You are on page 1of 4

Tarun Singh

Worked With:
Alexander Marcus and Alex Sloan
Economics 1123, Problem Set 4

1)
Table 1: Determinants of Terrorism, ctd.
(6) (7) (8) (9)
Dependent variable: lnftmpop lnftmpop lnftmpop lnftmpop
Regressor:
higdppc -2.8379** -3.8441 -.3837 -.5926
(.9987) (2.4057) (1.0421) (1.0844)
lackpf -.2482 .1312 1.6218** 1.9774**
(.1649) (1.0601) (.5245) (.5284)
lackpf2 __ -.0403 -.1886** -.2280**
(.1137) (.0615) (.0604)
higdppc × lackpf .6449** 1.8890 __ __
(.2377) (1.2161)
higdppc × lackpf2 __ -.1959 __ __
(.1407)
ethnic __ __ 1.4617 1.2365
(1.0450) (.9956)
religion __ __ -3.0074* -3.4056*
(1.4180) (1.6494)
higdppc × ethnic __ __ -1.2973 -.5015
(1.6976) (1.8009)
higdppc × religion __ __ 2.6547 2.5105
(1.8997) (2.0183)
Mideast __ __ __ -1.2532
(1.3117)
Other regional dummies (latinam, No No No Yes
easteurope, africa, eastasia)?
Intercept -.7172 -1.5174 -4.3760** -3.7822*
(.8423) (2.2381) (1.2595) (1.5458)
F-statistics testing the hypothesis that the population coefficients on the indicated
regressors are all zero:
higdppc × lackpf, __ 1.34 __ __
higdppc × lackpf 2 (.2689)
lackpf2, higdppc × lackpf2 __ 4.12 __ __
(.0204)
lackpf, lackpf2 __ 1.10 4.81 7.21
(.3390) (.0112) (0.0016)
higdppc × ethnic, higdppc × religion __ __ 1.24 0.79
(.2970) (0.4606)
ethnic, religion, __ __ 1.45 1.66
higdppc × ethnic, higdppc × religion (.2266) (0.1705)
Other regional dummies __ __ __ 0.63
(0.6422)
Regression summary statistics:
R2 0.0854 0.1311 0.1168 0.0773
R2 0.1220 0.1891 0.2015 0.2290
SER 1.8077 1.7619 1.7694 1.8085
n 76 76 74 74
Notes: Heteroskedasticity-robust standard errors are given in parentheses
under estimated coefficients, and p-values are given in parentheses under F-
statistics. The F-statistics are heteroskedasticity-robust. Coefficients are
individually statistically significant at the +10%, *5%, **1% significance level.
The “other regional dummies” included in regression (5) are latinam,
easteurope, africa, and eastasia (the omitted case is Western Europe
combined with North America)

2)
higdppc=1
lnftmpop = -3.555136 + .3966958(lackpf)
(.5366) (.1712)
2
0
-2
-4
-6

0 2 4 6 8
lackpf

Fitted values lnftmpop

higdppc=0
lnftmpop = -.717181 + -.248253 (lackpf)
(.8423) (.1649)
2
0
-2
-4
-6

2 3 4 5 6 7
lackpf

Fitted values lnftmpop


3)
a) The difference between the two slopes plotted in Question 2 is given by
higdppc x lackpf which has a t-value of 2.71 which is significant at the 5%
level meaning that the slopes are different.

b) When all else is constant, as countries with high gdp per capita
(higdppc=1) lose political freedoms, they experience more terrorism related
deaths, and as countries with low gdp per capita (higdppc=0) lose political
freedoms they experience more terrorism related deaths.

c) Null hypothesis: the coefficients on lackpf2 and higdppc x lackpf2 are equal
to 0.
Alternative hypothesis: at least one of the coefficients is not equal to 0.
We are testing whether or not the two quadratic fits, which fit lnftmpop and
lackpf for higdppc=1 and higdppc=0, are equal, meaning wheter or not they
are the same parabola.
Using an f-test we get an f-value of 1.34 and a p-value of .2689 which is
greater than .05 meaning we fail to reject the null hypothesis that all the
coefficients are equal to 0.

