February 2015
Draft prepared for 2015 MWIEG Conference
Productivity and Export Pricing Behavior: A Firm-level Analysis in India
Michael A. Anderson
Washington and Lee University
and
U.S. International Trade Commission
Martin H. Davies
Washington and Lee University
and
Center for Applied Macroeconomic Analysis
Australian National University
Jose E. Signoret
U.S. International Trade Commission
Stephen L. S. Smith
Gordon College
The views in this paper are strictly those of the authors and do not represent the opinions of the U.S.
International Trade Commission or any of its individual Commissioners.
Abstract. We examine the export pricing behavior of Indian firms in the early 2000s. Using a
unique data set that matches detailed firm, product, and destination-level trade data with firm
characteristics and destination characteristics, we find, in contrast to Harrigan, Ma and Shlykov
(2011) and Manova and Zhang (2012), that firm productivity is negatively associated with
prices, that distance to destination is unrelated to, or negatively associated with prices, and that
increases in export revenue have been driven by rising quantities rather than rising prices. These
points together are evidence that Indian exporters do not compete on the basis of quality. Indian
exporters do appear able to raise product prices in richer, larger markets, just as U.S. and Chinese
exporters do.
Acknowledgments. We thank participants at the 2014 Southern Economics Association
meetings, and especially Tom Prusa, for helpful comments. We also thank Richard
Marmorstein, Daniel Molon and Sommer Ireland (all from Washington and Lee University),
Nathanael Snow (Office of Economics, USITC), and Matthew Reardon (American University),
for indefatigable research assistance.
1
Introduction
Our paper contributes to the literature on firm-level productivity and exports by analyzing
the export pricing behavior of Indian firms. We use a new data set on Indian firms’ exports—
prices changed and quantities shipped abroad—to determine the sources of exporting
success. We make several contributions. First, ours is the first study to examine the pricing
behavior of Indian exporters, whose behavior should shed light on whether findings based on
U.S., European or Chinese firms should be considered universal. Toward that end, our study is
the first study to discover a negative association between firm productivity and the prices firms
charge in destination markets. This result, combined with a finding that distance to destination
market is not associated with product prices, is consistent with a no-quality-upgrading story,
again in contrast with what others in this literature have found for firms outside of
India. Second, we explain which current theories are capable of explaining the negative
productivity-price association. Our results are most consistent with Melitz (2003), and are in
contrast to the predictions of Baldwin and Harrigan (2011), where firms are competing on price
per unit of quality. Third, we perform a price-quantity decomposition of exporter revenue in our
descriptive analysis and find that exporters increase revenue by increasing quantities, and not by
increasing prices, a finding that broadly supports our first contribution.
Our results stand in contrast to recent findings for U.S. and Chinese exporters; several
papers suggest that firms tailor prices to market characteristics and, when firm productivity can
sustain it, upgrade quality within product categories to earn higher prices. For example Manova
and Zhang (2012, hereafter “MZ”), using 2005 Chinese firm and product data at the HS 8-digit
level, find that successful exporters earn more revenue in part by charging higher unit prices and
by exporting to more destinations than less successful exporters. Even within narrowly-defined
2
product categories, firms charge higher unit prices to more distant, higher income, and less
remote markets. MZ argue that firms’ product quality is as important as production efficiency in
determining these outcomes.1 Harrigan, Ma and Shlykov (2011, hereafter “HMS”), using 2002
U.S. data at the HS 10-digit level, make a similar finding: U.S. firms charge higher prices for
products shipped to larger, higher income markets, and to countries more distant than Canada
and Mexico, a result they attribute to higher quality. HMS also find that firms’ ability to raise
unit prices is positively affected by their productivity and the skill-intensity of production. On
their face, the results relating to distance and number of markets appear consistent with the
hypothesis first advanced by Alchian and Allen (1964), and developed by Hummels and Skiba
(2004), that “per unit” transport costs raises relative demand for high quality goods (the
“shipping the good apples out” hypothesis).
The paper proceeds as follows. Section II discusses data, where there are some
particularly difficult issues posed in matching firm-level data with trade and destination
information. These issues are significant enough that our results are drawn from several
different samples and should be considered preliminary. Section III presents our results in two
parts: descriptive statistics on successful exporters’ pricing patterns and regression results that
relate price decisions to firm and destination market characteristics. Section IV considers
possible theoretical reasons why our findings for India differ from those of other countries.
Section V concludes.
1
MHZ p. 2. Martin (2009) and Gorg et al. (2010) are further recent examples of firm-product level studies, for
France and Portugal, respectively. MZ present evidence that not only do successful exporters produce higher quality
goods (with higher quality inputs), but that firms adjust product quality according to characteristics of the
destination market. In particular, they find that the higher unit values associated with higher distance to destination
markets and with serving more destinations are due to compositional shifts within narrow product categories
towards higher product quality and higher quality inputs.
