IV.
Misallocation in Mexico: competitive frictions and wedges
Our results indicate that frictions and wedges are important in Mexico. In this section, we
consider some sharper tests by using more direct measures of these frictions.
IV.1 Proximity to the US Border
Why proximity matters (for manufacturing firms)
One striking finding in the data is the importance of proximity to the US for the management
practices in Mexican manufacturing firms. In particular, Panel (a) of Figure 6 highlights how
26
The equivalent increase between unweighted and size-weighted management mean is 0.091 in manufacturing.
14
the US bordering states, which contain
maquiladoras
– the Mexican firms particularly
focused on US exports - have the highest management scores.
There are several reasons why proximity to the US might matter. Our main hypothesis is that
being closer to Mexico’s main export market will strengthen competitive pressure on
Mexican firms, raising their management quality. In other words, proximity reduces market
frictions, raising management through a selection effect (the worst managed firms exit) and
potentially an incentives effect (e.g., fear of losing some or all market share increases
managerial effort). An implication of this hypothesis is that proximity should matter much
more for firms that are more exposed to international trade with the US. For example, it
should be much stronger for manufacturing firms (which generally trade internationally) than
for service firms (who generally trade domestically). A corollary of this is that within
manufacturing, the proximity effect should be more important for export-intensive industries
compared to those that do not.
An alternative hypothesis on why proximity might matter for management is that being closer
to the US raises management through learning. This might be because of faster flows of
information on managerial best practices, and/or through the managerial labor market with
American managers working in Mexican firms. Note that this story is based primarily on
labor markets that are geographical in nature – there is no obvious reason why it should be
different for sectors that are more or less exposed to trade. We shall test this idea directly by
looking for heterogeneity in the proximity to the US effect across industries with different
degrees of international exposure to trade.
A third hypothesis, popular in international economics, is that higher trade with the US could
benefit Mexican firms through the use of higher-quality intermediate inputs (e.g., De Loecker
et al., 2016). Consistent with this, we show that all our proximity results do also raise
productivity. But it is less clear why better intermediate inputs would raise management
quality.
27
This is one reason why having access to management data is an advantage over just
using productivity data: it enables us to help disentangle the mechanisms through which trade
effects on productivity may be occurring.
27
Service inputs could help as discussed in the previous mechanism – e.g., easier access to the managerial labor
market or better consultants. Trade and management are discussed in more detail in Bloom et al. (2018).
15
In Panel (b) of Figure 6, we reproduce Panel (a) for services. In contrast to manufacturing,
areas near the US do
not
have systematically higher management scores. Instead, the service
sector shows the highest management scores in locations close to the largest Mexican cities
(i.e., Mexico City, Guadalajara, Querétaro, and Monterrey). One interpretation of this, which
we will pursue in the next subsection, is that
local
market size and competition is much more
important for service firms (who typically sell non-tradeable outputs and are more locally
oriented) than manufacturing firms.
Evidence on proximity
To measure the geographical proximity of Mexican firms to US markets, we calculate the
drive time, measured in hours, between the municipality where the firm is located and the
closest of the three main border crossings between Mexico and the US (i.e., Tijuana, El Paso,
and Nuevo Laredo). Given that the exact location of each firm is not disclosed for
confidentiality reasons, we calculate the centroid of each municipality and then calculate the
time between this and each of the three border crossings.
28
Our results indicate that the relationship between management and proximity differs across
these two broad sectors, as the drive time to the border matters for firms in manufacturing,
but not for those in services. As shown in Figure 7, where we present the CDFs of
management across groups of drive time to the border, management score is higher among
firms with a drive time to the border below the median, that is, closer to the border.
29
Figure 8 Panel (a) shows that although there is a positive relationship between firm size and
management, the gradient is steeper for firms closer to the border, indicating stronger forces
of selection. The same is not true for services, where the slope is similar regardless of the
distance to the US border. This suggests proximity to US markets increases both the mean
management score and its covariance with size for Mexican manufacturing firms.
We show these results in a regression framework in Table 3. Column (1) for manufacturing
firms shows that for firms located close to the US border, management is positively and
28
We use Open Street Map to calculate the distances by using the command “osrmtime” in STATA. Similar
results obtain from using GoogleMaps. An analysis of the municipalities is shown in Appendix D.
29
Figure 7 is weighted by employment size. In Online Appendix Figure A10, we present the unweighted results,
which are very similar.
16
highly significantly related to size (as usual). However, we also identify a significantly
positive coefficient on the interaction between management and proximity to the border,
indicating that the covariance between size and management is much stronger for firms closer
to the border. In column (2), we repeat the same specification for service firms which shows
a positive relationship between size and management, but no significant interaction with
proximity. This is a placebo test, and is consistent with our priors that distance to the US
border should not matter for service firms, whose outputs are generally non-tradable.
Column (3) of Table 3 subjects the results to a tougher test. If it really is the closeness to the
US export market which drives the results in column (1), we would expect the interaction to
be stronger for export-intensive industries. Industries that are not export intensive are more
like the service firms of column (2). We build a NAICS-6 digit industry-level indicator of
the export share of sales and include the triple interaction of an industry level variable of
export intensity with the management*proximity variable (as well as the pairwise
interactions). Consistent with our main hypothesis, there is a positive and (weakly)
significant coefficient on this triple interaction.
Online Appendix C investigates a number of robustness tests. For example, we drop all firms
owned by foreign companies and further control for interactions between proximity and
market size with other variables correlated with management (capital, share of white-collar
workers, and education), and the results do not materially change. We also show how
closeness to the US border affects unweighted average management practices (reflecting
selection on the extensive margin) as shown in Figure 7 as well as discontinuity designs of
being right at the border.
Dostları ilə paylaş: |