The Economics of Crowding in Public Transport e de Palma a Robin - - PowerPoint PPT Presentation

The Economics of Crowding in Public Transport

Andr´ e de Palmaa Robin Lindseyb Guillaume Monchamberta,c,1

a´

Ecole Normale Sup´ erieure de Cachan

bUniversity of British Columbia cKULeuven

Railway Operations Research Seminar: ”Put Passengers first” KUL - May 3rd, 2016

1guillaume.monchambert@kuleuven.be

Introduction Literature review Hypothesis PTM Model Capacity Application to RER A

Motivation

2 / 37

Introduction Literature review Hypothesis PTM Model Capacity Application to RER A

Motivation

3 / 37

Introduction Literature review Hypothesis PTM Model Capacity Application to RER A

Modal split - Intuition 1/4

Generalized cost

• is the sum of the monetary and non-monetary costs of a

journey (travel time, waiting time...),

• may depend on the patronage.

4 / 37

Introduction Literature review Hypothesis PTM Model Capacity Application to RER A

Modal split - Intuition 1/4

Generalized cost

• is the sum of the monetary and non-monetary costs of a

journey (travel time, waiting time...),

• may depend on the patronage.

Figure 1: Generalized cost of using road as a function of the number of road users

4 / 37

Introduction Literature review Hypothesis PTM Model Capacity Application to RER A

Modal split - Intuition 2/4

The equilibrium on a network is reached when the generalized cost is the same for all modes (Wardrop equilibrium).

5 / 37

Introduction Literature review Hypothesis PTM Model Capacity Application to RER A

Modal split - Intuition 2/4

The equilibrium on a network is reached when the generalized cost is the same for all modes (Wardrop equilibrium).

Figure 2: Bi-modal equilibrium with economies of scale in public transport (Mohring effect)

5 / 37

Introduction Literature review Hypothesis PTM Model Capacity Application to RER A

Modal split - Intuition 3/4

Figure 3: Bi-modal equilibrium without economies of scale in public transport

6 / 37

Introduction Literature review Hypothesis PTM Model Capacity Application to RER A

Modal split - Intuition 4/4

Figure 4: Bi-modal equilibrium with diseconomies of scale in public transport (Crowding)

7 / 37

Introduction Literature review Hypothesis PTM Model Capacity Application to RER A

Characteristics of public transport crowding

Crowding in public transport have characteristics which are similar to congestion (traffic jam) on road:

• a negative externality,
• caused by an excess of demand,
• which increases with the demand,
• which degrades the individual experience of travel,
• and which raises the generalized travel cost.

8 / 37

Introduction Literature review Hypothesis PTM Model Capacity Application to RER A

In the literature

• There is an economic theory of road congestion (mainly

through the bottleneck model: Vickrey, 1969; Arnott et al., 1990, 1993),

• but few works on crowding in public transport.

Almost all works study competition between road and public transport including congestion on road but not on public transport (Tabuchi, 1993 ; Danielis et Marcucci, 2002 ; Mirabel et Reymond, 2011 ; Gonzales et Daganzo, 2013 ; Tian et al., 2013...). = ⇒ Fill this gap in the literature.

9 / 37

Introduction Literature review Hypothesis PTM Model Capacity Application to RER A

Goals of this work

• Develop a dynamic model of public transit usage that includes

crowding (PTM model).

• Derive:
• User equilibrium
• System (i.e., social) optimum
• Optimal fares: uniform and time-varying
• Optimal service
• Timetable, no. trains, train capacity
• Dependence on fare regime
• Comparisons with bottleneck model of road traffic congestion

(the most widely adopted road congestion model in the economic literature).

• Application to a segment of RER A transit line in the Paris

Region.

