Team Delta
What Drives Gold Returns?
– An empirical approach
____________________________________________________________________________
1.
Introduction
Does the real economy possess predictive power for Gold returns? While a sizeable
literature exists on this question, the evidence is mixed and continues to draw debate.
In this paper, we examine the issue of Gold return predictability over a range of time
horizons. A set of macro-economic, fundamental, and technical indicators is selected
for this purpose and based on a standard predictive regression framework; all
forecasts are generated by regressing the total returns of Gold on a constant and a
lag of the independent variables.
2. Literature Review
"Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor
incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud
exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure
dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur.
Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit
anim id est laborum."
3. Hypothesis
"Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor
incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud
exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure
dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur.
Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit
anim id est laborum."
1
Team Delta
4. Data Description
The data used came from three different sources: Bloomberg Terminals,
Ycharts, and Robert Shiller. From these sources, Team Delta accumulate sixty-six
variables, all of which were lagged one day. Each variable was collected in order to
predict the daily relative price difference of Gold from four different time periods:
1994-1999, 2000-2005, 2006-2011, and 2012-2017. Since our independent Y
variable is relative difference from the previous day, Team Delta decided it would be
best to take the relative difference for every X variable, if the data allowed. As
previously stated, our Y variable is the relative difference for daily Gold prices, which
indicates our data set is a Time Series one. Since our observations are daily and we
have 4, 5 year time periods, this means each time period has 1565 observations.
An important point, when the data was cleaned and merged Vlookup in excel
was used to match dates. If there was no data for a specific date, then the previous
value was chosen. This is especially apparent for any data that is not daily. Certain
data, mostly data from Bloomberg, contain NA values. These values were
transformed to a value of “-9999” in order to maintain numerical integral in R. Before
this procedure was done, Team Delta double checked the data to ensure “-9999” did
not previously exist, it did not. When the data was read in and split into 4 different
time periods, further cleaning ensued. For each time period, X variables that
contained NA’s for more than 20% of the data were removed. This would end up
helping with linear dependency. Going forward, Team Delta concluded weekly or
monthly data would be preferred over daily, mostly because economic data is
2
Team Delta
reported monthly or quarterly.
Talk about the descriptive statistics. Include data dictionary for each of these
variables. Should take about 4 sentences or so.
5. Econometric Model
You should write out the basic econometric specification first and explain each of the
variables and the parameters of interest. Why is this the correct specification for the
question you wish to address? Was it derived from theory and has it been used in
previous empirical work? Why are certain variables included and others not? Discuss
whether you are using basic OLS, IV, etc. and why this is appropriate. You should be
very clear about where identification is coming from and what assumptions you need
to make in order to interpret the parameters as you wish to ECON452 Econometrics 2
interpret them (e.g. discussing exclusion restrictions if you wish to interpret certain
parameters as causal). After discussing the basic specification, write out any
elaborations or additional tests you will perform and why.
6. Results
Here are some basic things to guide you in presenting your results:
3
Team Delta
1) You should present results in a way that develops your argument step-by-step.
For example, you may want to present your main results first, then break those
results down by subgroups and then perform robustness checks.
2) Any tables with parameter estimates should clearly state which dependent variable
you are using, which control variables are included and which specification you are
testing. Just discuss the most interesting and important estimates in your discussion
of the table. Make sure you report standard errors with your estimates. Just look at
some economics journals for a good table format.
3) Interpret the magnitude of your parameter estimates in an economically
meaningful way. For example: “we find that b=0.003, so that increasing X by one unit
increases y by 0.003. The implied elasticity is...”. This is particularly important if you
are not estimating a simple OLS regression. And even with OLS it is useful, especially
when the magnitudes of the variables are not immediately apparent, for example
when x is in logs.
4) Make sure you give your parameters the smell test. Are they a reasonable sign
and magnitude?
5) Graphs are worth a thousand words. Think about the most illustrative way of
presenting the results in a graph…this is a very convincing way to show your reader
that you have found something real.
6) Discuss whether the parameter estimates are statistically significant. If you don’t
get significance, why? Do you have enough data? Is your test strong enough to
detect effects below a certain magnitude (power tests are great for this sort of thing)?
Are the results still suggestive even if they are not estimated precisely?
7) Compare your results to what others have found or to your out-of-sample test. You
don’t need to worry if you don’t find anything significant as long as your methods are
sound and you have interpreted the results well. Discuss why your results may differ
from past research.
4
Team Delta
5
Team Delta
.
7. Conclusion
"Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor
incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud
exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure
dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur.
Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit
anim id est laborum."
6