Show Your Hand: The Impacts of Fair Pricing Requirements in Procurement Contracting

2.1 FEDERAL PROCUREMENT CONTRACTING

In 2019, procurement accounted for $586 billion of the U.S. federal government's $4.4 trillion budget (CBO [2020], GAO [2020]). Federal procurement is governed by the U.S. Code, a set of legal statutes devised by Congress, as well as various regulations. The Federal Acquisition Regulation (FAR) implements the U.S. Code and contains detailed policies and procedures for federal procurement of goods and services.5

Most federal procurement contracts are “negotiated” contracts. In awarding these contracts, the government considers not only price but also contract terms and the suppliers' capabilities and performance history. Negotiated contracts are the type of contracts affected by TINA's requirements.

Awarding a negotiated contract involves several steps. First, the CO makes a solicitation, which is typically posted online.6 The solicitation contains the government's requirements, including technical specifications, time frames, and the information suppliers must submit with the proposal.

Second, the government receives proposals from suppliers. The proposals describe their technical approach, price, fee structure, and any additional information required. In most cases, the government can require suppliers to submit any type or quantity of data to establish that the price is reasonable. However, COs are directed to request the minimum data necessary, in order to reduce preparation costs, award times, and government resources expended (FAR 15.402(a)(3)). The most comprehensive data COs can require are “cost or pricing data,” which FAR 2.101 defines as “all facts that, as of the date of price agreement[…] prudent buyers and sellers would reasonably expect to affect price negotiations significantly.” This information includes accounting data, vendor quotes, and any facts contributing to the supplier's judgment of future costs. Cost or pricing data can be a major advantage to the government during negotiations, as it can be used to establish reasonable prices and lower suppliers' profits.

Next, the CO analyzes the proposals to determine whether the prices are “fair and reasonable” as well as realistic (FAR 15.404-1).7 To do this, COs must consider several factors: their own analyses of historical bid prices and market research, whether they received multiple bids, the quality of the suppliers, any data provided by the bidders, and any applicable regulatory thresholds such as the TINA threshold (discussed in section 2.2). The government then selects the winning proposal. Finally, after the government and the supplier negotiate direct costs, indirect costs, and fees, the contract is signed (awarded).

COs manage contract contents and the procurement process; therefore, they play a key role in the competition for and performance of contracts. They are responsible for establishing that prices are fair and reasonable, lowering contract and procurement operating costs, and promoting competition, among other duties (FAR Part 15 and FAR 1.102). The FAR grants COs substantial leeway in determining preaward contract characteristics, such as specifications, contract type (e.g., fixed price, cost-plus-fixed-fee), and solicitation procedures (e.g., private negotiations, sealed bids). COs determine if only one supplier can reliably meet the specifications and requirements and thus whether to solicit a proposal only from that source (i.e., a “sole source” contract per FAR 6.302-1 and FAR 2.101). They also decide whether to seek additional suppliers to increase competition. COs review any data submitted by suppliers. After the contract is awarded, they monitor contract performance (FAR Part 16). Government managers and auditors oversee the COs, who can face punishment for not complying with regulations such as TINA.

By requiring adequate competition or additional data, policies such as TINA reduce the extent to which COs can rely solely on their own judgment in establishing reasonable prices. I discuss TINA's requirements and how it impacts supplier disclosures further in the following section.

2.2 TRUTH IN NEGOTIATIONS ACT (TINA)

In 1962, after several high-profile incidents of unfair competitive practices and price manipulation involving collusion between government officials and suppliers, Congress passed the TINA.9 Per TINA, suppliers bidding on a negotiated contract must disclose cost or pricing data whenever COs expect the contract price to surpass the TINA threshold, unless multiple bids are anticipated (other exceptions may apply, as discussed below). The threshold is currently set at $2,000,000. Further, for contracts with an agreed (i.e., realized) price above the TINA threshold, suppliers must certify that to the best of their knowledge, “the cost or pricing data provided were accurate, complete, and current” (FAR 15.403-3). COs and governmental auditors must ensure that suppliers comply with TINA.10 Figure 1 summarizes the application of the TINA threshold.