10 / 37

Outline

• 1. Literature review
• 2. Hypothesis of the model
• 3. PTM Model
• 4. Capacity adjustment
• 5. Application to RER A
• 6. Conclusions

Outline

• 1. Literature review

1.1 Empirical microeconomic evidences 1.2 Empirical macroeconomic valuations

• 2. Hypothesis of the model
• 3. PTM Model
• 4. Capacity adjustment
• 5. Application to RER A
• 6. Conclusions

Introduction Literature review Hypothesis PTM Model Capacity Application to RER A

Empirical microeconomic evidences

Mircoeconomic valuations of crowding cost:

• Fixed penalties, about 2,5$US/trip (Pepper et al., 2003 ;

Hensher et al., 2011)

• Time multipliers
• Whelan and Crockett, 2009 (next slide).
• Wardman and Whelan, 2011 : standing = 2,32 ; seating =

1,32.

• Kroes and al., 2013 :
• Seating when all seats are occupied = 1,1.
• Standing when the vehicle is full = 1,6.
• Haywood and Koning, 2015 (next slide).

12 / 37

Introduction Literature review Hypothesis PTM Model Capacity Application to RER A

Figure 5: Time multipliers as a function of the in-vehicle density - Source: Haywood and Koning, 2015.

13 / 37

Introduction Literature review Hypothesis PTM Model Capacity Application to RER A

Table 1: Times multipliers as a function of the trip characteristics - Source : Whelan and Crockett, 2009.

14 / 37

Introduction Literature review Hypothesis PTM Model Capacity Application to RER A

Empirical macroeconomic evidences

Valuations of crowding cost at a city scale:

• Prud’homme et al., 2012 : From 2002 to 2007, an 8%

increase in ridership on the Paris subway system caused a welfare loss due to extra crowding of at least e75M/year.

• Veicht et al., 2013 : In 2011, welfare loss of over-crowding in

Melbourne: $280M.

• de Palma, Monchambert and Picard, 2015 : crowding cost in

Paris Region public transport reaches from 11 to 15 millions euros per working day.

15 / 37

Outline

• 1. Literature review
• 2. Hypothesis of the model

2.1 Bottleneck theoretical framework 2.2 Specific to public transport 2.3 Crowding cost function

• 3. PTM Model
• 4. Capacity adjustment
• 5. Application to RER A
• 6. Conclusions

Introduction Literature review Hypothesis PTM Model Capacity Application to RER A

Bottleneck theoretical framework

Hypothesis from Arnott et al., 1990, 1993 :

• N identical individuals commute from A to B every day.
• All individuals have the same preferred arrival time, t∗.
• The strategic variable is the departure time, t : individuals

trade-off between schedule delay cost and congestion cost (extra travel time on road and crowding for public transport).

• Individuals incur a schedule delay cost if they do not arrive at

t∗, denoted δ (t) δ (t∗) = 0,

• δ′ (t) < 0

si t < t∗ δ′ (t) > 0 si t > t∗.

16 / 37

Introduction Literature review Hypothesis PTM Model Capacity Application to RER A

Specific to public transport

• A PT line connects A and B with no intermediate stops.
• Travel time is fixed (and normalized to zero).
• The public transport authority operates m trains, indexed by

departure time k = 1, ..., m (train m leaves last).

• Train k leaves (and arrives) at tk.
• Users know the timetable.
• Departure time decision is discrete.
• There is no upper limit to train capacity.

17 / 37

Introduction Literature review Hypothesis PTM Model Capacity Application to RER A

Crowding cost function

Linear form: g (n) = λn s where

• n is the number of users in the same train,
• s > 0 is a measure of the train capacity (might be m2),
• and λ is a scale parameter.

18 / 37

Outline

• 1. Literature review
• 2. Hypothesis of the model
• 3. PTM Model

3.1 Equilibrium - graphical intuition 3.2 Inefficiency of equilibrium 3.3 Social optimum

• 4. Capacity adjustment
• 5. Application to RER A
• 6. Conclusions

Introduction Literature review Hypothesis PTM Model Capacity Application to RER A

Equilibrium - graphical intuition

t7 t6 t∗ = t5 t4 t3 t2 t1 t Arrival time of PT User private cost

Figure 6: Equilibrium distribution of departure times when m = 7.