3
II.
Data
Our data comes from several sources. The first, TIPS, is a database of daily firm-level export
transactions collected by Indian Customs. TIPS contains the identity of the exporter, the date of
transaction, the product type by eight-digit HS code, destination country, exit and destination
port, and the quantity and the value of the export. TIPS provides useable data for four full fiscal
years, 2000-2003. (Indian fiscal years run from April 1 through March 31; the actual data run
from April 1999 through March 2003.) The data cover the transactions at eleven major Indian
seaports and airports. Summary statistics (Table 1) show that there are over 20,000 exporting
firms shipping approximately 80,000 types of goods to approximately 200 destination countries,
each year.
All told, TIPS records more than 5.8 million export transactions over 1999-2003. For the
purpose of our analysis we aggregate the data to fiscal-year average prices, by firm, product and
destination. We reduce the data set further with data cleaning, including the removal of firms
that appear to be wholesalers, and quantity unit harmonization, described below. Finally, we
perform our analysis on a subset of this data, the largest quantity units for each firm in each HS
line. All told, we use approximately 330,000 observations.
Tables 1 and 2 demonstrate that the most successful firms (in export value terms)
dominate the export sector in many dimensions. The top 10 percent of the firms account for
approximately 80 percent of exports by value, and export far more products per firm, and to far
more destinations per firm, than do other exporters. Interestingly, this is less concentrated than
the behavior of U.S. exporters in 2000, for which the top 10 percent account for 95 percent of the
4
value of all exports. The modal Indian firm exports 1 product to a single market; only 4 percent
of the firms in the sample export more than 10 products and to more than 10 markets (Table 3).
Again, this is different from the U.S. case; in India in 2003, 53 percent of all exporters sold to
one market only, whereas for the United States that same year the proportion was 65 percent.2
Finally, there is a lot of turnover among Indian exporters. Just in our small sample of years net
entrants into exporting careened between highly positive and highly negative values (Table 4),
and the two-year survival rate for exporters was never greater than 66 percent (Table 5). This
may be due to what at the time was India’s recent trade liberalization; the firms in our sample
behave differently than the perhaps internationally more seasoned firms in other countries.
The second data set is Prowess, a database of Indian firm characteristics. From this data
set we derive information on employment, labor skill, capital use, profitability, expenses, and
other firm-level variables. The dataset contains information on large- and medium-sized firms in
India, and includes all companies traded on India’s major stock exchanges as well as other firms,
including the central public sector enterprises. All told there are about 23,000 firms in the
database. When we merge this data with TIPS, we lose information about the informal sector as
well as most privately held firms. Still, Prowess contains a broad swath of Indian firms;
surveyed companies cover about 75 percent of all corporate taxes and over 95 percent of excise
duties collected. Merging the two databases presents a number of technical problems, but in an
initial match we paired 2,692 firms.
Finally, we have country characteristic data (income, population, and distance from
India) from the public online database maintained by CEPII (the Paris-based Institute for
Research on the International Economy).
2
U.S. information is from Bernard et al. (2007), p. 116 and p. 118.
5
In the work that follows, the dependent variable is the export product price, defined as an
export unit value and calculated as the relevant total value of exports divided by quantity. So,
for instance, a firm’s average price for selling a particular product to the United States in any
given year would be the value of sales divided by, say, the metric tons sold. The rich detail of
the TIPS data allows, in principle, calculation of each firm’s unit price for every product and
every market. The price of Tata Tea’s loose leaf Darjeeling sold in the UK can be distinguished
from its price in Italy. Likewise it is possible to calculate a firm’s mean price for a product
across all destinations, and the mean price across all firms for a product shipped to a single
destination. These details are essential for understanding how exporters behave and, in
particular, for understanding the role that firm and destination market characteristics play in
firms’ pricing and quality decisions.
Although the TIPS data are reported at the 8-digit HS level, we use the firm’s own
product labels to obtain the product prices used in this study. For example, instead of looking at
the unit value of 8-digit HS code 09101020 that includes a variety of spices, we are able to use
the product labels to obtain the unit value, or price, of “curry powder” and “ginger” and other
similar fine-grained prices. The result is something much more detailed than 8-digit data. In
brief, here is how we obtained that information: Within each of the 16,109 8-digit categories, the
median number of (reported) individual product lines is 8, and the mean is 166. In some cases
the product-level labels are variants of names for the same product, differing only in punctuation,
capitalization, or word order. Sometimes these differences are present along with changes in the
product description; thus we may see “Curry Powder” and “SPICE CURRYPOWDER”
describing what appear to be the same product. By contrast, in other cases the product names
reflect substantively different products within a particular HS line. We used a computerized
6
matching algorithm to match product names, to say (in the example above) that “Curry Powder”
and “ SPICE CURRYPOWDER” are the same product, but “Curry Powder” and “Ginger” are
different products, even though all of these are inside the same HS-8 code.3 We then aggregate
together the quantity and value information for those product labels that our algorithm deems as
the same product (from the same firm). When this process is complete the mean number of
individual product lines in an HS category is 11, with a median of 3.4 5
III.