TINA excepts from its requirements contracts whose prices are more likely to be reasonable. In such cases, requiring the data could unnecessarily drain procurement and supplier resources. These exceptions include contracts where COs anticipate adequate price competition (i.e., receiving two or more bids), contracts with prices set by law or regulation, and acquisitions categorized as “commercial items” for which there is already an established, competitive market.11 Regardless of TINA's requirements, the FAR enables COs to request uncertified cost or pricing data when necessary to ensure reasonable prices (see section A.1.1 of the online appendix and FAR 15.403-3 for additional details on certification). In certain cases, COs may also require certified data for below-threshold contracts if they obtain approval from their management (FAR 15.403-4(a)(2)).

In essence, TINA stipulates that for contracts above the threshold the government should protect its interest by requiring greater verification that prices are reasonable. For such contracts, procurement should rely less on COs' judgment and more on cost or pricing data and competition (FAR 15.403-4). COs can be subject to agency problems, such as colluding with bidders on contract awards and shirking. They can also make errors of judgment, especially as technology and production processes change. These problems can cause suppliers to perceive the process as unfair and reduce their desire to bid. For higher dollar contracts, these costs and the cost of an unreasonable price can increase substantially. TINA's requirements aim to alleviate these issues.12

Despite their intended benefits, TINA's requirements also impose various costs on suppliers and the government, which depend on whether COs require the data or encourage multiple bids. I begin by discussing the costs of COs requiring the data. First, suppliers must gather and prepare the cost or pricing data (using table 15-2 of FAR 15.408). If the data are to be certified, they must be the latest available across the supplier's entire enterprise (FAR 15.403-4(b)(2)). This requirement can involve time-consuming data collection “sweeps” that add 30 days on average to the award process according to one government estimate (Lorell, Graser, and Cook [2005]). Sweeps can disproportionately affect larger companies with multiple offices, though implementing a specialized certification system can be costlier for smaller companies. Second, cost or pricing data can be highly proprietary, deterring some companies who would otherwise like to contract with the government (Sheffner [2019], He et al. [2024]). Third, suppliers fear high pecuniary and reputational penalties for providing “defective” (i.e., incomplete, inaccurate, or outdated) cost or pricing data.13 Investigations can also be costly. Government claims against suppliers can also lower their supplier ratings in the Contractor Performance Assessment Reporting System and diminish their future prospects of contracting with the government (Heese and Pérez-Cavazos [2019]). Finally, the government must verify the additional information, which adds further time to the award process.

COs can decide to comply with TINA by encouraging bidding. To promote bidding, the CO can relax specifications (so a broader supplier base may bid), seek additional bidders, or make it easier to qualify new suppliers, each of which requires CO attention. Further, COs must verify any additional bids' prices are realistic. COs may also have less discretion to choose their favored suppliers who may perform better and require less oversight. Finally, involving more bidders increases the risk of bid protests and legal challenges.

Between October 1, 2015, and June 30, 2018, the TINA threshold was set at $750,000 per the FAR. However, on July 1, 2018, several agencies including the DOD raised their thresholds to $2,000,000. These “early adopter” agencies updated their thresholds before the FAR mandated it. They did so to reduce suppliers' data management costs and to align with the 2018 National Defense Authorization Act, which modified the U.S. Code and indicated an eventual FAR update. This threshold increase was the largest by dollar value in the history of TINA (see section B for details).

Table 1 presents the dollars obligated by the 15 largest federal agencies in fiscal years 2016–20, until emergency COVID procurement rules were instituted by the White House on March 13, 2020 (FR Proclamation 9994). The bottom row shows that early adopters' contracts comprised most federal contracts, in both number (81.1%) and dollar (84.6%) terms. The analyses in this paper thus focus on early adopter agencies' contracts around the $750,000 and $2,000,000 TINA thresholds (see section 4 for further discussion and a list of these agencies).14 In section 3, I discuss the conceptual underpinnings of this study and predict how the threshold affects contract bidding and execution.

3.1 COST OR PRICING DATA PREDICTIONS

TINA requires COs to request cost or pricing data for above-threshold negotiated contracts unless an exception applies (see figure 1). This requirement could lead COs to require cost or pricing data more often. Alternatively, COs may ignore TINA and request the data only when they believe the additional information will be useful, similar to below the threshold. However, I predict that TINA's requirements will produce a significant increase in the overall rate at which cost or pricing data are required.15

3.2 COMPETITION PREDICTIONS

The net effect of TINA's requirements on contract competition is not clear, as it depends on COs' choices between requiring data and taking measures to receive multiple bids. COs' decisions are influenced by personal factors (e.g., CO effort, discretion to choose favored suppliers, career concerns), taxpayer value considerations (e.g., government processing costs, award times, bargaining power, price impacts, performance, perceived fairness), and anticipated effects on bidding. Depending on COs' decisions, contract competition is subject to two countervailing effects, which I call the deterrence effect and the participation enhancement effect.