19 / 37

Introduction Literature review Hypothesis PTM Model Capacity Application to RER A

Equilibrium - graphical intuition

t7 t6 t∗ = t5 t4 t3 t2 t1 t Arrival time of PT User private cost SDC Scheduling cost

Figure 6: Equilibrium distribution of departure times when m = 7.

19 / 37

Introduction Literature review Hypothesis PTM Model Capacity Application to RER A

Equilibrium - graphical intuition

t7 t6 t∗ = t5 t4 t3 t2 t1 t Arrival time of PT User private cost SDC Scheduling cost ce Crowding cost

Figure 6: Equilibrium distribution of departure times when m = 7.

19 / 37

Introduction Literature review Hypothesis PTM Model Capacity Application to RER A

Inefficiency of equilibrium

In equilibrium, the pattern of departure times is sub-optimal (the marginal social cost of a trip is higher than the private cost). Crowding is a negative externality. It implies two distorsions:

• There are too many users using the facility (if the demand is

elastic).

• Ridership is not optimally distributed over trains.

(Presumption: riders are too concentrated on ”timely” trains).

20 / 37

Introduction Literature review Hypothesis PTM Model Capacity Application to RER A

Social optimum - Graphical intuition

Figure 7: Equilibrium cost and pattern of departure times

21 / 37

Introduction Literature review Hypothesis PTM Model Capacity Application to RER A

Social optimum - Graphical intuition

Figure 8: Optimal private and marginal social costs

22 / 37

Introduction Literature review Hypothesis PTM Model Capacity Application to RER A

Social optimum - Graphical intuition

Figure 9: Equilibrium and optimal patterns of departure times

23 / 37

Introduction Literature review Hypothesis PTM Model Capacity Application to RER A

Social optimum - Main result

The social optimum can be decentralized with a time-dependent fare: τ o

k = no kg′ (no k) .

Users are made worse of by the fare (proof: at least one train is more heavily loaded in the social optimum than in the equilibrium). In the bottleneck model, users are equally well off. The welfare gain from optimal pricing is independent of N, the patronage.

24 / 37

Introduction Literature review Hypothesis PTM Model Capacity Application to RER A

The Parisian subway case

Figure 10: Daily distribution of trips in the Parisian subway during a winter workday - Source: RATP 2008.

25 / 37

Outline

• 1. Literature review
• 2. Hypothesis of the model
• 3. PTM Model
• 4. Capacity adjustment

4.1 Three fare regimes 4.2 Results

• 5. Application to RER A
• 6. Conclusions

Introduction Literature review Hypothesis PTM Model Capacity Application to RER A

Three fare regimes

We maximize the social surplus with respect to the number of trains operated, m and the train capacity, s, under three pricing regimes:

• 1. No pricing (free regime) n,
• 2. Optimal flat pricing (uniform regime) u,
• 3. Optimal time-dependent pricing (optimal regime) o.

26 / 37

Introduction Literature review Hypothesis PTM Model Capacity Application to RER A

Results

Free regime:

• With underpriced usage, latent demand dilutes benefit of

capacity expansion.

27 / 37

Introduction Literature review Hypothesis PTM Model Capacity Application to RER A

Results

Free regime:

• With underpriced usage, latent demand dilutes benefit of

capacity expansion. Uniform regime:

• No welfare loss from latent demand.

27 / 37

Introduction Literature review Hypothesis PTM Model Capacity Application to RER A

Results

Free regime:

• With underpriced usage, latent demand dilutes benefit of

capacity expansion. Uniform regime:

• No welfare loss from latent demand.

Optimal regime:

• Generation of variable extra revenue effectively reduces

marginal cost of expanding s or m. Expanding capacity is more profitable in the optimal regime.

27 / 37

Introduction Literature review Hypothesis PTM Model Capacity Application to RER A

Results

Free regime:

• With underpriced usage, latent demand dilutes benefit of

capacity expansion. Uniform regime:

• No welfare loss from latent demand.