Results
We offer two sets of results. The first consists of descriptive findings about Indian exporters’
pricing behavior. The evidence here suggests that, at least in our sample period, Indian firms
may be achieving export revenue gains more through quantity increases than through price
increases. In the second set of results, we estimate the relation between prices and destination
market characteristics and firm characteristics. We are particularly interested in the relation
3
The algorithm creates matches based upon the textual similarity of the product names. Textual similarity was
calculated using a bigram specification of edit distance: each pair of labels was assigned a value between zero and
one, the number of common bigrams divided by the average total number of bigrams between the two. Labels whose
similarity according to this measure was above a certain threshold were placed in the same product-line. We
experimented with different thresholds; based upon some experimentation (examining random samples of groupings
that result from alternative thresholds) we chose a threshold of 0.35. In subsequent versions of this paper we will
examine how our results change when we slacken or tighten the threshold criterion.
4
A final cleaning step was also necessary. The TIPS data are reported in multiple quantity units. In many of the
single firm-product-destination categories, export values are reported for several different units, such as “buckles,”
kilos, pounds and “boxes,” the sum of which yields the total value of exports for that firm-product-destination
observation. We choose to aggregate and “harmonize” these values where there are well-established conversion
factors for the units. Therefore we convert pounds to kilos, and tons to metric tons, and so on, prior to calculating
unit values. However, there remain thousands of lines of data where the conversion factors are unknown, or for
which the reporting of separate lines based on different quantity measures strongly suggests that there are in fact
underlying differences between the goods reported in those lines (even when they are in the same 8-digit HS
category). It is not possible to make meaningful unit value comparisons, or aggregations, across different units in
these instances. (Is a good sold to France at $2 per buckle earning a higher price than that same 8-digit HS good
sold to France at $350 per ton?) Accordingly, for the regression results reported here we keep only the main unit in
each HS line, by value, and drop the others.
5
The data-cleaning issues relating to product descriptions and incompatible quantity units pose serious challenges to
work in this area. HMS, by using 10-digit U.S. data, may have sidestepped this problem; MZ, with 8-digit Chinese
data, may not have.
7
between firm productivity and export prices, and our preliminary finding is that firm productivity
is associated with lower prices, a result that stands in contrast to other results in the literature.
An increase of 10 percent in firm productivity is associated with a 2 percent decrease in product
prices.
A. Indian Export Price Patterns
Most generally, exporters’ revenue changes are the net effect of revenue changes on their
intensive and extensive margins. Changes on the intensive margin are driven by price and
quantity changes on shipments to (in some sense) “existing” markets, while revenues on the
extensive margin are driven by (in some sense) “new” sales. There is considerable latitude in
defining what constitutes “existing” or “new” markets, particularly when dealing with data as
fine grained as ours at the firm, product and destination level, by year. For instance, is it an
“extensive” or “intensive” change when a firm that has been exporting a good to, say, five
destinations, begins selling to a sixth? Or when a firm resumes sales to a destination after a
year’s hiatus, though it has been selling closely-related products to that destination continuously?
The trade literature offers no consensus methodology on decomposing revenue changes into their
intensive and extensive components, let alone decomposing changes on the intensive and
extensive margins into price and quantity components.6
As a first cut at assessing Indian exporters’ pricing behavior, we consider a narrow
definition of the intensive market in order to decompose revenue changes into their price and
quantity components. We restrict ourselves to the firm-product-destination combinations in the
TIPS data that continue for all four sample years, the narrowest possible definition of the
The literature on this topic is increasing, and “count” methods remain popular though not dominant. See
Besedes and Prusa (2011), Turkcan (2014), and Eaton et al. (2008).
6
8
intensive margin in firm-product-destination data. In an environment with high exit rates, these
firm-product-destination combinations can plausibly be taken to represent the behavior of some
of India’s most successful exporters.7
We adopt a simple decomposition, as follows, where f, p, and d (d = 1, 2, …D) subscript
firms, products and destination markets (and where D represents the total number of a firm’s
destinations for any given product p). Define Pfgd and Qfgd to be the unit value and quantity of
firm f’s shipment of product p to destination d. Suppressing time subscripts, in any given period
t the firm’s revenue from exporting a product, Rfp, is
(1)
𝑅𝑓𝑝 = ∑𝐷
𝑑=1 𝑃𝑓𝑝𝑑 𝑄𝑓𝑝𝑑 .
