Kuhn-Tucker and Multiple Discrete-Continuous Extreme Value (MDCEV) Model Estimation and Simulation in R: The rmdcev Package
The rmdcev R package estimate and simulates Kuhn-Tucker demand models with individual heterogeneity. The models supported by rmdcev are the multiple-discrete continuous extreme value (MDCEV) model and Kuhn-Tucker specification common in the environmental economics literature on recreation demand. Latent class and random parameters specifications can be implemented and the models are fit using maximum likelihood estimation or Bayesian estimation. All models are implemented in Stan, which is a C++ package for performing full Bayesian inference (see http://mc-stan.org/). The rmdcev package also implements demand forecasting and welfare calculation for policy simulation.
Development is in progress. Currently users can estimate the following models:
- Bhat (2008) MDCEV model specifications
- Kuhn-Tucker model specification in environmental economics (von Haefen and Phaneuf, 2005)
Models can be estimated using
- Fixed parameter models (maximum likelihood or Bayesian estimation)
- Latent class models (maximum likelihood estimation)
- Random parameters models (Bayesian estimation)
I recommend you first install rstan by following these steps:
https://github.com/stan-dev/rstan/wiki/RStan-Getting-Started
Once rstan is installed, you can install the released version of rmdcev from GitHub using devtools
if (!require(devtools)) {
install.packages("devtools")
library(devtools)
}
install_github("plloydsmith/rmdcev", build_vignettes = FALSE, INSTALL_opts="--no-multiarch")
If you have any issues with installation or use of the package, please let me know by filing an issue.
For more details on the model specification and estimation:
Bhat, C.R. (2008) “The Multiple Discrete-Continuous Extreme Value (MDCEV) Model: Role of Utility Function Parameters, Identification Considerations, and Model Extensions” Transportation Research Part B, 42(3): 274-303.
von Haefen, R. and Phaneuf D. (2005) “Kuhn-Tucker Demand System Approaches to Non-Market Valuation” In: Scarpa R., Alberini A. (eds) Applications of Simulation Methods in Environmental and Resource Economics. The Economics of Non-Market Goods and Resources, vol 6. Springer, Dordrecht.
For more details on the demand and welfare simulation:
Pinjari, A.R. and Bhat , C.R. (2011) “Computationally Efficient Forecasting Procedures for Kuhn-Tucker Consumer Demand Model Systems: Application to Residential Energy Consumption Analysis.” Technical paper, Department of Civil & Environmental Engineering, University of South Florida.
Lloyd-Smith, P (2018). “A New Approach to Calculating Welfare Measures in Kuhn-Tucker Demand Models.” Journal of Choice Modeling, 26: 19-27
As an example, we can simulate some data using Bhat (2008)‘s ’Gamma’ specification. In this example, we are simulating data for 2,000 individuals and 10 non-numeraire alternatives. We will randomly generate the parameter values to simulate the data and then check these values to our estimation results.
library(pacman)
p_load(tidyverse, rmdcev)
model <- "gamma"
nobs <- 2000
nalts <- 10
sim.data <- GenerateMDCEVData(model = model, nobs = nobs, nalts = nalts)
#> Sorting data by id.var then alt...
#> Checking data...
#> Data is good
Estimate model using MLE (note that we set “psi_ascs = 0” to omit any alternative-specific constants)
mdcev_est <- mdcev(~ b1 + b2 + b3 + b4 + b5 + b6,
data = sim.data$data,
psi_ascs = 0,
model = model,
algorithm = "MLE")
#> Using MLE to estimate KT model
#> Chain 1: Initial log joint probability = -102091
#> Chain 1: Iter log prob ||dx|| ||grad|| alpha alpha0 # evals Notes
#> Chain 1: Error evaluating model log probability: Non-finite gradient.
#> Error evaluating model log probability: Non-finite gradient.
