2.1 The institutional structure

Sweden’s public sector comprises three tiers with elected assemblies at both the national level, in 21 regions and in 290 local communities. Each of the (1+21+290=) 312 tiers raises taxes to finance their activities.

The government is responsible for policy-related aspects of infrastructure provision for all railways and most roads outside municipalities. Agencies are used to administer and implement investment in and maintenance of both modes on behalf of the government. The Government’s Offices handle the day-to-day communication between the government and its agencies. From an international perspective, Sweden’s Government Offices is small relative to the size of the agencies (Molander et al., Reference Molander, Nilsson and Schick2002).

The national railway monopoly was horizontally separated in 1988, making the Swedish National Railway Administration (Banverket) responsible for infrastructure, while train operations were operated by the incorporated SJ AB (Nilsson, Reference Nilsson2002). The Swedish National Road Administration (Vägverket) held the corresponding responsibility for national roads since the 1940s. The two administrations were merged in 2010 and now operate as Trafikverket, the Swedish National Traffic Administration, subsequently referred to as the infrastructure manager (IM). Its two main tasks are to prepare a draft program for investment and maintenance activities and - after that the government has established the priorities in these programs - to implement them by way of competitive tendering .

The public sector’s regional and local tiers handle public transport and local roads, respectively. While the central government pays for national infrastructure, regions and communities have an obvious interest in having national roads and railways built on their turf. One reason is that municipal masterplans for physical planning may be based on the construction of a new road or railway included in the national investment program. Deviations from that program may have knock-on consequences for projects in regions and communities. Moreover, national roads and railways provide valuable services not only for long-distance travel and transport but also for residents, intermingling the roles of the three tiers of policymakers. National infrastructure planning is therefore a component of a complex formal and informal interplay between different institutional and political tiers of society. This provides one background to the extensive lobbying for specific projects: compare Jussila Hammes and Nilsson (Reference Jussila Hammes and Nilsson2016).

2.2 The default process for planning and implementation of new infrastructure

The practical work of the physical planning – the specification of where and how new infrastructure is to be built – is handled by the IM’s regional offices. The physical planning is meant to account for a series of important concerns in society.

Measures in the infrastructure are triggered by shortcomings of an existing road or railway. A feasibility study is the first step of the process toward building a new road or railway (t ₁ in Figure 1). The main purpose is to assess whether a draft proposal would be able to handle the problems on the existing infrastructure that it is supposed to remedy. This includes a cost assessment based on experiences from previous similar projects. The next step is to prepare a Road or Railway Inquiry (t ₂). The resulting document is to be of sufficient quality to make it possible to take a position on the various alternatives to handle the problems that are at hand. In combination with inputs from a consultation process, the IM decides which alternative to proceed with.Footnote ³

Figure 1. Timeline for project planning and implementation.

The third step in the process is the preparation of a Train Plan or a Road Plan for each single project. The format is regulated in law (1971:948 and 1995:1649, respectively; cf. t ₃). The purpose is to streamline planning of national and local infrastructure and to make sure that relevant aspects of the environmental legislation are handled. Those affected are given the opportunity to express their views. The output is a scheme providing further details of the project’s location and design, including a cost estimate that is to provide a valid basis for subsequent budgeting.

While the IM handles the physical planning of single projects, the government manages the financial planning. The first step in this is to establish the budget for infrastructure construction and maintenance for an upcoming 12-year program which is updated every fourth year (t ₄). The first 8 years of a new program comprises the projects in the plan which is nearing its end. It is therefore the final 4 years where there is scope for new projects. The IM drafts a program proposal which inter alia is supposed to be based on BCA’s. The cost estimate marked by (a) in Figure 1 provides the input for the economic analysis. After sending the draft on a national round of consultations, the government establishes the 12-year program, that is, the priority order of projects that are to be implemented within the set budget limit (t ₅).

