Panel regression model / Multiple regression model?

[Dan VG]

Hi everyone,

I have a dataset available which links the the Covid-19 situation to the change in port activity (expressed in difference of port calls = ship arrivals in ports) for any specific date between 01/01/2020 - 31/08/2020 and any given country. An example of dataset for the countries China & Belgium is attached to see the data formulation yourself. In the end the data set is a list per country and any date, given the country linked to indicators summarising the Covid-19 situation in that country for that particular date. More information about the indicators: The policy indicator reports a number which reflects the severity of the corresponding containment policy. The stricter a government’s policy is, the higher the representing indicator will be. C1, C2, and C6 are all three reported on a scale between zero and three (0 – 1 – 2 – 3). On the other hand, C3, C5, and C7 use a scale between zero and two (0 – 1 – 2), while C4 and C8 make use of scale between zero and four (0 – 1 – 2 – 3 – 4). The stringency index is reported by a percentage where a value close to 100% indicates the strictest situation one can imagine.

First, the stringency index is linked to the change in port calls by a single regression model.
Second, the main purpose is to check how much of the change in port calls is explained by the indicators and even more crucial, which indicator has the declarative value.

Now, I had some questions related about the second research question.
- Would it be best to use a panel regression model, or make use of multiple regression model?
- There are three different scales of containment policies used in the data set, does Stata automatically recognise difference between these scales?

Thank you in advance!
Dan

[Carlo Lazzaro]

Dan:
- nobody on this list would ever download a spreadsheet coming fron an unknown source (see the FAQ and use CODE delimiters or -dataex-, indeed. Thanks);
- usually, panel data analysis are the first choice if the same sample is measured at (usually) equally spaced intervals in time;
- Stata does what you ask it for: if the three predictors you mentioned are coded differently, they will be recognised as such.

[Dan VG]

Hi Carlo,

The data looks like this:

CountryName	Date	C1_School closing	C2_Workplace closing	C3_Cancel public events	C4_Restrictions on gatherings	C5_Close public transport	C6_Stay at home requirements	C7_Restrictions on internal movement	C8_International travel controls	StringencyIndex	StringencyIndexForDisplay	Difference
Aruba	01/01/2020	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0,285714286
Aruba	02/01/2020	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0,285714286
Aruba	03/01/2020	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0,285714286
Aruba	04/01/2020	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0,285714286
Aruba	05/01/2020	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0,285714286
Aruba	06/01/2020	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	-0,142857143

...

Aruba	25/08/2020	3.00	2.00	2.00	4.00	0.00	1.00	1.00	3.00	69.44	69.44	0,285714286
Aruba	26/08/2020	3.00	2.00	2.00	4.00	0.00	1.00	1.00	3.00	69.44	69.44	-3,285714286
Aruba	27/08/2020	3.00	2.00	2.00	4.00	0.00	1.00	1.00	3.00	69.44	69.44	-3,285714286
Aruba	28/08/2020	3.00	2.00	2.00	4.00	0.00	2.00	1.00	3.00	71.76	71.76	-3,285714286
Aruba	29/08/2020	3.00	2.00	2.00	4.00	0.00	2.00	1.00	3.00	71.76	71.76	-3,285714286
Aruba	30/08/2020	3.00	2.00	2.00	4.00	0.00	2.00	1.00	3.00	71.76	71.76	-3,285714286
Aruba	31/08/2020	3.00	2.00	2.00	4.00	0.00	2.00	1.00	3.00	71.76	71.76	-0,571428571

Angola	01/01/2020	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	-0,571428571
Angola	02/01/2020	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	-0,571428571
Angola	03/01/2020	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	-0,571428571
Angola	04/01/2020	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	-0,571428571
Angola	05/01/2020	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	-0,571428571
Angola	06/01/2020	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0,142857143
Angola	07/01/2020	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0,428571429

...

Angola	25/08/2020	3.00	2.00	2.00	4.00	1.00	2.00	2.00	4.00	76.39	76.39	-0,571428571
Angola	26/08/2020	3.00	2.00	2.00	4.00	1.00	2.00	2.00	4.00	76.39	76.39	3,142857143
Angola	27/08/2020	3.00	2.00	2.00	4.00	1.00	2.00	2.00	4.00	76.39	76.39	3,142857143
Angola	28/08/2020	3.00	2.00	2.00	4.00	1.00	2.00	2.00	4.00	76.39	76.39	3,142857143
Angola	29/08/2020	3.00	2.00	2.00	4.00	1.00	2.00	2.00	4.00	76.39	76.39	3,142857143
Angola	30/08/2020	3.00	2.00	2.00	4.00	1.00	2.00	2.00	4.00	76.39	76.39	3,142857143
Angola	31/08/2020	3.00	2.00	2.00	4.00	1.00	2.00	2.00	4.00	76.39	76.39	0,285714286

And this repeats itself for more than 132 countries.

Can we consider this strange time serie set fit for a panel regression model?

Greetings
Daan

[Carlo Lazzaro]

Dan:
assuming that -stringency_index- is the continuous regressand, I would take a look at T>N panel data regression commands, such as -xtgls- and -xtregar-.