PUNTO C
par(mfrow = c(1, 2))
# Heterogeneidad individual
plotmeans(crmrte ~ county, main = "Heterogeneidad Individual",
data = crime4_filtrada, xlab = "Condado", ylab = "Tasa de delincuencia")
# Heterogeneidad temporal
plotmeans(crmrte ~ year, main = "Heterogeneidad Temporal",
data = crime4_filtrada, xlab = "Año", ylab = "Tasa de delincuencia")
# Crear un data frame con la tasa promedio de crimen por condado
crime_año_condado <- crime4_filtrada %>%
group_by(county, year) %>%
summarise(crmrte = mean(crmrte, na.rm = TRUE)) %>%
ungroup()
# Filtrar los condados con los niveles más altos de crimen en promedio
top_condados <- crime_año_condado %>%
group_by(county) %>%
summarise(prom_crimen = mean(crmrte, na.rm = TRUE)) %>%
top_n(5, prom_crimen) # Se seleccionan los 5 condados con mayor crimen
ggplot(crime_año_condado, aes(x = year, y = crmrte, group = county, color = as.factor(county))) +
geom_line(alpha = 0.5) +
gghighlight(county %in% top_condados$county, label_key = county) +
labs(title = "Evolución de la Tasa de Crimen por Condado (1981-1987)",
x = "Año",
y = "Tasa de Crimen per cápita",
color = "Condado") +
theme_minimal()
crime4_filtrada <- crime4_filtrada %>%
mutate(
log_crmrte = log(crmrte),
log_prbarr = log(prbarr),
log_prbconv = log(prbconv),
log_prbpris = log(prbpris),
log_avgsen = log(avgsen),
log_polpc = log(polpc)
)
# Convertir year en factor para las dummies
crime4_filtrada$year_factor <- as.factor(crime4_filtrada$year)
crime4_filtrada <- crime4_filtrada %>%
filter(!is.na(log_crmrte))
# Estimación por MCO (MÃnimos Cuadrados Ordinarios)
modelo_ols <- lm(log_crmrte ~ year_factor + log_prbarr + log_prbconv + log_prbpris + log_avgsen + log_polpc,
data = crime4_filtrada)
stargazer(modelo_ols, type = "html")
|
|
|
|
Dependent variable:
|
|
|
|
|
|
log_crmrte
|
|
|
|
year_factor82
|
0.005
|
|
|
(0.058)
|
|
|
|
|
year_factor83
|
-0.044
|
|
|
(0.058)
|
|
|
|
|
year_factor84
|
-0.109*
|
|
|
(0.058)
|
|
|
|
|
year_factor85
|
-0.078
|
|
|
(0.058)
|
|
|
|
|
year_factor86
|
-0.042
|
|
|
(0.058)
|
|
|
|
|
year_factor87
|
-0.027
|
|
|
(0.057)
|
|
|
|
|
log_prbarr
|
-0.720***
|
|
|
(0.037)
|
|
|
|
|
log_prbconv
|
-0.546***
|
|
|
(0.026)
|
|
|
|
|
log_prbpris
|
0.248***
|
|
|
(0.067)
|
|
|
|
|
log_avgsen
|
-0.087
|
|
|
(0.058)
|
|
|
|
|
log_polpc
|
0.366***
|
|
|
(0.030)
|
|
|
|
|
Constant
|
-2.082***
|
|
|
(0.252)
|
|
|
|
|
|
|
Observations
|
630
|
|
R2
|
0.570
|
|
Adjusted R2
|
0.562
|
|
Residual Std. Error
|
0.379 (df = 618)
|
|
F Statistic
|
74.485*** (df = 11; 618)
|
|
|
|
