Document Type

Dissertation

Degree

Doctor of Philosophy (PhD)

Department

Economics

First Advisor's Name

Hakan Yilmazkuday

First Advisor's Committee Title

Committee Chair

Second Advisor's Name

Cem karayalcin

Second Advisor's Committee Title

Committee Member

Third Advisor's Name

Hassan Zahedi

Third Advisor's Committee Title

Committee Member

Fourth Advisor's Name

Tobias Pfutze

Fourth Advisor's Committee Title

Committee Member

Date of Defense

6-29-2016

Abstract

This dissertation introduced a method to construct a new measure for trade flows within a region using nighttime lights. After analyzing the relation between lights data and other proxies of economic human activity, I employed light data and econometric techniques to estimate the bilateral trade between any two regions around the world. Using these estimations, I estimated the overall internal trade volume for all countries. Moreover, I estimated the effect of internal trade within a state of the United States on the state’s income. The first essay proposed nighttime lights as an alternative proxy for economic activity to be used in gravity regressions. Due to the well-known problems in the measurement of gross domestic or regional products, gravity regressions based on both international and intranational trade data suffer from potential biases. At both international and intranational levels, log nighttime lights positively and significantly enter the gravity regressions (with a coefficient of roughly one) that explain at least about half of the variance in exports. The results were shown to be robust to the inclusion of several control variables and the consideration of predicted trade flows.

Trade within a nation and internal distance are variables known to play key roles in explaining home bias and the distance puzzle in international trade literature, but data on these measures are limited to only a few countries. To address this

problem, in the second essay, I constructed micro-founded measures of internal trade and internal distances from satellite data on nighttime lights. By estimating the gravity equation coefficients using the simulated method of distance estimation, I constructed the bilateral trade flow at subnational scale and aggregated it to overall internal trade. I found my internal trade measure is highly correlated with its benchmark, the difference between GDP and total exports; however, I showed it has more information and is a more precise measure for developed countries, which have a large amount of non-tradable services included in their national income account data. The internal distance measure is generated as the lights-weighted average distances between the states within a country. While my internal distance measure is largely correlated with its alternative, which is constructed based on city-level population data, it does not suffer from the uncertainty surrounding population data.

Correlation between trade and income cannot identify the effect of trade because of the endogeneity problem. The third essay examined this relationship at subnational level and by focusing on instrumenting trade via time varying geographic factors. Proximity and economic size are determinants of trade that are uncorrelated with other income determinants. This experiment not only confirmed the effect of interregional trade, but also provided evidence that intraregional trade has a large and statistically significant impact on income. I found, however, that the effect of both trade measures is statistically similar; a one percentage point increase in the interregional and intraregional trade ratio increases income per person by 2 to 4 percent.

Identifier

FIDC000738

ORCID

orcid.org/0000-0003-0691-0737