A few years ago
I was in Cape Verde trying to help set up an evaluation of a matching grant program (which
ultimately failed). I remember one of the biggest complaints of firms was that in order to apply for the program, they had to be current on their taxes, yet many of them sold goods and services to the government and said it often took the government several months to pay. Receiving a government contract can be an important boost to firm growth (as
this work by Ferraz, Finan and Szerman in Brazil shows), yet this growth potential can be hampered if governments are not fast and reliable in paying for these services.
How important is quick payment? A new study by Jean-Noel Barrot and Ramana Nanda shows that, even in the U.S., where we might think credit markets work pretty well, paying firms 15 days sooner has a striking impact on firm employment growth.
The Quickpay reform
The U.S. Federal government procures $100 billion in goods and services directly from small firms each year. The typical contract involves the government paying the firm 1-2 months after approval of an invoice, meaning firms have to finance upfront the working capital and payroll needs for satisfying this contract, and then only get paid after all work is done.
The Quickpay reform, announced in September 2011 by President Obama, indefinitely accelerated payments to a subset of small business contractors of the U.S. federal government, cutting the time taken between invoice receipt and payment by half, from 30 to 15 days. The fact that this was an indefinite (“permanent”), rather than temporary, policy change is important, since otherwise firms would anticipate having to reduce size again in the future and not respond as strongly.
Another key point to note is that the threshold for being considered a small business in the U.S. varies by industry, with the upper limit varying from $750,000 to $38.5 million in revenue, or between 100 and 1500 employees. The median treated firm has 6 employees and $600,000 annual sales. So, these are firms which are larger than almost all firms in developing countries, and yet even at this size we will see that speeding up payment matters.
Identifying the impact
The authors estimate the impact of the program on employment at the county-sector level. They regress the change in log payroll between 2011 and 2015 on treatment (the share of government contracts in eligible firms in the pre-reform period standardized by payroll) and controls for average government spending at the county-sector level, employment growth, earnings, industry and county fixed effects, etc.
Conditional on total government contracting, variation in treatment comes from two types of government contracting that aren’t affected by the reform: small business contracts that were already getting paid within 15 days (mostly Department of Defense contracts to “disadvantaged small business contractors” and ones that were cost-plus rather than fixed price contracts); and contracts to larger businesses.
They find a positive and significant impact of speeding up payment on county-sector firm growth: payroll increases by 0.15% on average. This impact looks small, but reflects that only a small number of firms have government contracts: the average county-sector had 136 establishments, of which 15 had government contracts, and 8 had government contracts that had accelerated payments as a result of the reform.
At the firm level the impact is impressive:
They then look for general equilibrium impacts by considering commuting-zones, and seeing the effect of treatment in other county-sectors in the same commuting zone as a firm. They find a significant negative impact here, suggesting that part of the growth of treated firms comes from crowding out employment decisions of firms in the neighborhood of treated firms. This is complimented by analysis of job flow data which shows workers leaving sectors with low intensity and moving to sectors with high intensity.
Reflections
The paper does a lot of robustness checks and uses several different sources of data to convince the reader of the results. One disadvantage of the data is that it does not allow the authors to link treatment to individual firms: the authors have a nice appendix graph (below) that shows how the speed of payment changed for Department of Defense contracts. I would love to see a similar graph for employment or payroll, and then standard difference-in-difference analysis which compares changes in employment for treated firms to the two untreated groups. However, it seems the payroll data cannot be linked to the contract data at the individual firm level, only at the county-sector level.
Of course if there are spillovers then standard difference-in-difference analysis could be biased, since the untreated firms may lose employees to the treated firms. But it still would be a good starting point, and something to try and get should this be tested elsewhere.
Implications for our work
The fact that 15 days delay in payments can make such a difference to U.S. firms suggests that the impact of delays of several months until payment for firms in developing countries are likely to be much more severe. This strikes me as a fruitful area for World Bank and government efforts to help firms by just being more efficient.
It also causes me to reflect on the contracts we write for survey firms and other types of contracts used in conducting impact evaluations. There is always a tension between getting enough cash to firms to allow them to start paying enumerators and making arrangements, and wanting to make sure quality work is being delivered before paying for it. Our standard solution is to break payments up into tranches, so that we are effectively lending firms funding at the start of the project, and then they are effectively lending at the end while the wait for the final payment. This seems like a good compromise, but this paper reiterates how important even relatively short delays in payment can be.
