TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

Transport-, Bygnings- og Boligudvalget 2017-18
TRU Alm.del Bilag 215
Offentligt

Traffic Forecasts for

the Fehmarn Belt Fixed Link

Review of Critical Weaknesses

Knud Erik Andersen

February 2018

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

Knud Erik Andersen graduated with an MSc. in

Engineering from the Technical University of Den-

mark. For 39 years he has been employed by the

Danish Road Directorate, holding various man-

agement positions: Director, Roads and Traffic

2002 – 2010, Director Traffic and Transport 1998 –

2002, Head of Planning 1994 – 1998. From 2011 –

2013 he was employed as Chief Advisor by Tetra-

Plan A/S and since then, he has acted as a private

consultant.

From 1987 to 2001 he was a member of the Exec-

utive Committee of the Transport Economic Socie-

ty in Denmark (TØF) and from 1995 to 2001, he

acted as chairman. Throughout much of his ca-

reer, Knud Erik Andersen has been working with

traffic modelling and traffic forecasting including

forecasting for the major fixed links: Storebælt,

Øresund and Fehmarn Belt.

Since April 2016 Knud Erik Andersen has been

blogging in the Danish business newspaper Børsen

on topics covering the transport sector.

Mail: [email protected]

Phone: +45 40922561

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

List of Content

Page:

1. Preface

2. Summary

3. Background

4. Overview of the 2014 forecasts

5. Selected weaknesses

5.1 Missing Market Segmentation

5.2 Traffic Split between parallel Transportation Services

5.3 Ignorance of Competitive Reactions from Ferry Lines

5.4 Transferred Traffic from the Great Belt Bridge

5.5 Lorry Transport

5.6 No Handling of Risk and Uncertainties

6. Quality Assurance of Traffic Forecasts – Disqualified Consultant

Selected References

Technical note: Although this paper is written in English figures are presented with decimal points

and commas following the rules in Denmark and Germany.

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

1. Preface

On 18

April 2015 the Danish Parliament passed the law of construction for a fixed Feh-

marn Belt Link. The political process leading to that decisive milestone was highly depend-

ent of various traffic forecasts; in particular, of a traffic model that was set up in the late

90’s and then “re-activated” in 2014.

This author has in two books discussed the methods and trustworthiness of two of the

most important traffic forecasts used for a key political decision. The first one was a fore-

cast presented in 2008 when a majority of political parties agreed on constructing the fixed

link. The second one was the forecasts made in late 2014 used for the final parliamentary

approval of the project.

If the political agreement from 2008 consisted of only a decision to make further and more

detailed analyses, this would have been a sensible step by step process of decision making.

However, this was not the case. The political agreement was in fact formulated as an irre-

versible final decision to establish the fixed link.

To understand this peculiar political situation, it is necessary to explain the nature of Danish

‘political agreements’. Such agreements have formally no legal implications since they are

not part of an official parliamentary process. The agreements are in fact of a private nature.

However, their implications on the subsequent formal processes in Parliament are decisive.

This is because the political parties have agreed - as a self-chosen code of conduct - that the

agreements must be treated as unbreakable. Despite the provision of subsequent decision-

making material, at the end of the day individual MPs from the agreement parties are in

practice forced to vote in favour of the fixed link.

The consequence of this code of conduct was that as early as in 2008, it was agreed that

each and every MP belonging to the parties behind the agreement was destined many

years later to vote yes to whatever law of construction would be presented to Parliament.

Thousands of pages prepared between 2008 and 2015 for decision making might influence -

and did influence - what kind of a fixed link should be built. However, the final agreement

to any fixed link eventually presented to Parliament was, in reality, given as early as in

2008, and not in 2015, when the law of construction was passed formally in Parliament.

Such an upside-down decision process would naturally not welcome any new evidence

during the formal legislation preparing process. Accordingly, this setup has a built-in risk

that new evidence might be produced with a clear bias, might be mispresented, or even not

published in order to omit any public discussion that might question the original and in

practice final decision.

The forecast from 2008 stood publicly undisputed until June 2014 when my first book, “The

Fehmarn Belt Fixed Link – made in Denmark”, ref. [17], was published. The book came as a

clear surprise and Femern A/S was evidently not prepared for such an ‘ambush’ (this phrase

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

was used by a prominent MP in the media). Furthermore, Femern A/S was not staffed to

react professionally to criticism concerning their own forecasts. The company had been

organised to construct the fixed link and there was a binding political agreement to build no

matter what, so why staff for further documentation of the rationale behind the reason to

build? This seemed to be the thinking at Femern A/S.

However, another bolt of lightning flared down from the clear sky in November 2014 – and

this time ignited by Femern A/S. The date and time was presumably not accidental. A few

hours before I went to the podium to speak to an interested audience, Femern A/S an-

nounced a brand new traffic forecast documented in a report of more than 600 pages. No-

body, outside the Ministry and its subsidiaries, knew anything about the task, since contra-

ry to EU-regulation on public procurement, the task had not been exposed to competition.

At first glance, this new situation looked much like what I had argued and hoped for: A

comprehensive forecast made this time not by Femern A/S itself but by independent Ger-

man consultants. I really doubted whether it would be worthwhile for me to spend more

time on the subject. However, having learned by experience, I started reading, and what I

saw after a few pages, regrettably, did not support my initial optimism. An almost twenty

year old traffic model had been scantily dusted off without taking into account what basi-

cally had changed in transport supply and demand during all these years.

In January 2015 I published my second book,

“Analyse af nye trafikprognoser for Femern-

forbindelsen, 2014”.

(Analysis of new traffic forecasts for the Fehmarn Belt Fixed Link,

2014), ref. [18]. In the book I pinpointed severe drawbacks and misperceptions concerning

the new traffic forecasts.

The Danish State is currently waiting for the final German environmental approval, which

presumably will be issued by mid-2018, and it is time for me to close my own ‘case’ in rela-

tion to inadequate documentation of traffic demand for the Fehmarn Belt Fixed Link.

It is my hope that the documentation on the following pages might be used to raise a dis-

cussion of how to document, decide upon and organise the establishment of publicly fi-

nanced or guaranteed mega infrastructures.

It is my clear impression that most of the MPs are fully aware that had they known in 2008

what they know today, the political agreement would not have been signed. The main les-

sons to be learned from the past, according to my judgement, are the following:

−

In case of deciding infrastructures above a certain level of economic obligation, the

politicians should restrict themselves to making only conditional agreements in the

first place and then wait on the final agreement until the final planning documents

have been prepared. Such a change would reduce the risk of having biased docu-

ments produced in order to avoid public debate on the appropriateness of the orig-

inal and in fact final political agreement.

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

−

A state-owned organisation, such as Femern A/S, with only one single project to

construct and to operate, should not be asked to produce decision-material such as

estimates of traffic demand, socioeconomic and business economics. An organisa-

tion with only one single project should stick to its technical tasks and be supple-

mented by independent agencies to produce non-technical documentation needed

for political decision. By such limitations there would be much less risk that a tech-

nical oriented organisation produces biased documentation to safeguard its own

future survival.

Decisions in Parliament should, of course, be taken according to the members’ own will

and visions. However, documentation prepared for decision making should be in line

with best professional standards. Any attempt to organise analyses and conclusions to

cover inconveniences by previous decisions ought to be completely unacceptable.

February 2018

Knud Erik Andersen

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

2. Summary

The traffic forecasts for the Fehmarn Belt Fixed Link published in 2014 acted as the central

precondition for passing the law of construction in Folketinget – the Danish Parliament.

However, the forecasts were flawed by misperceptions; misunderstandings and maybe

even deliberate misrepresentations. The consequence was substantially overestimated

traffic forecasts for the fixed link and Parliament was regrettably misled.

800.000 cars not accounted for

A massive contingent of 800.000 cars per year was assumed to be diverted from the Great

Belt (Storebælt Bridge) to the Fehmarn Belt tunnel. This figure was not a result of the traffic

model run but quite extraordinarily, a manually added traffic demand on top of the calcu-

lated output. The client simply instructed the consultants to add an extra 800.000 cars per

year through the tunnel. Three years after the introduction of this astonishing and massive

diversion of traffic, no documentation whatsoever is publicly available.

Conclusion:

Until documentation might otherwise be produced, the 800.000 di-

verted cars can only be interpreted as a backward calculation of extra traffic de-

mand needed for keeping the payback period of the investment below a politically

defined pain threshold of about 36 years.

Violation of the model’s limits of usability

Based on traffic model forecasts, the consultants have concluded the demand for ferry traf-

fic would be too limited to make a profitable business case when the tunnel opens. Accord-

ingly, in forecast scenarios, the consultants have deleted the ferry line from the model’s

transport network and all ‘existing’ ferry traffic will be transferred to the tunnel.

However, the traffic model was not designed to run calculations on parallel traffic services.

The split of traffic between the two different modes of transport is not based on any docu-

mentable estimation of the customers’ trade-offs between such two modes. The output

from the model used to argue for closing down the ferry line has no traceable explanation

(causality). The model should have been substantially upgraded with specialised traffic

surveys if the output were to be used as it has been used. However, the model was set up

back in the 90’s when a scenario of parallel services was out of the question, and according-

ly, it would have been a waste of money to design the model and the data collection for a

scenario which at the time was excluded.

If private capital were to finance the tunnel, comprehensive, expensive and sophisticated

traffic surveys would necessarily have been carried out in order to be able to forecast al-

most all foreseeable competing business strategies between the two modes. Nothing of the

sort has been done.

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

Conclusion:

The traffic model was in its 2014-specification not usable for calculat-

ing the split of traffic between a parallel ferry line and the tunnel. The most im-

portant premise for the economics of the fixed link, i.e. closing down the Rødby –

Puttgarden ferry line, relied simply on applying a model totally outside its area of

validity.

Disregarding competitive behaviour

A substantial part of the so called generated traffic is traffic diverted from an array of ferry

routes from all over Scandinavian waters such as Skagerrak, Kattegat, Fehmarn Belt and the

Baltic Sea. These ferry routes are treated in the model’s transportation network, more or

less, as unresponsive road links but with a wider range of service parameters, such as: sail-

ing time, fares, frequencies, etc. However, once the model was calibrated to be in line with

base year transport statistics for the ferry lines, the model worked exactly as if these ferry

lines were just as commercially non-responsive as a simple road link.

If a stretch of road ‘experiences’ a loss of customers, it will of course not be able to ‘intro-

duce’ countermeasures. A ferry line, however, will naturally react with a variety of coun-

termeasures to try to hold on to its customers and to keep its market share. Such funda-

mental commercial responses are not at all accounted for in the current forecasts. The

model is erroneously used as if the private ferry operators were obliged to stick to original

levels of service defined in an earlier competitive setting.

The traffic model is conceptually designed as if it were a state

strategic

model, in which the

state controls or can forecast all supply variables: time consumption, km’s driven, fares,

frequencies, etc. However, the privately-operated ferry lines which are supposed to deliver

customers to the Fehmarn Belt call for

strategic

and

competition

modelling, which is com-

pletely absent in the forecasts.