The deterrence effect has a negative impact on competition because costly disclosures force suppliers to manage more detailed accounting data, assume greater liability for defective data, lose bargaining power, and potentially incur proprietary costs (Darrough [1993], Tomy [2019]). Information leaks, bid protests, legal challenges, hiring of federal employees, and Freedom of Information Act requests can increase the risk that competitors might learn suppliers' proprietary information (Sheffner [2019]). In contrast, suppliers might be more innovative and more willing to work with the government when they can disclose less information (He et al. [2024], Breuer, Leuz, and Vanhaverbeke [2019]). In these ways, TINA may discourage competition. In fact, the 2018 National Defense Authorization Act relaxed TINA's requirements for lower priced contracts to reduce suppliers' data management costs. In my discussions with practitioners, they tended to expect the deterrence effect to dominate.

On the other hand, the participation enhancement effect has a positive impact on competition whether COs decide to promote bidding or require the data. As the FAR stipulates, COs should avoid requesting cost or pricing data unless it is necessary to establish that the proposed price is reasonable. For contracts expected to surpass the threshold, the primary way for COs to bypass TINA's data requirement is by taking measures to expect multiple bids. For example, COs can avoid designating contracts as sole source (i.e., having only one supplier available). They can also seek out bidders, allow open competition, make it easier to qualify new suppliers, or loosen specifications.16 Of course, encouraging bidding has its own costs. It requires attention, increases the risk of bid disputes and legal costs, and requires COs to verify that the offered prices are realistic. However, promoting bidding is often still less costly than requiring cost or pricing data, which entails processing costs, potentially higher contract prices, and project delays as submission of certified cost or pricing data can add weeks or even months to award times.

Competition may also increase due to COs promoting bidding to avoid audit involvement. Audit organizations, such as the DOD's Defense Contracting Audit Authority, often have special requirements regarding data received for TINA-covered contracts. In such cases, COs may need to provide additional documentation. Similarly, incurred-cost audits, which are common for cost-plus contracts, require extra effort from COs when suppliers submit the data. To avoid audits, COs may prefer to encourage bidding.

Requiring additional data may identify cheaper vendors who can deliver adequate goods or services, making it tougher for COs to justify selecting their preferred suppliers.17 Further, in the case of a sole source contract, it may be difficult to argue that only a single supplier is capable of providing the good or service. Requirements to disclose cost data thus may lead COs to relax contract specifications and open competition to alternative suppliers.

Last, improving suppliers' perceptions of fairness may also increase competition. Imposing greater disclosure requirements on preferred suppliers, who may previously have faced limited scrutiny, could signal increased impartiality on part of the COs. In turn, this perception could encourage more bidders to compete.

In these ways, TINA's requirements could increase competition. I predict that the participation enhancement effect will dominate (i.e., TINA's requirements are likely to increase competition).

3.3 PERFORMANCE PREDICTIONS

In this section, I discuss two opposing channels by which TINA's requirements may impact contract performance, in terms of renegotiations and cost overruns. The first channel is the performance enhancement effect. TINA's requirements may lead COs to better monitor and manage above-threshold contracts, resulting in improved contract performance (Warren [2014], Decarolis et al. [2020]). Both obtaining data and encouraging bidding can enhance COs' familiarity with contracts, enabling them to draft contracts that have fewer errors and provide better protection for the government. In addition, to the extent that TINA increases competition, this could deter suppliers from delivering poor performance for fear of losing future contracts to competitors. Indeed, COs are instructed to consider suppliers' past performance when they make award decisions (FAR 12.206). Finally, COs often use cost-plus contracts to shift risk to the government. This approach is necessary when uncertainties prevent COs from adequately defining requirements or prevent suppliers from accurately estimating costs for fixed-price contracting. Improved CO familiarity with contracts can reduce uncertainty and help COs write fewer cost-plus contracts, leading to more fixed-price contracts, which are easier to monitor. If the performance enhancement effect dominates, I would expect TINA's requirements to improve performance.