How important is quick payment? A new study by Jean-Noel Barrot and Ramana Nanda shows that, even in the U.S., where we might think credit markets work pretty well, paying firms 15 days sooner has a striking impact on firm employment growth.
The Quickpay reform
The U.S. Federal government procures $100 billion in goods and services directly from small firms each year. The typical contract involves the government paying the firm 1-2 months after approval of an invoice, meaning firms have to finance upfront the working capital and payroll needs for satisfying this contract, and then only get paid after all work is done.
The Quickpay reform, announced in September 2011 by President Obama, indefinitely accelerated payments to a subset of small business contractors of the U.S. federal government, cutting the time taken between invoice receipt and payment by half, from 30 to 15 days. The fact that this was an indefinite (“permanent”), rather than temporary, policy change is important, since otherwise firms would anticipate having to reduce size again in the future and not respond as strongly.
Another key point to note is that the threshold for being considered a small business in the U.S. varies by industry, with the upper limit varying from $750,000 to $38.5 million in revenue, or between 100 and 1500 employees. The median treated firm has 6 employees and $600,000 annual sales. So, these are firms which are larger than almost all firms in developing countries, and yet even at this size we will see that speeding up payment matters.
Identifying the impact
The authors estimate the impact of the program on employment at the county-sector level. They regress the change in log payroll between 2011 and 2015 on treatment (the share of government contracts in eligible firms in the pre-reform period standardized by payroll) and controls for average government spending at the county-sector level, employment growth, earnings, industry and county fixed effects, etc.
Conditional on total government contracting, variation in treatment comes from two types of government contracting that aren’t affected by the reform: small business contracts that were already getting paid within 15 days (mostly Department of Defense contracts to “disadvantaged small business contractors” and ones that were cost-plus rather than fixed price contracts); and contracts to larger businesses.
They find a positive and significant impact of speeding up payment on county-sector firm growth: payroll increases by 0.15% on average. This impact looks small, but reflects that only a small number of firms have government contracts: the average county-sector had 136 establishments, of which 15 had government contracts, and 8 had government contracts that had accelerated payments as a result of the reform.
At the firm level the impact is impressive:
- A dollar of sales to the government requires 30 days of working capital, meaning 30/365 = 8.2 cents are tied up in accounts receivable at any point in time. The reform frees up 4.1 cents of cash, that leads to 5.6 to 7 cents higher payroll – an elasticity of 1.35 to 1.7.
- The estimated impact for a firm that is treated by acceleration of contracts equal to 100 percent of its payroll is a 5.7% increase in employment over the next four years.
They then look for general equilibrium impacts by considering commuting-zones, and seeing the effect of treatment in other county-sectors in the same commuting zone as a firm. They find a significant negative impact here, suggesting that part of the growth of treated firms comes from crowding out employment decisions of firms in the neighborhood of treated firms. This is complimented by analysis of job flow data which shows workers leaving sectors with low intensity and moving to sectors with high intensity.
Reflections
The paper does a lot of robustness checks and uses several different sources of data to convince the reader of the results. One disadvantage of the data is that it does not allow the authors to link treatment to individual firms: the authors have a nice appendix graph (below) that shows how the speed of payment changed for Department of Defense contracts. I would love to see a similar graph for employment or payroll, and then standard difference-in-difference analysis which compares changes in employment for treated firms to the two untreated groups. However, it seems the payroll data cannot be linked to the contract data at the individual firm level, only at the county-sector level.
Of course if there are spillovers then standard difference-in-difference analysis could be biased, since the untreated firms may lose employees to the treated firms. But it still would be a good starting point, and something to try and get should this be tested elsewhere.
Implications for our work
The fact that 15 days delay in payments can make such a difference to U.S. firms suggests that the impact of delays of several months until payment for firms in developing countries are likely to be much more severe. This strikes me as a fruitful area for World Bank and government efforts to help firms by just being more efficient.
It also causes me to reflect on the contracts we write for survey firms and other types of contracts used in conducting impact evaluations. There is always a tension between getting enough cash to firms to allow them to start paying enumerators and making arrangements, and wanting to make sure quality work is being delivered before paying for it. Our standard solution is to break payments up into tranches, so that we are effectively lending firms funding at the start of the project, and then they are effectively lending at the end while the wait for the final payment. This seems like a good compromise, but this paper reiterates how important even relatively short delays in payment can be.
Join the Conversation