Return to Answer

Your trade figures are hardly normally distributed, which makes your model very susceptible to a few outliers. The GDP data might have a similar problem. Try a log-log plot for instance. Looking at your assignment text, it doesn't seem to inquire/impose a linear relationship (between the raw values).

To summarize some stuff from the comments: two hypotheses are fairly common in the literature (but the first is much more common than the 2nd)

• that GDP growth depends on openness to trade. But as you said you are sure this wasn't your task.

• Wikipedia says "the trade-to-GDP ratio tends to be low in countries with large economies and large populations such as Japan and the United States, and to have a higher value in small economies". So if this was the intended causality to explore, then you have the axes swapped. One paper says this relationship is bogus, by the way, at least if interpreted as relating trade openness:

The trade intensity index (TII), constructed as exports plus imports divided by Gross Domestic Product (GDP), is the most commonly used measure for trade openness and increasingly for globalization as well.

The TII, however, often gives counterintuitive results when it comes to large countries. For instance, the U.S. is ranked way below Swaziland and Tajikistan by the index, and likewise China is ranked behind Cambodia and Laos [...]. These results are not surprising if the TII is (rightly) interpreted as a measure of trade dependency, as large countries are expectedly less reliant on international trade than the small ones. However, when the index is used as a measure of trade openness or globalization, the results become counterintuitive – considering the U.S. is a core nation in the world trade system while Swaziland and Tajikistan are far from that, and China is also way ahead of its two neighbouring countries as a trading power house. In short, the TII ‘appears’ to understate the degree of openness of large economies relative to small economies.

In fact, this issue has been noticed by many and there are some attempts in the current literature to ‘correct the size bias’ through modifying the TII [...]

Your trade figures are hardly normally distributed, which makes your model very susceptible to a few outliers. The GDP data might have a similar problem. Try a log-log (or a rank-rank) plot for instance. Looking at your assignment text, it doesn't seem to inquire about (or impose) a linear relationship between the raw values.

To summarize some stuff from the comments: two hypotheses are fairly common in the literature (but the first is much more common than the 2nd)

• that GDP growth depends on openness to trade. But as you said you are sure this wasn't your task.

• Wikipedia says "the trade-to-GDP ratio tends to be low in countries with large economies and large populations such as Japan and the United States, and to have a higher value in small economies". So if this was the intended causality to explore, then you have the axes swapped. One paper says this relationship is bogus, by the way, at least if interpreted as relating trade openness:

The trade intensity index (TII), constructed as exports plus imports divided by Gross Domestic Product (GDP), is the most commonly used measure for trade openness and increasingly for globalization as well.

The TII, however, often gives counterintuitive results when it comes to large countries. For instance, the U.S. is ranked way below Swaziland and Tajikistan by the index, and likewise China is ranked behind Cambodia and Laos [...]. These results are not surprising if the TII is (rightly) interpreted as a measure of trade dependency, as large countries are expectedly less reliant on international trade than the small ones. However, when the index is used as a measure of trade openness or globalization, the results become counterintuitive – considering the U.S. is a core nation in the world trade system while Swaziland and Tajikistan are far from that, and China is also way ahead of its two neighbouring countries as a trading power house. In short, the TII ‘appears’ to understate the degree of openness of large economies relative to small economies.

In fact, this issue has been noticed by many and there are some attempts in the current literature to ‘correct the size bias’ through modifying the TII [...]

Your trade figures are hardly normally distributed, which makes your model very susceptible to a few outliers. The GDP data might have a similar problem. Try log-log plot for instance.

Also, I'm not sure how much sense it makes to regress a per capita figure (your GDP is so) to a ratio of two absolute (not per capita) figures.

To summarize some stuff from the comments: two hypotheses are fairly common in the literature (but the first is much more common than the 2nd)

• that GDP growth depends on openness to trade. But as you said you are sure this wasn't your task.

• Wikipedia says "the trade-to-GDP ratio tends to be low in countries with large economies and large populations such as Japan and the United States, and to have a higher value in small economies". So if this was the intended causality to explore, then you have the axes swapped. One paper says this relationship is bogus, by the way.

Your trade figures are hardly normally distributed, which makes your model very susceptible to a few outliers. The GDP data might have a similar problem. Try a log-log plot for instance. Looking at your assignment text, it doesn't seem to inquire/impose a linear relationship (between the raw values).

To summarize some stuff from the comments: two hypotheses are fairly common in the literature (but the first is much more common than the 2nd)

• that GDP growth depends on openness to trade. But as you said you are sure this wasn't your task.

• Wikipedia says "the trade-to-GDP ratio tends to be low in countries with large economies and large populations such as Japan and the United States, and to have a higher value in small economies". So if this was the intended causality to explore, then you have the axes swapped. One paper says this relationship is bogus, by the way, at least if interpreted as relating trade openness:

The trade intensity index (TII), constructed as exports plus imports divided by Gross Domestic Product (GDP), is the most commonly used measure for trade openness and increasingly for globalization as well.

The TII, however, often gives counterintuitive results when it comes to large countries. For instance, the U.S. is ranked way below Swaziland and Tajikistan by the index, and likewise China is ranked behind Cambodia and Laos [...]. These results are not surprising if the TII is (rightly) interpreted as a measure of trade dependency, as large countries are expectedly less reliant on international trade than the small ones. However, when the index is used as a measure of trade openness or globalization, the results become counterintuitive – considering the U.S. is a core nation in the world trade system while Swaziland and Tajikistan are far from that, and China is also way ahead of its two neighbouring countries as a trading power house. In short, the TII ‘appears’ to understate the degree of openness of large economies relative to small economies.

In fact, this issue has been noticed by many and there are some attempts in the current literature to ‘correct the size bias’ through modifying the TII [...]

Your trade figures are hardly normally distributed, which makes your model very susceptible to a few outliers. The GDP data might have a similar problem. Try log-log plot for instance.

Also, I'm not sure how much sense it makes to regress a per capita figure (your GDP is so) to a ratio of two absolute (not per capita) figures.

To summarize some stuff from the comments: two hypotheses are fairly common in the literature (but the first is much more common than the 2nd)

• that GDP growth depends on openness to trade. But as you said you are sure this wasn't your task.

• Wikipedia says "the trade-to-GDP ratio tends to be low in countries with large economies and large populations such as Japan and the United States, and to have a higher value in small economies". So if this was the intended causality to explore, then you have the axes swapped.

Your trade figures are hardly normally distributed, which makes your model very susceptible to a few outliers. The GDP data might have a similar problem. Try log-log plot for instance.

Also, I'm not sure how much sense it makes to regress a per capita figure (your GDP is so) to a ratio of two absolute (not per capita) figures.

To summarize some stuff from the comments: two hypotheses are fairly common in the literature (but the first is much more common than the 2nd)

• that GDP growth depends on openness to trade. But as you said you are sure this wasn't your task.

• Wikipedia says "the trade-to-GDP ratio tends to be low in countries with large economies and large populations such as Japan and the United States, and to have a higher value in small economies". So if this was the intended causality to explore, then you have the axes swapped. One paper says this relationship is bogus, by the way.