Conclusion:

Compared to typical applications of traffic models in other contexts,

the Fehmarn Belt case is extremely atypical. The generated traffic is to a large ex-

tent supposed to be diverted traffic, not from pure road transport corridors, but

from corridors dominated by privately operated ferry lines with strategic strate-

gies in contrast to road stretches. Each of these ferry lines will naturally react to

even a risk of loss of customers to Fehmarn Belt. Such commercial and competitive

reactions from a series of ferry lines have not been properly dealt with in any way,

leading to a substantial overestimation of diverted traffic to the tunnel.

Averaging into meaninglessness

The Fehmarn Belt traffic model was basically structured and estimated to reflect the

transport market as it looked back in 1996 when a huge data collection was carried out. At

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

that time the market for passenger car traffic crossing Fehmarn Belt was very simple: cars

driving between all over Eastern Denmark / Scandinavia to and from all over the Continent.

Accordingly, the model was set up to handle only one segment of passenger car traffic.

However, this market segment of traditional ‘Europe-traffic’ has been slightly decreasing

during the last twenty-five years – mainly because of deregulation of air traffic leading to

markedly cheaper air tickets. As a consequence, the ferry line Rødby – Puttgraden, even

when it was state-owned by Germany and Denmark, began to develop a demand for a

completely new low-price market for short time border shopping in Puttgarden.

This special segment currently comprises about one third of the total passenger car market

on the ferry line, which offers fare discounts of some 75 pct. or more compared to the fare

for ordinary ‘Europe-traffic’. Such two completely different market segments can, of

course, not be handled accurately in a model capable of handling only one passenger car

segment. However, such was the case in the 2014- forecasts. The model was only “re-

activated” and not re-estimated. Therefore, a Europe-traffic-market with a fare level of

index 100 and a low price and time restricted border-shop market with a fare level of index

25 pct. (or below) is averaged in the model to only one segment, simply because the model

back in the late 90’s was and still is not structured and estimated to forecast two complete-

ly different car traffic segments.

Conclusion:

Due

to an outdated model specification, the two completely different

traffic segments with current indexed fare level 100 and 25 have been merged into

one single “average” segment with an indexed price level of about 80. However, if

the fare for the border shoppers in real life rises from index 25 to index 80, these,

by nature, cost-sensitive customers will simply disappear or stay with a ferry line.

Nevertheless, they have all been “forced” down into the tunnel and assumed to

pay the average fare and with no influence at all on traffic volumes. On the other

hand, non-shoppers have been attracted to the tunnel with a discount of about 20

pct. and thereby being over-attracted to the tunnel compared to the current situa-

tion. In real life Femern A/S will be forced to abandon the flat fare strategy and

turn to a more differentiated one which should have and might have been evalu-

ated using a traffic model suited for the purpose. Currently estimated revenues will

naturally not hold in a real fare scenario – most likely they will be substantially re-

duced.

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

Lorry Transport

Unlike the situation with passenger cars, the forecasts for lorry traffic in the Fehmarn Belt

corridor seem reasonable compared to the observed traffic – maybe even a bit conserva-

tive, judged from a short-term perspective.

The main problem with the forecast for lorry traffic through the tunnel is that the traffic

model was neither estimated nor calibrated to forecast a scenario with a possible parallel

ferry service. At a very late stage, Femern A/S asked KPMG to make an analysis on the busi-

ness economics of a potential parallel ferry service. But the key problem with the forecasts

from Femern A/S and the report from KPMG is the same: they are not based on traceable

customer analyses of trade-offs in the choice of either transport above or beneath Fehmarn

Belt. Two of the leading hauliers have made a clear statement that they would make their

choice of passage mode crossing the Fehmarn Belt based on tight cost evaluations.

Conclusion:

The far most potentially dangerous threat to the business economics

of the fixed link has not in any way been substantiated by in depth analyses of the

hauliers’ own preferences in a realistic competitive setting. The market for trans-

porting lorries is extremely cost sensitive. This has not been reflected in the fore-

casts.

Closing remarks

The aforementioned problematic preconditions and additional worrisome matters shown

on the following pages have contributed to a major overestimation of the traffic forecasts

and thereby to an unrealistic economic outlook for the fixed link. It would require access to

the traffic model and most importantly additional specially designed traffic surveys if a well-

documented forecast estimate was to be made. This is regrettably not possible.

Based on past and long-time experience with traffic forecasting and modelling, this author

believes that Femern A/S should expect less than half of the generated traffic currently

foreseen and a slower general growth in person traffic due to the fact that long distance

trips have shifted from car to air and short-range traffic will not be able to fill the gap simp-

ly due to the sparsely populated nearby catchment areas.

Lorry traffic however will with little doubt continue to increase in the future. However, if a

low-price ferry line continues with parallel services, a substantial part of the road-based

goods transport will have an economically relevant choice.

Even under the current extremely optimistic presumptions, the fixed link is still a high-risk

investment. Such a situation would normally call for analysis of statistical uncertainty of the

total outcome of the investment, but such an analysis has not been demanded by Danish

authorities.

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

The decision to build the fixed link is political and cannot as such be discussed. However,

what can be discussed is that the documentation behind the forecasts does not meet even

the simplest requirements for professional standards. Normally an independently prepared

quality assurance would pinpoint such grave deficiencies and suggest corrections before

further activities were taken. However, much to the surprise of professionals in the

transport sector, the Ministry of Transport appointed a quality assurance consultant who at

the time by no means was independent but indisputably disqualified by being prequalified

to make bids for three out of four tunnel construction contracts and shortly thereafter in a

joint venture won all three at a sum of 3,4 billion €.

This choice of quality assurer was naturally neither a coincidence nor a regrettable mistake.

It was a deliberate ministerial action designed to obtain a predetermined external approval

of the forecasts.

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

3. Background

The Role of the Øresund Fixed Link

After decades of asynchronous Danish and Swedish interests in establishing a fixed link

across Øresund, the two states finally signed a treaty in 1991 concerning planning and con-

struction. The Danes were primarily interested in linking southern Sweden with eastern

Denmark in order to develop a Swedish-Danish regional economic potential. The Swedes

were mainly interested in improving access for Swedish industrial interests into Western

European markets. As a precondition for the Swedish signature, Denmark accepted pursu-

ing the possibilities of establishing a fixed link across the Fehmarn Belt.

Planning activities for the Fehmarn Belt fixed link were initiated in 1995 and lasted in its

first phase until 1999.This planning was to a large extent carried out in order to comply with

the Swedish conditions. However, later on the process was driven mainly by Danish political

interests in building the third and last leg in the league of mega strait crossings. After 1999

Denmark adopted the Swedish interest in a fixed Fehmarn Belt Link but with a much weak-

er formulated goal than the Swedish one.

Even several years before signing the Øresund treaty back in 1991, Swedish and Danish

industrialists had formed a lobby group called “Scandinavian Link” with former CEO of Vol-

vo, Pehr G. Gyllenhammar as one of

the driving front figures. These

early ideas, long before the fall of

the Berlin Wall in 1989, seem to

have been long-lived. The opening

up of Eastern Europe in the years

after 1989 seems not to have had

any implication at all as to what

kind of a project should be consid-

ered to link Scandinavia with the

continent. The plan was, and still is,

a fixed link between Rødby and

Puttgarden linking Scandinavia with

Western Europe as we knew it from the times during the cold war, before the fall of the

Berlin Wall (See illustration above)

The practical aspect of the Swedish Øresund condition was implemented by a mutual major

planning task between Germany and Denmark to investigate the feasibility of a fixed link

across the Fehmarn Belt. One of several technical / economic tasks was to set up a traffic

model that would be used to forecast traffic demand. As for the task of transport model-

ling, a steering committee was formed with German and Danish governmental representa-

tives. This author had the privilege of being a member of the Danish delegation to this

committee.

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

To implement the modelling tasks, a group of five international consulting companies were

hired in order to set up a forecast model and as part of this, to prepare and conduct a com-

prehensive data collection for estimating the transport model.

Back then the ferry line Rødby – Puttgarden was owned mutually by the Danish and the

German states. However, years later in 2007 shortly before signing the treaty of a fixed link

between Germany and Denmark, the two states decided to sell Scandlines A/S to private

capital investors.

Selected Previous Traffic Forecasts

The 1999 traffic model

As mentioned above, a comprehensive governmental cooperation between Germany and

Denmark took place in the late 90’s in order to analyse the economic feasibility of a fixed

link across the Fehmarn Belt. As for the traffic forecasts, results were published in the fol-

lowing reports: Ref.[1], “Fehmarnbelt Traffic Demand Study – Final report January 1999”,

and ref.[2],“Femer Bælt forbindelsen, forundersøgelser – resumérapport”, 1999.

The findings of the 1999 study will not be presented here. However, some important as-

pects of the model characteristics will nevertheless be highlighted here because the model

version used in 2014 is almost identical to the original 1999 version except for the calibrat-

ing / updating to a later year of reference.

The figure to the right shows the cordon lines (red fully drawn double lines) used to define

which crossing ferry lines were

subject to various forms of inter-

views describing passenger and

goods transport. Air traffic be-

tween major Scandinavian air-

ports and major airports on the

continent was subject to inter-

views as well, but the geograph-

ical details about these surveys

are not shown on the map. For

unknown reasons, the northern

cordon line in Skagerrak is not

extended southward into the Kat-

tegat. However, the ferry lines

crossing Kattegat between Jutland

and West-Sweden were subject to interview surveys as well.

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

In relation to the 2014 forecasts, an interesting missing survey object emerges clearly from

the map: no cordon line passes down the strait of Storebælt. At the time, the Storebælt

Fixed Link was under construction, which the consultants naturally were well aware of.

The judgement at the time was that the consultants did not expect any significant competi-

tion between Storebælt and Fehmarn Belt. This judgement was confirmed many years later

in a report from 2014, ref. [8] “Ex post samfundsøkonomisk vurdering af Storebæltsforbin-

delsen”, quotation (in Danish):

“Det er primært danskere, der bruger Storebæltsforbindelsen. Kun en mindre del af

brugerne er udlændinge af den simple grund, at det sjældent er den oplagte vej at

rejse, hvis man kommer fra Tyskland, Sverige eller Norge.”

Which translated says:

“It is primarily Danes that use the Storebælt fixed link. Only a minor part of the us-

ers are foreigners simply because it is rarely the obvious route for traffic coming

from Germany, Sweden or Norway.”

That statement goes also for traffic between major parts of Zealand and Germany since the

transport corridor from Sweden passes south of Greater Copenhagen.

In the 2014 forecasts, this judgement was totally abandoned and instead the Storebælt

bridge was considered as a major route for traffic between Zealand/ Scandinavia and Ger-

many with a big potential for diversion of car traffic to The Fehmarn Belt Fixed Link.