The second effect on contract performance is the price pressure effect. Increased competition may lead to lower offer prices. However, heightened competition may also diminish CO discretion and increase lowball bidding, both of which can result in poorer performance. First, if COs have less discretion to choose preferred (higher quality) suppliers because sufficiently capable, alternative suppliers submit lower bids, then contract performance may suffer. Second, lowball bidding, also called “buying in” (FAR 3.501-2), is a strategy by which suppliers lower their prices to undercut competitors; then, after winning the contract, they attempt to compensate for diminished profits through contract renegotiations (Decarolis [2014]).18 To the extent that TINA's requirements heighten competition, reduced CO discretion and lowball bidding may lead to more contract modifications and cost overruns (Alesina and Tabellini [2008]). Therefore, if the price pressure effect dominates, TINA's requirements should have a negative effect on performance.

I predict that TINA's requirements are likely to improve performance (i.e., the performance enhancement effect will dominate). I also predict that the threshold will have a negative effect on cost-plus contracts as long as the performance enhancement effect plays a role (regardless of whether it dominates the price pressure effect), because the price pressure effect should not affect cost-plus pricing. Figure 2 summarizes the predictions in this section.

6.1 BUNCHING BELOW THE TINA THRESHOLD

An assumption underlying my RDD analyses is that suppliers and COs do not engage in threshold gaming to avoid treatment to a significant degree (Saez [2010], Duguay, Rauter, and Samuels [2023]). In this section, I assess whether suppliers and COs violate this assumption by strategically manipulating contract prices to undercut the TINA threshold (i.e., “bunching” below the threshold). Following Palguta and Pertold [2017], I carry out two sets of tests to examine changes in the distribution of contracts below the $750,000 and $2,000,000 cutoffs due to the 2018 shift in the TINA threshold: graphical analyses and difference-in-differences estimations of the excess mass below the threshold.33

The extent to which threshold gaming occurs should be limited for several important reasons. First, the requirement for cost or pricing data under TINA is based on the prices COs expected before bidding, which are challenging for suppliers to manipulate. Second, any data required from suppliers are typically specified in the solicitation (prior to bidding) and are therefore largely independent of realized bid prices. COs may manipulate contract sizes to evade TINA's requirements by splitting contracts into smaller contracts. However, this should not occur often as it is explicitly prohibited (FAR 13.003(c)(2)(ii)), and COs are monitored by their management and government auditors.34

To the extent that suppliers or COs manipulate contract prices to undercut the threshold, I would expect a higher concentration of contracts at or just below the threshold value when the threshold is in effect compared to when it is not in effect. At the same time, when the threshold is in effect, I would expect an accompanying dip in the concentration of contracts just above the threshold where the manipulated contracts' prices would have been, compared to when the threshold is not in effect.

Figures 3 and 4 examine the price distributions of contracts from the Full Baseline sample around the $750,000 and $2,000,000 thresholds, respectively. Panel (a) of each figure shows the distribution of contracts before the 2018 TINA threshold change, panel (b) shows the distribution after the change, and panel (c) shows the pp differences in the distribution between the two time periods (i.e., after minus before). Figures 3 and 4 use right-endpoint inclusive price-bin widths of $10,000 and $20,000, respectively.

At first glance, the masses of contracts just below the effective TINA thresholds in figures 3(a) and 4(b) may seem to indicate contract bunching. However, these masses may occur for behavioral reasons rather than manipulation to circumvent any rules. In contracting, suppliers frequently bid at or just below round numbers, even when no policy-based threshold is in effect. Consistent with a round number preference, there are large masses of contracts at and below both TINA thresholds, even when the thresholds are not in effect (figures 3(b) and 4(a)). Similarly, there are relatively large concentrations of contracts at the round numbers $1,900,000 and $2,100,000 in figure 4(b), even when the $2,000,000 threshold is not in effect. To gain insight into how often suppliers and COs engage in manipulation to circumvent the TINA threshold, I exploit the timing of the threshold by focusing on the difference in distributions between the periods when the threshold is and is not in effect (i.e., panel (c) of each figure). This approach assumes that the density distribution around each threshold value would look the same before the threshold moved as after the threshold moved.

In figure 3(c), the concentration of contracts at and just below the $750,000 cutoff is no less prominent after the threshold changes to $2,000,000. Though there is a minor decrease in contracts in the few bins just below this cutoff, this evidence is inconsistent with prevalent bunching to avoid the $750,000 threshold. In section A.2.1 of the online appendix, I estimate the upper bound of the missing mass in the three bins below this threshold per Palguta and Pertold [2017] and find approximately 65 fewer contracts than predicted (p-value = 0.026), equivalent to only 0.15% of the $750,000 threshold Baseline sample.