Optimal regime:

• Generation of variable extra revenue effectively reduces

marginal cost of expanding s or m. Expanding capacity is more profitable in the optimal regime. → Congestion pricing and capacity expansion are complementary.

27 / 37

Outline

• 1. Literature review
• 2. Hypothesis of the model
• 3. PTM Model
• 4. Capacity adjustment
• 5. Application to RER A

5.1 RER A, some facts 5.2 Data 5.3 Results 5.4 Short term vs long term 5.5 Many to one network

• 6. Conclusions

Introduction Literature review Hypothesis PTM Model Capacity Application to RER A

RER A, some facts

Figure 11: RER A map

28 / 37

Introduction Literature review Hypothesis PTM Model Capacity Application to RER A

RER A, some facts

RER A, it is:

• More than 300 millions trips per year.
• More than one million trips per working day.
• 60% concentrated on the central segment.

La D´ efense station:

• Every working day, 32 600 users stop at La D´

efense station between 8:25am and 9:25am.

• Average travel time in RER of these users: 40 minutes.

29 / 37

Introduction Literature review Hypothesis PTM Model Capacity Application to RER A

Data

Calibration of the model to describe RER A - heading to La D´ efense station during the morning peak hour:

• K (m, s) = (ν0 + ν1s) m + ν2s (Kraus and Yoshida, 2002)
• β = 7, 4e/h, γ = 17, 2e/h, λ = 4, 4e/user
• h = 2, 5 min
• Nu = 32 600, mu

∗ = 24 et su ∗ = 1 733

• Price-elasticity of the demand -1/3
• Cost recovery ratio = 5/6

− → N0 = 69 003 potential users, ν0 = 936.7 e/train, ν1 = 0.1344 e/train/user, and ν2 = 61.63e· usager.

30 / 37

Introduction Literature review Hypothesis PTM Model Capacity Application to RER A

Results

Free Uniform Optimal Number of trains 25.26 24 26.70 Trains capacity (m2) 1,762 1,733 1,710 Patronage 37,173 32,600 32,907 (-12%) (-11%)

• Ave. scheduling cost (e)

2.1 1.9 2.4

• Ave. crowding cost (e)

4.3 4.1 3.4

• Ave. fare (e)
• 3.5

3.4 Private cost (e) 6.4 9.5 9.2

• Priv. cost - ave. fare (e)

6.4 6 5.8 Social surplus (ke) 1,735 1,743 1,749 Gain/user (e)

• 0.27

0.45

Table 2: Individual costs, fares and surplus in three pricing regimes on RER A during morning peak.

31 / 37

Introduction Literature review Hypothesis PTM Model Capacity Application to RER A

Short term vs long term

Regime change n → u u → o n → o Long term gains +8 714 e +6 076 e +14 788 e (Flexible supply) Short term gains +8 336 e +5 273 e +14 589 e (Fixed supply) Difference ST → LT +4,5% +15,2% +1,2%

Table 3: Social gains due to a change in the pricing regime at long and short terms (n = free, u = uniform and o = optimal).

32 / 37

Introduction Literature review Hypothesis PTM Model Capacity Application to RER A

Many to one network

Figure 12: Departure pattern from three departure stations in equilibrium

33 / 37

Introduction Literature review Hypothesis PTM Model Capacity Application to RER A

Many to one network

Figure 13: Application to the West branch of RER A: from Saint Germain-en-Laye to La D´ efense

34 / 37

Introduction Literature review Hypothesis PTM Model Capacity Application to RER A

Many to one network

Figure 14: Distribution of equilibrium arrival times at La D´ efense

35 / 37

Introduction Literature review Hypothesis PTM Model Capacity Application to RER A

Many to one network

Figure 15: Distribution of equilibrium and optimal arrival times at La D´ efense

36 / 37

Outline

• 1. Literature review
• 2. Hypothesis of the model
• 3. PTM Model
• 4. Capacity adjustment
• 5. Application to RER A
• 6. Conclusions

Introduction Literature review Hypothesis PTM Model Capacity Application to RER A

Conclusions

• 1. A general tractable model of crowding in public transport.
• 2. Microeconomic analysis of this externality.
• 3. Dynamics of crowding differ from the dynamics of road
• congestion. Crowding pricing seems less profitable.
• 4. However, the revenue generated can be used to improve

capacity and service quality.

