January 2021 · Number 9L The Impact of COVID-19 on Formal Firms in Honduras: Evidence from Monthly Tax Returns Pierre Bachas, Anne Brockmeyer and Camille Semelet1 S UMMARY We measure the impact of the COVID-19 crisis and the re- service sector was the most severely affected, experiencing a 45 sulting lockdown on formal firms in Honduras, using monthly percent revenue loss. Larger firms experienced smaller revenue value-added tax records for January 2018 to August 2020. losses than smaller firms, even when accounting for the sectoral Firms’ revenue fell by 26 percent, or 342.6 billion lempiras composition of firm-size groups. A non-negligible number of (USD 14.3 billion), in real terms between March and August firms remained shut down until the end of available data in Au- 2020 and the same period in 2019. Sectors subject to stricter gust 2020. containment measures experienced larger revenue losses. The 1 The World Bank; with substantial inputs from Jose Carlo Berm´ udez Sanchez, Roldan Manuel Enamorado Irias, Pedro Rafael Zuniga Figueroa, David Fernando Pineda Pinto (Servicio de Administraci´ on de Rentas, SAR), Thiago Scot (University of California, Berkeley) and Vedanth Nair (Institute for on Fiscal Studies). The findings and conclusions are those of the authors; they do not represent the views of the World Bank or the Servicio de Administraci´ de Rentas in Honduras. We thankfully acknowledge funding by the World Bank through the Knowledge of Change Trust Fund and the Fiscal Policy and Sustainable Growth Unit, and by the UK Foreign, Commonwealth and Development Office (FCDO) through the Centre for Tax Analysis in Developing Countries (TaxDev) at the Institute of Fiscal Studies. C ONTEXT struction. The lockdown was further eased-up on June 8th for In response to the COVID-19 health crisis, the govern- most of the country, and on August 3rd for areas with the high- ment of Honduras declared a national state of emergency est COVID-19 incidence. To mitigate the impact of the crisis and implemented a strict lockdown beginning March 16, on businesses, the government took several fiscal measures 2020. The containment measures included a nation-wide cur- aimed at supporting businesses and maintaining employment.2 few and the closing of national borders. In the beginning, Most importantly for the purpose of this paper, firms were al- only essential businesses were allowed to remain open. Non- lowed to defer their VAT filing and payments if they stopped essential activities were gradually allowed to re-open in mid- operating during the lockdown. April, starting with hardware stores, food delivery, and con- Figure 1: Changes in the Number of Firms Filing VAT (a) Temporary Exit (b) Re-Entry (c) Exit without re-entry Panel (a) shows the number of firms that exited the VAT panel each month (by not filing for at least one month), but subsequently re-entered before the end of the year (or before August for 2020 data). Panel (b) shows the number of firms rejoining the panel each month, after having temporarily exited. Panel (c) shows firms which left the panel and have not re-entered until August 2020 (the last month for which data is available). See Figure 7 in the Appendix for first entry. Figure 1 examines changes in the number of firms filing to the economy. While the data currently does not allow us to VAT, suggesting that the administrative data provide a reli- fully distinguish between firms that temporarily stopped their able picture of the impact of the crisis. First, the peak in firm operations and firms that have closed down, the extension of exits in March (Panel (a)) corresponds to the beginning of the the VAT panel data to the end of the 2020 would allow us to lockdown in Honduras. Panel (b) shows that firms started re- identify firms that have re-opened after August 2020. opening in April, when restrictions were eased, and continued to re-open in larger numbers until August. It is likely however that some businesses took advantage of the exceptional VAT OVERALL I MPACT deferral decree and did not report their sales even if they had The large rise in the number of firms temporarily ex- some, which would lead to an overestimation of the true num- iting the VAT system in Figure 1 motivates the use of an ber of firms with zero