Before a project can be started, formal parliamentary approval is required. This is done in the annual budget decision (t ₆). Practical implementation starts with the IM tendering a consultant to prepare a project plan (t ₇) which includes a Bill of Quantities. This document enumerates all construction activities that are to be implemented to have the road or railway built and in which quantities. Appendix describes the nature of this document which is part of the subsequent Quote for Bids for project implementation (t ₈). The builder willing to implement activities at lowest cost is allocated the contract (t ₉). The winning bid provides the first outside indication of the realism of the initial cost estimate. During implementation, the contractor is paid for work performed and the final cost is the sum of invoices paid by the client (t ₁₀).

3.1 Planning and implementation of six metropolitan projects

To handle both the need for additional funds and the overlapping responsibilities of the public sector’s three tiers, the government appointed a negotiator for the country’s three metropolitan areas in March 1990. The envoys were instructed to draft agreements with representatives for the respective regions and local communities. Each settlement was to include measures to improve the environmental situation and accessibility and to improve preconditions for regional development through measures in the overall traffic system. In addition, each negotiation was supposed to stimulate the planning and implementation of regional and local activities complementary to national responsibility.

After one year, the proposals from the three negotiators were published as SOU 1991:19. Later that year, the parliament approved a subsequent government bill by allocating SEK 10 billion for investments in traffic infrastructure over and above regular appropriations (1990/91: TU 24). This is the point of time when the actual decision to start these projects were made. This section presents the four projects emanating from the negotiations in Stockholm (Section 3.1.1) and one each in Malmö and in Gothenburg (Section 3.1.2).

3.2 Planning and implementation of the Bothnia Line

Since traffic started at the end of the 19th century, trains between the northern and southern parts of Sweden use a single-track line situated long away from the cities along the bay of Bothnia. The Bothnia Line is a greenfield single-track railway parallel to the existing north-south track but close to the sea. The city of Umeå is at the line’s northern end with the greenfield section proceeding 200 km along the coast. It links to the existing Ådalsbanan and to the city of Sundsvall, an additional 120 km to the south. The stated purpose of the Bothnia Line was to provide a railway benefitting both passenger and freight traffic along the line as well as the region at large.

The project was not included in the investment program established at this time but was seen to be a long-term option. Both the six projects described in Section 3.1 and most of the other projects shortlisted in the program were situated in densely populated parts of the country, while few projects in the sparsely populated northern half were included. The decision to proceed with the Bothnia Line therefore included a regional equity dimension.

The first official document referring to the Bothnia Line was published in 1992 (Bill 1992/93: 176). The cost for the entire Umeå–Sundsvall project was estimated to be SEK 7.9 billion. In 1994, the government commissioned a study, and the subsequent report estimated costs to be SEK 9.8 billion SOU(1996, p95)… Soon after, Bill 1997/98: 62 was sent for approval to the parliament.

Botniabanan AB, a limited company owned by the state and the municipalities along the line, was to handle the financing, design, procurement of construction, and the subsequent leasing of the line’s greenfield section. Restoration of the existing Ådalsbanan was to be financed by appropriations.

The section immediately south of Umeå gave rise to substantial delays relative to the original schedule. Reasons included both the formal administrative admissibility process, design issues, and court proceedings triggered by environmental concerns. In addition, the European Commission saw a risk that the project would not fulfill obligations according to the Species and Habitats Directive and the Birds Directive. The Swedish government was therefore requested to clarify the environmental protection status of the river delta close to the line’s northern endpoint. Much of the legal proceedings were in progress at the same time as construction of non-controversial sections had commenced. At a late stage of the construction process, it was decided to install the new ERTMS signaling system. This further delayed the start of traffic, which came to be in August 2012. The final cost for the Bothnia Line is estimated to be SEK 25.2 billion.

4.1 Challenges in defining costs

The article seeks to quantify and explain the difference between the projects’ final costs and the initial cost estimate. The final project cost at (c) in Figure 1 is straightforward to define but may be difficult to pinpoint in practice. The Bothnia Line provides an illustration. Neither the upgrading of Ådalsbanan nor the construction of a marshalling yard in Umeå was part of the formal definition of the project. These costs have been included in the estimate of final costs since it would not be feasible to operate traffic in the way it is now done without these extra activities.

The starting point for a project, (a) in Figure 1, is both conceptually and empirically challenging to pinpoint. Much of the literature in the field, including Flyvbjerg et al. (Reference Flyvbjerg, Ansar, Budzier, Buhl, Cantarelli, Garbuio and Glenting2018), use the point of time when a project is first included in an investment program as its starting point. They note that the real decision to build a project often has been made earlier.

The seven projects in the present study were, however, initiated outside the ordinary planning process. Specifically, the government asked the parliament to accept bills in 1991 and 1993 (cf. Sections 3.1 and 3.2, respectively). Five of the seven projects later came to be parts of investment programs, but not until long into the planning process.

Even though the parliament gave the go-ahead for the seven projects in the early 1990s, the analysis in Section 4.2 is cautious in the choice of starting year for the ex ante–ex post cost comparison. For the City Tunnel and the Bothnia Line, the government asked for extra cost information before formally approving them. It is therefore the cost at the time when the final design was approved that is used. In Table 1, the City Tunnel’s first estimate is therefore SEK 4.8 rather than SEK 3 billion and SEK 9.8 rather than SEK 7.9 billion for the Bothnia Line. Similarly, the Third Track was part of the 1991 agreement between the different public sector tiers in Stockholm. The first and formally approved surface solution was however abandoned, and the City Line came to be built. It is therefore the estimate for the latter design that is used as the project’s baseline cost. Similarly, it is the formally accepted cost for v. 2 of the Bypass that is used as a baseline, even though this version is very much like v. 1. If the very first cost estimates had been used, the cost increases estimated in the next section would have been larger.

The seven projects do not represent a random selection of infrastructure schemes. Jäderholm (Reference Jäderholm2020) is a consultancy report which was to focus on projects that include more tunneling than average. Moreover, the projects are larger than average-sized projects in Swedish infrastructure planning. This means that the numerical results in the next section may not be generalized to average infrastructure projects. Instead, the discussion in the article provides indications of the type of situation that may appear in infrastructure planning. As more and more projects in metropolitan regions are built underground to avoid situations with competing interests for access to land on the surface, it is particularly important to pay attention to these challenges for tunneling projects.

Several other large projects have, however, been implemented outside the ordinary planning process. One spectacular example is an 8.7 km-long double-track tunnel through the Hallandsås ridge. Planning commenced in 1975, but the line was not opened for traffic until 2015 with huge cost overruns.Footnote ⁵ In Gothenburg, another large tunneling project is under construction, seeing substantial cost increases before construction started. Moreover, the ongoing discussion about building a high-speed railway provides higher construction costs for every new update of the physical planning.

4.2 From nominal to real cost increases

Before establishing the size of cost overdraws, it is necessary to transfer cost information in Table 1 from current to fixed prices. Two different techniques are used. The first uses the consumer price index (CPI), which controls for the way in which prices in a basket of consumer goods and services change over time. This is referred to as the taxpayer perspective since it explains if, and by how much, tax revenue must be raisedFootnote ⁶ to pay for increasingly expensive construction. The second means for disentangling nominal from real cost increases is to use a sector-specific index.

Figure 2 shows CPI and sector-specific indices for railway and road construction. Prices for inputs used by the construction industry obviously increase much faster than consumer costs do. Specifically, costs for building new railways trebled over this 30-year period, road construction costs increased by a factor of 2.4, and consumer prices increased by a factor of 1.6.Footnote ⁷

Figure 2. Price index changes between 1990 and 2020. Sources: CPI from Statistics Sweden and investment indices from the IM’s home page.

These indices are applied to distinguish between nominal and real cost increases for the seven projects in Table 2. As an example, the cost for the City Line tunnel increased by 168 per cent in nominal terms. In real terms, this corresponds to costs increasing by 133 per cent from a taxpayer perspective, that is, using CPI, while costs increased by 63 per cent when using the railway-specific index.

Table 2. The seven projects.

Note: Nominal cost increases in Table 1 converted to real cost increases for each project using the consumer price index and specific road/rail index.

Abbreviation: CPI, consumer price index.

^a ((Final cost/estimated cost)−1) × 100.

Taken together, the table shows that the nominal costs on average are 169 per cent higher than the initial estimate for the six concluded projects. In real terms, and using CPI, costs increased by 122 per cent and by 48 per cent when using the sector-specific index. These differences follow directly from the indices illustrated in Figure 2.

