Trends in borrowings of Indian Households from 2003-2019

The credit market plays a crucial role in economic growth and development by helping ease access of funds for borrowers. As one of the fastest growing economies, India’s credit market has expanded rapidly over the last decades with rise in demand for consumption, expansion of businesses and government interventions. However, structural barriers to access continue to persist with small borrowers finding it difficult to obtain loans and inadequate infrastructure in rural areas. This article explores the evolution of credit and its sources in India from 2003-2019 to better understand the process and the current state of the country.

In India, the Central government since the 1970s has undertaken extensive efforts to deepen credit markets, particularly by expanding the network of state-owned banks, and attempting to re-direct credit to rural households and marginalized groups (RBI, Govt. of India). In 2014, the central government through its network of state-owned banks directed the opening of over 280 million no-frills bank accounts, providing basic financial services to the poor (Govt. of India). This piece attempts to provide a broad descriptive overview of how access to household credit has changed  between 2003 and 2019. Along with overall credit access, we also focus on how household credit is distributed across India’s major communities and regions. 

Data

We have used data from the All India Debt and Investment Survey (AIDIS) for this exercise. The AIDIS is a nationally representative household survey, conducted periodically by the National Sample Survey Organisation. The first survey was undertaken in 1992, and the last in 2019. Each AIDIS survey interviews approximately 100,000 households, with the primary objective of ascertaining the household’s financial assets and liabilities. The AIDIS obtains information on household savings, bank deposits, deposits in long-term financial instruments, and the holding of assets subject to market risk such as mutual funds, equities or bonds. They also collect information on ownership of land, real estate, livestock, and transport assets.

In terms of liabilities, the AIDIS enquires about every single outstanding loan which the household has at the time of survey. Details on the source of credit, cost of credit, whether the loan is secured or unsecured, and the purpose of borrowing is also provided. There is also self-reported information on repayments made towards each loan. Additional data on household demographic and economic characteristics such as the geographical location, household size, education of household members, and per capita monthly consumption is also provided. 

For this piece, we restrict ourselves to the three most recent survey rounds – Rounds 59 (2003), 70 (2013) and 77 (2019) – and to households residing in 19 of India’s largest states. The total sample exceeds 300,000 households. Since it has been over five years since the collection of the last available data, it is possible that the trends might have changed since, however we are unable to comment on the same due to lack of current data. 

We study access to credit both at the extensive and intensive levels. The extensive level refers to whether one has any access – it measures who all is included in the system and is able to use it to access resources. The intensive measure studies the extent of access more precisely by analysing the amounts borrowed amongst households who have borrowed and how they differ amongst various demographic groups. 

Access to Credit

We begin by comparing how households’ credit access evolved across regions. Overall, household indebtedness has remained fairly similar between 2003 and 2019, with 40 percent of households having some outstanding loan. The primary difference pertains to the source of credit: the share of households with at least one loan from a commercial bank has increased from 7 percent in 2003 to 24 percent in 2019. This has been accompanied by a decline in the share of households with a loan from co-operative banks (10 to 5 percent), and informal sources (25 to 19 percent) – namely friends, relatives, money-lenders, input suppliers, landlords, employers and so forth. Both the decline in the share of households with a loan from co-operative banks, and loans from informal sources occurred primarily between 2013 and 2019. 

Figure 1a and 1b depicts the variation in credit access across geographical regions. The top panel presents the share of households with a bank loan in each state; the bottom panel shows the share of households with a loan from any source in each state. Darker shades reflect a higher share of households with an outstanding loan. Across the three figures in the bottom panel, we see that the share of households with some outstanding loan has been consistently higher in southern India, with over half the households in Kerala, Andhra Pradesh and Telangana having some outstanding loan over this period. In recent years, half the households in Rajasthan and Orissa also have some outstanding loan. From the top panel, we see that while there has been secular growth in the share of households with a bank loan across all geographical regions, this has seen a significant increase in the states of Andhra Pradesh and Telangana. The state of Gujarat (along with Jharkhand) has the lowest share of households (27 percent) with outstanding loans. 

Figure 2a disaggregates credit according to its major sources, and also by urban and rural areas. Overall, urban households have a higher propensity to have an outstanding loan – almost half of urban households in 2013 and 2019 had some outstanding loan, while a third of rural households had an outstanding loan. There has been a steady rise in the share of households with bank credit between 2003 and 2013, but this was the most marked in rural areas. While banks were the primary source of credit for urban households since 2003, professional money-lenders supplied rural credit till 2013, before being supplanted by commercial banks. Thus, a fifth of rural households had access to a bank loan in 2019, as opposed to a quarter of urban households. This assuages concerns that the expansion in bank credit was just an urban phenomenon. The expansion in bank credit was accompanied by a steady reduction in credit obtained from professional money-lenders, especially in rural areas. 

Figure 2b compares loan amounts across various sources. As opposed to the current amount outstanding, we use initial loan values for this exercise. The current amount outstanding is affected by the interest incurred on the loan which will differ across loans based on when they were taken and the interest rate charged. For example, when measuring the amount outstanding in 2019 for a loan taken in 2017 and 2018, the 2017 loan will have incurred interest for two years while the 2018 one will only have incurred one year of interest and hence, the 2017 loan will appear to be a larger loan even when the sum borrowed was the same. Thus, initial loan value is unaffected by the capitalization of interest payments and reflects the total value of credit initially received by the household and is more easily comparable. We adjust for inflation and ensure comparability over time by scaling the loan values using the wholesale price index. This results in all loan amounts being expressed in INR 2019 values. The comparison of loan sizes is also limited to households with some outstanding loan. 