Data collection 1996

An enormous amount of survey data was made available:

Ferry lines

Passenger interviews

Stated Preference (SP) survey

Interview of lorry drivers

(SP) cargo handling agents

Grand total

Table 1

Traffic surveys, 1996

This is an impressive amount of data compared to current standards for designing traffic

surveys. The passenger surveys were primarily used for estimating flows of traffic between

geographical traffic zones in Scandinavia and the Continent. Stated preference interviews

however had quite a different purpose. They were used to extract detailed knowledge on

individual choices. Stated preference survey data are collected by designing a series of stat-

ed choice experiments where an interviewer presents different sets of mode characteristics

13.620

449

1.553

Data from

Øresund Fixed Link

3.294

4.123

Air lines

1.515

306

Total

18.429

805

5.676

392

25.302

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

to an interviewee. During the interview, these characteristics are varied by the interviewer

to learn under which circumstances the interviewee would change his preferred mode or

route choice. Stated preference interviews are expensive and take a very long time, but

they are essential for creating a trustworthy forecasting model reflecting complex real-life

trade-offs.

An important limitation of the use of such Stated Preference (SP) Analyses is that they can-

not be used for setting up a model that can handle almost any future scenario. This is be-

cause the interviewee cannot answer questions unless the choices presented are precisely

described, limited in numbers, and most importantly, relevant for the interviewee in an

actual choice context. This means that you can only make SP trade-off interviews for pre-

cise and on earlier specified future scenarios.

The key challenge of using traffic models for traffic demand forecasting is that the user of a

model can feed it with almost any future scenario and get an output forecast. The problem,

however, is that this output is useless if the model has been forced to run outside its limits

of validity. In such cases the causality does not exist, leaving the output more or less as a

random number but regrettably with no flashing red lights. Traffic models are rarely pro-

vided with a guide presenting predesigned and limited use. As a consequence, it takes a

professional with deep insight into the design of the model to know what scenarios lie with-

in and what scenarios lie outside the limits of the model. Neither the client nor the consult-

ants have apparently been aware of this crucial limitation in using the traffic model in 2014.

Back in the late 90’s it made of course no sense to make SP-interviews about the choice

between continued ferry services and a fixed link. In case of a fixed link, the two states

would, as owners, shut the ferry line down. Therefore, there was no need at the time for

sophisticated modelling to cover such an irrelevant future scenario. However, the ferry line

was sold in 2007 and the previously irrelevant scenario became suddenly extremely rele-

vant in relation to designing basic traffic surveys (SP).

The 2003 Forecast

An updated forecast report, ref.[3]: “Fehmarn Belt Forecast 2002, Final Report”, was pub-

lished in 2003. In the preface the following interesting information is given:

“The work has been done by the four FTC partners:

BVU – Beratergruppe für Verkehr und Umwelt GmbH, Freiburg (BVU)

Carl Bro a/s, Glostrup (CB) – leading partner

Institut für Seeverkehrswirtschaft und Logistik, Bremen (ISL) and

Intraplan Consult GmbH, München (ITP).

The fifth FTC partner, the former Hague Consulting Group (HCG), now merged into

RAND Europe, Leiden, has not been involved in this project as HCG contributed to

the forecast model construction and calibration but not in the forecasting work.”

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

The message is a clear statement that the original partner HCG (now RAND Europe) was the

partner that had structured, built and validated the core causal elements of the model and

that the technical qualifications of the rest of the partners were primarily concentrated on

general administration tasks (Carl Bro A/S), shipping know-how (ISL) and operational skills

connected with running the traffic model and first-hand experience with the German Na-

tional Traffic Model (BVU & ITP). It is noteworthy that precisely the latter two companies

BVU and ITP were picked by Femern A/S only to “re-activate” the model and not to re-

estimate the model which in fact was needed.

According to the 2003-report, there were several reasons for making an updated forecast.

The forecast year of the 1999-report was 2010. But the general forecast year in the German

Bundesverkehrswegeplanung, BVWP (German federal investment-plan for infrastructure)

was 2015, so changing the forecast year was one of the reasons. Another reason was that

during 2001 to 2002 an enquiry of commercial interest had been conducted with the aim to

check the interest of private capital to finance the investment in a fixed link and to take the

commercial risk. The enquiry (ref. [5] and [6]) gave a clear negative result since capital in-

vestors required an unconditional minimum advance of one third of the construction costs

before they would consider the investment at all. Furthermore, these capital investors were

worried by the prospect of:

−

Parallel ferry service close to the fixed link and

−

Competition from the Great Belt and

−

Competition from other ferries across the Baltic Sea.

The 2003 report had the forecast year changed accordingly to 2015 to be in line with the

BVWP. Ferry services across the Baltic Sea were updated and two sections about the

threats from a parallel ferry service and competition from the Great Belt Bridge were add-

ed.

The 2003 forecast report encompasses, quite surprisingly,

trend forecasts

for 2025. The

purposes of updating the 1999-forecasts are given in detail in chapters 2.1.1 “Study Objec-

tives” and in chapter 2.1.3 “Need for Updated Forecasts”. These specifications of the pur-

pose do not mention any demand for expanding the forecasts further to 2025. Had that

been the case, it would of course have been mentioned in either chapter 2.1.1 or in 2.1.3.

Nevertheless, chapter 8 deals with trend forecasts for 2025. The beginning of the chapter

states as follows:

“Two trend forecasts for the year 2025 have been carried out for each of

the base cases A and B”.

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

Figure 1

From ref. [3], figure 8.3.2 page 131.

The result is shown in figure 1 above. This 2025 trend-forecast was clearly not part of what

the client had ordered. It was, more or less, a speculation made by the consultants them-

selves. That was unimportant at the time. The problem arose in connection with the 2014-

forecasts where the same two consultants presented the

Trend Forecasts

from 2003 as if

they were the official

Forecasts

from 2003. This was simply not the case. In 2014 the client

was a complete newcomer and had had nothing to do with the forecast work done more

than a decade ago and had accordingly no insight into, and perhaps no interest in, the con-

tinuities and discontinuities between 2003 and 2014. So, the erroneous interpretation of

the 2003 forecast of being forecasts not only for 2015 but for 2025 as well was accepted

without protests, only to distort the benchmark in the 2014 forecast relative to the 2003

forecast. But the 2025 forecasts in the 2003-report was pure speculation.

In figure 1 the so-called “low” scenario represents the same yearly average growth in traffic

from 2015 – 2025 as in the years 2001 – 2015. So here the consultants have rephrased

“trend” to “low” and thus sending a value-laden message, that a trend is something that is

“low” of nature. However, a trend is a trend, and a trend forecast is the most information-

empty forecast that can be made. The “high” forecast is simply a case where the trend rep-

resents a yearly growth that “is at least twice as high as in the low forecasts”. Therefore,

the “high” forecast is simply introduced by the consultants as the consultants’ own sugges-

tion with no solid argumentation.

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

Parallel Ferry Line

As for the issue of competition from a parallel ferry line and from the Storebælt bridge, the

conclusive remarks in the 2003-report, ref. [3], are respectively:

Parallel ferry line

“… A decisive conclusion cannot be drawn out of the national experiences. A paral-

lel ferry service very close to the fixed link on the Great Belt and Øresund has

shown not to be able to survive, contrary to the situation on the Channel. As stat-

ed, there are great differences between these three situations, which make it ra-

ther difficult to transfer the experiences directly to the Fehmarn Belt.”

It is noteworthy that the model in 2002/2003 was not able to make any decisive conclu-

sions about this subject.

Competition from Storebælt

“… Furthermore, evaluations and model calculations have shown that the amount

of traffic that was transferred from the ferries Rødby - Puttgarden to the Great

Belt fixed link after opening in 1998 was approx. 2 %. Correspondingly, this

amount can be expected to be transferred back to a Fehmarnbelt fixed link after

opening.

The major part of the existing road traffic between Scandinavia (east of the Great

Belt) and Northwest Germany passing through Denmark uses the considerably

shorter route via Rødby-Puttgarden, because this route is much more cost-

effective.

Unless the toll rates on the two fixed links will differ substantially in favour of the

Great Belt, this will also be the case after establishment of a fixed link across the

Fehmarn Belt.

From these conclusions, it is clear that neither in 1999 nor in 2002/2003 did forecasts show

any evidence at all that Storebælt was considered to be an important competitor to Feh-

marn Belt. The volatility was limited to only 2 pct. of the Fehmarn Belt traffic. In the 2014-

forecasts however, this traffic was estimated orders of magnitude higher at 8 pct. However

it was worse than that because the 2 pct. was to be calculated as 2 pct. of a very small

number of cars passing across the Fehmarn Belt. Quite opposite the 8 pct. was to be calcu-

lated as 8 pct. of the much larger number of cars crossing The Great Belt. The percentage

figure had not just jumped from 2 to 8 pct. In real terms the volatility between the two

traffic corridors had risen from 31.000 to 713.000 cars/year. (Both figures represented in

traffic level 2011).

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

Not surprisingly, this figure of 713.000 cars/year (increased to 800.000 in forecast years)

now in its third year from publication is still subject to “further investigations” with no ex-

pected end time.

Lack of Documentation

The two consultants regrettably have the habit of not documenting the reference case or

the “do nothing scenario”. For a traffic planner this scenario might even be judged to be the

most important of all scenarios, because this scenario is needed to evaluate the conse-

quences of all other scenarios being evaluated. The absolute figures resulting from a sce-

nario forecast are in many aspects not of major interest. Oftentimes it is the difference

between key figures from the reference case and the same key figures from scenario case

that forms the result. The reference case is accordingly extremely important.

In the 2003 forecasts, the client observed this important lack of documentation and the

consultants were forced to complete their task and document the reference case in a sup-

plementary report ref. [4]

“Fehmarn Belt Forecast 2002 Reference Cases, Supplement to

Final Report of April 2003”.

Regrettably, this habit of not documenting the reference case was continued by the same

consultants in 2014 and the new client did presumably not notice this important lack of

documentation. The result was that spotting the flaws in the forecasting was made unnec-

essarily complicated.

The 2008 Forecast

The 2008 forecast was probably the most important one because it was used in the process

leading to the Danish political agreement to build the fixed link. As described earlier, politi-

cal agreements can in practice not be rolled back, so it might be argued that the initial

agreement from 2008 was in fact the final decision as well, leaving passing the law of con-

struction in 2015 as a simple follow up to the 2008 agreement.

The forecast was not a standalone publication but just a chapter of 2�½ pages out of a total

of 27 pages in the Financial Analysis from 2008, ref. [7]:

Chapter 4 begins with the following text (translated from Danish):

“The expected revenues for the coast to coast link has been calculated on the basis

of the traffic forecast prepared by FTC (Fehmarnbelt Traffic Consortium) and pub-

lished in the report: “Fehmarn Belt Forecast 2002, Final Report, April 2003” ”

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

(Annual average daily traffic, vehicles)

[Realised ferry traffic]

[Traffic demand presupposed in the financial analysis]

Figure 2.

Traffic forecast 2008. Illustration from ref. [7] page 12

The main problem with this forecast is that the reference to the 2003-forecast is, at best,

extremely weak. As mentioned above, the official forecast year in the 2003-publication was

2015. The trend extension unto 2025 shown in chapter 8 was not a forecast but an adden-

dum with a simple trend line combined with pure guesswork.

The client had not commissioned the expressions in chapter 8. This can be seen from the

fact that the chapters:

Preface, Background of the Study, Forecast preparation, Base case

forecasts, Scenario forecasts 2015

and

Forecast comparison,

do not contain one single word

of forecasting further on than 2015. A trend line – not described in the terms of reference -

cannot be viewed as being a formal forecast.

