Mexico's Toll Road Collapse: What the 1989-1997 Concession Crisis Teaches Investors Today

The most ambitious road privatization program in the developing world

In 1989, Mexico launched what would become one of the most closely studied infrastructure privatization experiments in history. Under President Carlos Salinas de Gortari, the government set out to double the national toll road network by awarding concessions to private operators, primarily construction companies, to finance, build, and operate new highways in exchange for the right to collect tolls for a fixed number of years.

The ambition was extraordinary. Between 1989 and 1994, Mexico awarded 52 concessions covering more than 5,300 kilometers of toll roads, with a total investment of approximately 13 billion dollars financed through local commercial bank debt, concessionaire equity, and federal and state government contributions.

By 1997, 23 of those concessions had collapsed. The government was forced to take them over in a bailout that left local commercial banks with non-performing loans estimated at 4.5 to 5.5 billion dollars. Users were left paying some of the highest toll rates in the world for roads that, in many cases, carried a fraction of the traffic that had been projected.

It remains one of the most instructive infrastructure failures in Latin American history, and its lessons are directly applicable to any foreign company or investor evaluating concession opportunities in Mexico today.

The program design: where the problems were built in

The failures of the 1989-1994 program were not primarily the result of bad luck or unforeseeable events. They were the result of structural design choices that made the concessions financially fragile from the moment they were awarded. The peso crisis of December 1994 accelerated the collapse, but the foundations had been cracked years earlier.

Concession terms were too short

The standard concession period in the Mexican program was 12 years, with some as short as 8 years. These terms were far shorter than the 30 to 75 year periods typical of toll road concessions in other countries. The short duration meant that concessionaires had to recover their entire investment, including construction costs that frequently exceeded original estimates, within a very compressed window.

The only mechanism available to compress the recovery period was to raise tolls. The result was toll rates that were, in many corridors, three to five times higher than the rates on comparable public roads. High tolls depressed traffic. Depressed traffic reduced revenue. Reduced revenue made it impossible to service the debt. The structure was self-defeating.

Traffic and cost forecasts were systematically optimistic

The demand projections used to justify the concessions were, in hindsight, significantly overstated. Concessionaires and their banks accepted traffic forecasts that assumed users would pay premium tolls for time savings on routes where free alternatives existed. In many cases, drivers chose the free alternative, regardless of the time cost.

On the cost side, construction cost overruns were common and substantial. The concession contracts placed construction risk on the concessionaire, but the short terms and high leverage left no financial buffer to absorb those overruns. When construction costs exceeded projections, the concessionaire had two options: raise tolls further or default. Many chose default.

The academic analysis by Carpintero and Gomez-Ibanez of Harvard Kennedy School identified this as a central design defect: concessions were awarded to investors who lacked strong incentives to perform rigorous due diligence on demand and cost projections, because the concession structure transferred the upside to the concessionaire while the downside was effectively absorbed by the banking system and, ultimately, the government.

Financing was entirely in local currency with local banks

The 13 billion dollar program was financed almost exclusively through Mexican commercial banks lending in pesos. This created two compounding vulnerabilities. First, it concentrated credit risk in institutions that were themselves structurally fragile. Second, it exposed the entire program to Mexican macroeconomic conditions, with no international capital markets participation that might have imposed more rigorous financial discipline on the deal structures.

When the peso devalued by approximately 50 percent in December 1994, the real cost of debt service on peso-denominated loans increased sharply for concessionaires whose revenues, already below projections in traffic terms, were simultaneously worth less in real terms. The combination was fatal for the weakest concessions.

Concessions were awarded primarily to construction companies

A structural feature of the program that received less attention at the time was the profile of the concessionaires. Most of the 52 concessions were awarded to Mexican construction companies, not to infrastructure operators with experience in traffic risk management, toll collection, and long-term asset maintenance.

Construction companies had strong incentives to win concessions: doing so guaranteed them construction contracts on the roads they would then operate. But their expertise was in building, not in operating. Traffic risk management, demand forecasting, and long-term maintenance planning were not their core competencies. The program was designed in a way that selected for the wrong operator profile.

The peso crisis: accelerant, not cause

The December 1994 devaluation of the peso is often cited as the cause of the toll road program’s failure. This framing is misleading. The devaluation was the accelerant, not the cause.

By 1993, before the currency crisis, the government was already renegotiating concessions that had proven financially unviable. The fundamental problems, short terms, optimistic projections, excessive leverage, high tolls, and traffic shortfalls, were visible within three years of the program’s launch.

What the peso crisis did was collapse the timeline. Concessions that might have limped along for several more years of renegotiation were pushed into immediate default by the combination of increased real debt service costs, reduced real revenues, and a severe economic recession that further reduced traffic. The crisis that would have unfolded over a decade happened in two years.

For infrastructure investors, this distinction matters. The lesson is not “beware of currency crises,” though that is prudent advice. The lesson is that a concession program with structural design flaws will fail. The currency crisis simply determined when, not whether.

The 1997 bailout and its costs

By 1997 the government had no viable alternative to a direct takeover of the failing concessions. Twenty-three concessions were transferred to a government trust, the Fideicomiso de Apoyo al Rescate de Autopistas Concesionadas (FARAC), which assumed responsibility for their debts and operations.