d) Null hypothesis: the coefficients on lackpf2 and higdppc x lackpf2 are


equal to 0.
Alternative hypothesis: at least one of the coefficients is not equal to 0.
We are testing whether or not the relationship between lnftmpop and lackpf is
linear and not quadratic.
Using an f-test we get an f-value of 4.12 and a p-value of .0204 which is less
than .05 meaning we reject the null hypotheses that the coefficients on
lackpf2 and higdppc x lackpf2 are equal to 0.

e) No, the coefficient on the dummy variables is not statistically significant at


the 5% significance level.
Null hypothesis: all the coefficients are equal to 0.
Alternative hypothesis: at least one coefficient is not equal to 0.
Using an f-test we get an f-value of .63 and a p-value of .6422 which is
greater than .05 meaning we fail to reject the null hypothesis that all the
coefficients are equal to 0.

4) Although not statistically significant, regression 8 shows that the


coefficient on ethnic is positive and the coefficient on hi_ethnic is negatve
(but not statistically significant.) Since ethnic + hi_ethnic is positive, we can
say that for countries with a high gdp, ethnic diversity has a less positive
effect on terrorism. The same conclusion can be reached for regression 9
since the coefficient for ethnic is positive (but not significant) and the
coefficient on hi_ethnic negative (but not statistically significant.) This is the
same as the hypothesis which stated that in countries with a high gdp per
capita ethnic would have less of an effect in increasing terrorism.

In regression 8 we see that the coefficient on religion is negative, and is


statistically significant at the 5% level. Meaning that religious diversity in
countries with a low gdp has a negative effect on terrorism. We also see that
the coefficient on hi_religion is positive (but not statistically significant.) The
same can be said when looking at regression 9 since the coefficient on
religion is also negative and significant at the 5% level and the coefficient on
hi_religion is positive (but not statistically significant.) Although not
statistically significant, it appears that in poor countries increasing religious
diversity has a greater effect lowering terrorism than it does in rich countries
which is opposite of the hypothesis.

5) Looking at the data from problem sets 3 and 4, we can see that political
freedom appears to be the most consistent predictor. This reaffirms what we
saw in problem set 3 where we saw a quadratic relationship between lackpf
and lnftmpop and that countries with medium levels of political freedoms had
more terrorism than those with low or high levels of political freedoms.

This conclusion was reaffirmed in regression 6 where we found a negative


correlation of -.5232 between lackpf and higdppc. This explains why countries
with higdppc=1 had a positive correlation with lackpf and those with
higdppc=0 had a negative correlation with lackpf. This is also in line with our
conclusion that countries with medium levels of political freedoms have more
terrorism than those with low or high levels of political freedoms.

By including the interaction term between religion and ethnic in regression 8


we saw that religion became statistically significant. Though the religion
interaction term was not significant at the 5% level it appears as if a lack of
religious diversity leads to an increase in terrorism.

6)
a) Omitted variable bias is a possible reason to doubt this study. For example,
if we had considered income inequality and not just GDP per capita, we could
have seen a correlation between terrorism and income inequality that may
not be accounted for in GDP per capita since income inequality would be
relative to each country. Further, if we looked at the actual religions in
present in each country it could be possible that countries with a specific
religion are more likely to have more terrorism related deaths and that these
same countries have little religious diversity. Therefore, it is possible that
there was omitted variable bias in this study.

b) There is most likely no functional form misspecification because we found


that the lackpf relationship is best estimated with a quadratic so
misspecification in this relationship is unlikely and it wouldn’t make sense to
change the functional form of dummy variables either. Although unlikely, the
only place where we could change the functional form and see a different
result is with respect to religion. Thus, functional form misspecification is
most likely not a problem in this data.