But the forecast reference in ref. [7] is even more peculiar. In the Financial Analysis, the

construction costs of the fixed link are calculated to be paid back by user charges after 26

years of operation. This implies that in 2008 there was some kind of an undocumented traf-

fic forecast at least until the year 2041.

To make a reference to a documented forecast from 2003 looks at first glance to be reas-

suring in relation to transparency of the basic presumptions. But to misuse chapter 8 in ref.

[7] and to make undocumented forecasts at least all the way until 2041 reveals a weakness

in traceability and in documentation of key figures.

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

4. Overview of the 2014-forecast

Car traffic forecasts

The forecasts are documented in three reports from November 2014. Two technical reports

were prepared by the German consultants BVU and ITP:

“Fehmarnbelt Forecast 2014 – Up-

date of the FTC-Study of 2002”,

ref. [9] and

“Fehmarnbelt Forecast 2014 – Update of the

FTC-Study of 2002, ANNEX”,

ref. [10]. The last of the three reports is a summary report from

the Danish Ministry of Transport

“Trafikprognose for en fast forbindelse over Femern Bælt”,

ref. [11]

Figure 3

2014-forecasts compared to 2002-forecasts for year 2015, supplemented with the consult-

ants’ choice of possible but undocumented extensions to 2025. Ref. [9]: Figure 9.2 p. 190

The above figure shows three different categories of traffic developments:

•

Observed traffic on the ferry line Rødby – Puttgarden (“real development 2001 –

2011”)

Traffic forecasts referring to the 2003-report (“FTC 2002” - three elements)

Forecast 2014 (“FTC 2014”, Case A: German assumptions referring to BVWP, Case

B: Danish assumptions)

As described in the preceding chapter, the reference back to the 2002-forecasts ( ‘low’ and

‘high’) is misleading since the 2002-forecast covered only one forecast year 2015 . The line-

ar extensions were not based on professional methods and not part of the client’s terms of

reference. Moreover, the terms “low” and “high” cannot meaningfully be interpreted as

low and high. The “low” is simply the trend line and there is no explanation as to why the

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

trend between 2001 and 2015 should represent something low – low compared to what?

The “high” is simply a non-contested suggestion by the consultants.

However, presenting the 2002-forecasts together with the 2014 forecasts in the misleading

way shown in figure 3 gives the reader the impression that the 2014-forecasts are totally in

line with the 2002-forecasts, i.e. something between ‘high’ and ‘low’ compared to 2002.

Ferry Traffic

As for the observed traffic on the ferry line shown in figure 3, it is obvious that demand has

not recovered after the financial crisis in 2008. This in turn has led to the need for a lower

starting point for the 2014-forecast than that of the 2008-forecast which was 1.89 m cars

and busses (level 2007). In the 2014-forecast the base year is 2011 when the observed traf-

fic had dropped from 1,89 m cars and busses in 2007 to 1,59 m. in 2011. Four years of ex-

pected traffic growth had been replaced by a drop of 16 pct.

Figure 3 shows that this worrying reduction of 16 pct. in realised ferry traffic seems to have

been counteracted by a surprising and suddenly identified much larger generated (diverted)

traffic than ever seen before. The undocumented influx of 800.000 cars per year claimed to

be diverted from Storebælt to Fehmarn Belt was politically needed in order to meet the

requirements not to exceed a politically defined pain threshold of a 36-year payback time.

The base reference in the 2014-forecast was determined by the consultants to be 2011. But

since then, transport statistics for the Rødby - Puttgarden ferry line do not detect any sign

of recovery with only 1,56 m vehicles in 2016 compared to 1,59 m in 2011 . Furthermore,

the three first quarters of ferry traffic statistics for 2017 reveal a reduction of 2,2 pct. It

seems that even 800.000 more or less imaginary diverted vehicles from Storebælt are still

not enough. Traffic demand has been decreasing since 2007 and even now when European

economies have substantially recovered after the financial crisis, traffic in this corridor

seems to be stagnating.

million vehi-

cles/ year

Passenger

cars & bus-

ses

2001

1,39

2002

1,62

2007

1,89

2011

1,59

2012

1,58

2013

1,57

2014

1,57

2015

1,57

2016

1,56

Table 2

Transport statistics for the ferry line Rødby – Puttgarden. Source: Danish Bureau of

Statistics (Statistikbanken)

There is no sign whatsoever that car traffic in the corridor of Rødby – Fehmarn will follow

the same general growth in traffic volumes that is seen on the motorway network in Den-

mark and in Germany as well. The reasons are mainly:

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

•

The Fehmarn Belt corridor has for decades been servicing long distance person

transport between Scandinavia and the Continent. Due to dramatic decreases in air

fares - as a consequence of air deregulation - this market has changed markedly

from car to air transport. The markedly general growth in car traffic on motorways

all over Europe cannot be used to argue for a similar growth in the Fehmarn corri-

dor. Passenger car trip purposes like commuting, vans with various service purpos-

es, etc., currently dominating general growth in car traffic, have never affected the

ferry line and the former key customer car traffic segment is now passing ten kilo-

metres above the ferry line.

Long distance car traffic experiences rising congestion problems especially on Ger-

man motorways.

The Fehmarn Belt region on both sides of the belt is characterised by a relative low-

density population with no major population and business centres that might fuel

short range traffic development.

•

The figure below is an update of the above figure 3: The ferry line statistics have been up-

dated to cover the period until 2016 and the “Case A” with German planning presumptions

has been removed leaving only the “Case B” with Danish planning presumptions.

Figure 4

Traffic forecasts 2014. This figure is based on figure 3 but updated with the latest

available ferry traffic statistics and the ‘case A’ (German planning) forecast has been re-

moved.

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

With reference to the first 9 months of 2017, the blue line might even with very little uncer-

tainty have been extended to cover 2017 as well and with a slight decrease relative to 2016

making the gap between forecast and observed traffic even worse.

Lorry Transport

As for lorry transport, the forecasted volumes are not in conflict with other sources of in-

formation.

The key question however is – as it is for passenger car traffic – will there be a parallel ferry

service taking its share of the total market? For further discussion see chapter 5.2.

Figure 6

Forecast

for lorries crossing Fehmarn Belt. Ref [9], Figure 9-3 page 193.

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

5. Selected Weaknesses

5.1 Missing Market Segmentation

Background Statistics

Before year 2000 the ferry line Rødby – Puttgarden transported passenger cars coming to

and from all over the European continent travelling predominantly for leisure, holidays and

business.

However, this demand for car traffic began to stagnate around the year 2000; as a coun-

termeasure, the ferry line began to develop a new low fare market with border shopping in

Germany.

Wage levels and VAT were and still are considerably lower in Germany than in Denmark,

resulting in lower retail prices, which attract Danish and Swedish customers to border shops

in Germany. To ensure that normal Europe-traffic did not take advantage of markedly lower

fares, Scandlines A/S introduced cheap round trip tickets for border shopping with restrict-

ed time limitations on the stay on German grounds.

At its website, Scandlines A/S has published table 3 showing traffic statistics for cars during

the years 1998 – 2016. These figures are shown in a diagram on the next page.

Table 3

Rødby-Puttgarden

car

traffic statistics. Reference: scandlines.dk

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

Figure 7

Development in yearly car ferry traffic in two catategories: shopping and Europe-

traffic. Reference: scandlines.dk.

If we look at the official transport statistics for passenger cars via Rødby – Puttgarden in a

longer perspective from 1990 – 2016 and subtract the shopping segment published by

Scandlines A/S, we get the following curve for ordinary Europe bound cars using the Rødby

– Puttgarden ferry line:

Figure 8

Rødby-Puttgarden, passenger cars per year excl. shopping.

Sources: Danish Bureau of Statistics (Statistikbanken) and scandlines.dk

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

As can be seen from figure 8, the ordinary Europe-bound car traffic crossing Fehmarn Belt

has declined by 15 pct. since 1990 and is currently only seven pct. higher than during the

low period in the mid 90’s when Germany was reuniting after the fall of the Berlin Wall.

During this same period, the number of passenger cars in Denmark has risen by 51,6 pct.

and GDP has likewise risen by 51,7 pct. – two indicators that one way or the other should

be a driving factor for road traffic in general.

It is astonishing that despite this exceptional track record of more than 25 years with de-

crease and stagnation in a supposedly key customer segment for the largest transport in-

vestment in Denmark ever, no need for further analyses has been addressed. Had private

capital been involved in financing this project, such information would have been most

alarming and called for thorough investigations including risk analyses long before any bind-

ing steps would have been taken.

Making Average where Average Distort Evaluations

The forecast calculations by Femern A/S are based on an average fare for a round trip of

968 DKK (130 €) for all passenger cars. However, since 2000 two much differentiated fare

levels have split the market of passenger cars in two: Shopping and Europe. By making av-

erages for demand and fares for these two very different markets, Femern A/S has simply

averaged explanatory power away leaving the old traffic model incapable of describing the

competition between an array of ferry lines and the fixed link.

The concept of the shopping

market is driven by a combina-

tion of very low fare tickets for

the ferry line and substantially

lower retail prices in Germany

compared to Denmark. This

low fare market covers about

35 pct. of the passenger car

market. The German destina-

tions for such round trips are

situated rather close to the

Puttgarden harbour because the fare discount is offered on condition of a time limited stay

on German soil.

This is completely opposite to the other car market that has substantial higher fares and of

course no time limit. Origins and destinations for these trips are spread out on an array of

European countries. The fare structure on the ferry line varies over time but under the as-

sumption of a discount fare of 299 DKK for a time restricted shopping ticket and a distribu-

tion of 35 pct. shopping and 65 pct. ordinary traffic, we get an estimate for the latter of

1328 DKK for a round trip ticket.

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

By using an average segment of just “cars” no matter whether we are talking about low fare

shopping tickets or full fare Europe traffic tickets, the forecasts from Femern A/S simply

violate fundamental economic mechanisms of supply and demand.

Several important conditions must be met to justify making averages of two very different

market segments. One of these is that by making an average, demand and fare competition

to other markets-- i.e. in this case ferry lines in the rest of Scandinavia-- is not distorted. But

that is exactly the case. The perception is that with an average fare much lower than the

real fare for Europe-traffic, the fixed link will attract much more traffic from alternative

routes compared to what will happen in real-world competition.

This of course leaves a deficit in revenues. However, by using average fares this deficit is

counteracted by “hand-moving” all shoppers down into the tunnel at the average fare level

of 968 DKK instead of 299 DKK or less and making no correction in the demand resulting

from a three time more expensive fare. Obviously, the reason people take the short-term

shopping trip is that discount benefits in border shops outweigh the fare for ferry transport.

With a round trip fare of DKK 968, the demand for such trips would in actuality disappear.

When confronted with this inconsistent way of traffic forecasting, Femern A/S claims that

at the opening date, Femern A/S will adopt its own price differentiated fare strategy with

the implicit understanding that such a strategy might turn out to encompass a substantial

discount to border shoppers. However, it should be clear that such alternative strategies

should be designed under the condition that the total revenue should be as close to, or

higher, than the revenue presented to Parliament and was used as a basis for passing the

law of construction. If half a million border shoppers – plus the forecasted growth in this

segment – do not pay more than about one third of what was presupposed when revenues

were calculated, then a considerable amount of money is simply missing in revenues.