turnover. Second, the crisis and resulting unbalanced and a semi-balanced panel3 for the rest of our lockdown had an immediate impact on the economy, with analysis. The true impact of the crisis should fall in between Panel (a) showing that more than 4000 firms did not file for these two estimates. A strictly balanced panel is not considered VAT in March 2020. Third, despite the easing of the lockdown as it would exclude the large number of firms that did not file by August 2020, Panel (c) shows that about 2600 firms have in March or April 2020 because of the lockdown, but resumed not resumed filing by then, indicating possible lasting damage their operations in the following months. A semi-balanced 2 In particular, companies that maintain their pre-crisis employment level during the 2020 financial year will benefit from a one-off income tax credit (10 percent of salary expenses). Second, advance payments for corporate income tax were reduced, and personal income tax payments and social security contributions were deferred. Third, cleaning products and medicines was fully exempt from the VAT. See the IMF Policy Responses to COVID-19. website for an updated list of fiscal measures adopted, and the Decree No. 24-2020 on VAT payment deferral. 3 The unbalanced panel includes all firms, independently of their lifetime in the panel. Firms in the semi-balanced panel must appear at least once in every quarter of 2018 and 2019, and in the second or third quarter of 2020. 2 panel on the other hand includes firms that re-appeared in the disappeared in March. In all analyses, sales are adjusted for second or third quarter of 2020, even though they may have inflation. Figure 2: Aggregate Impact on VAT Sales (a) Seasonal Plot of Log Sales, Unbalanced (b) Monthly Relative Change in Adjusted Sales Note: Panel (a) shows the seasonal trends of aggregate log(sales) for all firms filling VAT in 2018, 2019 and 2020 (unbalanced panel, sales are adjusted for inflation). Panel (b) shows the monthly relative change in aggregate adjusted sales, between 2019 and 2020, for a semi-balanced and an unbalanced panel. For example, the monthly change in September 2019 is computed by comparing sales in September 2018 and in September 2019. The black dashed vertical line marks February 2020, the last month before the lockdown was introduced on March 16. Figure 2 shows that the crisis led to a large and sharp to the end of summer. In comparison to Figure 1, we see that drop in aggregate reported sales. Panel (a) compares sales sales were most severely hit in April when the lockdown was in 2020 to sales in 2019 and 2018, for an unbalanced panel fully in place, and the gradual resumption of activities started to of firms. In 2018 and 2019, the log of aggregate sales fol- show its effect from June. The difference between the balanced lowed a similar trend: sales were lower than average in Jan- and the semi-balanced data in Panel (b) is almost impercepti- uary and February, and higher than average in November and ble during the months of the crisis, which suggests that most of December. In 2020, the same pattern appeared in January and the firms that stopped operating during these months are small- February, but sales dropped dramatically in March. Panel (b) size firms which only make up a small proportion of aggregate shows that sales dropped by almost 40 percent in April 2020, sales. and remained below 2019 levels by more than 15 percent up I MPACT BY I NDUSTRY 3 Figure 3: Impact on VAT Sales by Sector Note: These graphs show the monthly year-on-year change in aggregate sales by sector, for the last year of the panel. The percentage above each graph corresponds to the relative change in aggregate sales, between March-August 2020 and March-August 2019, for the unbalanced panel. The black dashed vertical line marks February 2020, the last month before the lockdown was introduced on March 16. The category “Primary/Other Secondary Sectors” includes primary activities and secondary activities excluding Manufacturing (A, B, D, F). The category “Other services” includes services activities (R, I, L, P, S, Q) but excludes Retail, and the category “Knowledge-Based Activities” includes (J, M). Financial and administration activities are not represented here. See Table 