文献来源
- Fleten, S.-E., et al. (2024). The reliability pricing model and coal-fired generators in PJM
- Appendix B. Supplementary data【数据+Stata】
示例代码
cd C:\Download\1-s2.0-S0140988324002512-mmc1\ffuDataCodeEE
clear
import delim using kmeans_input_nrg_cap_Coal.csv
replace cappmt = cappmt*(365/1000)
replace cappmtdiff = cappmtdiff*(365/1000)
replace demgth = demgth*10000
gen coalng = ngfd - p_f
egen min_inyr = min(inyr)
gen inyr2 = inyr
replace inyr = inyr-min_inyr
encode state, gen(state2)
tsset plantgen year
gen abandon6 = 1 if (decision==5 | decision==6) & naics==22
replace abandon6 = 0 if (decision==1 | decision==2 | decision==3 | decision==4) & naics==22
gen capf=f.capacity
replace capf=0 if decision==6 | decision==5
keep if year<2019 & year>2007
gen cappmtdp=0
gen cappmtdn=0
gen pickp = (cappmtdiff>0)
gen pickn = (cappmtdiff<0)
replace cappmtdp = cappmtdiff if cappmtdiff>0
replace cappmtdn = cappmtdiff if cappmtdiff<0
egen mn_xtr = mean(xtreal), by(year zone)
gen dev_xtr = xtreal - mn_xtr
melogit abandon6 mn_xtr dev_xtr cappmtdiff inyr capacity demgth newent || pcode:, vce(robust) or
得到结果
Fitting fixed-effects model:
Iteration 0: Log likelihood = -411.32349
Iteration 1: Log likelihood = -367.45373
Iteration 2: Log likelihood = -366.14862
Iteration 3: Log likelihood = -366.13745
Iteration 4: Log likelihood = -366.13745
Refining starting values:
Grid node 0: Log likelihood = -323.46369
Fitting full model:
Iteration 0: Log pseudolikelihood = -323.46369
Iteration 1: Log pseudolikelihood = -302.15001
Iteration 2: Log pseudolikelihood = -295.03556
Iteration 3: Log pseudolikelihood = -293.39972
Iteration 4: Log pseudolikelihood = -293.3419
Iteration 5: Log pseudolikelihood = -293.34723
Iteration 6: Log pseudolikelihood = -293.34822
Iteration 7: Log pseudolikelihood = -293.34835
Iteration 8: Log pseudolikelihood = -293.34837
Mixed-effects logistic regression Number of obs = 1,701
Group variable: pcode Number of groups = 97
Obs per group:
min = 1
avg = 17.5
max = 77
Integration method: mvaghermite Integration pts. = 7
Wald chi2(7) = 46.27
Log pseudolikelihood = -293.34837 Prob > chi2 = 0.0000
(Std. err. adjusted for 97 clusters in pcode)
------------------------------------------------------------------------------
| Robust
abandon6 | Odds ratio std. err. z P>|z| [95% conf. interval]
-------------+----------------------------------------------------------------
mn_xtr | .9601638 .0080005 -4.88 0.000 .9446104 .9759732
dev_xtr | .9830503 .0117958 -1.42 0.154 .9602007 1.006444
cappmtdiff | 1.067458 .0221167 3.15 0.002 1.024978 1.111698
inyr | .9334783 .0186032 -3.45 0.001 .8977195 .9706614
capacity | .997196 .001381 -2.03 0.043 .9944929 .9999063
demgth | .9775999 .0087958 -2.52 0.012 .9605117 .9949922
newent | 1.000881 .0008661 1.02 0.309 .9991852 1.00258
_cons | .1197382 .0946848 -2.68 0.007 .0254175 .5640703
-------------+----------------------------------------------------------------
pcode |
var(_cons)| 9.909341 3.781994 4.69003 20.93697
------------------------------------------------------------------------------
Note: Estimates are transformed only in the first equation to odds ratios.
Note: _cons estimates baseline odds (conditional on zero random effects).
期刊排版
（完）