No matter how this problem is considered, there is no practical way to rectify traffic vol-

umes and revenues from passenger car traffic with what was presented to Parliament be-

fore the passing of the law of construction. At least two thirds of the assumed revenues

from border shoppers only exist in spreadsheets; thus, the diverted traffic from other

transport corridors has been calculated from an unrealistic assumption of tunnel fare levels.

So why has Femern A/S made such a mistake? The answer is that the traffic model was

designed and estimated several years before the concept of low-fare border shopping was

initiated. The model was therefore not designed to handle two very different car traffic

markets. The model was only set up to handle one single car segment and the consultants

had no other choice than to short-circuit the markets, leaving the model unable to handle

neither the low fare nor the high fare market.

As in a supermarket, there can be no sensible strategy for selling bananas and pineapples

based on an average unit price calculated as the sales weighted average of previously dif-

ferent unit prices. For obvious reasons it doesn’t work in the supermarket and for exactly

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

the same reasons, it will not work in the Fehmarn Belt Tunnel.

The lack of professional forecast handling of two currently very different car segments us-

ing the ferry line is alarming in relation to estimates of future tunnel traffic demand and

revenues as well.

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

5.2 Traffic Split between Parallel Transportation Services

As clearly stated in ref. [9]

“Fehmarnbelt Forecast 2014 – Update of the FTC-Study of 2002”

the forecast work was just the result of an update of traffic statistics, but the core mecha-

nisms of the traffic model remained unchanged leaving the model in a state from which

answers to old questions could be found but not necessarily to new ones:

Figure 9

Quotation from Ref. [9] page 82 [with added underscores in red]

A Short Introduction to Modelling

The above important statement can be seen as kind of an indirect disclaimer. A transport

model is in no way an “oracle” that can calculate consequences for almost any relevant

future scenario. A transport model calculates consequences for clearly specified types of

scenarios that were designed for and data collected for. It would be technically impossible

to set up a transport model that can give answers to almost any thinkable, albeit, relevant

question. More importantly: the more designers attempt to create a model to cover wide

ranges of different types of scenarios, the more expensive the model will be, and the de-

signers risk losing track of things due technical complexity.

Before a transport model is set up, it is necessary to define which types of tasks the model

should be able to evaluate. Such tasks or group of tasks define rather precisely how the

model should be structured e.g. into sub-models, and how these sub-models should work

and interact with each other. This design phase encompasses also requirements for data

collection in order to estimate parameters in the model, i.e. to develop causal mathemati-

cal expressions that describe traffic volumes as a function of input data defining a scenario.

Thus, when a model has been set up, it has automatically built in limits for the use of the

model.

Regardless of these limitations, it is nevertheless operationally possible to run a transport

model completely outside its boundaries of validity. This can be the case if the scenario

differs markedly from the standard types it was designed for, or it might be because the

scenario is based on extreme levels of service that have not been reflected in the data sets

that the model was estimated on. Likewise, the model can be forced to work outside its

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

boundaries in case of scenarios with complicated trade-offs that were not reflected in the

data sets used for estimating the model.

Route choice in the Fehmarn model

The map to the right shows the ferry lines

from which the model originally was de-

signed to calculate diversion of traffic to

the fixed link.

From each traffic zone in Scandinavia (east

of Storebælt) to each traffic zone in the

rest of the continental Europe a total ‘gen-

eralized cost’ was calculated. This cost

function reflected driving time, driving

distance and a broad set of characteristics

for each separate ferry line such as: fares,

waiting time at terminals, sailing time,

sailing frequencies, etc. By using various

weighing factors (or unit prices) all these

level of service parameters were added

into on single ‘cost’ estimate for the given

route via a given ferry link.

After ensuring that total volumes of traffic had been adjusted to the observed level in the

base year, the model was run with these cost functions in order to allow the model to cal-

culate the distribution of car traffic between the different ferry lines. This calculation would

not totally reflect the actual volumes on each ferry line. The process will be finalised by a

‘calibration’ in order to balance the model calculated ferry traffic volumes with observed

ferry traffic statistics in the base year.

The Parallel Problem

Seen from a “model perspective”, the change from ferry services to a fixed link is rather like

introducing a new high-speed ferry line with no waiting time at terminals, much faster ‘sail-

ing time’ and the introduction of an extra cost equivalent to 18 km’s drive.

In the main forecast scenario, Femern A/S claims that there is no business case in a parallel

ferry service and the existing ferry line will have to close down and all ‘existing traffic’, i.e.

all forecasted ferry traffic just before opening of the fixed link would be transferred to the

fixed link. As previously described, such reasoning violates basic economic principles of

supply and demand: One third of all existing car traffic – the border shoppers – will by this

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

decision be ‘forced’ to pay a fare of about three times more than in the case of ferry

transport but with no implication on demand.

The argumentation for closing down the existing ferry line was based on a model run in

which parallel services are available by the ferry line and the fixed link as well. In this mod-

el, according to the consultants, the resulting demand for ferry transport was too limited to

keep the ferry line running profitably.

The key problem in that argumentation is that the model was never designed to forecast

traffic in such a highly competitive situation. In the phase of setting up the model back in

the late 90’s, the types of scenarios the model should be able to evaluate were sharply de-

fined, and based on that, traffic surveys were designed to estimate the model. Parallel ser-

vices by tunnel and ferry were simply not relevant at that time because the two states

would naturally close down their mutually owned ferry line if a fixed link were established.

Accordingly, in the 90’s no data was collected showing trade-offs between two geograph-

ically identical choices which differed markedly from each other in respect to level of ser-

vice expressed in a variety of level of service variables. Should a traffic model be able to

handle such a competitive situation, it would have been necessary, for statistical reasons,

to have access to data showing travellers’ real life trade-offs between different types of

supply for competing modes. Such data has not been collected and used for model estima-

tion; therefore, the model does simply not contain causal explanations to handle this very

special and specific scenario.

Nevertheless, there are no process-related barriers against letting the model run such a

scenario of two parallel services. However, the output from the model cannot be traced

back to the result of statistically analysed real life choice behaviour. Such output is accord-

ingly more or less arbitrary.

What should have been done if the goal were to evaluate parallel services was to perform a

series of stated choice experiments where an interviewer presented interviewees with dif-

ferent sets of varied characteristics for each of the two possible choices and given these

presumptions, asked the interviewee to give his/her preference for one or the other. Such

series of interview data are used to estimate a probabilistic model that under given choice

characteristics can make a statistical traceable distribution of traffic between the two alter-

natives. Such interviews are expensive and take a very long time, but they are necessary if

central estimates are to be produced.

However, Femern A/S and its consultants have ruled out that a competing ferry will have

any chance of surviving economically after the opening. This argumentation relies on a

clearly outdated traffic model that was not upgraded to handle this complicated competi-

tive situation, which was unforeseen when the model was set up and estimated back in the

late 90’s. The rejection of the scenario was not even supported by analysing the ferry line

business case.

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

If private capital should have financed the fixed link, this issue of a competing parallel ferry

line would have attracted huge attention regarding traffic forecasts and business econom-

ics. A competing ferry line is one of the foremost threats against traffic demand for and

business economics of the fixed link. Several million DKK would have been spent to come as

close as possible to a best estimate of the traffic split between the two alternatives and

evaluate the respective risks. Furthermore, as revealed in the next chapter, a private inves-

tor would have evaluated all sorts of possible competitive strategies between the two ser-

vice providers.

Nothing of the kind has been done even though back in 2002 strong warnings on the matter

were conveyed by potential private investors, see ref. [5] and [6].

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

5.3 Ignorance of Competitive Reactions from Ferry Lines

Every forecasted future customer of the Fehmarn Belt tunnel - except those who might be

diverted from the Storebælt Bridge, if any - has been attracted away from one of more than

twenty privately operated ferry companies. These ferry companies will naturally – if they

experience any loss of traffic to the tunnel - act with all kinds of commercial countermeas-

ures in order to minimize, or even to avoid, any loss of customers to the fixed link. Regard-

less of this fact, the forecasts have been made as if these ferry lines were frozen in an earli-

er setting of competitive balances between ferry lines and after the opening of the tunnel,

would be completely paralysed.

Back in 1999 and likewise in 2002, neglecting such an obvious competitive behaviour when

making traffic forecasts was a serious shortcoming. But to repeat such a blunder in 2014 is

not acceptable.

Public vs. Private Infrastructure

In mainstream transport modelling, the

transport network is usually provided by a na-

tional or local government and use of the infra-

structure is normally free of charge. In such a

planning environment, a transport model can

be used to calculate the effect of changes in e.g.

infrastructure such as a new road link or poten-

tial changes in the future level of service for an

existing road link. As a result, traffic on various

road links in the influenced area will increase,

decrease or stay constant. An example of out-

put from a traffic model in such a planning situation is shown to the right with green signa-

ture showing links with decrease in traffic and red signature showing increase in traffic in

case of a certain scenario relative to a base scenario. The “attractiveness” of each road link

depends only on road geometry and traffic

volumes. This is an ideal situation for automat-

ic calculation of attractive routes from A to B.

In the Fehmarn Belt case, the transportation

network of the traffic model consists not only

of road links but of ferry links as well. The “at-

tractiveness” of each ferry link is calculated as a

function of its main service variables: Fare, de-

parture frequencies, sailing time, waiting time

at terminals etc. All these ferry line qualities are

by use of statistical methods weighed together

in order to bring the ferry line “attractiveness”

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

in line with the “attractiveness” of road links on a common and comparable standard. The

traffic model can now work with all network links on an equal basis and calculate the distri-

bution of traffic over the complete network including traffic on the special links represent-

ing ferry lines.

In the base case scenario, the traffic model was fed with the ferry line characteristics pre-

vailing at the time of setting up or calibrating the model. So, the model was fed with what

could be described as a certain snapshot of ‘competition armistice’ between ferry lines.

In the forecast situation, this ‘armistice’ between ferry lines and more importantly relative

to the fixed link is frozen in the traffic model to its original levels of service. This will of

course not be the case when the fixed link opens. Each ferry line in the Scandinavian waters

which might experience any loss of customers to the fixed link will of course improve its

competitive status and fight to keep its customers.

This blunder of neglecting ferry lines as fiercely competitive players in the market was

made in the forecasts made in 1999 and in 2002 as well. But to repeat this blunder as late

as in 2014 seems unacceptable. In the

“Enquiry of commercial interest for the proposed

fixed link across the Fehmarnbelt”,

ref.[5] and [6], it was clearly stated that serious econom-

ic competitive risks existed between the fixed link and the ferry lines. This warning should

have been sufficient for Femern A/S to ensure that detailed competitive analyses would be

carried out. Regrettably, this did not happen.

As a first step it was justified to run the model on previous competitive balanced levels of

service for each ferry line. However, this first step should naturally have been followed by

an in-depth analysis of which profitable countermeasures each ferry line might adopt in

order to keep its customers if the model results showed a loss of customers.

If the forecasting task from the start had been offered by use of a public tender process,

consultants with experience in private competition analyses might have provided the nec-

essary knowledge in cooperation with more classic traffic planners like the chosen German

companies.

The consequence of not assessing competitive countermeasures has led to substantial

overestimation of diverted traffic from a long array of competing ferry routes. This overes-

timation comes on top of the overestimation due to using average fares instead of segment

defined fares for shopping trips and Europe trips.

Special case: Rødby – Putgarden

Chapter 5.2 explained why the traffic model used for the 2014 forecasts was not able to

evaluate the split of traffic between a fixed link and a potential parallel ferry line service.

Even if we ignore this fundamental lack of ability, Femern A/S and affiliated consultants

ought to have been worried by technical requirements that exclude competitive strategies

in relation to introducing time dependant departure frequencies:

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

Figure 10

Due to model restrictions dated back to the late 90’s, the consultants had to accept that a

competing ferry line service should run 24/7 with constant departure frequencies.

Not precisely the prescription for competition.

The consultants could choose between departures each half hour, hourly, every second

hour or any other 24/7 service with constant departure frequency all day and night. This

feature is one more sign of how unsuitable the model was for evaluating parallel services. If

a ferry line was to make a service strategy, it might be relevant to evaluate concentrated

services during busy hours and accept the tunnel take over during hours of low traffic de-

mand. However, the model back in the 90’s was pre-coded to have fixed frequencies, which

of course is not a wise strategy for a competitor.

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

5.4 Transferred Car Traffic from the Storebælt Bridge

The 2014-forecasts, predicted, without documentation, that 800.000 cars per year would

be diverted from the Storebælt Bridge. Before 2014 such markedly volatility in traffic de-

mand between the two transport corridors had been rejected. The German consultants

behind the 2014 forecast had back in 2002 made an update of the original 1999 forecast. In

this update: “Fehmarn Belt Forecast 2002 Final Report, April 2003”, ref. [3], the following

conclusion was made about competition between Storebælt and Fehmarn Belt:

“Competition from the Great Belt Link

According to the conclusion in Chapter 10, the competition relationship between the Great

Belt link and a fixed link across the Fehmarn Belt is rather modest.

In a recent survey performed by Sund & Belt Ltd., it was found that only 3 percent of the

present Great Belt traffic has either destination or origin in Germany; 97 percent is nation-

al Danish traffic. Hence, only the 3 percent could consider to use the Fehmarn Belt link in

the future.

This result confirms previous FTC forecasts, which showed that only 1,9 % of car traffic and

0,8 % of lorries on the Great Belt link would be attracted by the Fehmarn Belt link in 2010.

The above shows that a[t] Fehmarn Belt link will only be an attractive alternative for a

small share of the existing traffic across the Great Belt.

On the other hand, the Great Belt link might be an attractive alternative for some of the

travellers that could use a Fehmarn Belt link. This will depend entirely on the difference in

the toll levels at the two fixed links. The transport route via Rødby-Puttgarden is approxi-

mately 150 km shorter, than the route via the Great Belt. The current cost of travelling via

this route including the cost associated with travelling a longer distance is 60-80 Euros,

which is substantially higher than the ferry fare at Rødby-Puttgarden of 46 Euro. Unless,

there are significant changes in relationship between the tolls at these crossings, the Great

Belt link, will not be a significant competitor to a Fehmarn Belt link.”

As late as in a report from 2014 ref. [8]: “Ex post samfundsøkonomisk vurdering af Sto-

rebæltsforbindelsen” (Ex post socio economic assessment of the Great Belt Fixed Link) it

was stated:

“Det er primært danskere, der bruger Storebæltsforbindelsen. Kun en mindre del af bru-

gerne er udlændinge af den simple grund, at det sjældent er den oplagte vej at rejse, hvis

man kommer fra Tyskland, Sverige eller Norge.”

Which translated says: “it is primarily Danes that use the Storebælt fixed link. Only a minor

part of the users are foreigners simply because it is rarely the obvious route for traffic com-

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

ing from Germany, Sweden or Norway.”

That conclusion goes for major parts of the traffic between Zealand and Germany as well,

since the transport corridor from Sweden through Denmark passes south of Greater Co-

penhagen.

It is well documented ref. [2], [3] that this enormous amount of claimed diverted traffic

simply was not part of the traffic model. So, the big question is, on what grounds was this

suddenly discovered massive traffic demand based?

Danish media tried to examine the issue and this interest from the media initiated a ques-

tion from MP Henning Hyllested (Enhedslisten) to the then Minister of Transport, Magnus

Heunicke (Social Democrat). The following text is a quote from the official answer, ref. [13]

from the Minister of Transport to Henning Hyllested:

”…Intraplan og BVU oplyser, at de 713.000 per-

sonbiler er fremkommet med basis i en række

forskellige analyser, herunder analyser af trafi-

kanternes startsted og slutdestination, deres ru-

tevalg, nummerpladescanninger samt analyser af

transaktions-data på Storebælt. Sidstnævnte er

udarbejdet af A/S Storebælt, mens de resterende

analyser i udgangspunktet er udarbejdet af DTU

med basis i blandt andet dataindsamling udført

af de to danske konsulentfirmaer COWI og Te-

traplan. Disse oplysninger er tilgået Intraplan og

BVU, der har indarbejdet dem i deres trafikmo-

del.”

Translation:

Intraplan and BVU have stated that the 713,000 passenger cars have been established on

the basis of a variety of analyses, including analyses of the drivers' origin and end destina-

tion, route choice, license plate surveys and analyses of transaction data on the Great Belt.

The latter has been prepared by A/S Storebælt, while the remaining analyses were original-

ly prepared by DTU

[Danish Technical University],

based on data collection by the two Danish

consultancy firms COWI and Tetraplan. This information has been passed on to Intraplan

and BVU, which have incorporated these findings into their traffic model.

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

Not only was it unprecedented that a client instructs a consultant to incorporate specific

client produced extra traffic demand for a specific transport link, but also it turned out that

the references were incorrect.

The unusual behaviour by a state authority was given seemingly credibility by referring to

the best Danish university compe-

tences. But the references turned out

to be a clear-cut falsification. On 15

May 2015 the Danish technical jour-

nal Ingeniøren published an inter-

view with one of the leading senior

researchers at DTU. The headline -

shown to the right - in the journal

was: “DTU is surprised: Our studies

do not say anything concerning the Fehmarn forecast”

It must have been embarrassing for the consultants in Munich and Freiburg to receive and

to obey instructions by the client to insert Danish fabricated figures into the forecast docu-

mentation delivered by November 2014. By accepting this highly unusual behaviour by a

client, the consultants made themselves vulnerable and a sign of that was the memo:

“Fehmarn Belt Forecast 2014 – Update of FTC-Study of 2002 – Treatment of Great Belt in

Forecast (Passenger Traffic)”

prepared by the German consultants and dated 2015. This

memo seems to have been intended as a damage control by the consultants. Instead it

turned out to unveil more embarrassing details on an inconceivable incompetent traffic

survey made by the Danish client. From the note to be found in ref. [14]:

“…Samples of plate counts in summer indicate a share of 5 to 6 % of foreign cars on GBFL.

Foreigners use the GBFL mainly on North-South

traffic. In winter the share is lower, 2 to 3 %, which

is logical as in winter the price differences be-

tween GBFL (no seasonal or daytime variation of

toll rates) and the ferry line Rødby-Puttgarden

(much higher prices in peak periods) is lower than

in summer.

“…Assuming a similar number of Danish cars using

GBFL in north-south traffic instead of domestic

traffic between east and west Denmark, in total a

share of 8 % of the GBFL traffic can be as-signed

to international resp. north-south traffic on GB.

This is a quite reliable figure for 2011. In 2014/15

this share should be higher due to a growing

“price gap” between GBFL (stable toll rates and

decreasing fuel costs for longer distance) and Rødby-Puttgarden (raising prices).

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

The numbers for the GB transports in the north-south traffic (713.000 cars in 2011) have

been added to the transports on the relevant ferry lines (around 2.9 million cars per year in

2011) to get the totals of north-south traffic. At the same time, it has been used in the

same way as the ferry statistics: to calibrate the route choice model for the forecasting….”

A simple licence plate count was carried out and all cars were classified into just two cate-

gories: Danish and foreigners. In the above quotations, one short sentence discloses an

unbelievable misunderstanding: “Foreigners

use the GBFL mainly on North-South traffic”.

This assumption is however as erroneous as can be – more on that issue in following para-

graphs.

The average share of cars with foreign licence plates, estimated by using simple licence

plate surveys, was 4 pct. These cars were all erroneously judged to be future customers in

the Fehmarn Belt Tunnel. One should think that such a huge analytical blunder would

stand alone but there was more to come, as the 4 pct. of foreigners were assumed to have

Danish ‘mirror effects’, i.e. with an equal number of Danish registered cars making the trips

in the opposite direction. So, one further huge blunder doubled the figure from 4 to 8 pct.

Based on two grave misinterpretations of the simplest thinkable traffic surveys, the share of

the Storebælt car traffic forecasted to be diverted to the Fixed Fehmarn Belt Link ended at

a level of 8 pct. To put this figure in perspective, former studies had shown 1 – 2 pct. of the

Storebælt traffic to be border crossing D/DK. Furthermore the forecasts in 1999 and 2002

did not incorporate any diversion of cars at all because the potential was insignificant. In

2014 however, one out of each twelve cars passing Storebælt was claimed to be diverted to

Fehmarn Belt.

In the following paragraphs these misperceptions are explained in more detail.

Misperception no 1

“Foreigners

use the GBFL mainly on North-South traffic”

There is of course no relation at all between the nationality of a car crossing Storebælt and

the origin and destination for the trip of that car. So, it is complete nonsense to claim that

foreign cars passing Storebælt are mainly North-South bound. There is no documentation

whatsoever for such a claim.

A foreign car crossing Storebælt can have all sorts of destinations / origins that would not

include the Fehmarn Belt Fixed Link as a future potential route choice. Some examples:

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

•

A Swedish family visiting the Legoland amusement park in Jutland

•

A Dutch salesman visiting several Danish customers on both sides of Storebælt

•

A German tourist staying in a

summer cottage in West Den-

mark making a round trip visit to

Copenhagen

•

A Polish guest worker visiting

customers or colleagues in

Denmark

•

A German citizen living in South

Schleswig (the German region

just south of the green border

D/DK) with an errand in East

Denmark. Fehmarn Belt will

never be the first choice for such

a traveller despite German li-

cence plates on the car

•

A Danish expatriate living in

South France but at the time of

the licence plate survey, visiting friends, family or business contacts in Denmark.

It is astonishing that such a clear-cut blunder can be made by the client in the first place,

and in the second place, was not rejected by the German consultants. Nobody, even with

the faintest idea of how traffic surveys are carried out, would make such a grave misper-

ception.

Misperception no 2

After having estimated foreign cars passing Storebælt at an average of 4 pct. and errone-

ously presuming these cars to be future customers of Fehmarn Belt, misperception 1 has

led to misperception 2: Each foreign car passing Storebælt is anticipated to have a Danish

“counterpart” driving in the opposite direction. Such “opposite” Danish trips makes no

sense either. The mirror image of a Danish business man visiting various customers in The

Netherlands would not cross the Storebælt Bridge while travelling internally in The Nether-

lands. Likewise, a Danish technician doing service on technical installations in South Ger-

many would not use the Storebælt Bridge while serving his clients in South Germany.