Taking the total differential of Rfp with respect to Pfpd and Qfpd, and defining fpd to be the firm’s
product p revenue share attributable to sales in destination d, we can decompose intensive
𝐼𝑀
margin export revenue growth for product p ( 𝑅̂𝑓𝑝
) into the (destination-weighted average)
contributions of price and quantity changes across destinations:
(2)
𝐼𝑀
𝐷
̂
̂
𝑅̂𝑓𝑝
= ∑𝐷
𝑑=1 𝜃𝑓𝑝𝑑 𝑃𝑓𝑝𝑑 + ∑𝑑=1 𝜃𝑓𝑝𝑑 𝑄𝑓𝑝𝑑 .
7
For the purpose of these decomposition results we performed the analysis at the 8-digit HS level; we did
not make use of the written product descriptions. We do however use the full product detail for the
regression results in the subsequent section.
9
𝑃̂𝑓𝑝𝑑 and 𝑄̂𝑓𝑝𝑑 represent the rate of change in unit value and quantities, respectively, for
destination d, the weighted average of which are the intensive margin’s price change
𝐷
̂
̂
(∑𝐷
𝑑=1 𝜃𝑓𝑝𝑑 𝑃𝑓𝑝𝑑 ) and quantity change (∑𝑑=1 𝜃𝑓𝑝𝑑 𝑄𝑓𝑝𝑑 ).
Results are in Table 7, all at the HS 8 level. The top panel summarizes the weighted
price and quantity changes found in our 9,834 unique firm-product combinations; note that each
such combination yields an observation regardless of the number of destinations to which the
firm ships the product. The table reports median price and quantity changes by fiscal year (note
that 1999 drops out as the base year for calculating rates of change). Strikingly, revenue changes
appear to be dominated by quantity changes. In the pooled data across all years, the median
firm-product annual revenue change was 29.8 percent, while the median price change was -0.3
percent and the median quantity change was 30.2 percent. A similar pattern emerges when the
firm-product price and quantity changes are aggregated across products (to yield each firm’s
weighted average price and quantity changes, the data in the middle panel) or across firms (to
yield each product’s weighted average price and quantity changes, the bottom panel). 8
On their face, these descriptive results suggest that successful Indian exporters, at least in
this period, raise revenue by raising quantity, not price.
MZ offer an alternative approach to assessing the contribution of price changes to export
revenue growth. MZ use simple linear double-log regressions with fixed effects to explore the
relation between revenue and unit values; this essentially correlative approach allows use of all
observations (i.e. from the extensive and intensive margins, however defined), as here:
(3)
ln(Pfp) =  + ln(Rfp) + p + fp
8
In results not reported here, we applied this decomposition to goods based on their Rauch (1999) categorization
(homogenous, reference priced, or differentiated). We use a concordance between the SITC and the HS to mark
those Indian exports that are differentiated. Results were qualitatively unchanged.
10
(4)
ln(Pfpd) =  + ln(Rfpd) + pd + fpd
Equation (3) is run on observations aggregated to the firm-product level and includes product
fixed effects p. Equation (4) is run on more disaggregated observations—that is, at the firmproduct-destination level, with product-destination fixed effects pd. Equation (5) replicates the
more disaggregated approach of (4) but with product-firm, rather than product-destination, fixed
effects:
(5)
ln(Pfpd) =  + ln(Rfpd) + pf + fpd.
With Chinese data MZ find strongly statistically significant positive results, and interpret
this as evidence that higher revenue is driven by higher unit prices, an effect they attribute
product quality upgrades. Our results for India are in Table 6, where columns (1)-(3) correspond
to equations (3)-(5) above. Again, our estimates are based on four full fiscal years of pooled data
across 2000-2003. The coefficients in columns (1) and (2) are estimated to be 0.13 and 0.11,
respectively, and highly significant; these estimates are comparable to MZ’s 0.09 and 0.08 for
China. Controlling for product and destination variation, firms that achieve higher revenues do
so with higher prices (equation 4, column 2); and when controlling for product and firm
variation, firms raise prices across destinations to raise revenues (equation 5, column 3). Finally,
in column (4) we regress price on the number of destinations a firm serves. The coefficient is
negative and strongly statistically significant, in stark contrast to MZ’s finding of a strong
positive relationship between prices and the number of destination markets for Chinese firms.
We emphasize that these coefficients are evidence of correlation rather than proof of
causation. They suggest that Indian firms are able to raise prices to raise revenues in some
situations, such as in higher income destination markets. Are these results inconsistent with the
findings from our decompositions? Possibly not. For instance, it is possible the Indian firms can
11
successfully price-to-market, an effect embedded in the level of their prices for certain goods in
certain markets, even as revenue growth on those markets over time is driven by quantity
changes, not price changes. That Indian firms that export to the largest number of destination
markets do not raise prices is also consistent with the decomposition results above.
