xi
An additional concern is that the results might be driven by the importance of two main states:
Mexico City, which is located in the central region of the country, and Nuevo Leon, which
is located near the Northern border of the country. According to the 2014 Economic Census,
these two states account for 17% of Manufacturing Gross Product and 34% of Services Gross
Product. The results do not change significantly when we estimate the same equations but
exclude these states.
City Density and Reallocation in the Service Sector: Robustness Tests
Testing the robustness of our results on market size, we estimate the relationship between
management and market size in a regression framework. The coefficient on the linear city
size variable is small, negative,
and insignificant, whereas the interaction between the
services dummy and market size is positive, large, and statistically significant.
A similar
pattern holds when we condition on other covariates or split the sample by manufacturing
and service sectors. We further assess the robustness of these results by using alternative
proxies for market size based on population size. We use a dummy that indicates whether the
municipalities are small urban areas, medium,
metropolitan areas, or large metropolitan
areas. As the municipalities grow in market size, the management score for the services sector
improves, but the same is not observed for manufacturing.
In an alternative specification using the TFP index as the dependent variable, we find that
local city size matters for productivity in services, but not in manufacturing. The absence of
a city-size effect on productivity may appear surprising,
as there is a vast economic
geography literature that argues for higher productivity effects in large cities. It is worth
noting, however, that most individuals and firms are not in manufacturing, so some of the
existing empirical studies are likely driven by the services sector.
3
Furthermore, most of the
studies are in high-wage countries where agglomeration effects may be stronger than in a
middle-income country like Mexico.
One concern with our results is that the average income could reflect the presence of more
skilled potential employees. We disaggregate our market size measures by including income
and population density as separate variables to address this. We find that our local size effects
are driven by population density that has a similar statistically
significant coefficient, but
income is insignificant.
4
It is possible that these results are driven by reverse causality or omitted location-specific
confounders. To partially address this, we use population density and income in the 1990
Population and Housing Census as an instrument for the 2010 market size data. The results
do not change much.
Finally, we estimated the same equations again, excluding Mexico City and Nuevo Leon,
and confirm that our results are robust to excluding these two main cities.
3
A smaller literature uses plant and firm level data in manufacturing. Some of these studies do find significant
and positive city size effects, but all the ones that we know of are in high wage countries (e.g., Combes et al.,
2012).
4
As before, the variables are insignificant in the manufacturing sector
.
xii
Institutional Frictions and Misallocation: Robustness Tests
We analyze the robustness of the results presented in Figure 12 and Table 4, by using a
different threshold (5%) to define a high level of contract enforcement problems, kidnapping,
corruption, and our business crime composite index. Our results
are robust to using this
alternative threshold.
Dostları ilə paylaş: