able to reason effec-
tively about it’s condition and performance. Moreover,
if for some reason the technology becomes temporar-
ily unavailable, either due to power outages, technical
glitches, budget limitations, etc., a paper-based system
can continue to operate without a significant interrup-
tion in protocol.
Flexible The system leaves the specific operational proto-
col for data collection relatively open to local control.
Even though an example protocol is described above,
there is much room for variation. One can imagine tak-
ing a digital camera or small scanner to the villages,
and collecting digital images directly from the field.
This could be advantageous if the field worker will be
out for an extended period of time, and does not want
to transport a lot of physical documents. Also, this
saves the local users from duplicating data from their
local records.
In some conditions the institution may opt to tran-
scribe data manually from some groups. This may be
for cases of remote groups with infrequent communica-
tion, or poor handwriting. The paper user interface for
data correction could still be deployed as before, be-
cause even in manual data entry there can be frequent
errors.

Some mixture of methods can be used based on the
location of each client, the scale of the operation, the
technology and infrastructure available, etc. Since mi-
crofinance institutions operate in diverse contexts and
operating conditions, this type of flexibility is invalu-
able in allowing an institution to function effectively
in different situations and to serve diverse clients.
5. RELATED WORK
Several microfinance institutions have experimented with
technology-based solutions to their data collection require-
ments. PRODEM has successfully deployed ATM-based ac-
cess points for their clients in Bolivia, including support for
illiterate users through features such as a natural language
interface and digital fingerprint recognition. However, this
was only possible in areas with dependable electricity and
communication lines, and with a huge investment in tech-
nology infrastructure that was supported by grants and gov-
ernment support.
Other organizations have experimented with technologies
such as Palm Pilots, hand-held computers and smart card
devices[6]. Unfortunately most of these efforts have met
with limited success, primarily because the overhead costs
of deploying technology in rural locations could not be sus-
tained by the local economies. Our approach seeks to avoid
this pitfall by judiciously deploying technology only at a
central location, so that thereafter it can be supported by
increasing economies of scale.
There are other forms of data collection from rural areas
that have been used by the development community in their
monitoring and evaluation activities. These include social
science methods such as ethnography, participatory need as-
sessment, case studies, contextual inquiry, qualitative and
quantitative surveys, etc., supported by audio and video
recordings. However, these largely qualitative methods are
generally not appropriate for the well-defined and well-ordered
data collection requirements in microfinance institutions.
6. CURRENT STATUS
The project is currently only in the conceptual stage. There
has been little work thus far in actually implementing or
deploying the system. One problem with the system as de-
scribed in this paper is that it allows for a lot of temporary
data inaccuracies between the central and remote locations.
This may not be an acceptable circumstance for many or-
ganizations, so we are looking at ways to reduce the level of
possible inconsistency.
Right now we are investigating options for the design or
use of a suitable automated forms processing software suite.
This is typically a large, complicated software system, with
state-of-the art character recognition and image processing
components. It would not be a simple solution to implement
from scratch, so we are seeking to leverage open source tools
and partnerships with existing automated forms processing
service providers to expedite the implementation process.
A large concern is that the hardware and software overheads
to be incurred could only be made sustainable through huge
economies of scale, even with the proposed centralization of
technology and hardware. One option is through the initial
donation of hardware, software and technology by civil soci-
ety actors such as donor agencies and national and regional
governments. These types of infrastructure inputs could do
a lot for the long-term sustainability and efficiency of micro-
finance service providers. We are currently exploring such
funding options for the development of open source software
and purchase of hardware, both in this and other projects.
Another concern is in the long-term service and maintenance
of the hardware and software required for such a system. We
are hoping that socially-oriented technology service compa-
nies, along with support from volunteers, students, univer-
sities and private companies, have a role to play in such
situations.
