Political Factors
Why Politics Matters for Procurement
The efficiency comparison in Results shows that Type 1 firms (government-selected) are more cost-efficient for complex contracts, but less efficient for simple ones. This raises a natural follow-up question: what drives the government’s allocation decisions? If governments were purely optimizing for cost, we would expect discretionary mechanisms to be used only when selected firms have genuine cost advantages — i.e., mainly for complex contracts. Instead, we observe that the majority (~62%) of public auction contracts are for simple projects where Type 1 firms are actually more expensive.
The political economy literature offers an alternative explanation: governments use procurement discretion to reward politically connected firms, especially ahead of elections. This section tests for two specific political mechanisms: the electoral cycle and party alignment.
The Electoral Cycle
Mechanism Choice and Election Timing
In a first-price auction, firms compete on price to win contracts. In a discretionary mechanism (I3P, direct), the government selects which firms are allowed to compete (or simply awards directly). This gives incumbents the ability to direct resources to politically affiliated contractors. The incentive to do so is strongest in the period just before an election, when incumbents need resources for campaign activities and when payoffs to connected firms are most politically valuable.
The probit regression on mechanism choice includes an indicator for the year before a municipal election. Municipal elections in Mexico follow a fixed, staggered schedule, as municipalities do not all vote in the same year, providing variation in the electoral pressure variable across municipalities and over time.
Being in the year before a municipal election increases the probability that a contract is allocated through I3P or direct award (rather than public auction) by 6.2 percentage points, controlling for project size, complexity, state fixed effects, and year fixed effects.
This is a large effect relative to the unconditional probability of discretionary allocation. It is consistent with the hypothesis that governments direct contracts to connected firms ahead of elections, using the broad exceptions allowed under Mexico’s procurement law.
Bidding Behavior Near Elections
The electoral cycle affects not just the mechanism choice but also behavior within public auctions. In the SFA, the mean of the inefficiency term \(\mu(W)\) is modeled explicitly as a function of political variables — election-year proximity and municipal–state party alignment. The marginal effect of being in the year before a municipal election on the mean inefficiency term is:
- +13.8% increase in the average overuse of inputs, all else equal
What could explain higher excess costs in competitive auctions during election years? One interpretation: firms that are part of a political network bid less aggressively in public auctions, knowing that they may receive direct contracts regardless of their performance in competitive bidding. This reduces competitive pressure even within the public auction mechanism.
If connected firms expect to receive I3P invitations or direct awards in the future, their opportunity cost of not winning a public auction is lower. This weakens their incentive to submit aggressive bids, which is observationally consistent with higher excess costs near election time. The structural model cannot distinguish this channel from a simple behavioral response to political uncertainty, but both interpretations imply that public auction outcomes deteriorate near elections.
The Political Business Cycle in Procurement
This pattern connects to a broader literature on political business cycles in procurement. The finding here adds a new dimension: the distortion operates not only through the level of spending but also through the efficiency of procurement — the government buys the same roads but at higher cost near elections.
Party Alignment
The Alignment Mechanism
Beyond the electoral cycle, political alignment between levels of government may shape procurement choices. When the municipal and state governments are controlled by the same party, they may coordinate more effectively on infrastructure priorities, or alternatively, the shared political network facilitates the flow of contracts to affiliated firms.
PAN vs. PRI municipalities. The probit model includes an indicator for PAN-governed municipalities (relative to PRI). The large partisan difference in mechanism choice is one of the strongest findings in the paper.
This is a very large effect. One standard deviation in the party alignment variable corresponds to nearly twice the effect of a one-standard-deviation increase in project size.
Several interpretations are possible: (1) PAN municipalities face stronger internal party discipline or external accountability that discourages discretionary contracting; (2) PAN municipalities have less-developed networks of politically connected contractors; (3) PAN municipalities are more likely to be in urban areas where the contractor market is thicker and public auctions are more feasible; (4) PAN governments face greater scrutiny from civil society and media.
Party Alignment and Efficiency
Beyond the mechanism choice, party alignment also affects efficiency outcomes within public auctions:
- When the municipal and state government share a party, firm excess costs in public auctions fall by approximately 9.8%
- When party alignment coincides with an election year, the two effects partially offset each other: alignment reduces costs, but the election-year effect increases them
Regression Specifications
Mechanism Choice Probit
\[\Pr(D_i = 1) = \Phi\!\left(\gamma_0 + \gamma_1 \ln z_i + \gamma_2 \text{sewage}_i + \gamma_3 \text{election}_i + \gamma_4 \text{PAN}_i + \gamma_5 \text{aligned}_i + \boldsymbol{\delta}' \mathbf{x}_i + \alpha_s\right)\]
where \(D_i = 1\) for I3P or direct allocation; \(z_i\) is cubic meters of concrete; \(\text{election}_i\) is the year before municipal elections; \(\text{PAN}_i\) is the governing party indicator; \(\text{aligned}_i\) indicates municipal–state party alignment; \(\mathbf{x}_i\) includes contract-level controls; and \(\alpha_s\) are state fixed effects.
Efficiency SFA with Political Controls
The SFA model from Methods is augmented with political variables:
\[\ln \hat{c}_{ji} = \ln C^*(w_{ji}, z_{ji}) + \eta_{ji}(\text{election}_i, \text{aligned}_i) + \varepsilon_{ji}\]
where the inefficiency term \(\eta\) is modeled as a function of election-year and alignment indicators (following Wang, 2002, who allows the mean of the truncated-normal to depend on covariates):
\[\eta_{ji} \sim \mathcal{N}^+\!\left(\mu_0 + \mu_1 \text{election}_i + \mu_2 \text{aligned}_i,\; \sigma_u^2\right)\]
This specification directly tests whether the mean inefficiency level varies with political conditions, rather than estimating reduced-form correlations.
Full Results Summary
| Variable | Mechanism Choice (Probit) | Excess Cost in Public Auctions (SFA) |
|---|---|---|
| Concrete volume (1 SD) | −11.4 pp | — |
| Sewage work | −9.3 pp | — |
| Election year | +6.2 pp | +13.8% excess cost |
| PAN municipality | −18.9 pp | — |
| Municipal–state alignment | — | −9.8% excess cost |
Policy Implications
The political economy results reinforce the efficiency findings:
The majority of discretionary contracts are misallocated. Simple pavement projects, where Type 1 firms are less efficient, account for most discretionary allocations, and their frequency is amplified near elections.
Institutional reform should target the exceptions. The broad language in Mexico’s procurement law allows governments to bypass public auctions on vague grounds. A tighter legal standard, requiring genuine technical justification for I3P contracts, could reduce political discretion without preventing legitimate use.
Monitoring during election years is especially important. The election-year spike in both discretionary allocations and excess costs within public auctions suggests that standard monitoring frameworks that treat all periods equally will miss the highest-risk interval in the political calendar.
The results show that political factors affect not only which mechanism is used, but also how firms behave within competitive auctions. A reform that mandates public auctions without addressing the underlying political incentives would reduce discretion mechanically but might not fully restore competitive intensity.