We emphasize however that the results so far are unconditioned. In the next section we
examine how firm prices are associated with a range of firm and destination characteristics.
B. Controlling for Destination Market and Firm Characteristics
In Table 8 we consider the relationship between export unit values and a set of firm and
destination-market characteristics. The firm characteristics include the capital to labor ratio,
labor usage (which we take as a proxy for size), and TFP9; destination-market characteristics
consist of distance, remoteness, GDP, and GDP per capita.
We follow HMS by using product fixed effects, and destination standard-error clusters.
Column (1) is a baseline regression with no selection correction. Columns (2) through (4)
employ, successively, three different selection correction techniques. We must consider the
possibility of selection bias because firm prices are only observed if firms choose to export to
particular destinations. To account for this, we implement Wooldridge’s (1995) sample
correction procedure also suggested in HMS. Specifically, the procedure first regresses log
export values by firm, product, and destination on all exogenous market characteristics Xd, using
a left-censored tobit model. Residuals from this first-stage regression are then used as an
additional regressor in our log price equation. Results in column (2) come from this two-stage
Here capital is measured as the size of a firm’s gross fixed assets, and labor is measured as the wage and salary
bill. We calculate TFP using the Levisohn-Petrin (2003) technique, following Topolova and Khandelwal’s (2011)
approach (see pages 998-999) to calculating each firm’s productivity index. One distinction in our approach from
Topolova and Khandelwal’s is that their measure of firm output is sales, ours is value-added.
9
12
Wooldridge technique. Because the first stage is estimated over the sample with positive exports
(i.e., there are no zeros), it is not our preferred technique. Column (3) is the two-stage
Wooldridge correction with the first stage estimated over an expanded sample of all possible
firm-destination-year combinations; that is, it is applied to a “rectangularized” data set with zeros
added.10 Finally, column (4) reports a regression which implements a three-stage correction used
by HMS, with the first stage estimated over the expanded sample with zeros. Product fixed
effects are used in all models.
These results are drawn from a sample that has been trimmed based upon the 5th and 95th
percentiles of the distribution of the TFP variable. This variable has some extreme values,
possibly related to data reporting errors for the firms in the sample. We observe the outliers for
firms in particular years, and not typically by firms across all four years, which increases our
doubts about the validity of the outliers. In results not reported here we estimated our equation
across two other trim thresholds (90/10, 80/20) and on the entire sample.
The results are robust
to the various trims, but the results for TFP change and are insignificant in the entire sample.
Several interesting generalizations emerge when considering Table 8. Similar to HMS
we find a positive association between unit values and destination GDP and GDP per capita.
The distance between the Indian exporter and the destination country is negative and significant
when (in column 4) we apply the Harrigan selection correction, and not significant in the other
columns. In like manner remoteness is statistically significant only when we apply the Harrigan
correction. Our measure of firm size, labor (the wage bill), is positively associated with unit
values, though the capital to labor ratio is significant in only one equation. The strongest results
in term of statistical significance are obtained in with the Harrigan correction; the weakest
10
This is feasible with the smaller matched data set used for the TIPS-Prowess estimations; to do this with the full
TIPS data base is computationally prohibitive.
13
performance is the Wooldridge correction (column 3) on the “rectangularized” data. We
consistently find over all specifications however that firm TFP is negatively associated with
export prices. Moreover and in contrast to others in the literature we find no positive association
between export prices and distance. Overall our model performs well; R-squared values are
about 86 percent in each case.
Consider next the TFP variable. We find a negative association between firm
productivity and firm unit values. The coefficient is about -0.17 in all three of the four
regressions, which means that firms with a ten percent higher productivity charge about 2
percent lower prices. The result for firm productivity is in contrast to HMS and MZ who both
find a positive association (for U.S. and Chinese firms) between the two variables. We note that
our negative coefficient on productivity is robust to how this variable is measured. When we use
value-added per worker as a measure of firm productivity we again obtain a consistently negative
(and statistically significant) association with firm prices. 11
Overall our results are quite surprising—HMS find the capital intensive U.S. firms charge
lower prices, and more productive firms charge higher prices. Distance in their study is
positively associated with price increases. Their interpretation of the coefficient on productivity
is that more productive firms engage in quality upgrading, a result we do not find for Indian
exporters. Our finding that distance is not positively associated with higher prices is further
evidence that Indian exporters are not engaging in quality upgrading. From Hummels and Skiba
(2004) we expect exporters to ship the higher quality products (the “good apples”) to the more
distant locations, which would be reflected in a positive and significant distance coefficient.