Everybody understands that a commuter trip from a suburb to a business centre does not

necessarily have a “mirror commuter trip” from the business centre to the suburb. But this

simple fact is apparently not understood by Femern A/S.

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

It is alarming that about 25 pct. of the forecasted car traffic demand on the Fixed Fehmarn

Belt Link is based on sheer incompetence and the figure is dramatically overestimated

compared to earlier analyses based on best practice methods.

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

5.5 Lorry Transport

Unlike the car market, the forecasts for lorry traffic in the Fehmarn Belt traffic corridor

seem to be far more realistic or even conservative of nature. During recent years, realised

traffic demand seems to have exceeded forecasts.

Figure 11

Ref [9], Figure 9-3 page 193. Forecast for lorries crossing the Fehmarn Belt. The following changes to

the original illustration have been made: Case A has been removed, figures for certain years have

been removed; observed traffic for year 2013 has been updated and observed traffic for years 2014 –

2016 have been added.

On the face of it, this development in observed traffic seems reassuring for the fixed link to

have revenues from lorries at least as forecasted. However, this development is on the oth-

er hand perhaps even better news for a potential continued ferry service because its core

customers are expected to be lorries, border shoppers and travellers who dislike long dis-

tance tunnel driving or simply travellers who fear tunnels.

As earlier described, in its present version, the traffic model is not able to evaluate the dis-

tribution of traffic between the tunnel and a parallel ferry service. That applies to all kinds

of vehicles, but this weakness is indeed most significant in the case of lorries.

Since the model was estimated in the late 90’s, marked changes have taken place in the

road haulage market. Today there is hardly any business sector as cost sensitive as road

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

haulage. The profit in this sector is made by small margins. One sign of this strong focus on

controlling costs can be seen from the fact that big haulage companies are flagging out

from their home countries and/or are hiring lorry drivers from countries with low wages.

The traffic model does not take this increased cost sensitiveness into account.

With its state guaranteed loans, Femern A/S will always be able to outcompete any parallel

ferry line by lowering fares beneath the pain threshold of such a ferry line and thereby clos-

ing it. But that would probably be a violation of EU regulations. If Femern A/S have to stick

to fares which would provide revenues of a magnitude as stated in the background papers

for the decision of the investment – then there will be a clear competitive choice between

low cost / high sailing time and high cost / fast driving time.

In this respect, high cost sensitiveness will be an important parameter in hauliers’ decision

making. However, the current version of the traffic model is not based on detailed analyses

on competitive trade-offs and has, accordingly, not given trustworthy outputs.

Another important change since the time of setting up the model is the focus on mandatory

rest hours for lorry drivers. Back in the 90’s there was only a very inefficient technology

available to check whether or not a driver had taken his rest periods as demanded. It was

easy to remove the cardboard disc showing recent driving hours and replace the disc with a

new, empty one. This situation has dramatically changed since then but with no corre-

sponding changes in the parameters that controls the trade-off calculations in the model.

In a future scenario with parallel ferry services, the sailing time is not necessarily just a

waste of time as is the premise for the model forecasts. For each 4�½ hours of driving, the

lorry driver will have to rest for 45 minutes which is exactly the sailing time. Depending on

the precise logistics for each lorry trip, the negative effect of the sailing time can vary be-

tween 0 and 45 minutes. This effect will naturally be incorporated in hauliers’ evaluation of

trade-offs between the two different modes of transport. However as mentioned above,

this is not reflected in the traffic model.

The risk that parallel ferry service might undermine the business economics of the fixed link

was presented early in the process as part of the conclusions in the

Enquiry of commercial

interest in 2002

(Ref. [5] and [6]). The risk presented by potential financial investors has

been ignored by Danish politicians throughout their handling of the project. But the risk has

proven real by statements from haulage market leaders using the Fehmarn Belt transport

corridor.

On 10

June 2014, the Danish business newspaper Børsen interviewed two of the largest

Danish hauliers, DSV and HCS, concerning their interest in using the fixed link. The answers

had clear messages:

Ole Bolm,

CEO, HCS:

"International road transport is extremely price sensitive and it will always choose

the cheapest solution - even if it may take a little longer”.

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

Simon H. Galsgaard,

CEO, DSV Road A/S

"When crossing Øresund, we frequently use the ferries, although it may sometimes

be faster via the bridge. If the tunnel becomes significantly more expensive than

the ferries, then the ferry service is definitely a future for us. "

Figure 12

Headline in the business newspaper Børsen: Price war tempting lorries away from expen-

sive tunnel fares

Such worrisome market reactions seem to have had no reactions at all, neither from

Femern A/S nor from the political backers in Parliament. This message from important mar-

ket players did even not lead to a raised eyebrow at COWI, when the company undertook

the quality assurance of the traffic forecasts.

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

5.6 No Handling of Risks and Uncertainties

It comes as a surprise that a project worth DKK 62,1 bn (equivalent to € 8,3 bn) has been

approved by the Danish Parliament without evaluations of risks and uncertainties in traffic

forecasts - and thereby revenues from users.

However, partial sensitivity analyses of certain figures and parameters have been carried

out. But such analyses cannot give a realistic picture because only one parameter has been

varied at a time. In the real world all parameters can and will vary simultaneously, some

dependently – some independently.

One of the largest risks for the business economics for Femern A/S will become reality if the

existing ferry line continues to operate after the opening of the tunnel. This possibility has

not been handled as a risk but has simply been rejected based on a deterministic working

model outside its limits of validity.

The law of construction was passed in May 2015; nevertheless, the Danish Ministry of

Transport waited for about half a year before in total secrecy, hiring the consultancy KPMG

to make a business review of future ferry economics under parallel services. To request

such a report in secrecy can only be interpreted as an attempt to keep a potential negative

result - for the Ministry - secret, because a documentation that a ferry line would be able

to exist parallel to the tunnel would have made the Ministry look incompetent or even

guilty in misleading the Parliament. However, a leak revealed that a secret report from

KMPG, dated 14

January 2016 existed, but this version has never been published.

This disclosure forced the Ministry to publish a revised version by 14

March 2016. Two

months elapsed from the disclosure of the existence of a secret report until the publication

of a new version. This is a very long time given the embarrassing situation for the Ministry

attempting to keep important information from Members of Parliament. The first version

must have contained conclusions that would have caused severe problems for the Ministry.

Chapter 6 reveals how The Ministry, in another case, handles consultants to ensure desired

outcomes in accordance with predetermined political expectations.

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

6. Quality Assurance by a Disqualified Consultant

A Message from the Transport Minister

Just before the quality assurance of the traffic forecasts were made public, former Minister

of Transport, Hans Chr. Schmidt wrote a letter to the political parties behind the original

2008-agreement on establishing the Fehmarn Belt Fixed Link:

Dear parties behind the agreement,

The Ministry of Transport and Building has

conducted an independent quality assurance

of traffic forecast for the Fehmarn Belt Fixed

Link.

The quality assurance, prepared by COWI,

shows, among other things, that:

•

The traffic forecast provides a realistic es-

timate for the expected road traffic on

the Fehmarn Belt Fixed Link.

There is nothing to suggest that the fore-

cast systematically overestimates the to-

tal traffic volume.

The assumptions for the forecast are rea-

sonable, and the forecast models are in line with professional standards for traffic

forecasting.

The traffic forecast is cautious in assessing induced travel because of lack of tools to

forecast dynamic effects of the Fixed Link.

•

COWI finds difficulties in documenting international traffic on the Great Belt Bridge and

thereby estimation of diverted traffic to the Fehmarn Belt. However, this does not change

the above conclusions by COWI.

I enclose for your information the Quality Assurance Report prepared by COWI.

Yours sincerely

Hans Chr. Schmidt

There are several interesting messages from the Minister to the majority of members of the

Parliament. The minister starts off making it clear that this is an independently conducted

quality assurance. However, the Minister knew very well that COWI – as part of a joint ven-

ture - at the time had been prequalified to make bids for the construction work. It was be-

yond discussion, and the Minister was of course fully aware that worldwide no company

could be more disqualified for the job than precisely COWI.

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

The next very interesting message is that the Minister in his letter finds it necessary to em-

phasize to his colleagues in Parliament that they should focus on the blank Quality Assur-

ance given by COWI and not that COWI had waved a red flag because one quarter of all

passenger car traffic in the tunnel was not in any way accounted for in the forecasts. The

discrepancy between the given blank quality assurance and the red flag shows more than

anything what was at stake between the Ministry and COWI and the massive pressure put

on COWI.

The Settings for the Quality Assurance

The traffic forecasts published in 2014 were written in

English. The geographical market for hiring consultants to

perform a quality assurance of the Fehmarn Belt fore-

casts was accordingly almost worldwide.

Nevertheless, the Ministry of Transport limited its search

for a consultant to the Danish market of consultancies.

The Ministry concluded that only two Danish companies

were qualified to do the job: One however was disquali-

fied because of another forecast job for Femern A/S.

Back in the basket was not surprisingly the Danish com-

pany COWI. This company was at the time prequalified to

several of the construction jobs – se ref. [12]. COWI was

accordingly with no doubt disqualified for the task. In a

letter of 15

March 2016 (J. nr. 2016-40) to the newspaper Weekendavisen (see ref. [14]),

the Danish Ministry explained the choice of COWI in the following most peculiar way:

Text in Danish:

”Valget af COWI til at gennemføre den eksterne kvalitetssikring af trafikprognosen var alene

fagligt baseret. Således var vurderingen, at kun to rådgivere i Danmark ville have den nødven-

dige kompetence til at gennemføre en ekstern kvalitetssikring af trafikprognosen. Alternativet

var imidlertid engageret af Femern A/S til at levere et bidrag til trafikprognosen og var der-

med inhabil.

Transport- og Bygningsministeriet valgte COWI til at gennemføre den eksterne kvalitetssikring

af trafikprognosen i maj 2015. På det tidspunkt var det kun navnene på de virksomheder, der

indgik i de af Femern A/S prækvalificerede entreprenørkonsortier på de fire store anlægsen-

trepriser, der var kendt. Femern A/S offentliggjorde navnene på de prækvalificerede entre-

prenørkonsortier den 27. maj 2013.

Translation:

"The choice of COWI to perform the external quality assurance of traffic forecast was based

only on a judgement of professional skills. Based on that condition the assessment was that

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

only two consultancies in Denmark would have the necessary skills to perform an external

quality assurance of the traffic forecast. However, the other possible consultant had been

hired by Fehmarn A/S to deliver a contribution to the traffic forecast and was thus disquali-

fied.

The Ministry of Transport and Building chose COWI to perform the external quality assurance

of the traffic forecast in May 2015. At that time, only the names of the companies that were

prequalified by Femern A/S to each of the four big construction contracts were known.

Femern A/S published the names of the prequalified construction consortia on May 27, 2013.

The statements from the Ministry can be commented as follows:

•

No explanation is given as to why the consultant had to be a Danish company while

the forecast itself had been made by two foreign companies. Logically, this makes

no sense.