In the meantime, we are making progress in designing and
testing appropriate form layouts for use in the field. We
have commenced a project to design a complete Paper-based
MIS for those organizations that do not have the resources
or capacity for any kind of computerization. This project
will include the development of paper forms accessible to
semi-literate users, as well as a complete data collection,
management and analysis system based only on paper data
formats and well-documented operational protocols. This
project can show that computers are as much of a design
tool as an implementation tool in the development of new
technologies for rural, under-served users.
In this way, we are conceptualizing several approaches that
differently leverage the paper / digital interface, and are
estimating the trade-offs in terms of overhead, efficiency and
infrastructure for each solution. We hope that this will give
valuable insights about sustainable, appropriate information
and communication technologies (ICTs) for serving these
areas.
7. CONCLUSION
This project could be a compelling example of how to in-
troduce digital technology into rural villages in a graceful,
sustainable manner, without disturbing local life or making
people overly dependent on unforeseen externalities. It re-
mains true to Mahatma Gandhi’s vision of gram swarajaya,
a locally self-sufficient, independent and empowered village,
by relying on locally available technologies and resources
(i.e. paper) wherever possible. We hope that by using this
system microfinance institutions can more effectively col-
lect, process and store data, allowing communities to make
more informed financial decisions and better manage their
community capital.
In the future we hope to add other features and capabilities
to the system. Paper need not only be used as a mecha-
nism for data collection, it can be used as a way to transfer
data into villages as well. We plan to experiment with novel
reporting and data visualization techniques that allow semi-
literate users to more effectively understand their financial
positions and performance. Eventually this kind of exper-
tise can be leveraged for local business management as well.
That is another application domain we hope to tackle in the
near future.
This project is just a small part of the overall vision of using
scalable, flexible digital technologies for the empowerment

of local people and communities. While there are still a lot
of questions yet to be answered, in the long term we fore-
see many applications of Information and Communications
Technologies - implemented in locally appropriate ways - to
make a difference in the lives of poor, disadvantaged peoples
around the world.
8. ACKNOWLEDGMENTS
We would like to thank Kaushik Ghosh of Media Lab Asia
and Apala Chavan of Human Factors International for help
in earlier design studies. We would especially like to thank
the staff of CCD (including John, Annachi, Bala and Natara-
jan) and the women of the Mahakalasm Self-Help Group
Federation for the many hours of discussion and observa-
tion that resulted in these conclusions.
9. REFERENCES
[1] K. Ghosh, T. Parikh, and A. Chavan. Design
considerations for a financial management system for
rural, semi-literate users. In Proceedings of the ACM
CHI 2003 Conference on Human Factors in
Computing Systems, pages 824–825, 2003.
[2] S. Gopisetty, R. Lorie, J. Mao, M. Mohiuddin,
A. Sorin, and E. Yair. Automated forms-processing
software and services. IBM Journal of Research and
Development, 40(2), 1996.
[3] Grameen Bank home page.
http://www.grameen-info.org, 2003.
[4] IFP Success Stories.
http://www2.clearlake.ibm.com/GOV/ifp/success stories.html,
2003.
[5] W. Johnson, H. Jellinek, K. J. L., R. Rao, and
S. Card. Bridging the paper and electronic words: The
paper user interface. In Proceedings of ACM InterCHI
’90 Conference on Human Factors in Computing
Systems, pages 507–512, 1993.
[6] Microfinance Gateway - technology highlights.
http://www.microfinancegateway.org/highlight tech.htm,
2003.
[7] T. Parikh, K. Ghosh, A. Chavan, P. Syal, and
S. Arora. Design studies for a financial management
system for micro-credit groups in rural India. In
Proceedings of the ACM CUU 2003 Conference on
Universal Usability, 2003.
[8] A. Sellen and R. Harper. The Myth of the Paperless
Office. The MIT Press, first edition, 2002.
[9] S. Srihari, S. Lam, V. Govindaraju, R. Srihari, J. Hull,
and E. Yair. Document understanding: Research
directions. Technical Report CEDAR-TR-92-1, SUNY
Buffalo - CEDAR, May 1992.
[10] C. Waterfield and N. Ramsing. MIS for microfinance
institutions: A handbook, 1998.