Agent based modelling of technological diffusion and network influence

Size: px

Start display at page:

Download "Agent based modelling of technological diffusion and network influence"

Linda Kelly
5 years ago
Views:

1 Agent based modelling of technological diffusion and network influence May 4, 23 Martino Tran, D.Phil.(Oxon) Senior Researcher, Environmental Change Institute, University of Oxford

2 Outline Outline. Study motivation 2. Analytical framework & model description 3. Sample results (agent & systems behaviour) 4. Discussion & next steps 2

3 Techno-behavioural dynamic framework Demand Low. Status Quo Extrapolation of current macro and micro level trends in demand and supply. 2. Technological-optimism Radical departure from current trends led by accelerated technological diffusion Supply Low Conservative policy deployment leading to incremental change. 3. Behavioural change Focuses on changing individual behaviours and lifestyles (consumption, practices, norms) Partly reactionary to the conventional focus on technological solutions without social context. Backed by government and industry investment. 4. Techno-behavioural Dynamics Strategic decision-making and adoption behaviour as a mechanism for behavioural change. Technology performance and behaviour as co-evolving coupled dynamical system. Supply High Demand High 3

4 Implementing the approach general model description System Behaviour S = ƒ (Φ, λ, T, P) Technology Change T = ƒ (Ω, λ T, X) Decision Analysis P = ƒ(x, β, N) Network Influence N = ƒ (Q, K) Agent Behaviour A = ƒ (P, N, Ψ) 4

5 Conventional technology diffusion models mean field approaches dp(t) dt = αp(t) ( P(t) K ) Feedback key but no other explanatory variables dn(t) dt = M n t p + q M n t, n = Additional parameters important but need to disaggregate, p & q Diffusion = ƒ(a, T, P, N) 5

6 ABM derivation need to account for variation in individual behaviour and balancing between internal and external influence ABMs with a general binomial form*: Prob(t) = ( p) * ( q) ^ k(t) t() = no adopt t + = prob(t) > U adopt Re-specify model to account for technology trade-off behaviour i.e. index into matrix of options, j and influences (P, Q, K) specific to each individual, i : Prob(t) = ( P ij ) * ( Q ij ) ^ K ij (t) *Source: Goldenberg & Shapira, 29 6

7 Monte Carlo simulations used for assessing = unknown individual preferences (P ij ) Static approach, Dynamic approach, P ni = e β X ni J e β X nj j= + P ni θ = L ni β ƒ β θ) d β, HEV s outcompete ICE s, survey bias? L ni β = exp(β x ni )/ exp( β x ni ) j Accounts for heterogeneity, better approximation 7

8 Co-evolution of technological performance and agent behaviour Feedback effect between technological change and agent preferences over time How do different feedback effects scale up to impact the larger technological system? 8

9 Simple assumptions made on network influence (Q ij, K ij ) Indirect influence, Q ij, based on value estimate, V of previous adopters, n k out of a local population, n, Q ij = V(p k ), p k = n k /n Direct influence, K ij based on previous adopters out of personal contacts, w ij that has influence, y i on an individual, Therefore the master function is, K ij (t) = w ij y i / w i Prob(t) = ( [ R R r= L ij ( r )]) * ( [n k /n]) ^ ( w ij y i / w i ) See Tran, 22 for full derivation 9

10 Technology data and assumptions Technologies Purchase Price (2 USD) Fuel Price (USD/ 6 km) Fuel Consumption (L/ km) Performance (Acceleration - km/hr in seconds) Range (Km on tank/charge) Environment (Annual WTW GHG CO 2 -eq emission, metric tonnes) Petrol Diesel HEV PHEV BEV FC Refuelling Availability (%) Rewire for random and small world network influence and sensitivity analysis Simulations P Q K Sources: See Tran, M. 22 for full references

11 Probability Probability Probability Probability Captures heterogeneous behaviour, distribution and trend effects over time A Petrol Diesel HEV PHEV BEV FC B Petrol Diesel HEV PHEV BEV FC C Petrol Diesel HEV PHEV BEV FC D Petrol Diesel HEV PHEV BEV FC A) Mass market (price, reliability), B) Early adopter (CO 2, fuel economy) C) Trade-offs (performance vs. environment) D) Exogenous effects (refuelling infrastructure)

12 Total Total Total Total High variability during early phases of diffusion, better approximation to empirical evidence A C y =.e-6*x *x 3 +.4*x 2 -.2*x +.57 Norm of residuals = 6.6 Data 4th degree y = - 6.7e-7*x 4 +.4*x 3 -.*x *x -.7 Norm of residuals = B D y = - 4.9e-7*x e-5*x 3 -.5*x *x -.99 Norm of residuals = Data 4th degree y = -.7e-7*x e-5*x *x 2 +.4*x - Norm of residuals = Data 4th degree Data 4th degree Battery electric vehicle preferences: A) Mass market (price, reliability), B) Early adopter (CO 2, fuel economy) C) Trade-offs (acceleration) D) Exogenous effects (increasing refuelling). Network effects held constant 2

13 Total Total Total Total ABM can capture a range of outcomes more consistent with empirical evidence showing multiple trajectories A 35 3 LNG B C 4 x 4 2 CNG C B 6 x 4 ELE D 5 x 5 E Cumulative adoption for A) Liquefied natural gas (LNG), B) Compressed natural gas (CNG), C) Full battery electric, D) Ethanol-85 (E85), US (DOE, 28) 3

14 Total Total Accounting for evolving social network influence clustering effects for indirect and direct network influence A 5 A B QA =.4 QB =.5 QC =.6 QD =.7 KA =.4 KB =.5 KC =.6 KD =.7 A) Increasing indirect influence (Q), direct influence held constant (K=.), low individual preference (P=.2) held constant B) Assume personal contacts (K) reflect local population behaviour and exerts influence on individual Combined influence can have non-linear effects on individual adoption 4

15 Total Sensitivity analysis on all parameters 3 Parameters S S2 S3 S4 S5 2 P Q K Cumulative adoption range S = Ref, S2 = Pref, S3 = Indirect, S4 =Direct, S5 = Combined High variability in short term but clear diffusion pattern over longer term Combined network effects can have as much influence as individual preference Indirect influence can have greater effect than direct personal contacts ( influence through numbers effect ) Indicates a mechanism to change risk averse behaviour? 5

16 Constraints: Discussion and next steps Lack of network data (esp. on high investment technologies e.g. energy technologies) Potential: Framework has wide ranging applications for technological and behavioural interactions ABM can give a distribution of possible outcomes instead of locking into single trajectories ( magic number solutions ), can deal with future uncertainty rather than prediction How do agents balance internal strategy/preferences (profit maximization for service provision) and external influence (networked behaviour of competitors, changing consumer preferences, macrolevel signals e.g. investment climate, policy incentives, etc.) How do personal (or firm-level) networks evolve with population level behaviour (larger network of service providers)? What is the direction and magnitude of the feedback effect? Can individual behaviour (habits, risk aversion, preferences, etc.) be outweighed by external network influence over time? What network topologies amplify behavioural signals at the individual, local and population levels? How does agent behaviour scale up to impact the larger system e.g. disruptive technological change? 6

17 Thank you! Questions? Contact: 7

Competitive Equilibrium

Competitive Equilibrium Econ 2100 Fall 2017 Lecture 16, October 26 Outline 1 Pareto Effi ciency 2 The Core 3 Planner s Problem(s) 4 Competitive (Walrasian) Equilibrium Decentralized vs. Centralized Economic