Figure 4(c) examines the $2,000,000 cutoff value. As this panel shows, following the 2018 threshold change, there is an increase in contracts priced within the bin that includes contracts just below and at the new threshold of $2,000,000. The two bins below that one experience small positive changes after the threshold moved. Contracts priced just above the $2,000,000 threshold become somewhat less frequent. By themselves, these findings could be consistent with manipulation to avoid exceeding the $2,000,000 threshold, albeit only to a very small extent. Thus, following a similar approach as earlier, I estimate the upper bound of the excess mass to be approximately 172 contracts (p-value $$<$$ 0.001), or 1.6% of the $2,000,000 threshold Baseline sample.35 I cannot completely rule out that these masses of contracts just below either TINA threshold are driven partly by strategic manipulation. Although the evidence in this section suggests that the degree of change in bunching following the threshold increase is small and economically insignificant, the motives for bunching may change for some unspecified reason.

As an additional robustness check, I examine whether there is bunching of contracts requiring cost or pricing data below the TINA threshold. Given that TINA's requirement for certified cost or pricing data is based on realized prices, suppliers and COs may have more incentive to manipulate contract prices to undercut the threshold for solicitations requiring the certified data. That way, suppliers can avoid any additional liability posed by having to certify that the data are complete, accurate, and timely. In section A.2 of the online appendix, I restrict my bunching analyses to contracts requiring cost or pricing data, which yields even weaker evidence of manipulation.

6.2 COST OR PRICING DATA AND COMPETITION RESULTS

In this section, I present the results of RDD estimation of the effect of the TINA rule on both the requirement of cost or pricing data (i.e., comprehensive accounting information) and contract competition. TINA mandates that COs either require suppliers to disclose cost or pricing data or take measures to receive multiple independent bids for contracts above the threshold. See section 3 for the predicted effects of TINA's requirements on how often the data are required and on contract competition.

6.2.1 Cost or Pricing Data Results

In this section, I assess the effects of TINA on how often COs require suppliers to submit cost or pricing data with their bids. My empirical approach includes graphical and RDD analyses around the TINA threshold while it is in effect.

Figure 5 demonstrates the effect of TINA's requirements on whether cost or pricing data are required by threshold period. The y-axis is CP Data, the share of contracts requiring cost or pricing data. The x-axis is the contract price. In panel (a), as the $750,000 cutoff is crossed from the left, a sharp increase occurs in the fraction of contracts providing the data during the $750,000 threshold period. That fraction remains higher for all prices above $750,000. As shown in panel (b), during the period after the threshold increases to $2,000,000, the discontinuity shifts to $2,000,000. These results indicate that TINA's requirements significantly impact how often cost or pricing data are required for bids. Supplemental graphical analyses in section A.4 of the online appendix show that, as expected, these findings are even more pronounced for definitive and purchase order contracts (such contracts tend to be subject to less competition and are therefore less likely to be exempt from TINA's data requirement). They also show that commercial contracts, which are exempt from TINA, do not exhibit a discontinuity in the requirement of data across either threshold (see section 7.3 for further discussion on definitive, purchase order, and commercial contracts).

Columns 1 and 4 of table 3 present RDD analyses exploring the effects of TINA's requirements on the rate at which the data are required around each active threshold. Column 1 uses the $750,000 threshold sample and column 4 uses the $2,000,000 threshold sample. The dependent variable is CP Data. The coefficient of interest is $$\alpha _2$$ (on $$Treat_{TINA}$$), which captures the effect of the TINA threshold on how often the data are required (see equation (1)). Columns 1 and 4 demonstrate that the threshold statistically and economically significantly increases the frequency with which cost or pricing data are required by 2.81 pp (p-value = 0.054) and 3.85 pp (p-value = 0.043) for contracts above the $750,000 and $2,000,000 thresholds, respectively, compared to below-threshold contracts. These estimated effects constitute 17.2% (= 2.812/16.32) and 21.9% (= 3.851/17.55) of the below-threshold means, respectively. Together, these results complement the visual evidence in figure 5 and provide evidence that TINA's requirements increase the rate at which cost or pricing data are required.

