Some data issues

[vibhuti6]

Hi all, I am observing some data challenges as we merge firm information into our sample. I am posting it here for documentation purposes.

It looks like the classification of a firm as a large or a small business varies across contracts. This was not much of a problem with our analysis thus far because we were only concerned about categorization in a specific contract -- and that is more or less constant. I think we may ok going forward as well but we should be aware of these issues.

To illustrate, suppose there is a firm A holding contracts C1, C2, C3, and C4. It could be classified as a small business in C1 and C2, and as a large business in C3 and C4.

This variation can be due to a few reasons:

1) Contracting officer's discretion

• each contract may be handled by a different officer and their decision may be playing a role.
• Recall that our data variable is "contracting officers determination of business size"

2) Firm size within a given Naics code

• It looks like the government allows each firm to enlist in multiple naics codes (Source)
• The classification of firms as small and large businesses depends on firm size within a naics code. E.g., see the 2019 classification: https://www.sba.gov/sites/default/files/2019-08/SBA%20Table%20of%20Size%20Standards_Effective%20Aug%2019%2C%202019_Rev.pdf

3) The variation could result from the effect of Quickpay

• Suppose C1 and C2 were signed before Quickpay, and C3 and C4 were signed after the policy came into effect
• Barrot and Nanda paper shows that Quickpay increased employment
• Then, it could be that Firm A hired more employees and exceeded the threshold for being classified as a small business for contracts C3 and C4

Please let me know if you have any comments or suggestions about the above factors. Thanks!

[JNing0]

Thanks, Vibhuti. In my mind, we are not interested in "small business" per se. We are interested in contracts that are treated under QuickPay. As long as we can establish the link that if the contract is categorized as a small business, then it is treated under QuickPay initially, then I think we are OK.

We can control for business characteristics such as sales and workforce explicitly.

[vibhuti6]

Thanks, Jie. I agree -- I think this is just something to keep in mind in terms of how we write the paper.

[vob2]

Do we know the scale of this issue? Out of all firms, how many have contracts that classify firms as both small and not small?

I guess, it is OK. As both Vibhuti and Jie said, we care if the contract receives expedited payment. But if we arguing that small or large firms behave differently when receiving payments sooner, i.e., we need to think at the firm level, not contract level, this can become an issue.

[vob2]

By the way, do we know the ID of the officer who exercises discretion?

[vibhuti6]

Thanks, Vlad. Yes, we need to be careful in how we interpret the results. To answer your specific questions:

• Only 10% of the firms in our sample are assigned to both categories, but these firms hold around 60% of the contracts in the sample.
• We do not observe the ID of the contracting officer unfortunately.