Even if there were reasons to limit the field to Danish consultancies, it is surprising

that only two were believed to have the needed qualifications. Had the Ministry

followed the principles of public tendering and specified the qualifications, then a

much larger group of potential consultancies - Danish and Internationals as well -

would have had a chance to show their capabilities. By identifying only two poten-

tial consultancies, of which one in advance was known by the Ministry to be dis-

qualified, leaves the impression that the Ministry had designed a process to ensure

that precisely COWI got the job simply because the prequalification to the much

larger construction jobs would make it impossible for COWI to deliver nothing but

green lights for the quality of the traffic forecasts.

COWI was as part of a consortium prequalified for major parts of the construction

jobs as early as 27

May 2013. Femern A/S decided by 4

March 2016 which con-

struction consortiums should be assigned to which technical construction jobs.

COWI ended up winning three out of four construction tasks. COWI was according-

ly and indisputably disqualified during the period from May 2013 until March 2016

concerning evaluation of the fundamental reasons to build the tunnel: traffic de-

mand.

The message: “At that time, only the names of the companies that were prequali-

fied by Femern A/S to each of the four big construction contracts were known” is

formulated as if this clears the responsibility of assigning the quality assessment to

COWI. But this makes no sense. The one and only piece of information needed to

disqualify COWI was precisely that the names of the prequalified companies had

been known since 27

May 2013.

•

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

Quality Assurance – Conclusions by COWI

The Ministry of Transport achieved the desired conclusion because COWI had no other

choice but to deliver a blank quality assurance given the fact that COWI had been prequali-

fied to huge construction tasks.

Danish version first:

Translated:

Conclusion

The overall conclusion is that COWI has assessed the traffic forecasts to be

a realistic assessment of how much road traffic can be expected on a fixed

Fehmarn Belt fixed link. There is nothing that indicates that the forecasts

systematically overestimate total volume of traffic. There are uncertainties

in the forecasts, partly because of data concerning the reference year

2011 are difficult to document especially concerning traffic crossing the

Great Belt. Finally, there are elements in the forecast which appear to be

underestimated, primarily in relation to dynamic effects and newly gener-

ated traffic. Uncertainty can be reduced by better documentation of data

and possibly with additional traffic surveys on the Great Belt. It will, how-

ever, require significant calendar time and substantial resources to imple-

ment.

This is a blank approval of the traffic forecasts.

The issue of the huge diversion of traffic from Storebælt to Fehmarn Belt is in the text pre-

sented as an ‘uncertainty’ and uncertainties can, if not described in further depth, work

either way. Consequently, COWI has presented a rather neutral statement in that respect.

Professionals can always desire more accurate data, but the key question that should have

been addressed by COWI was whether the amount of transferred traffic was a central best

estimate – albeit with uncertainties - or whether it was a highly biased estimate. It must

have been crystal clear to an otherwise professional company as COWI that the massive

diversion of traffic was based on, at best, embarrassing misperceptions.

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

Moreover, the potentially highly dangerous uncertainty in forecasting traffic for the fixed

link is the risk of continued ferry services. This risk has not been mentioned at all in the

summary of conclusions. The report from COWI has two and a half pages in chapter five

titled, “Calculations on Sensitiveness with continued ferry services”.

This chapter reveals that the German consultants presented their forecasts at a meeting

with COWI in Copenhagen 21

August 2015. COWI was appointed to do the job in May

2015. Presumably, this meeting was focussed on the issue of continued ferry services –

otherwise, why not make this statement in the beginning of the report and not first in

chapter five?

The final report from COWI ref. [15], was submitted by 12

November 2015. In a standard

case of a genuine independent quality assurance, such a meeting should only – if at all -

take place late in the process in order not to influence the quality assurer in an early phase.

A meeting may take place late in the process to give the forecaster a chance to defend him-

self. By organising such an early meeting - presumably primarily concerning continued ferry

services – the Ministry ensured that the quality assurer was not only presented to the fore-

cast as such, but also, more importantly, coached in due time on the big issue of continued

ferry services. This strategy seemed to have worked well. COWI did not question the abil-

ity of the model as such to run forecasts for parallel services or whether the model was able

to forecast traffic under fierce competition between a private ferry line and a publicly run

fixed link. Neither did COWI question whether a 24-hour service with constant departure

frequencies day and night would be the most probable way to compete with the fixed link.

Many other critical assumptions ought to have been dealt with by the quality assurer. In

chapter 1.1 COWI made a description of how the company viewed its methodical approach:

“COWI har gennemført en kvalitetssikring af trafikprognoserne I overensstemmelse med

tænkningen i eksterne kvalitetssikring af anlægsprojekter”

“COWI has conducted a quality assurance of the traffic forecasts in accordance with the

thinking behind external quality assurance of construction projects”

This interpretation of the task is peculiar. To make a quality assurance of a traffic forecast

has very little to do with the quality of technical processes in relation to construction jobs.

Quality Assurance – A Ministerial Procedural Blunder

On the very same day, 12

November 2015, when the Ministry of Transport received an

unconditional quality approval by COWI, the Ministry secretly asked Sund & Bælt Holding

A/S (the state-owned parent company of Femern A/S) to analyse the magnitude of interna-

tional traffic crossing Storebælt in order to further document the undocumented diversion

of traffic from Storebælt to Fehmarn.

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

This request to dig further into the quality assurance is totally out of line with the focused

ministerial strategy to appoint COWI to carry out the quality assurance job to ensure the

desired and expected approval.

It is very unusual by Danish standards that such a request would be sent on the very day of

receipt of the COWI report. In ref. [16] Hans Schjær-Jacobsen has given a comprehensive

description of how this request - now in its third year - is being sent around in closed circles

in order - presumably – to postpone the result until the construction work has been initiat-

ed and the point of no return has been passed.

In ref. [16] there is a reference to e-mails showing that Sund & Bælt A/S had got the im-

pression from the Ministry that “..der er ikke et overvældende tidspres på” (“there is no

overwhelming time pressure”). The discrepancy between the initial hasty request to Sund &

Bælt Holding A/S and the current deliberate delaying of the process supports the theory

that the initial request was made by mistake by a person in the Ministry not totally aware

of the strategy of having COWI do the quality assurance.

Today, more than two years later, the task to document the huge transfer of traffic from

Storebælt to Femern has still not been carried out and there is clearly no incentive for the

Ministry or its subsidiaries to close the case before the final environmental approval from

Germany, and consequently, the implementation of construction. At that time, the result

will be completely unimportant.

It was important for the Ministry and Femern A/S to have the quality assurance of the traf-

fic forecasts in place as the first of several subsequent quality assurances of other matters.

These quality assurances would thereby be safeguarded against discussions of uncertainties

about traffic forecasts and revenues from users.

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

References:

[1]

Fehmarnbelt Traffic Demand Study – Final Report

Fehmarn Belt Traffic Consortium (FTC) – 1999

[2]

Femer Bælt-forbindelsen, forundersøgelser – Resumérapport

Trafikministeriet - 1999

https://www.trm.dk/da/publikationer/1999/femer-baeltforbindelsen-forundersoegelse-resumerapport

[3]

Fehmarn Belt Forecast 2002 Final Report,

Fehmarn Belt Traffic Consortium (FTC) - April 2003

https://www.bmvi.de/SharedDocs/DE/Anlage/VerkehrUndMobilitaet/Schiene/2003/fehmarn-belt-forecast-

2002-final-report-april-2003.pdf?__blob=publicationFile

[4]

Fehmarn Belt Forecast 2002 Reference Cases, Supplement to Final Report of April

2003,

FTC Fehmarn Belt Traffic Consortium - November 2003

[5]

Enquiry of commercial interest for the proposed fixed link across the Fehmarnbelt

Olaf Meyer-Rühle, Prognos AG

[6]

Finansiering og organisation Interessetilkendegivelsesrunden Resumé

Fehmarnbelt Development Joint Venture June 2002

Commissioned by: Trafikministeriet Danmark, Trafik-, bygge- og boligministeriet Tyskland

[7]

Finansiel analyse, September 2008.

Femern Bælt A/S (later Femern A/S) for the Ministry of Transport

https://www.femern.info/sites/default/files/library/finansiel_analyse_2008.pdf

[8]

Ex post samfundsøkonomisk analyse af Storebæltsforbindelsen

Incentive & Tetraplan – August 2014

Commisioned by Ministry of Transport and Sund & Bælt

http://www.incentive.dk/wp-content/uploads/2017/03/Incentive-Ex-post-samfunds%C3%B8konomisk-analyse-

af-Storeb%C3%A6ltsforbindelsen.pdf

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

[9]

Fehmarnbelt Forecast 2014 – Update of the FTC-Study of 2002

Intraplan & BVU Beratergruppe – 2014

Commisioned by Femern A/S

[10]

Fehmarnbelt Forecast 2014 – Update of the FTC-Study of 2002 – ANNEX

Intraplan & BVU Beratergruppe – 2014

Commisioned by Femern A/S

[11]

Trafikprognose for en fast forbindelse over Femern Bælt

Femern A/S – November 2014

[12]

Femern A/S prækvalificerer internationale entreprenører

[Prequalification of international contractors]

Femern A/S (Press release 27

May 2013)

https://femern.com/da/News-and-press-search/2013/May/Femern-AS-praekvalificerer-internationale-

entreprenoerer

[13]

Transportudvalget 2014-15 L141 endeligt svar på spørgsmål 13

Ministry of Transport – 30

March 2015

http://www.google.dk/url?sa=t&rct=j&q=&esrc=s&source=web&cd=1&cad=rja&uact=8&ved=0ahUKEwi4gpyg4

PvYA-

hUP_qQKHQkfAcQQFggmMAA&url=http%3A%2F%2Fwww.ft.dk%2Fsamling%2F20141%2Flovforslag%2Fl141%2

Fspm%2F13%2Fsvar%2F1240922%2F1514506.pdf&usg=AOvVaw1Uott46QfZ1wSiSCTVi9JE

[14]

Fehmarn Belt Forecast 2014 – Update of the FTC-Study of 2002 – Treatment of Great

Belt in the Forecast (Passenger Traffic)

https://ing.dk/artikel/se-hele-ingenioerens-aktindsigt-om-femerns-trafikprognose-176064

[The documents are found by scrolling down]

[15]

Ekstern kvalitetssikring af den opdaterede trafikprognose af Femern Bælt-projektet

COWI, November 2015

Commissioned by the Ministry of Transport

https://www.trm.dk/da/publikationer/2015/ekstern-kvalitetssikring

[16]

Opdatering: Ekstern kvalitetssikring af business case for den faste Femernforbindelse

Hans Schjær-Jacobsen

http://www.trafikdage.dk/papers/soeg/Paper.asp/?PaperID=1921

TRU, Alm.del - 2017-18 - Bilag 215: Henvendelse af 1/3-18 fra Knud Erik Andersen, Taastrup om Femern

[17]

The Fehmarn Belt Fixed Link – made in Denmark.

Knud Erik Andersen - June 2014.

Publisher and sale

www.praxis.dk

[18]

Analyse af nye trafikprognoser for Femern-forbindelsen, 2014

Knud Erik Andersen - January 2015.

Publisher and sale