TABLE 3. Regression Discontinuity Design: Effect of the TINA Threshold on Cost or Pricing Data Being Required and Competition

	$750,000 Threshold			$2,000,000 Threshold
	(1)	(2)	(3)	(4)	(5)	(6)
	CP Data (%)	Sole Source (%)	Multiple Offers (%)	CP Data (%)	Sole Source (%)	Multiple Offers (%)
$${\bm{Treat}}_{\bm{TINA}}$$	2.812*	−3.893***	4.388***	3.851**	−3.927**	6.306**
	(1.460)	(1.092)	(1.136)	(1.899)	(1.992)	(2.625)
Distance	0.005**	0.006**	−0.007**	−0.001	0.003*	−0.006**
	(0.002)	(0.003)	(0.003)	(0.002)	(0.002)	(0.003)
$$Treat_{TINA}$$ * Distance	−0.002	−0.002	0.005	0.006**	−0.002	<0.001
	(0.004)	(0.004)	(0.005)	(0.003)	(0.003)	(0.004)
Controls	Yes	Yes	Yes	Yes	Yes	Yes
Product-service code fixed effects	Yes	Yes	Yes	Yes	Yes	Yes
Year-quarter fixed effects	Yes	Yes	Yes	Yes	Yes	Yes
Bandwidth (h, $1,000's)	500	500	500	1,000	1,000	1,000
Mean outcome (below threshold)	16.32	26.51	58.46	17.55	30.20	55.50
Std. dev. outcome	36.96	44.14	49.28	38.04	45.92	49.70
Observations	41,993	41,993	37,769	10,492	10,492	6,024

• This table reports results for analyses of the effect of the TINA threshold on cost or pricing data (i.e., comprehensive accounting data) and competition variables. All columns present results from estimating equation (1) in section 5 by ordinary least squares. The coefficient of interest, on $$Treat_{TINA}$$, represents the discontinuity (i.e., change in level) in the dependent variable at the threshold indicated. The dependent variables are defined as follows: CP Data equals 100 if cost or pricing data were required to be submitted with the contract proposal, and zero otherwise; Sole Source equals 100 if the contract was awarded as part of a sole source solicitation (i.e., where all but one bidder is excluded), and zero otherwise; Multiple Offers equals 100 if the contract received multiple bids and zero otherwise. The regressions in this table include the following controls: $$Distance$$ is the difference (in dollars) between the contract price and the threshold value shown for each column; Initial Duration is the expected duration of the contract at signing; a vector of dummies controlling for (1) set asides for small businesses and (2) all other set asides (e.g., women and minority-owned businesses). Contracts are from the threshold-specific Baseline sample indicated (i.e., negotiated, noncommercial, prime contracts awarded by early adopter agencies in the threshold period and within the bandwidth indicated) in each column. The $750,000 threshold period is between October 1, 2015, and June 30, 2018. Early adopters are defined as agencies that deployed the $2,000,000 TINA threshold in 2018. See section 4 for the list of early adopter agencies and when the $2,000,000 TINA threshold applied for each agency. Standard errors (in parentheses) are clustered by agency-contracting-office. *, **, *** denote statistical significance at the 10%, 5%, and 1% levels (two-tailed), respectively.

6.3 PERFORMANCE AND COST-PLUS PRICING RESULTS

In this section, I examine the effects of the TINA threshold on contract performance and cost-plus pricing. As described in section 3.3, I test whether the performance enhancement effect or the price pressure effect dominates at each value of the TINA threshold. Table 4 presents the results. This table features the coefficient $$\alpha _2$$ on $$Treat_{TINA}$$ from estimating equation (1) with performance and cost-plus variables as the dependent variables, using two types of samples. First are the threshold-specific samples, which are for the focal results, shown in the top row. Examining the $750,000 threshold, we see that the coefficient on $$Treat_{TINA}$$ in column 1 indicates a modest improvement of 3.0% in the rate of renegotiations for affected contracts. The statistically and economically significant negative coefficient on $$Treat_{TINA}$$ in column 2 indicates that affected contracts are 12.6% (= 2.98/23.63) less likely to experience a cost overrun. Column 3 shows a small but statistically insignificant decrease in the use of cost-plus pricing for affected contracts.

Next, I examine the $2,000,000 threshold. Columns 5 and 6 show that the threshold's effects on contract performance are statistically insignificant. As column 6, shows, the cost-plus type becomes 9.2% (= 2.612/28.39) less common among affected contracts (significant at the 10% level). This finding suggests that TINA's requirements may help COs to write fewer cost-plus contracts as the government's information environment improves. Figures A.9 and A.10 in the online appendix provide the corresponding RDD plots, which align with these results.