1 for the full description of the sector categories and the breakdown of the Retail sector. Figure 3 examines the impact on VAT sales by sector, faster recovery. The timing of the worst part of the shock (max- showing that the service sector, which was most affected by imum monthly drop in sales) differed by a sector’s position in social distancing measures, experienced the largest drop in the value chain. The primary and secondary sectors experi- sales. Sales in services fell by about 44 percent from March enced their largest drop in sales in May, compared to April to August 2020, compared to the same period in 2019. Within in the other sectors, suggesting a one-month lag in the shock. the service sector, the arts and entertainment subsector as well This is likely driven by the extractive industry which faced a as the accommodation and food services subsector were by far fall in demand for raw materials from the manufacturing sector. the most severely affected, experiencing a drop of 84 and 74 percent respectively.4 Perhaps surprisingly, the retail sector has Despite the disparities across industry groups, it is been impacted evenly across essential trade (which includes su- worth noting that all sectors still faced a significant shock permarkets and grocers; pharmacies and health providers; gas compared to the previous years and that this shock is long- stations; and vehicle repair shops) and non-essential trade.5 lasting for most firms. By the end of the summer, only the Both essential and non-essential retail experienced a 30 per- mining industry and health activities overcame the shock and cent drop in sales on aggregate between March and August managed to exceed 2019 levels in aggregate sales.6 2020 compared to the previous year, though non-essential re- tail experienced a larger drop in sales in April and enjoyed a I MPACT BY F IRM -S IZE 4 Table 1 in the Appendix lists all sectors by decreasing order of the overall shock faced from March to August 2020, and Figure 5 shows similar graphs as Figure 3 for all subsectors. 5 See full description of essential retail activities in Appendix Table 3. 6 See Figure 6 in the Appendix for the detail by sectors. 4 Figure 4: Impact on VAT Sales by Firm Size (a) Quartiles (b) Top 5 Deciles Note: Panel (a) presents the monthly relative change in adjusted sales between 2019 and 2020, by quartiles of average 2018-2019 sales, for a semi-balanced panel of active firms only (see Figure 8 in the Appendix for the unbalanced panel figure). Firms that have no sales in 2018 and 2019 are dropped. See Table 2 for the quartile thresholds. Panel (b) shows the relative change in sales for the top 5 deciles, also based on the average 2018-2019 sales distribution. The black dashed vertical line marks February 2020, the last month before the lockdown was introduced on March 16. Firms in the semi-balanced panel must appear at least once in every quarter of 2018 and 2019, and in the second and/or third quarter of 2020. In Figure 4 we study the impact of the shock by firm size, showing that small firms experienced a significantly Figure 5 shows that this result is not due to the sectoral larger drop in sales than large firms. We plot the relative composition of firm-size groups, i.e. the fact that smaller change in aggregate adjusted sales by quartiles and deciles of firms are over-represented in retail. When we reproduce Fig- 2018-2019 average sales, for the semi-balanced panel.7 Panel ure 4 (a) for the manufacturing and retail sectors separately, the (a) shows that the size of the shock decreases with firm size, difference in the size of the shock across the firm-size distribu- with the smallest quartile of firms experiencing a 65 percent tion becomes even more striking. Within the manufacturing sales drop, compared to a 35 percent sales drop for the largest sector (Figure 5, Panel a), smaller firms experienced a shock quartile. After the initial shock had passed, the smallest quar- almost 50 percentage points greater than the shock experienced tile of firms experienced the same sales drop as mid-sized by larger firms. Within the retail sector (Panel b), the picture is firms. Panel (b) breaks down the change in sales for the top 5 very similar although large retailers still faced a bigger shock