The immediate financial cost was substantial. Local commercial banks were left with non-performing loans of 4.5 to 5.5 billion dollars. The government assumed liabilities that would take years to restructure. Users continued to pay elevated tolls, in some cases for roads that were now publicly operated again, with the toll revenues going to service the restructured debt rather than to fund new investment.

The political cost was equally significant. The program had been presented as a model of private sector efficiency replacing government infrastructure provision. Its collapse reinforced skepticism about infrastructure privatization that would take nearly a decade to overcome.

What the program got right

Academic analysis of the program, including the retrospective by Carpintero and Gomez-Ibanez, argues that the conventional verdict of total failure is too harsh. Many of the roads built under the program were socially worthwhile investments. Mexico needed the infrastructure, and the concession model, despite its flaws, delivered roads that would not otherwise have been built within the fiscal constraints of the period.

The 1993-1997 renegotiations, which extended concession terms and restructured debt in exchange for toll rate reductions, converted many of the failing concessions into financially sustainable projects. The 1997 bailout, which appeared catastrophic at the time, ended up costing the government less in budget terms than initially feared, as restructured toll revenues eventually covered much of the assumed debt.

The program’s legacy is therefore not simply one of failure. It is one of a program that achieved real infrastructure goals through a flawed financial structure, paid a substantial short-term price for those flaws, and eventually produced assets that continued to generate value for decades.

The second program: lessons applied

After a hiatus of nearly a decade, Mexico relaunched its toll road concession program in 2003 with a fundamentally different design. The new program incorporated the lessons of the first:

Concession terms were extended to 25 to 30 years, giving concessionaires adequate time to recover investment without resorting to prohibitive toll rates. Traffic risk was partially shared between the government and the concessionaire through minimum revenue guarantees on selected projects, reducing the incentive for optimistic demand projections. International operators and financial institutions participated alongside Mexican construction companies, bringing toll road management expertise and more rigorous financial discipline. Construction cost risk was more carefully allocated, with better project preparation before concessions were awarded.

The second program succeeded in attracting investment and expanding the network through the mid-2000s, until the global financial crisis of 2008 dampened appetite for infrastructure assets worldwide.

What this means for investors evaluating Mexico today

Mexico’s toll road history is not ancient history. The structural risks that brought down the first program are present in any infrastructure concession market, including Mexico’s current framework under the APP Law.

Traffic risk is the central variable

The first program failed primarily because traffic projections were wrong and the concession structure placed all traffic risk on the concessionaire without adequate compensation mechanisms. Any infrastructure concession evaluation must stress-test traffic projections against realistic downside scenarios, not just base cases. A concession that is viable at projected traffic levels but insolvent at 70 percent of projections is not a sound investment regardless of how attractive the base case looks.

Concession term determines toll rate, which determines traffic

The relationship between concession term, required toll rate, and actual traffic is not linear. Short terms require high tolls. High tolls depress traffic. Depressed traffic requires even higher tolls to meet revenue targets. The feedback loop can be vicious. Investors evaluating Mexican concessions should model the toll rate trajectory over the full concession term and assess its sustainability against realistic traffic elasticity assumptions.

The concessionaire’s profile matters as much as the financial structure

The first Mexican program selected for construction companies rather than infrastructure operators. The result was concessions managed by entities whose incentives and competencies were misaligned with the demands of long-term toll road operation. Foreign investors entering Mexican infrastructure through joint ventures or consortiums should assess not just the financial structure of the deal but the operational capabilities of all parties involved in execution and long-term management.

Government rescue is not a reliable backstop

The 1997 bailout was real, and it protected some investors from the worst outcomes. But it was not designed to protect investors, it was designed to protect the banking system and maintain the infrastructure. Investors who price Mexican concession risk with an implicit assumption of government rescue are mispricing the risk. The rescue in 1997 came at a cost to the concessionaires, including loss of the asset and in many cases loss of invested equity.

Renegotiation is a structural feature, not an exception

The history of Mexican infrastructure concessions is a history of renegotiation. The first program saw systematic renegotiations from 1993 onward. The second program has seen renegotiations in the energy, water, and transport sectors. This is not unique to Mexico, it is a global pattern in infrastructure concessions. Investors who structure their returns on the assumption that original concession terms will hold for 25 to 30 years without modification are making an assumption that the historical record does not support. Renegotiation risk should be explicitly priced, and concession contracts should include clear mechanisms for managing it.

How Construbufete can assist

Construbufete advises foreign companies and investors on infrastructure concession opportunities in Mexico, drawing on deep knowledge of the legal, regulatory, and political dimensions of the APP framework:

Legal due diligence on concession contracts: term structure, risk allocation, renegotiation provisions, termination mechanisms
Traffic and revenue risk assessment in the context of Mexican concession law
Consortium and joint venture structuring for concession participation
Regulatory mapping for transport, energy, and water concessions
Representation in concession renegotiation and dispute proceedings

If you are evaluating an infrastructure concession in Mexico, contact us before the financial model is finalized.