Taken together, this evidence suggests that the performance enhancement effect dominates the price pressure effect, as monitoring and the subsequent performance of contracts improve due to TINA's requirements. This effect may occur at least partly due to COs originating fewer cost-plus contracts. Figure 2 summarizes the observed effects on cost or pricing data, competition, cost-plus pricing, and performance.38

As mentioned in section 6.2.2, my competition results may have an alternative explanation. COs may promote bidding to avoid requiring their favored suppliers to provide data; however, they may directly or indirectly signal to those suppliers before bidding that they will select them regardless.39 Such coordination is unlikely due to institutional factors (e.g., illegality, monitoring by management, bid protests), but if it occurred, it could negate any potential benefits from increased bidding (e.g., reduced prices). Such collusion could also harm competitors and the government by wasting resources. Of course, one cannot observe the counterfactuals: We do not know the number of bidders and who the winners would have been if TINA's requirements were not in effect. Studying the effect of increased bidding on equilibrium prices presents its own estimation challenges as well. However, if such collusion occurs—if COs simply select the same suppliers and do not require any additional data as they would have if the TINA threshold were not in effect—the threshold should not affect contract performance or cost-plus pricing.

Using this reasoning, I investigate this alternative hypothesis by testing whether my performance and cost-plus type results hold for the subsample of contracts that do not require cost or pricing data. The coordination hypothesis predicts insignificant effects, whereas if COs meaningfully increase bidding, then my results should still hold. As shown in the second row of table 4, my previous results are unchanged. This finding suggests that the competition results are not primarily driven by COs increasing bidding “on paper.” Rather, COs meaningfully encouraging bidding appears to play a role in my results.40

7.1 PLACEBO TESTS

In this section, I perform placebo RDD tests to further validate my findings in section 6. For these tests, I examine the effect of the $750,000 and $2,000,000 cutoffs on each outcome variable from section 6 when these thresholds are not in effect. These tests use samples obtained by applying the same filters as the threshold-specific Baseline samples, but they use alternative (placebo) time periods.41 In contrast to my main results in tables 3 and 4 from the time periods during which the thresholds were in effect, in these alternative time periods I expect the coefficient $$\alpha _2$$ on $$Treat_{TINA}$$, which now captures the effect of the placebo threshold on the contracting outcome, should by and large be insignificant.

The placebo results are presented in tables A.7 and A.9 in the online appendix. Importantly, all estimated coefficients on $$Treat_{TINA}$$ for the Baseline samples (the first row of each table) are statistically and economically insignificant, as expected (the remaining rows in these tables relate to heterogeneity tests discussed in section 7.3). These findings provide additional evidence that my primary results are driven by TINA's requirements.

7.2 ALTERNATIVE RDD SPECIFICATIONS

In this section, I test the robustness of the main RDD results from section 6 to different specifications.

Table 5 presents RDD results using alternative specifications for the dependent variables CP Data, Multiple Offers, and Log(Number of Modifications). Columns 1 through 3 of table 5 contain the results of estimating equation (1) around the $750,000 threshold, adding a quadratic $$Distance$$ term and its interaction with $$Treat_{TINA}$$ to the model. My findings are unchanged. Columns 4 through 6 show the results of estimating equation (1) using a bandwidth of $300,000 instead of $500,000. My main findings are also robust to this alternative specification.42

7.3 HETEROGENEITY TESTS

In this section, I devise a set of heterogeneity tests to further explore whether the results in section 6 are driven by TINA's requirements. These tests analyze how the effects of the threshold vary across different samples of contracts that are likely to experience stronger or weaker effects due to TINA's requirements (see section A.5.1 of the online appendix for details and the full results).

First, I compare the effects of TINA's requirements on IDVs and non-IDVs (i.e., definitive and purchase order contracts). Relative to non-IDVs, IDVs entail more competition: They are associated with fewer sole source contracts and more bids (per table A.4 in the online appendix). Therefore, IDVs should more often be exempt from the requirements, relieving COs from having to take any additional actions to comply with TINA. Consistent with this prediction, I find that non-IDVs exhibit statistically stronger effects from the threshold than IDVs in terms of data required and performance.