deciles of firm size and confirms that the largest firms faced a than large manufacturers. Small retailers (in the first quartile) smaller shock overall. recovered faster than the small manufacturing firms. D ISCUSSION 40% in April 2020, with a gradual yet incomplete recovery un- til August 2020. Service sector firms and small firms in all This note is a first attempt to measure the impact of the sectors have been hit hardest and will likely constitute the ma- COVID-19 crisis and the lockdown on firms’ revenues and as- jority of firm closures. Our results also highlight the need for sess heterogeneity in the size of the revenue shock. Our re- further research on these issues. Extended data for the sec- sults suggest that a large number of firms, possibly up to 2600 ond half of 2020 would provide a better pictures of the num- (9.4% of the pre-crisis firm population) may have shut down ber of firms that ceased their activity completely, and on the completely. The lockdown caused a drop in sales of almost 7 Figure 8 in the Appendix plots the percent change in sales for quartiles for the unbalanced panel and shows a similar trend. 5 Figure 5: Impact on VAT Sales by Firm Size Within Sector (a) Manufacturing (b) Retail (Essential & Non-Essential) Note: These graphs present the monthly relative change in adjusted sales between 2019 and 2020, by quartiles of average 2018-2019 sales, for a semi-balanced panel of active firms, by sectors (see Figure 9 in the Appendix for the unbalanced panel equivalent). Firms that have no sales at all from 2018 to 2019 are dropped. Panel (a) shows quartiles for the manufacturing sector and panel (b) for the retail sector. The black dashed vertical line marks February 2020, the first full month before the lockdown was introduced on March 16th. Firms in the semi-balanced sector must appear at least once in every quarter of 2018 and 2019, and in the second and/or third quarter of 2020. pace of the recovery. Data on personal income tax and pay- tags (and ideally with some measure of consumer spending) as-you earn filings would allow studying how firms respond to would allow for a more finely-grained analysis of the demand the revenue shock, in particular whether they reduced employ- and supply response to the crisis, as in the analysis conducted ment and wages. Linking these various datasets with location by Chetty et al. (2020)8 . 8 Chetty,R., Friedman, J.N., Hendren, N. and Stepner, M., 2020. How did covid-19 and stabilization policies affect spending and employment? A new real-time economic tracker based on private-sector data (No. w27431). National Bureau of Economic Research. 6 A PPENDIX Figure 6: Relative Change in Sales 2018-2020, from September 2019 to August 2020, All Sectors Note: These graphs show the monthly relative change in aggregate sectors’ adjusted sales, between September 2019 and August 2020, by sectors. The black dashed vertical line marks February 2020, the first full month before lockdown was introduced on March the 16th. Table 2: Average Firm Sales and Value-Added (in thousand) in 2019, by Quartiles and Panels Unbalanced Balanced Quartiles Sales Value-Added N Sales Value-Added N 1 9.7 2.9 6,788 21.4 10.0 4,462 2 73.0 38.9 7,067 135.3 73.8 4,461 3 330.3 189.6 7,128 555.7 307.0 4,461 4 13,113.2 5,505.5 7,165 18,660.0 7,529.6 4,461 Notes: Quartiles are based on 2018-2019 average adjusted sales for active firms only, i.e. firms with null reported sales in 2018 and 2019 are not included. Firms in the semi-balanced must appear at least once in every quarter of 2018 and 2019, and in the second and/or third quarter of 2020. 7 Table 1: Sectors, by Decreasing Order of Shock Size during March-August 2020 Unbalanced Panel Balanced Panel Sales Sales N N Sector Full Sector Name Shock (% Agg. (2019) Shock (% Agg. (2019) Sales) Sales) R Arts, Entertainment and Recreation -84.7 0.1 277 -78.9 0.1 176 I Accommodation and Food Services -75.0 1.4 918 -74.4 1.5 628 F Construction -63.1 2.2 1221 -62.4 1.9 598 Activity of Extraterritorial Organizations and U Bodies -56.2 0 7 -41.5 0 2 Z Undefined Category -52.1 0.1 281 -52.2 0.1 210 Activity of Households As Employers, Activities of T Households for Own Use -51.2 0 4 -54.0 0 3 L Real Estate Activities -48.9 1.3 1454 -47.5 1.2 1025 H Transportation and