Table 3 provides additional information about the cost escalation by introducing the tendered cost, that is, at (b) in Figure 1. There is a degree of uncertainty about the tendered cost, not least since most projects have been tendered in several separate contracts. The estimate is, however, sufficiently precise to lend itself to an obvious conclusion: While the average construction cost is 124 per cent higher than the first estimate for the six projects that have opened for traffic, only 12 per cent units are due to changes during the final stage of the process, that is, between t ₉ and t ₁₀ in Figure 1. It should be noted that both the total cost increase and the share thereof emanating from the construction phase will increase once the Bypass is opened for traffic.

Table 3. Nominal costs (a) at the start of the process, (b) when projects are tendered and (c) final.

Note: Invoiced costs. Real cost increase using consumer price index.

5.1 Engineering and conservatism

Procurement and contracting within the construction industry are constrained by rules and procedures established and upgraded over long periods of time by representatives of both buyers and sellers of the industry’s services. The Unit Price Contract format (UPC, Bajari et al., Reference Bajari, Houghton and Tadelis2014), by engineers referred to as Design-Bid-Build,Footnote ⁸ is frequently used. Rather than specifying the output of an assignment – a new road or railway with specified qualities – these contracts include a Bill of Quantities indicating precisely which activities are to be implemented and in which quantities. Once all activities have been completed, the road or railway is supposed to be ready for use.

A consultancy is tendered to undertake engineering, inter alia to prepare the Bill of Quantities, using its experts in the different types of activities that are necessary for preparing for the upcoming project. An example of a Bill is given in Appendix. A builder’s bid for this type of contract comprises one unit price for each quantity – which explains the UPC acronym – and the vector product of prices and quantities generates the total bid.

While this contract format has several advantages, one drawback is that it makes the implementation process rigid. The builder is not at liberty to improve the way the work is done compared to the specification in the Bill of Quantities without negotiations with the buyer. If a firm comes up with suggestions for truly innovative ways to do the job, the IM may have to reject that proposal or re-tender the project.

Restrictions on adapting the way work is done may be one reason for the sector’s large cost increases relative to consumer prices described in Figure 2. Contrary to most industrial production, it is, for instance, not feasible to substitute inputs which become more expensive. The same rigidity is introduced by the fact that most multiyear project contracts are indexed to changes in the price of inputs. This is a way to let the IM take the exogenous price risk, eliminating builder incentives to take measures during the construction phase to avoid cost increases. These features of the UPC contracts may also contribute to that the construction industry is a productivity laggard compared to other sectors of the economy (Salomonsson et al., Reference Salomonsson, Nilsson and Nyström2019).

For projects that include tunneling, the engineering tasks specified in the Bill of Quantities must be based on correct information about the ground at the construction site. Since many years, below-surface engineering studies use geotechnical techniques for this purpose, meaning that holes are drilled. Originating in the benefits of identifying the most likely sites to find oil when drilling wells, a set of techniques referred to as geophysical engineering has gradually been developed. This is complementary to geotechnical surveys. Geophysical data may, for instance, identify places where boreholes seem to be most desirable. The high local quality of the geotechnical data helps calibrate the low-resolution geophysical images and proves large-scale images of the ground. In combination with geological data, these results are valuable for evaluating the quality of the rock and to suggest how subsurface work should be implemented to reduce the risk of unexpected challenges both during and after construction.

Although geophysical methods are involved along with geology and geotechnics when investigating the underground in many countries (Reynolds, Reference Reynolds2011), Malehmir et al. (Reference Malehmir, Zhang, Dehghannejad, Lundberg, Döse, Friberg, Brodic, Place, Svensson and Möller2015) provides one of the few examples of its use in Sweden. The Bypass project illustrates the consequences of inadequate pre-studies. That project was stopped for several months during 2019 because of problems pertaining to rock quality. Better ex ante engineering of the ground may have reduced the risk for subsequent delays and cost increases. Even if the inclusion of geophysical techniques would increase the costs for contract preparation, it may be balanced by cost savings during implementation.