This for example is what HMS found for the United States. In India however the distance
11
These results are not reported here, but are available from the authors.
14
coefficient is negative and significant when we apply the Harrigan correction, and statistically
insignificant in all other equations.
If our results are correct it raises the question of why Indian exporters are not competing
on the basis of product quality to the extent found in other countries. This result may reflect how
early our data are in India’s timeline of market opening (1999-2003); perhaps during the early
reform years firms prefer to increase market share rather than engage in upgrading product
quality. Another possibility may be that high-productivity firms in India enjoyed substantial
economies of scale in this period, which allowed them to lower prices in the context of a rapidly
increasing demand.
We intend to explore why the Harrigan correction yields such strong results in contrast to
the weaker results for the Wooldridge correction. We further intend to explore our results for
productivity and distance by looking at product and geographic subsamples. A number of other
extensions are planned to explore these preliminary findings from the matched firm characteristic
and trade data. We lack a measure of labor skill, and there are important missing firm-level
variables (like private domestic, private foreign, and state ownership) that need to be included in
the analysis. All these will be added as we exploit the Prowess database. Finally, we intend to
explore how alternative fixed effects and standard-error clusters affect results.
IV.
Theoretical Discussion
We briefly examine two theoretical stories which help to elucidate both the state of the literature
and the place of our paper in it. In the first, firms are competing only on price as goods are not
differentiated by quality. Melitz (2003) introduces firm level marginal cost heterogeneity with
beachhead costs in a model of monopolistic competition. The model finds that export prices are
15
decreasing in distance, and increasing in the size and remoteness of export markets. Melitz and
Ottaviano (2011) use a linear demand system in a model of monopolistic competition. The
model predicts that more productive firms will set lower prices, that export prices will decline
with distance and market size, and increase with remoteness.
These predictions run counter to a number of empirical studies, as noted above, which
have found a positive correlation between productivity and price. The common explanation is
that competition depends on both quality and price. Baldwin and Harrigan (2011) modify the
Melitz (2003) model to include a quality dimension. In their model firms with higher unit costs
produce the higher quality goods and, because quality is increasing in cost as a sufficient rate,
they charge a lower price per unit of quality than lower cost firms. Here the higher cost (lower
price per unit of quality) firms charge higher prices, and the model predicts that export prices are
increasing in distance, and decreasing in market size and remoteness, the opposite to Melitz
(2003).
V.
Conclusion
We find a negative association between firm productivity and export prices for Indian firms.
This suggests that firms are not competing based upon quality. Moreover we also find either an
absence of association, or a negative association, between distance to market and export prices.
These two results stand in contrast to what others have found for exporting firms for other
countries. In comparing our results to the theoretical predictions our preferred model is column
(4) with the Harrigan selection correction; it performs well empirically and is a less restrictive
correction than that applied by Wooldridge (our columns (2) and (3)). Our results in equation (4)
16
are consistent with the predictions of Melitz (2003) in which export prices are increasing with
market sizes and remoteness, and decreasing with distance.
17
Table 1.
Indian Exporters
Year
Totals:
Exporters
Destinations
Sectors
Products
2000
2001
2002
2003
12,657
272
98
51,668
22,941
228
98
76,450
20,891
209
99
79,374
25,212
223
98
67,947
Means:
Destinations
Sectors
Products
Value
3.19
2.14
32.69
283,280
2.87
2.15
30.70
376,283
3.02
2.24
33.62
372,596
2.90
2.11
29.96
314,848
Means, Top 1%
Destinations
Sectors
Products
Value
Means, Top 5%
Destinations
Sectors
Products
Value
Means, Top 10%
Destinations
Sectors
Products
Value
Means, Top 20%
Destinations
Sectors
Products
Value
Means, Bottom 80%
Destinations
Sectors
Products
Value
16.95
6.09
288.61
9,685,884
10.58
4.25
149.30
3,598,157
8.46
3.68
110.66
2,177,746
6.60
3.19
84.23
1,259,511
2.33
1.88
19.81
39,270
15.96
5.25
255.76
14,579,223
17.11
5.55
280.09
15,202,731
19.28
6.05
333.57
14,435,995
10.01
3.99
129.75
4,983,483
10.68
4.10
154.88
5,084,675
11.84
4.40
178.74
4,625,873
7.94
3.59
99.68
2,961,283
6.02
3.13
75.40
1,686,531
2.08
1.91
19.53
48,738
8.43
3.68
116.62
2,990,010
9.04
3.91
133.50
2,669,991
6.45
3.26
86.88
1,686,205
6.81
3.45
95.48
1,475,646
2.16
1.98
20.31
44,213
1.92
1.77
13.59
24,678
Source: Authors’ calculations.
18
Table 2.
Year
2000
2001
2002
2003
Table 3.
