Emergency loans and consumption: Evidence from COVID-19 in Iran
Across the globe, the COVID-19 crisis has hit poorer population segments more heavily, especially in developing markets (Furceri et al. 2020). Working in the informal economy, primarily in services, most low-income workers are not able to work from home or benefit from the employment benefit protection of large formal enterprises. The high degree of informality also makes public health-oriented containment and their enforcement less effective, while limited fiscal space and limited access to international financial markets make economic support policies more difficult to implement (Djankov and Panizza 2020). Nevertheless, many developing country governments implemented support programmes for households and firms and an evaluation of whether these programmes were successful in reaching the most affected in the economy and what support payments were used for is thus important. In a recent paper, we offer such an assessment for emergency household loans in Iran (Hoseini and Beck 2020).
Our study is part of a rapidly increasing literature on consumption that uses transaction data for impact assessment of COVID-19, most of which are on advanced countries, including on Portugal (Carvalho et al. 2020), Denmark (Andersen et al. 2020), Japan (Watanabe and Omori 2020), UK (Hacioglu et al. 2020), the US (Baker et al. 2020) and Mexico (Campos-Vazquez and Esquivel 2020).
COVID-19 in Iran and emergency loan programme
Iran was the first country in the region to be hit by COVID-19, with the first confirmed case reported on 19 February 2020. In response to the pandemic, the government on 22 February announced the cancellation of all cultural and religious events as well as closure of schools, and universities in the affected provinces, extended to all provinces on 4 March. However, it was not until 21 March (right before the start of the Persian holiday Nowruz) that the government announced a ban on travel between cities as well as closure of shopping centres and bazaars across the country with exceptions for pharmacies and grocery stores.
As the number of new cases started to fall, restrictions were gradually relaxed starting in April. Also, in April, the government announced that eligible households can apply for an emergency loan (≈ 54% of the minimum wage). This loan of 10 million IRR is based on eligibility for a monthly cash transfer that the government has been paying to every Iranian above 18 supported by oil income, with the exception of the top 5% income earners. The loan is to be repaid out of future cash transfers, starting in July-August 2020. Out of 25.6 million Iranian households, 24.2 million are eligible for this monthly cash transfer and among them, 21 million applied for the loan. The loans were paid out in four waves, with 17.1 million households being paid on 23 April, 2.3 million on 30 April, 775,000 on 7 May, and 867,000 on 11 June. Hence, over 80% of 83.5 million Iranian individuals are covered by the emergency loan programme.
We use payment transaction data to proxy for high-frequency changes in consumption patterns across provinces and across different goods and services. This follows the approach by Aladangady et al. (2019) who show that aggregating anonymized transactions data from a large electronic payments technology company to the national level provides similar patterns of monthly consumption growth rates as the Census Bureau’s Monthly Retail Trade Survey.
Our monthly and daily transaction data are from Shaparak, a company belonging to Iran’s Central Bank that acts as the clearinghouse for all transactions done via point of sale (POS) and online terminals using Iranian rial. While we do not capture cash purchases, this includes only a small bias as according to CBI (2018), 97% of Iranian households use electronic cards as the main payment method for their purchases. We have daily data for POS (in-store) and online transactions for each of the 31 provinces for April-May 2019 and April-May 2020. In addition to data on the province level, we distinguish between durable, semi-durable and non-durable goods, 12 different groups of goods and services and 18 different retail segments. All values are in real terms, i.e. we adjust data for inflation using province-level monthly price index.
We also have data on the value of the emergency loans for each round and province and use both total loans relative to total monthly transactions and loans per household (in million IRR) in our regression analysis.
In order to estimate the effect of the emergency loans on consumption across different provinces and categories, we use a difference-in-differences set-up, which stacks daily province-level transaction data for April-May 2019 and 2020. We assume that the treatment days are from 23 April to 13 May, between the day of the first loan payment and six days after the third loan payment, while 20 to 22 April and May 14 to 20 are the control dates. We also use April-May 2019 as control period. We saturate our model with province, day, weekday and holiday fixed effects. In our regression analysis we focus on the first loan wave, as (i) we cannot distinguish between transactions of households who received loans in the first, second and third week and since the effect of loans on consumption could go beyond one week; and (ii) the first loan wave is by far the largest.