5.2 The (im-)possibility of following up on construction costs

Compared to the massive cost increases before tendering, the average 12 per cent increase in real costs during construction seems modest. Moreover, the reality of digging and blasting may by nature bring unpleasant surprises. This observation does, however, make it even more relevant to try to establish if some of these surprises can be avoided. For this purpose, follow-ups are inevitable.

However, today it is not feasible to follow up on the micro components of projects tendered by Sweden’s IM.Footnote ⁹ While there are records of all contracts that are signed, as well as meticulous handling of incoming invoices, there is at present no way to link prices and quantities ex ante (the contract) to ex post (the invoices) without manual detective work. Moreover, large projects typically comprise several separate contracts, making ex post reviews even more challenging.

Road and railway engineers frequently assert that all projects are unique, meaning that there is nothing to be learned from follow-ups. This may be a valid argument on an aggregate level, for instance if the seven projects in this article would be used for developing proposals for better implementation approaches. However, the claim is not correct on the micro level. The detailed nature of the Bill of Quantities in UPCs establishes the way in which an entrepreneur is supposed to implement projects. Subsequent invoices and payments are not made against estimated but the actual quantity of work, which must therefore be specified in each invoice. Moreover, the IM’s project leader monitors the invoices on a random basis, meaning that the buyer and seller of works typically agree about total costs.

Taken together, this means that the scope to follow up on performance is vast. Since this is currently not feasible, it is, for instance, not possible to establish the significance of tunneling costs for overall cost overdraws in the seven projects.

6.1 Cost increases generated by external forces

Not least during long periods of gestation, changes which are external to the parties in the process may have consequences for project costs. One example is that the focus on the environmental side effects of construction projects grew during the planning of the seven projects, subsequently expressed in a new environmental code (Bill 1997/98: 45). Moreover, rules and regulations for tunnel construction were tightened. Bill 2005/06:168 (“Safety in tunnels”) affected costs both because of new tunnel standards and due to stricter requirements on disturbances from traffic in tunnels.

Another change occurred when the Act that triggered a change in policy (cf. Section 3.1) was reversed. Bill 1995/96: 220 established that the benefits of new roads and railways do not generate public sector revenues. A new Act therefore sent the formal responsibility for financing infrastructure investment back to the parliament’s annual appropriations Bill. In this way, lawmakers are supposed to be given a better overview and greater influence over priorities than when complementary funding from regions and communities, ad hoc loans, or other financing mechanisms is used. One consequence was that responsibility for planning and implementation of some of the seven projects gradually came to be incorporated into ordinary long-term infrastructure planning. This may have had consequences for at what time allocations were made available and for the time it took to build new infrastructure.

What here is labeled external changes have some resemblance to what was referred to as evolutionist explanations in the initial review of literature. At least some of these consequences could have been foreseen. Public protests during the period, for instance, signaled that the public’s position toward trade-offs between “hard” and “soft” policy aspects was changing.

6.2 An external review of the Bothnia Line

The description of the six metropolitan projects is based on official documents and interviews. The Bothnia Line has also been subject to a formal ex post review; compare Riksrevisionen (2012). One topic for that examination was the project’s cost increase from SEK 8.2 (1997) to SEK 16.8 billion (2010) (Riksrevisionen, 2012, Table 5.1).Footnote ¹⁰ The audit concluded that the IM’s original estimate omitted necessary upgrades and alterations in adjacent networks, first and foremost on the existing southern part of the system, Ådalsbanan. Furthermore, the length of bridges and tunnels increased significantly between the original assessment and the line that was built. The audit concluded that initial cost estimates were deliberately kept low by the IM.

The government’s overriding purpose with the project was to improve the efficiency of traffic to and from the country’s northern parts, thereby increasing the competitiveness of the region’s business community. The review concluded that this purpose was not met. One reason was that the initial target for travel time – 5 h between Umeå and Stockholm – neither was nor will be achieved. The IM did not provide appropriate documentation of this aspect before the decision to build the Bothnia Line was made. This is said to be particularly remarkable considering that the IM shortly after the decision to build the line radically revised its travel time forecasts.

