Download as docx, pdf, or txt
Download as docx, pdf, or txt
You are on page 1of 4

Time Series Questions

N u m b e r o f C h ild r e n / U n e m p lo y e d
1. Plot the time series for ’children’ and ’unemployed’ over time.. Interpret your findings.

Time series plot of Children and Unemployed

.1 .2 .3
0

1990 2000 2010 2020


Time

(mean) children (mean) unemployed

A simple regression of children on unemployed that could approximately capture the relationship between
fertility and the business cycle ..

. reg children unemployed

Source SS df MS Number of obs = 33


F(1, 31) = 2.96
Model .001112029 1 .001112029 Prob > F = 0.0953
Residual .011644739 31 .000375637 R-squared = 0.0872
Adj R-squared = 0.0577
Total .012756767 32 .000398649 Root MSE = .01938

children Coef. Std. Err. t P>|t| [95% Conf. Interval]

unemployed .3724816 .2164865 1.72 0.095 -.0690456 .8140088


_cons .2660623 .0118803 22.40 0.000 .2418322 .2902924

The coefficient estimate for 'unemployed' is 0.3725, which indicates that, on average, for every one unit
increase in unemployment, the number of children increases by 0.3725 units. However, the p-value for
'unemployed' is 0.095, which is greater than the conventional significance level of 0.05. Therefore, we
cannot conclude with 95% confidence that the relationship between unemployment and fertility is
statistically significant.

The intercept, or constant term, is 0.2661, which represents the expected number of children when
unemployment is equal to zero. The p-value for the intercept is very small (less than 0.001), indicating that
the intercept is statistically significant.

The R-squared value is 0.0872, which means that only 8.72% of the variability in 'children' can be explained
by changes in 'unemployed'. This suggests that other factors not included in the model may be important in
explaining fertility

2. A regression of ’children’ on ’unemployed’ including ’share_married’ as a control variable.


. regress children unemployed share_married

Source SS df MS Number of obs = 33


F(2, 30) = 98.01
Model .011063475 2 .005531737 Prob > F = 0.0000
Residual .001693293 30 .000056443 R-squared = 0.8673
Adj R-squared = 0.8584
Total .012756767 32 .000398649 Root MSE = .00751

children Coef. Std. Err. t P>|t| [95% Conf. Interval]

unemployed .3892996 .083927 4.64 0.000 .2178978 .5607014


share_married .7461724 .0561955 13.28 0.000 .6314059 .8609389
_cons -.1489748 .0315946 -4.72 0.000 -.2134995 -.0844501

The coefficient estimate for 'unemployed' is 0.3893, which indicates that, on average, for every one unit
increase in unemployment, the number of children increases by 0.3893 units, holding 'share_married'
constant. The p-value for 'unemployed' is very small (less than 0.001), indicating that the relationship
between unemployment and fertility is statistically significant, even after controlling for 'share_married'.

The coefficient estimate for 'share_married' is 0.7462, which indicates that, on average, a one unit increase
in the share of married individuals in the population is associated with an increase in the number of children
by 0.7462 units, holding 'unemployed' constant. The p-value for 'share_married' is very small (less than
0.001), indicating that the relationship between marriage and fertility is statistically significant, even after
controlling for 'unemployed'.

The intercept, or constant term, is -0.1490, which represents the expected number of children when
'unemployed' and 'share_married' are both equal to zero. The p-value for the intercept is very small (less than
0.001), indicating that the intercept is statistically significant.

The R-squared value is 0.8673, which means that 86.73% of the variability in 'children' can be explained by
changes in 'unemployed' and 'share_married'. This suggests that the addition of 'share_married' as a control
variable has improved the model's ability to explain fertility.

3. Investigating whether there are trends in the variables we are using.

To investigate if there are trends in our variables, we will use a time series plot to visually investigate
if there is any patterns of up and downs in our dataset.
.6
.4
.2
0

1990 2000 2010 2020


Survey year

(mean) children (mean) unemployed


(mean) share_married
From the above plot we can identify patterns of up and downs from the three variables. Thus we should
come up with a model which includes time as a additional control variable.

4. The regressions in (2) and (3) might still give us biased and inconsistent coe"cients and/or standard
errors. It could be due to the presence of nonstationary dynamics (e.g. unit roots) and stationary
dynamics (e.g. autocorrelation) in the error terms. Investigate this issue and run regressions that
address this concern if you find that it could be important. Discuss your results.

You might also like