Our regression results show:
Emergency loans are positively related with higher consumption of non-durable and semi-durable goods, while there is no significant effect on the consumption of durables or asset purchases, suggesting that the emergency loans were predominantly used for their intended purpose.
These results hold when we focus only on the first week after the first loan wave as well as when consider the first three weeks after the first loan wave.
The coefficient estimates suggest that two thirds of the emergency loans went into non-durable rather than semi-durable consumption, with the largest increase in absolute value in consumption of food and beverages.
The effects were strongest in the first few days and then dissipated over time, as shown in Figure 1.
We find effects only for in-store but not online transactions and in poorer rather than richer provinces, suggesting that it is the poorer who reacted more strongly with higher consumption to the emergency loans.
Figure 1 Day effects of the first round of loans
Notes: The graphs show the estimated coefficients δ2i of the regression log(Ypt)=∑iδ1i +∑iδ2i × Loan1 + Dayt + Wdayt + Yeart + Holidayt + Provincep + ϵpt, which gives the effect of loan in Di days after the first round (23 April) of emergency loans. The 2nd, 9th, and 16th days are Friday. Loan1 is loan volume relative to total monthly transaction in the provinces. Day, weekday, year, holiday, and province fixed effects are included in the regressions.
Our findings are line with theory and previous studies on the impact of temporary income shocks in the presence of credit and liquidity constraints. (see Jappelli and Pistaferri 2010 for a literature survey), which suggest that consumers respond to negative shocks by reducing spending, especially in the presence of liquidity and credit constraints. Iran shows a high degree of financial inclusion (94% account ownership and 79% of adults with a debit card in 2017, according to Global Findex), but with large parts of the population facing liquidity and credit constraints (only 38% had emergency funds available in 2017). While in 2017 (2014), 24% (32%) borrowed from a financial institution, 40% did so in 2014 from stores and 49% from friends and families. An unanticipated and symmetric negative income shocks such as the COVID-19 shock can thus result in substantial consumption declines even if seen only as transitory and support payments by the government resulting in consumption increases, even if this support is in the form of loans and has to be repaid.
While our paper provides a snapshot of the COVID-19 crisis and government support measures in a developing country, there are further important questions that will arise in the near future. First, as these support payments are in the form of loans, to be repaid starting in July-August 2020 there are concerns of repayment burdens on the lower income segments, which calls for assessing the effect of repayments (out of income subsidies) on consumption patterns. Second, will there be a permanent shift towards online transactions away from POS transactions in store? As data become available over time, we will be able to answer these questions.
Aladangady, A, S Aron-Dine, W Dunn, L Feiveson, P Lengermann, and C Sahm (2019), “From Transactions Data to Economic Statistics: Constructing Realtime, High-frequency, Geographic Measures of Consumer Spending”, NBER Working Paper 26253.
Andersen, A, E T Hansen, N Johannesen and A Sheridan (2020), “Consumer Responses to the COVID-19 Crisis: Evidence from Bank Account Transaction Data”, Covid Economics 7: 88-114.
Baker, S R, R A Farrokhnia, S Meyer, M Pagel and C Yannelis (2020), “How Does Household Spending Respond to an Epidemic? Consumption During the 2020 COVID-19 Pandemic”, Covid Economics 18: 73-108.
Campos-Vazquez, R and G Esquivel (2020), “Consumption and geographic mobility in pandemic times: Evidence from Mexico”, Covid Economics 38: 218-252.
Carvalho, B P, S Peralta and J Pereira (2020), “What and How did People Buy during the Great Lockdown? Evidence from Electronic Payments”, Covid Economics 28: 119-158.
Furceri, D, P Loungani, J D Ostry and P Pizzuto (2020), “Pandemics and inequality:Assessing the impact of COVID‑19”, in S Djankov and U Panizza (eds), Covid-19 in Developing Economies, a VoxEU.org eBook, CEPR Press.
Hacioglu, S, D Känzig and P Surico (2020), “Consumption in the time of Covid 19: Evidence from UK transaction data”, CEPR Discussion Paper 14733.
Hoseini, M and T Beck (2020), “Emergency loans and consumption: evidence from COVID-19 in Iran”, Covid Economics 45.
Jappelli, T and L Pistaferri (2010), “The Consumption Response to Income Changes”, Annual Review of Economics, 479–506.
Watanabe, T and Y Omori (2020), “Online consumption during the COVID-19 crisis: Evidence from Japan”, Covid Economics 38: 218-252.