Several previous audits have pointed to the problem with cost increases. The most recent audit, Riksrevisionen (2021), testifies that the repeated critique has not had any discernible consequences. One possible reason is that neither the auditors nor anyone else have had the competence, or had access to relevant information, to see through the details of construction contracts. The Bill of Quantities and the subsequent invoices are fully controlled by the IM’s project manager and the firm doing the job. Importantly, the few people at the Government Office who are operationally liaisoning with the IM are not hired to have this type of competence. Ex ante cost estimates are a black box for outside parties.

6.3 The role of BCA for cost overruns

Flyvbjerg and Bester (Reference Flyvbjerg and Bester2021) show that the results of ex ante BCAs frequently are in error; costs are systematically underestimated, and benefits overestimated. With this as the empirical basis, they define “the cost-benefit fallacy” as a situation in which individuals behave as if benefit–cost estimates are largely accurate and unbiased, when in fact they are not.

This observation would provide a relevant background for the present study if BCA results were instrumental in the decision-making process. The only BCA that was formally part of any decision process was, however, included in the Bothnia Line inquiry (SOU, 1996, p. 95). Costs and benefits were then approximately equal; final costs came to be higher and benefits lower than anticipated.

A BCA published in 2006 established that the net present value of the Third Track was negative (Citybanan, 2006). In 2001, an audit reviewed the use of BCA in infrastructure planning. Nilsson (Reference Nilsson2000) produced a background report discussing the social costs for and benefits of the City Tunnel. The main conclusion was that BCA results did not feature in the stated motives for the tunnel and that the project, at the cost level at that time, was not socially profitable. These assessments were made after the decision to build was taken. It has not been possible to identify BCAs of the Southern or the Northern Link nor Götaleden.

A possible motive for policymakers to accommodate increasing costs during projects’ gestation periods is that benefits are thought to exceed construction costs. There is, indeed, one example of this interpretation that may be relevant. In approximately year 2000, 17 years after the first proposal, a BCA of the Bypass indicated that project benefits exceed costs. That first assessment is not in the public domain, but an updated BCA from 2012 estimated investment costs to be SEK 31.9 billion and the present value of benefits SEK 81 billion; cf. Förbifarten (2012). While this indicates that even a very costly Bypass is worthwhile to build, the assessment was prepared long after that the original decision was made.

The discussion about the relevance of BCA for decision-making is also related to the role of different parties involved in the process. The initial phase of the seven projects, which is the stage when projects are given the green light and in which the largest cost increases occur, is driven by representatives for the electorate. Texts and cost estimates are prepared by road and railway engineers officials at the IM’s regional offices while no economists have been part of these provisions.

In their critique of BCA, Flyvbjerg and Bester (Reference Flyvbjerg and Bester2021) show that cost underestimation is more common than benefit overestimation. This observation should be cast against the fundamental transparency differences between the costs and benefits in a BCA. Transport sector benefits are based on traffic forecasts, and there is an extensive literature on traffic forecasting; compare Eliasson et al. (Reference Eliasson, Börjesson, van Amelsfort, Brundell-Freij and Engelson2013) and Andersson et al. (Reference Andersson, Brundell-Freij and Eliasson2017) for the Swedish context and a recent survey published by NCHRP (2020). In contrast, Section 5 established that construction cost estimates are notoriously impenetrable, at least in Sweden. The systematic differences in benefit and cost bias may therefore derive from that it is not possible to scrutinize and question cost estimates while the analytical quality and transparency of benefit estimates are high.

Irrespective of BCA results, elected decision-makers are commissioned to prioritize projects according to their political beliefs. Nellthorp and Mackie (Reference Nellthorp and Mackie2000) provide an example where project priorities in Britain are not in conflict with BCA results. But starting with Nilsson (Reference Nilsson1991), several articles have shown that projects are typically prioritized irrespective of their social rate of return; Eliasson et al. (Reference Eliasson, Börjesson, Odeck and Welde2015) is a recent application to the Scandinavian context. The prime function of BCA today is therefore to function as an instrument for transparency, making it clear that some decisions are irresponsible. Cost overruns appear even if projects are not preceded by a BCA and if a BCA is prepared, the economist becomes the slave on the triumph chariot.