#>
#> Chain 1: 19 -36191.9 0.238697 170.87 1 1 33
#> Chain 1: Iter log prob ||dx|| ||grad|| alpha alpha0 # evals Notes
#> Chain 1: 39 -36099.7 0.09316 104.396 1 1 54
#> Chain 1: Iter log prob ||dx|| ||grad|| alpha alpha0 # evals Notes
#> Chain 1: 59 -36075.1 0.0173156 53.9469 0.3078 0.3078 80
#> Chain 1: Iter log prob ||dx|| ||grad|| alpha alpha0 # evals Notes
#> Chain 1: 79 -36072.3 0.0278811 12.2672 1 1 103
#> Chain 1: Iter log prob ||dx|| ||grad|| alpha alpha0 # evals Notes
#> Chain 1: 99 -36072.2 0.00180446 0.959796 1 1 124
#> Chain 1: Iter log prob ||dx|| ||grad|| alpha alpha0 # evals Notes
#> Chain 1: 107 -36072.2 0.000747187 0.522778 0.9515 0.9515 135
#> Chain 1: Optimization terminated normally:
#> Chain 1: Convergence detected: relative gradient magnitude is below tolerance
Summarize results
summary(mdcev_est)
#> Model run using rmdcev for R, version 1.2.1
#> Estimation method : MLE
#> Model type : gamma specification
#> Number of classes : 1
#> Number of individuals : 2000
#> Number of non-numeraire alts : 10
#> Estimated parameters : 18
#> LL : -36072.19
#> AIC : 72180.38
#> BIC : 72281.2
#> Standard errors calculated using : Delta method
#> Exit of MLE : successful convergence
#> Time taken (hh:mm:ss) : 00:00:2.85
#>
#> Average consumption of non-numeraire alternatives:
#> 1 2 3 4 5 6 7 8 9 10
#> 7.18 0.95 16.85 11.57 8.81 58.15 18.51 17.74 0.30 6.56
#>
#> Parameter estimates --------------------------------
#> Estimate Std.err z.stat
#> psi_b1 -5.012 0.126 -39.78
#> psi_b2 0.484 0.082 5.91
#> psi_b3 2.107 0.074 28.47
#> psi_b4 -1.513 0.055 -27.52
#> psi_b5 2.015 0.047 42.86
#> psi_b6 -1.027 0.053 -19.38
#> gamma_1 5.673 0.480 11.82
#> gamma_2 1.799 0.207 8.69
#> gamma_3 2.237 0.140 15.98
#> gamma_4 2.440 0.147 16.60
#> gamma_5 6.482 0.527 12.30
#> gamma_6 8.840 0.512 17.27
#> gamma_7 2.684 0.156 17.20
#> gamma_8 8.299 0.573 14.48
#> gamma_9 6.507 1.892 3.44
#> gamma_10 3.867 0.313 12.36
#> alpha_num 0.501 0.008 62.67
#> scale 1.006 0.015 67.04
#> Note: All non-numeraire alpha's fixed to 0.
Compare estimates to true values
coefs <- as_tibble(sim.data$parms_true) %>%
mutate(true = as.numeric(true)) %>%
cbind(summary(mdcev_est)[["CoefTable"]]) %>%
mutate(cl_lo = Estimate - 1.96 * Std.err,
cl_hi = Estimate + 1.96 * Std.err)
head(coefs, 200)
#> parms true Estimate Std.err z.stat cl_lo cl_hi
#> 1 psi_b1 -5.000000 -5.012 0.126 -39.78 -5.25896 -4.76504
#> 2 psi_b2 0.500000 0.484 0.082 5.91 0.32328 0.64472
#> 3 psi_b3 2.000000 2.107 0.074 28.47 1.96196 2.25204
#> 4 psi_b4 -1.500000 -1.513 0.055 -27.52 -1.62080 -1.40520
#> 5 psi_b5 2.000000 2.015 0.047 42.86 1.92288 2.10712
#> 6 psi_b6 -1.000000 -1.027 0.053 -19.38 -1.13088 -0.92312
#> 7 gamma1 5.880124 5.673 0.480 11.82 4.73220 6.61380
#> 8 gamma2 2.037718 1.799 0.207 8.69 1.39328 2.20472
#> 9 gamma3 2.326599 2.237 0.140 15.98 1.96260 2.51140
#> 10 gamma4 2.378944 2.440 0.147 16.60 2.15188 2.72812
#> 11 gamma5 6.188133 6.482 0.527 12.30 5.44908 7.51492
#> 12 gamma6 9.298275 8.840 0.512 17.27 7.83648 9.84352
#> 13 gamma7 2.845774 2.684 0.156 17.20 2.37824 2.98976
#> 14 gamma8 7.553436 8.299 0.573 14.48 7.17592 9.42208
#> 15 gamma9 8.992650 6.507 1.892 3.44 2.79868 10.21532
#> 16 gamma10 3.494510 3.867 0.313 12.36 3.25352 4.48048
#> 17 alpha1 0.500000 0.501 0.008 62.67 0.48532 0.51668
#> 18 scale 1.000000 1.006 0.015 67.04 0.97660 1.03540
Compare outputs using a figure
coefs %>%
ggplot(aes(y = Estimate, x = true)) +
geom_point(size=2) +
geom_text(label=coefs$parms,position=position_jitter(width=.5,height=1)) +
geom_abline(slope = 1) +
geom_errorbar(aes(ymin=cl_lo,ymax=cl_hi,width=0.2))
Create policy simulations (these are ‘no change’ policies with no effects)
npols <- 2 # Choose number of policies
policies <- CreateBlankPolicies(npols, mdcev_est)
df_sim <- PrepareSimulationData(mdcev_est, policies, nsims = 1)
Simulate welfare changes
wtp <- mdcev.sim(df_sim$df_indiv,
df_common = df_sim$df_common,
sim_options = df_sim$sim_options,
cond_err = 1,
nerrs = 15,
sim_type = "welfare")
#> Using general approach in simulation...
#>
#> 6.00e+04simulations finished in0.4minutes.(2477per second)
summary(wtp)
#> # A tibble: 2 x 5
#> policy mean std.dev `ci_lo2.5%` `ci_hi97.5%`
#> <chr> <dbl> <dbl> <dbl> <dbl>
#> 1 policy1 4.19e-11 NA 4.19e-11 4.19e-11
#> 2 policy2 4.19e-11 NA 4.19e-11 4.19e-11