In terms of average loan sizes, Figure 2b shows rural loans are between 30 and 42 percent of the size of loans for urban households. This could reflect either the rationing of credit for rural borrowers owing to lower collateral and higher costs of screening and monitoring, or lower demand for credit. While there has been steady credit growth between 2003 and 2019, average urban loan growth was faster in real terms than rural growth (7.4 percent per annum vs 6.5 percent per annum). The overall growth rates are not too dissimilar from the overall growth rate of the Indian economy during this period. 

Conditional on having an outstanding loan, the average loan size for urban households in 2019 equaled INR 405,245, while that for rural households equaled INR 153,726. For reference, average annual household consumption in 2019 for rural households equaled INR 115,449; for urban households, INR 200,155. Unsurprisingly, bank loans are significantly larger than loans from other sources, followed by loans from co-operative banks. In 2019 though, the value of rural loans from professional money lenders almost equaled that from co-operative banks. 

Figures 2c-2f explore the distribution of credit across India’s five broad communities – namely higher ranked Hindu castes, Other Backward Classes (OBCs), Dalits (Scheduled Castes or SCs), Adivasis (Scheduled Tribes or STs), and Muslims. These groups are mutually exclusive and non-Muslim religious minorities are excluded from this exercise. Across all four major communities, Figure 2c shows a steady rise in the share of households which have access to some bank loan over time. The gap in bank credit access between socially marginalized groups and Hindu dominant castes have somewhat declined over this period. Conditional on having an outstanding bank loan, Hindu dominant castes continue to have substantially larger volume of bank credit. In 2019, total bank credit for Hindu dominant castes exceeded INR 400,000; twice the magnitude for OBCs, and thrice the magnitude for Dalit and Adivasis. Hindu dominant castes also saw a higher rate of credit growth over this period. This can reflect both credit rationing from banks towards these communities, or lower demand for credit. 

Figure 2g shows the distribution of overall credit across the five major social groups over time. Other Backward Classes accounted for 40-45 percent of aggregate credit, irrespective of source, over this period. This was followed by Hindu dominant castes, whose share in credit hovered between 30 and 35 percent over this period. Dalit, Adivasi and Muslim households accounted for less than 20 percent of aggregate credit, despite accounting for a third of the households in the sample. 

A similar trend holds for bank loans too, seen in the right-hand panel of Figure 2g. Muslim, Dalit and Adivasi households received 17 percent of bank credit. The share of bank credit received by Hindu dominant castes has declined from 40 percent to 35 percent. Other Backward Classes as a group received the highest share of bank credit in 2019 at 39.5 percent. This is quite comparable to their population share of 45 percent in the survey. 

Figure 3a shows that households primarily borrowed for consumption purposes. This is true across both urban and rural areas, and for both bank and non-bank sources. In fact, the share of households with a consumption loan from a bank has been steadily increasing over time. Less than four percent of households borrowed for the purpose of non-farm businesses, and less than two percent of households had such loans from banks. This has been a constant feature in the data over the three survey rounds. 

The last two AIDIS survey rounds disaggregated expenditure loans into housing loans, loans for the purpose of health or education spending, and other consumption loans. From these, it can be seen that consumption loans were the major component of household expenditure loans. Less than 10 percent of households took loans for the purpose of health and education. The fraction of households with a housing loan was also very similar and this pattern was also observed for bank loans. Along the intensive margins, household expenditure loans from banks were the largest component of household borrowings. Based on the surveys undertaken in rounds 70 and 77, this was primarily due to loans taken for the purpose of housing. 

Figure 4 shows the cost of credit, i.e., interest rates, across various sources. Interest rates from banks were the lowest in 2019 across both urban and rural areas, with a slight decline from about 13 percent per annum to 11 percent over these 16 years. Co-operative banks too charged a rate of interest very comparable to banks. Predictably, money-lenders charged the highest rate of interest, exceeding 30 percent per annum in both urban and rural areas. There has however been a steady decline in the interest rates charged by money-lenders, from 38 percent in 2003, to 31 percent, possibly due to higher competition from commercial banks. The vast majority of loans from friends and family were provided interest-free. However, it is possible that there were other accompanying unobservable informal arrangements which compensated for the lack of interest. 

Figure 5 and 6 shows regional variations in the cost of credit. While bank interest rates were very similar across regions, overall borrowing costs declined across both north and south India. Credit costs were the highest in the state of Bihar, and the lowest in the states of Himachal Pradesh, Uttarakhand and Gujarat. These states also had a relatively lower share of loans being issued by professional money-lenders. 

Figure 7 compares the cost of credit across communities. Households from marginalized communities faced a substantially larger cost of credit in the informal credit market. Despite the reduction in interest rates charged by money-lenders, Dalit borrowers on average were charged 9 percentage points higher,  Adivasi borrowers on average were charged four percentage points higher in the informal credit market, while Muslim borrowers were charged 12 percentage points higher. The difference in interest rates however was less than two percentage points in commercial and co-operative banks, suggesting that these marginalized households were not inherently riskier borrowers, but were treated differently in the informal credit market.  

Conclusion

Overall, access to credit has increased and become more accessible with lower charges over the years. Further, there has been an increase in borrowing from banks as well as the amount borrowed from them. The main reasons for borrowing have however remained the same – consumption purposes. The data paints a picture of increasing financial inclusion across social groups and better access to credit from formal sources over the last 16 years. The past five years have seen a great deal of changes with the pandemic and it is hard to ascertain whether the patterns would have remained the same. Availability of more recent data post-2022 would help us establish how the trends have changed since.