Products
1
2-5
6-10
>10
Total
Distribution of Export Values, Percent Shares
Top 1%
34.04
38.68
40.62
45.83
Top 5%
63.42
66.22
68.20
73.43
Top 10%
76.83
78.69
80.24
84.80
Top 20%
88.91
89.64
90.51
93.73
Bottom
80%
11.09
10.36
9.49
6.27
Firms’ Export Market and Product Diversification, 2003
Destinations
1
2-5
5,437
59
4,334
2,538
1,457
1,747
2,102
4,654
13,330
8,809
6-10
0
13
72
1,882
1,967
>10
0
1
2
1,103
1,106
Total
5,496
6,706
3,278
9,732
25,212
Source: Authors’ calculations.
19
Table 4.
Year
2000
2001
2002
2003
Export Entry and Exit
Exits
0
4,342
9,644
9,103
Table 5.
Year
2000
2001
2002
2003
Entries Net Entry
12,657
12,657
14,626
10,284
7,594
-2,050
6,538
-2,565
Total
Firms
12,657
22,941
20,891
18,326
%
Entries
100%
64%
36%
36%
% Net
% Exits
Entries
0% +100.00%
19%
+45%
46 %
-10%
50%
-14%
Export Survival Rates by Cohort
2000
100%
66%
54%
49%
2001
2002
2003
100%
49%
34%
100%
38%
100%
Total
12,657
22,941
20,891
25,212
20
Table 6.
Indian Exporters’ Pricing Behavior
All observations: not restricted to continuing firm-product-destinations
VARIABLES
logrev
(1)
logprice
(2)
logprice
(3)
logprice
0.132***
(0.006)
0.109***
(0.008)
0.071***
(0.004)
lognumdest
Observations
R-squared
Fixed effects
SE clusters
238,915
0.820
Prod
Firm
329,459
0.917
Prod-Dest
Prod-Dest
329,459
0.964
Prod-Firm
Prod-Firm
(4)
logprice
-0.035***
(0.014)
238,915
0.814
Prod
Firm
Note: Pooled annual data for fiscal years 2000-2003. All regressions include year fixed effects.
Wholesalers excluded. Quantity units are harmonized and each regression is run on the dominant
quantity unit for each product. Regressions (1) through (4) can be contrasted to China results in
MZ tables III, IV, VI, and V(a), respectively.
21
Table 7. Price and Quantity Contributions to Changes in Revenue, 2001-2003
Annual median percent changes in price, quantity and revenue; continuing firm-productdestinations only, at HS 8.
By Firm-Product (n = 9834)
Year
Price
2001
-2.3
2002
0.8
2003
0.7
Pooled, all years
-0.3
Quantity
60.6
1.1
38.3
30.2
Revenue
55.2
3.6
40.0
29.8
By Firm (n = 5787)
Year
2001
2002
2003
Pooled, all years
Price
-1.8
2.0
0.8
0.1
Quantity
88.5
17.1
58.8
50.0
Revenue
80.9
18.9
60.4
48.1
By Product (n = 1878)
Year
Price
2001
0.2
2002
-0.5
2003
1.1
Pooled, all years
0.3
Quantity
157.7
28.2
91.0
79.6
Revenue
136.8
23.7
86.9
75.4
22
Table 8.
Firm-Product Pricing by Destination and Firm Characteristics
VARIABLES
loggdppc
loggdp
logdist
logremoteHMS
logtfp
logklabor
loglabor
selection0
(1)
logprice
0.0892***
(0.0209)
0.0366***
(0.0133)
-0.0242
(0.0473)
0.00446
(0.0432)
-0.171**
(0.0827)
0.0823
(0.0554)
0.0645*
(0.0345)
Trimming at percentiles 5 and 95
(2)
(3)
logprice
logprice
0.0792***
(0.0211)
0.0318**
(0.0148)
0.00650
(0.0496)
-0.0302
(0.0469)
-0.202**
(0.0844)
0.187***
(0.0686)
0.0859***
(0.0310)
0.211***
(0.0464)
selection1
0.0118
(0.0483)
-0.148
(0.102)
0.256
(0.180)
-0.290
(0.183)
-0.175**
(0.0841)
0.0763
(0.0524)
-0.0244
(0.0689)
0.177***
(0.0277)
0.271***
(0.0540)
-0.373***
(0.0661)
0.361***
(0.0779)
-0.162**
(0.0816)
0.0933*
(0.0560)
0.181***
(0.0334)
-0.0403*
(0.0223)
selection2
Observations
20,850
20,850
R-squared
0.862
0.871
Fixed effects
Prod
Prod
SE clusters
Dest
Dest
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
(4)
logprice
0.211***
(0.0464)
20,850
0.862
Prod
Dest
20,850
0.871
Prod
Dest
Notes: Pooled annual data for fiscal years 2000-2003. All regressions include year fixed effects.
