Does a greater degree of integration into
world markets lead to a more elastic demand for labor? Often
referred to as the Rodrik hypothesis, this question, despite
being important from both theoretical and applied policy
standpoints, is still not satisfactorily answered. Several
studies that examine data for different countries, periods and
aggregation levels, have failed to draw uniform conclusions.
Likely reasons for this fact include poor and aggregated data,
a lack of large enough shocks in tariffs, an inability to
separate the effect of trade from that of other reforms, and
the absence of rich sets of predictions from models. In this
paper, I avoid these concerns by using a unique firm-level
panel data set for several thousand manufacturing firms in
Russia during a period of significant and rapid currency
devaluation. This event made imports of final and intermediate
goods more expensive while making Russian exports less
expensive and acted like a tariff on imports and / or a
subsidy on exports. In a simple, but yet revealing model, I
specify the implications of a devaluation for the labor demand
and derive a set of testable predictions. I then use the data
and reduced form regressions to test the essence of these
predictions and show that trade barriers affect labor demand
elasticities. In particular, I find that a 25-30% drop in the
conditional labor demand elasticity can be safely attributed
to the devaluation of 1998.
"Firm Heterogeneity and Costly Trade:
A New Estimation Strategy and Policy Experiments"
This paper studies how heterogeneous firms
respond to trade policies in their export markets. Firms are
heterogeneous in two ways: they face different demand
conditions in each market and have different productivities.
We thus augment the standard Melitz framework and embed it in
a partial equilibrium small country setting. The model is
designed to deal with real world trade policy variations, like
tariffs, preferences and the fixed and variable costs of
obtaining them. Our estimation procedure can be used on purely
cross sectional data, and requires only commonly available
variables at the firm level such as sales, and the number of
firms, which significantly extends its empirical
applicability. The application is to Bangladeshi apparel
exports to the US and EU. We are able to recover all the
parameters of the model, including the multiple levels of
fixed costs that firms incur in producing and exporting,
preference parameters, and the parameters of the distributions
that underlie the multidimensional heterogeneity.
The estimated model implies that there is a very large
response of exports to changes in fixed or variable costs.
This occurs primarily through changes in the mass of firms
entering the industry while the sales of incumbent firms
change by little. As a result, trade liberalization by one
country, say the EU, results in an increase in exports of
Bangladesh to all its export markets, rather than diversion
from these markets to the EU. Our work suggests that
developing countries could significantly raise their exports
by investing in ways of reducing the costs of exporting. This
includes improving infrastructure at all levels: from building
ports and roads to providing expertise and access to critical
factors like refrigeration, irradiation, and certification of
produce to meet international standards.