With that statement in mind, it is still obvious that the huge cost increases established in Section 4 have severe consequences for the results of infrastructure BCA at large. Economic analysis is supposed to provide a guide for project prioritization which happens at t ₄ in Figure 1. Cost estimates of new projects are based on an even earlier phase of the planning process when many design issues remain to be considered. The risk that BCA results overestimate project rate of return is therefore obvious. The differences of cost overdraws between projects that have been established also introduce an obvious risk that project priorities turn out to be biased.

6.4 Cost increases and the planning process

Flyvbjerg (Reference Flyvbjerg2021) asserts that political bias drives cost overruns. But how reasonable is it that decision-makers seek to trick the rest of society into believing that projects are worthwhile and that trusting early cost estimates is reasonable? And if the political bias hypothesis does not explain why costs increase for the present seven cases, what would rather rationalize projects being implemented despite costs increases? What makes governments of both center-right and center-left majorities carry on?

I suggest that the logic of the planning process provides an under-estimated reason for cost overdraws. The critical decision in this process is the prioritization of projects in the investment program, represented by the arrow at (a) in Figure 1. The IM’s regional officers know that the lower the initial estimate of construction cost, the higher is the chance that a project is included in the program. Furthermore, with low costs, more projects can be squeezed into the pre-set budget for the 12-year program.Footnote ¹¹ The government of the day will then be able to posture with their grand ambitions for the future when making the program public.

This logic is further supported by the poor understanding of cost drivers and pitfalls of completed projects discussed in Section 5. 2. In the absence of systematic information about what historically has been working well or not, the early-stage preparation of new projects is fully reliant on the IM’s engineer(s).

This logic would not provide a sufficient explanation if future program revisions would change the priority order in the face of increasing costs. There are, however, few if any examples of increasingly costly projects that have been excluded or substantially delayed relative to the program’s priorities. Being aware of this, neither politicians nor IM officials have reasons to provide prudent cost estimates at the early stages of the process.

The seven projects in this review deviate from projects included in the regularly updated infrastructure programs. The incentives generated by the planning process are however the same as for the extraordinary measures. Once the government has sent a bill to the parliament in the way described in Section 3, the chance – or risk – that subsequent cost increases may result in the original decision being canceled is slim.

From a political perspective, regions and communities regard a project in an investment program on “their” turf to be a gift from the national government, which foots the bill. The program is seen to be a promise and inevitable protests from the beneficiaries makes any backtracking by the government politically costly. Cost increases therefore become an integral part of the planning of new infrastructure, not primarily as a means for deception but because of the way in which the process is organized.

Social psychology offers a model with some resemblance to this assertion. The “foot in the door technique” is a compliance tactic that assumes that agreeing to a small request increases the likelihood of agreeing to subsequent requests. If an initial small request is accepted, the person who agrees to this finds it more difficult to refuse a bigger one (Freedman & Fraser, Reference Freedman and Fraser1966).

The process also involves a degree of dynamic inconsistency. This refers to a situation in which a decision-maker’s preferences change over time in such a way that one belief can become inconsistent with the opinion at a later point in time (Loewenstein & Prelec, Reference Loewenstein and Prelec1992). The projects’ extended gestation periods provide some substance to this explanation in that high priorities in today’s investment program may not materialize in costly(ier) spending for many years yet. It is not even certain that today’s political majority will survive and pay the cost once implementation starts.

Anecdotes about two of the projects cast additional light on these aspects. Allegedly, the Bothnia Line was approved as part of political horse trading: The ruling party gave the project a green light in return for an opposition party voting for the government on a completely different issue. Similarly, the City Tunnel was said to be a price paid by a senior central government figure for the support of the same party’s representatives in the southern region in a different context.

While it is not feasible to corroborate the validity of these stories, they provide an additional indication of that costly investment projects may be politically cheap. Moreover, the fact that the first part of the quid pro quo has been delivered – the political majority has had its issue approved by the opposition party – is more difficult to back on the “quo” because of increasing construction costs. After a long period of time with central-left political majority, the last 50 years has seen several shifting majorities, today with a politically weak minority government. This strengthens the reasons for any government to use “gifts” of this nature for pushing through important changes in other areas.