Note:
|
p<0.1; p<0.05;
p<0.01
|
PUNTO D
modelo_fd <- plm(log_crmrte ~ year_factor + log_prbarr + log_prbconv + log_prbpris + log_avgsen + log_polpc,
data = crime4_filtrada,
model = "fd")
stargazer(modelo_fd, type = "html")
|
|
|
|
Dependent variable:
|
|
|
|
|
|
log_crmrte
|
|
|
|
year_factor82
|
0.014
|
|
|
(0.016)
|
|
|
|
|
year_factor83
|
-0.072***
|
|
|
(0.019)
|
|
|
|
|
year_factor84
|
-0.106***
|
|
|
(0.021)
|
|
|
|
|
year_factor85
|
-0.096***
|
|
|
(0.020)
|
|
|
|
|
year_factor86
|
-0.055***
|
|
|
(0.015)
|
|
|
|
|
log_prbarr
|
-0.327***
|
|
|
(0.030)
|
|
|
|
|
log_prbconv
|
-0.238***
|
|
|
(0.018)
|
|
|
|
|
log_prbpris
|
-0.165***
|
|
|
(0.026)
|
|
|
|
|
log_avgsen
|
-0.022
|
|
|
(0.022)
|
|
|
|
|
log_polpc
|
0.398***
|
|
|
(0.027)
|
|
|
|
|
Constant
|
-0.006
|
|
|
(0.007)
|
|
|
|
|
|
|
Observations
|
540
|
|
R2
|
0.433
|
|
Adjusted R2
|
0.422
|
|
F Statistic
|
40.318*** (df = 10; 529)
|
|
|
|
Note:
|
p<0.1; p<0.05;
p<0.01
|
PUNTO F
# Efectos fijos
modelo_fe <- plm(log_crmrte ~ year_factor + log_prbarr + log_prbconv + log_prbpris + log_avgsen + log_polpc,
data = crime4_filtrada, model = "within")
stargazer(modelo_fe, type = "html")
|
|
|
|
Dependent variable:
|
|
|
|
|
|
log_crmrte
|
|
|
|
year_factor82
|
0.013
|
|
|
(0.022)
|
|
|
|
|
year_factor83
|
-0.079***
|
|
|
(0.021)
|
|
|
|
|
year_factor84
|
-0.118***
|
|
|
(0.022)
|
|
|
|
|
year_factor85
|
-0.112***
|
|
|
(0.022)
|
|
|
|
|
year_factor86
|
-0.082***
|
|
|
(0.021)
|
|
|
|
|
year_factor87
|
-0.040*
|
|
|
(0.021)
|
|
|
|
|
log_prbarr
|
-0.360***
|
|
|
(0.032)
|
|
|
|
|
log_prbconv
|
-0.286***
|
|
|
(0.021)
|
|
|
|
|
log_prbpris
|
-0.183***
|
|
|
(0.032)
|
|
|
|
|
log_avgsen
|
-0.004
|
|
|
(0.026)
|
|
|
|
|
log_polpc
|
0.424***
|
|
|
(0.026)
|
|
|
|
|
|
|
Observations
|
630
|
|
R2
|
0.434
|
|
Adjusted R2
|
0.327
|
|
F Statistic
|
36.911*** (df = 11; 529)
|
|
|
|
Note:
|
p<0.1; p<0.05;
p<0.01
|
# Efectos aleatorios
modelo_re <- plm(log_crmrte ~ year_factor + log_prbarr + log_prbconv + log_prbpris + log_avgsen + log_polpc,
data = crime4_filtrada, model = "random")
stargazer(modelo_re, type = "html")
|
|
|
|
Dependent variable:
|
|
|
|
|
|
log_crmrte
|
|
|
|
year_factor82
|
0.014
|
|
|
(0.023)
|
|
|
|
|
year_factor83
|
-0.075***
|
|
|
(0.023)
|
|
|
|
|
year_factor84
|
-0.113***
|
|
|
(0.023)
|
|
|
|
|
year_factor85
|
-0.106***
|
|
|
(0.023)
|
|
|
|
|
year_factor86
|
-0.080***
|
|
|
(0.023)
|
|
|
|
|
year_factor87
|
-0.042*
|
|
|
(0.022)
|
|
|
|
|
log_prbarr
|
-0.425***
|
|
|
(0.032)
|
|
|
|
|
log_prbconv
|
-0.327***
|
|
|
(0.021)
|
|
|
|
|
log_prbpris
|
-0.179***
|
|
|
(0.034)
|
|
|
|
|
log_avgsen
|
-0.008
|
|
|
(0.028)
|
|
|
|
|
log_polpc
|
0.429***
|
|
|
(0.026)
|
|
|
|
|
Constant
|
-1.673***
|
|
|
(0.175)
|
|
|
|
|
|
|
Observations
|
630
|
|
R2
|
0.426
|
|
Adjusted R2
|
0.416
|
|
F Statistic
|
459.169***
|
|
|
|
Note:
|
p<0.1; p<0.05;
p<0.01
|
stargazer(modelo_ols, modelo_fd, modelo_fe, modelo_re, type = "html")
|
|
|
|
Dependent variable:
|
|
|
|
|
|
log_crmrte
|
|
|
OLS
|
panel
|
|
|
|
linear
|
|
|
(1)
|
(2)
|
(3)
|
(4)
|
|
|
|
year_factor82
|
0.005
|
0.014
|
0.013
|
0.014
|
|
|
(0.058)
|
(0.016)
|
(0.022)
|
(0.023)
|
|
|
|
|
|
|
|
year_factor83
|
-0.044
|
-0.072***
|
-0.079***
|
-0.075***
|
|
|
(0.058)
|
(0.019)
|
(0.021)
|
(0.023)
|
|
|
|
|
|
|
|
year_factor84
|
-0.109*
|
-0.106***
|
-0.118***
|
-0.113***
|
|
|
(0.058)
|
(0.021)
|
(0.022)
|
(0.023)
|
|
|
|
|
|
|
|
year_factor85
|
-0.078
|
-0.096***
|
-0.112***
|
-0.106***
|
|
|
(0.058)
|
(0.020)
|
(0.022)
|
(0.023)
|
|
|
|
|
|
|
|
year_factor86
|
-0.042
|
-0.055***
|
-0.082***
|
-0.080***
|
|
|
(0.058)
|
(0.015)
|
(0.021)
|
(0.023)
|
|
|
|
|
|
|
|
year_factor87
|
-0.027
|
|
-0.040*
|
-0.042*
|
|
|
(0.057)
|
|
(0.021)
|
(0.022)
|
|
|
|
|
|
|
|
log_prbarr
|
-0.720***
|
-0.327***
|
-0.360***
|
-0.425***
|
|
|
(0.037)
|
(0.030)
|
(0.032)
|
(0.032)
|
|
|
|
|
|
|
|
log_prbconv
|
-0.546***
|
-0.238***
|
-0.286***
|
-0.327***
|
|
|
(0.026)
|
(0.018)
|
(0.021)
|
(0.021)
|
|
|
|
|
|
|
|
log_prbpris
|
0.248***
|
-0.165***
|
-0.183***
|
-0.179***
|
|
|
(0.067)
|
(0.026)
|
(0.032)
|
(0.034)
|
|
|
|
|
|
|
|
log_avgsen
|
-0.087
|
-0.022
|
-0.004
|
-0.008
|
|
|
(0.058)
|
(0.022)
|
(0.026)
|
(0.028)
|
|
|
|
|
|
|
|
log_polpc
|
0.366***
|
0.398***
|
0.424***
|
0.429***
|
|
|
(0.030)
|
(0.027)
|
(0.026)
|
(0.026)
|
|
|
|
|
|
|
|
Constant
|
-2.082***
|
-0.006
|
|
-1.673***
|
|
|
(0.252)
|
(0.007)
|
|
(0.175)
|
|
|
|
|
|
|
|
|
|
Observations
|
630
|
540
|
630
|
630
|
|
R2
|
0.570
|
0.433
|
0.434
|
0.426
|
|
Adjusted R2
|
0.562
|
0.422
|
0.327
|
0.416
|
|
Residual Std. Error
|
0.379 (df = 618)
|
|
|
|
|
F Statistic
|
74.485*** (df = 11; 618)
|
40.318*** (df = 10; 529)
|
36.911*** (df = 11; 529)
|
459.169***
|
|
|
|
Note:
|
p<0.1; p<0.05;
p<0.01
|
PUNTO G
options(scipen = 999)
hausman_test <- phtest(modelo_fe, modelo_re)
print(hausman_test)
##
## Hausman Test
##
## data: log_crmrte ~ year_factor + log_prbarr + log_prbconv + log_prbpris + ...
## chisq = 36.924, df = 11, p-value = 0.0001187
## alternative hypothesis: one model is inconsistent