37 / 37

Thank you!

References

• Arnott, R., De Palma, A., & Lindsey, R. (1990). Economics of a bottleneck. Journal of urban economics, 27(1), 111-130.
• Arnott, R., De Palma, A., & Lindsey, R. (1993). A structural model of peak-period congestion: A traffic bottleneck with

elastic demand. The American Economic Review, 161-179.

• Danielis, R., & Marcucci, E. (2002). Bottleneck road congestion pricing with a competing railroad service. Transportation

Research Part E: Logistics and Transportation Review, 38(5), 379-388.

• de Palma, A., Kilani, M., & Proost, S. (2015). Discomfort in mass transit and its implication for scheduling and pricing.

Transportation Research Part B: Methodological, 71, 1-18.

• Gonzales, E. J., & Daganzo, C. F. (2013). The evening commute with cars and transit: duality results and user equilibrium

for the combined morning and evening peaks. Transportation Research Part B: Methodological, 57, 286-299.

• Haywood, L., & Koning, M. (2015). The distribution of crowding costs in public transport: New evidence from Paris.

Transportation Research Part A: Policy and Practice, 77, 182-201.

• Hensher, D. A., Rose, J. M., & Collins, A. T. (2011). Identifying commuter preferences for existing modes and a proposed

Metro in Sydney, Australia with special reference to crowding. Public Transport, 3(2), 109-147.

• Huang, H. J., Tian, Q., & Gao, Z. Y. (2005). An equilibrium model in urban transit riding and fare polices. In Algorithmic

Applications in Management (pp. 112-121). Springer Berlin Heidelberg.

• Kraus, M., & Yoshida, Y. (2002). The commuter’s time-of-use decision and optimal pricing and service in urban mass
• transit. Journal of Urban Economics, 51(1), 170-195.
• Kroes, E., Kouwenhoven, M., Debrincat, L., & Pauget, N. (2013). On the value of crowding in public transport for

ˆ ıle-de-France. International Transport Forum Discussion Paper.

• Mirabel, F., & Reymond, M. (2011). Bottleneck congestion pricing and modal split: Redistribution of toll revenue.

Transportation Research Part A: Policy and Practice, 45(1), 18-30.

• Mohring, H. (1972). Optimization and scale economies in urban bus transportation. The American Economic Review,

591-604.

• Pepper, J., Spitz, G., & Adler, T. (2003). Customer perspectives on multilevel coaches for increasing rail system capacity.

Transportation Research Record: Journal of the Transportation Research Board, (1838), 19-29.

• Prud’homme, R., Koning, M., Lenormand, L., & Fehr, A. (2012). Public transport congestion costs: The case of the Paris
• subway. Transport Policy, 21, 101-109.
• Tabuchi, T. (1993). Bottleneck congestion and modal split. Journal of Urban Economics, 34(3), 414-431.
• Tian, L. J., Yang, H., & Huang, H. J. (2013). Tradable credit schemes for managing bottleneck congestion and modal

split with heterogeneous users. Transportation Research Part E: Logistics and Transportation Review, 54, 1-13.

• Veitch, T., Partridge, J., & Walker, L. (2013, October). Estimating the Costs of Over-crowding on Melbourne’s Rail
• System. In Australasian Transport Research Forum (ATRF), 36th, 2013, Brisbane, Queensland, Australia.
• Vickrey, W. S. (1969). Congestion theory and transport investment. The American Economic Review, 251-260.
• Wardman, M., & Whelan, G. (2011). Twenty years of rail crowding valuation studies: evidence and lessons from British
• experience. Transport Reviews, 31(3), 379-398.
• Whelan, G. A., & Crockett, J. (2009, March). An investigation of the willingness to pay to reduce rail overcrowding. In

International Choice Modelling Conference 2009.