Storage -39.7 2.6 1068 -40.1 2.7 758 P Education -39.5 0.5 478 -38.7 0.3 186 G1 Essential Wholesale and Retail Trade -31.8 20.5 2118 -31.7 22.3 1539 S Other Service Activities -31.8 2.2 4785 -28.4 1.9 3356 G2 Non-Essential Wholesale & Retail Trade -29.2 13.6 5751 -28.4 14.2 4149 M Professional, Scientific and Technical Activities -24.2 1.4 1786 -25.8 1.4 1202 C Manufacturing -23.3 34 2729 -21.6 32.5 1997 Public Administration and Defence, Compulsory Social O Security -22.9 0.2 66 -15.3 0.2 39 Energy (Electricity, Gas, Steam and Air conditioning D supply; Water supply Sewerage, Waste Management -20.7 7.1 334 -21.3 7 236 and Remediation Activities) N Administrative Activities -20.4 1.7 1203 -21.1 1.7 817 J Information and Communication -11.2 3.1 869 -11.3 3.5 638 Q Human Health and Social Work Activities -10.3 0.7 720 -12.5 0.6 362 K Financial and Insurance Activities -8.2 1.9 568 -8.7 2 282 A Agriculture, Forestry and Fishing -4.5 4.8 1024 -12.3 4.4 521 B Mining and Quarrying 3.6 0.5 60 4.2 0.6 42 Notes: Sectors follows the NACE Rev. 2 classification. Sector D includes both “D” - Electricity, Gas, Steam and Air conditioning supply” - and “E” - Water supply, Sewerage, Waste Management and Remediation Activities - original NACE Rev. 2 categories. Original sector “G” - Wholesale and Retail Trade - is split between essential and non-essential activities. Essential activities have the following 4 digit codes: 4530, 4772, 4730, 4661, 4712, 4711, 4520, 4721, 4620, 4645, 4781, 4653, 4630, 4722, 4791 (see Table 3 for details); the rest is classified as non-essential. Columns ”shock” present the percentage change of aggregate adjusted sales from March-August 2019 to March-August 2020 by sectors, for both panels. 8 Figure 7: VAT Payments – Number of First Entry Note: This graph shows the number of firms that first entered the VAT panel, from February 2018 to July (or August) 2020. Figure 1 presents exit and re-entry patterns. Figure 8: Impact on VAT Sales by Firm Size – Unbalanced Panel (a) Quartiles, based on 2018-2019 average sales (b) Top 5 Deciles, based on 2019 Sales Note: Panel (a) presents the monthly relative change in adjusted sales between 2019 and 2020, by quartiles of average 2018-2019 sales, for a semi-balanced panel of active firms only (see Figure 4 for the semi-balanced panel figure). Firms that have no sales at all from 2018 to 2019 are dropped, see Table 2 for the quartile thresholds. Panel (b) shows the relative change in adjusted sales for the top 5 deciles, also based on the average 2018-2019 sale distribution. The black dashed vertical line marks February 2020, the first full month before the lockdown was introduced on March 16. 9 Figure 9: Impact on VAT Sales by Firm Size Within Sector – Unbalanced Panel (a) Manufacturing (b) Retail (Essential & Non-Essential) Note: These graphs present the monthly relative change in adjusted sales between 2019 and 2020, by quartiles of average 2018-2019 sales, for a semi-balanced panel of active firms only, by sectors (see Figure 5 for the unbalanced panel figure). Firms that report zero sales from 2018 to 2019 are dropped. Panel (a) shows quartiles for the manufacturing sector and panel (b) for the retail sector. The black dashed vertical line represents the monthly change marks February 2020, the first full month before the lockdown was introduced on March 16. Table 3: Retail and Wholesale Trade: Businesses Considered Essential (G1) Reference to ISIC Rev. 4 Description 4530 Retail trade of motor vehicle parts and accessories 4772 Dispensing chemist in specialised stores 4730 Retail sale of automotive fuel in specialised stores 4661 Wholesale of solid, liquid and gaseous fuels and related products 4712 Retail sale in non-specialised stores 4711 Retail sale in non-specialised stores with food, beverages or tobacco 4520 Maintenance and repair of motor vehicles 4721 Retail sale of food, beverages and tobacco in specialised stores 4620 Wholesale of agricultural raw materials and live animals 4645 Wholesale of household goods 4781 Retail sale via stalls and markets 4653 Wholesale of agricultural machinery, equipment and supplies 4630 Wholesale of food, beverages and tobacco 4722 Retail sale of beverages in specialised stores 4791 Retail sale via mail order houses or via Internet 10