Second, I compare the effects of the threshold on contracts for supplies (i.e., goods) versus contracts for services and works. Supplies are often more standardized, whereas services and works contracts tend to have more ex post cost uncertainty, multidimensional quality heterogeneity, and limited contractibility (Tadelis [2002]). Contracts for supplies also face less competition (see table A.5 in the online appendix) and thus are less likely to be exempt from TINA's requirements without action from COs. I expect that it is easier for suppliers to provide cost or pricing data and for COs to promote bidding for supplies contracts than for services and works contracts. Further, services contracts are less straightforward to monitor than contracts for supplies, as evidenced by more modifications for services (per table A.5 in the online appendix). Therefore, I also expect the performance of service contracts to benefit more from additional monitoring due to TINA's requirements. I find statistically significant evidence in favor of each of these predictions.

Third, contracts expected to receive multiple offers are exempt from TINA's requirements. Therefore, I examine contracts of PSCs that historically tend not to receive multiple offers. I predict that such contracts will exhibit a significantly stronger effect on competition due to the threshold than contracts of PSCs that tend to have multiple offers. My findings confirm this prediction. In addition, placebo tests using each subsample described above show that these results hold at the threshold values only while the TINA threshold is in effect.

I also compare the effects of the threshold on commercial contracts, which are exempted from TINA's requirements, to the effects on noncommercial contracts. As anticipated, I find that only noncommercial contracts exhibit a discontinuity at the threshold solely while it is in effect. Together, the results of these heterogeneity tests suggest that my primary results in section 6 are indeed driven by TINA's requirements.43

7.4 SUBSTITUTION BETWEEN COST OR PRICING DATA AND COMPETITION

The FAR outlines three ways for COs to establish price reasonableness below the TINA threshold: expecting multiple (independent) bids, relying on the CO's judgment, and requiring additional data. A substitution effect, or a negative relationship, might therefore occur between the frequency of requiring cost or pricing data and the frequency of contracts with multiple bids. Furthermore, above the threshold, the options reduce to only two: expecting multiple bidders or requiring the additional data. Therefore, any substitution effect between multiple bids and comprehensive data submission should be stronger above the threshold. To test these substitution predictions, I modify the right-hand side of equation (1) from section 5 to include an indicator variable for multiple bids and its interactions with $${Treat}_{TINA}$$ and $$Distance$$ (see section A.5.2 of the online appendix for details on the implementation and results). The dependent variable is CP Data.

The results of this estimation suggest that there is a substitution effect below the threshold: Below-threshold contracts with multiple offers have a 52.9% (=11.746/22.19) lower likelihood of requiring the data than below-threshold contracts without multiple offers (i.e., the base group). I also find that the TINA threshold significantly enhances the substitution effect. In addition, the results indicate that my finding from section 6.2.1—that the TINA threshold increases the rate cost or pricing data are required—is concentrated among contracts without multiple offers for both the $750,000 and $2,000,000 thresholds. This finding is consistent with expectation. Together, these results suggest that competition and accounting information can act as substitutes in reducing information asymmetry in supply contracting. They also provide evidence that COs can avoid requiring cost or pricing data by taking measures to receive multiple bids.

These results suggest that contract competition and the requirement of cost or pricing data (as well as other contract outcomes) are likely simultaneously determined. However, it becomes prohibitively difficult to obtain multiple exogenous instruments to jointly estimate all contractual terms. Thus, I estimate the cost or pricing data, competition, cost-plus, and performance regressions from section 6 as a system of equations using a seemingly unrelated regression model, allowing the error terms to be correlated in all regressions. My primary findings are robust to this estimation (untabulated).

7.5 DIFFERENCE-IN-DISCONTINUITIES DESIGN

In this section, I discuss the results of a difference-in-discontinuities (“Diff-in-Disc”) design I perform as an additional robustness check of my main findings from section 6 (see section A.5.3 of the online appendix for details on the specifications and results of these tests). This design exploits the 2018 change in the TINA threshold from $750,000 to $2,000,000 as an additional source of exogenous variation to analyze the impact of TINA's requirements on contract outcomes. A difference-in-discontinuities design has an important advantage over an RDD: If a persistent confounding factor similarly affects the variable of interest around the threshold both before and after a policy change, then under certain assumptions (see Grembi, Nannicini, and Troiano [2016]), that effect can be “differenced out” across the two periods.44 The results of these difference-in-discontinuities tests echo those of my RDD analyses in section 6, as anticipated.