Wholesalers excluded. Quantity units are harmonized and each regression is run on the dominant
quantity unit for each product. Compare with HMS Table 4.
23
References
Alchian, A. and W. Allen. 1964. University Economics. Belmont, CA: Wadsworth.
Bernard, A., B. Jensen, S. Redding, and P. Schott. 2007. “Firms in International Trade.”
Journal of Economic Perspectives, 21 (3), 105-130.
Bernard, A. and B. Jensen. 1999. “Exceptional Exporter Performance: Case, Effect, or Both?”
Journal of International Economics, 47(1), 1-25.
Besedes, T. and T. Prusa. 2011. “The Role of Extensive and Intensive Margins and Export
Growth.” Journal of Development Economics, 96, 371-379.
Eaton, J., M. Eslava, M. Kugler, and J. Tybout. 2008. “Export Dynamics in Colombia: FirmLevel Evidence,” chapter 8 in E. Helpman, D. Marin and T. Verdier, eds., The
Organization of Firms in a Global Economy, Cambridge, MA: Harvard.
Gopalan, R., V. Nanda and A. Seru. 2007. “Affiliated Firms and Financial Support: Evidence
from Indian Business Groups.” Journal of Financial Economics, 86, 759-795.
Gorg, H., H. Laszlo, and B. Murakozy. 2010. “Why Do Within Firm-Product Export Prices
Differ Across Markets?” Kiel Institute for the World Economy, Working Paper 1596
(February).
Harrigan, J., X. Ma and V. Shlychkov. 2011. “Export Prices of U.S. Firms.” NBER Working
Paper No. 17706.
Hummels, D. and A. Skiba. 2004. “Shipping the Good Apples Out? An Empirical Confirmation
of the Alchian-Allen Conjecture.” Journal of Political Economy, 112 (6) 1384-1402.
Khanna, T. and K. Palepu. 2000. “Is Group Affiliation Profitable in Emerging Markets? An
Analysis of Diversified Indian Business Groups.” The Journal of Finance, 55 (2), 867891.
Kotwal, A., B. Ramaswami, and W. Wadhwa. 2011. “Economic Liberalization and Indian
Economic Growth: What’s the Evidence?” Journal of Economic Literature, 49(4), 11521199.
Levinsohn, J. and A. Petrin. 2003. “Estimating Production Functions Using Inputs to Control
for Unobservables.” Review of Economic Studies, 70, pp. 317-341.
Manova, K. and Z. Zhang. 2012. “Export Prices Across Firms and Destinations.” The
Quarterly Journal of Economics, 127, 379-436.
24
Martin, J. 2009. “Markups, Quality and Transport Costs.” Institut National de la Statistique et
des Etudes Economiques, Centre de Recherche en Economie et Statistique Working
Paper 2010-17 (September).
Melitz, M. 2003. “The Impact of Trade on Intra-Industry Reallocations and Aggregate Industry
Productivity.” Econometrica, 71(6), 1695-1725.
Topalova, P. and A. Khandelwal. 2011. “Trade Liberalization and Firm Productivity: The Case
of India.” The Review of Economics and Statistics. 93(3), pp. 995-1009.
Turkcan, K. 2014. “Investigating the role of Extensive Margin, Intensive Margin, Price and
Quantity Components on Turkey’s Export Growth During 1998-2011.” MPRA Paper No.
53292.
Wooldridge, J. 1995. “Selection Corrections for Panel Data Models Under Conditional Mean
Independence Assumptions.” Journal of Econometrics 68(1), 115.132.
25
Appendix
Figure A.1
Log Export Price
0
.1
Density
.2
.3
Full and analytical datasets
-20
-10
0
logprice
10
20
kernel = epanechnikov, bandwidth = 0.1094
Table A.1
2000
2001
2002
2003
Dataset
Total firms
Full
Analytical
23,314
7,248
22,668
12,589
22,368
11,602
28,652
13,615
Total destinations
Full
Analytical
204
171
180
154
197
174
211
187
Total products
Full
Analytical
10,250
3,986
10,161
6,509
10,345
6,248
9,329
5,506
Av. destinations
Full
Analytical
2.09
2.29
2.72
2.38
2.68
2.34
2.79
2.62
Av. products
Full
Analytical
4.38
2.85
4.65
2.85
4.63
2.82
4.66
2.96
Av. export value
Full
Analytical
521.15
47.72
43.35
64.39
46.13
69.99
6.65E+3
0
55.57
Med. export value
Full
Analytical
6.73
8.90
5.44
10.31
4.78
9.12
3.18
5.72
26
Notes: Years are fiscal years. Export values are in thousand dollars.
27