Please answer the questions in Bold

Researchers have been interested for a while in developing measures of the level of uncertainty in the economy regarding economic policies. One measure is the news-based economic policy uncertainty (EPU) index of Baker, Bloom and Davis. Information about their project and their data for a number of countries can be found at https://www.policyuncertainty. com/. Their main measure is based on the frequency of newspaper articles that mention combinations of keywords relating to economic policy uncertainty (e.g., words like "uncertainty" together with words like "economy", "policy", "legislation", "congress", etc).

When looking at the global and country-specific time series available on the home page of their website, you can clearly see spikes in EPU around major events, such as 9/11 and the Covid-19 pandemic. We also see that important votes such as the 2015 Brexit referendum in the UK and the 2016 US presidential election led to spikes in news-based EPU indices for those countries. Those were quite unusual elections, however, and it's not clear, more generally, to what extent typical elections are associated with increased EPU. So let's empirically investigate the relationship between elections and EPU in Canada.

[Note: you do not need to submit your Excel spreadsheet or tables. Just the numbers requested are sufficient. Answer the open ended questions briefly. No answer required for (i) and (iii).]

(i)Go to https://www.policyuncertainty.com/ and download the EPU data series for Canada. Only use the News-Based Index, ignore the other sheet with Old EPU and Forecast Data. Our sample period will be January 1985 to February 2020, inclusive.

(ii)Do you expect federal elections to impact EPU? If so, in what way?

(iii)Using Wikipedia or another reliable (at least for dates of events) source, record the month of each Canadian federal election from 1985 through to February 2020. Form a new column in the Excel spreadsheet, FederalElection, which will be a dummy variable indicating election months. That is, FederalElection = 1 if there is a federal election during that month and FederalElection = 0 if there is not a federal election that month.

(iv)What is the average level of EPU during an election month? What is the average level of EPU during a non-election month? [Hint: in Excel the easiest way is with a Pivot Table. Go to "Insert", then "Recommended Pivot Table", then "Blank Pivot Table", then drag your dummy variable of interest to the Rows field, in the Values field put EPU, and change the statistic

(v)Is the difference in average EPU between election and non-election months consistent with your hypothesis in (ii)? Would you say it is a big difference in magnitude? [Hint: compare to the overall standard deviation of EPU. Note: don't worry about formal tests of statistical significance.]

(vi)The above assumes the impact of a federal election is restricted to just the one month during which the election takes place. It's possible, however, that elections have effects over longer periods of time, say, a number of months leading up to and after the election. Moreover, the effect of the election on EPU may be different before an election compared to after an election. Let's define the pre-electoral period as the three months leading up to and including the election month. Let's define the post-electoral period as the three months immediately following the election (excluding the election month). For example, if an election takes place in November 2000, the pre-electoral period is September-October-November 2000 and the post-electoral period is December 2000 and January-February 2001. In two new Excel columns, create these pre- and postelection dummy variables, PreFederalElection and PostFederalElection.

(vii)What is the average EPU during the pre-electoral period? What is the average EPU during the post-electoral period? What is the average EPU during months that are outside the pre-electoral period and the post-electoral period?

(viii)Can we conclude that elections cause differences in EPU? If so, why. If not, give one alternative reason for the differences we see.

(ix)The Canadian EPU index is based on the frequency of articles mentioning economic policy uncertainty related keywords in several Canadian newspapers. It assumes that the number of such newspaper mentions accurately reflects the true level of economic policy uncertainty. If media outlets tended to be biased in favor of one party, how might this affect the general accuracy of the EPU index?

(x)Even if media outlets are not biased in favor of any particular party, what else might lead the number of EPU mentions to not track the true level of economic policy uncertainty. (xi) Given the close relationship between the US and Canada, policies implemented in the US may also have spillover effects on Canada. US policies are determined by US elections. Let's look at the relationship between US federal elections and Canadian EPU. Record the month of each US federal election over the sample period. By Federal elections, I am referring to the major national elections that take place once every two years, in which all the seats in the US House of Representatives are contested, in 3 addition to various Senate seats and, every four years, the presidency. Calculate the average Canadian EPU in US federal election months compared to months where there is no US federal election.

(xi)How does the difference in (xi) compare in magnitude to the difference in (v)?

(xii)The Canadian EPU index is based on the frequency of articles in Canadian newspapers mentioning economic policy uncertainty related keywords. Canadian newspapers frequently write stories about other countries, particularly the US. From what I can tell, the algorithm used to calculate the EPU index wouldn't distinguish between a story about the US and a story about Canada in the same newspaper. Could this potentially be a problem if we want to conclude that US federal elections have a real impact on Canadian economic policy uncertainty?