Economics Letters 123 (2014) 37–41

Contents lists available at ScienceDirect

Economics Letters

journal homepage: www.elsevier.com/locate/ecolet

Financial reforms, product differentiation, and trade

Abdul Munasib a,1, Devesh Roy b,∗

a Department of Agricultural and Applied Economics, University of Georgia, 213 Stuckey Building, 1109 Experiment St, Griffin, GA 30223, USAb Markets, Trade and Institutions Division, International Food Policy Research Institute, 2033 K. St., N.W., Washington, DC 20006-1002, USA

h i g h l i g h t s

• Differentiated product exports is associated with larger share in world trade.• We show impacts of financial reforms on exports of differentiated products.• Reforms of greater intensity increase exports of differentiated products more.• Reforms also differ in terms of time taken to affect exports.

a r t i c l e i n f o

Article history:Received 14 August 2013Received in revised form9 January 2014Accepted 12 January 2014Available online 21 January 2014

JEL classification:F10F14

Keywords:Product differentiationExporter–time fixed effectsExporter–industry fixed effectsExports

a b s t r a c t

We hypothesize that exports of differentiated products, which entail greater upfront costs, increase moreas financial reforms take place. We find strong and robust empirical support of this hypothesis with acomprehensive set of measures of reforms encompassing the banking sector, interest rates, equity andinternational capital markets.

© 2014 Elsevier B.V. All rights reserved.

1. Introduction

Adistinguishing feature of the process of industrialization in thelast century has been a rapid increase in exports in countries suchas Korea and Taiwan (Artopoulos et al., 2011). Another stylizedfact is a near-monotonic increase in world trade over the last fewdecades that is rooted in a similar increase in the trade of differenti-ated products (Fig. 1). Exporting differentiated products, however,involves higher upfront fixed costs related to the identification anddevelopment of market opportunities; with many varieties, dif-ferentiated products often do not have established prices (Rauch,1999). A specific example of greater costs in exporting differenti-ated products is explained in Fink et al. (2005) who find greatereffect of communication costs on exports of such products.

∗ Corresponding author. Tel.: +1 202 862 5691; fax: +1 202 467 4439.E-mail addresses:munasib@uga.edu (A. Munasib), d.roy@cgiar.org (D. Roy).

1 Tel.: +1 770 228 7231x113.

0165-1765/$ – see front matter© 2014 Elsevier B.V. All rights reserved.http://dx.doi.org/10.1016/j.econlet.2014.01.006

In general, search costs, product design costs and othermarket-ing and distribution costs being higher compared to homogeneousproducts, credit is comparativelymore important for exports of dif-ferentiated products (Becker et al., 2013). Indeed, the state of finan-cial development – measured by accounting standards and by theratio of bank credit to GDP – is positively associated with exportsof differentiated products (Becker et al., 2013). Financial reformsby relieving credit constraints are, therefore, expected to have dif-ferential impacts on exports by product type.

Using the Rauch (1999) classification we test the hypothesisthat improved credit availability leads to relatively greater increasein exports of differentiated products.2 We exploit shocks to avail-ability of finance due to reforms in 88 countries for the period

2 Rauch (1999) defined homogeneous goods as products whose prices are seton organized exchanges. Goods that are not traded on organized exchanges, butpossess a benchmark price, were defined as reference priced. Products whoseprices are not set on organized exchanges and/or lack a reference price were

38 A. Munasib, D. Roy / Economics Letters 123 (2014) 37–41

Fig. 1. World trade.

Fig. 2. Industry-level trade by financial reform and product differentiation. Notes:(a) In both figures trade values are in ‘000 2005 US GDP, (b) In Fig. 2, trade valuesare those in SITC 4 classification industries, (c) In Fig. 2, conservative Rauch productdifferentiation classification is used. Exports are on the vertical axis, z. The financialreform index is on x axis while the product differentiation is on y axis. Sincethe product differentiation variable is a binary indicator variable, all observationsappear either at y = 0 (non-differentiated) or at y = 1 (differentiated).When y = 0,moving along the x axis from 0 to 1 – i.e., higher levels of reforms – exports increase.Similarly, at y = 1 moving along the x axis from 0 to 1, again, exports increase.Comparing the scatter points at y = 0 with those at y = 1, exports increase morefor differentiated products as the level of reforms increases.

1988–2005 based on the recently developed financial reforms databyAbiad et al. (2010). Combining this datawith industry level tradeflows, the 3-dimensional plot in Fig. 2 provides the illustrative sup-port for our main hypothesis: the mean levels of exports of dif-ferentiated products are higher under reformed financial markets.Empirical results based on a fixed effectsmodel and an event studyapproach provide strong evidence in support of our hypothesis.

2. Data on financial reformmeasures and exports of differenti-ated products

Abiad et al. (2010) provide a measure of liberalization for eachof the seven different aspects of financial reforms: directed credit,interest rate controls, entry barriers, banking supervision, pri-vatization, international capital, and equity markets. Each mea-sure comprises four categories indicating different intensity of

labeled as differentiated. Rauch assigns each SITC industry to one of threecategories: differentiated products (e.g., apparel), goods with reference prices (e.g.,polymerization products) and exchange-traded goods (e.g., lead).

Table 1Descriptive statistics of the key variables.

Mean s.d.

Log trade (in ‘000 2005 US GDP) 7.43 3.81Reform index 0.71 0.22Product differentiation dummy 0.62 0.49

Notes: (a) N = 753, 166. (b) Year range is [1988, 2005]. (c) Only non-zero SITC-4digit industry log trades are in the sample with the maximum value of 18.5. (d) Thereform index is a continuous variable with range [0, 1]. (e) Product differentiationdummy: 0 = homogeneous, 1 = differentiated (Rauch classification, conservative).

reforms: fully repressed = 0, partially repressed = 1, partially lib-eralized = 2, and fully liberalized = 3. Overall financial reform isobtained by summing over seven individual scores (each coded as0, 1, 2 or 3) and dividing it by 21, hence normalizing it to a [0, 1]interval.

The industry level trade data comes fromComtrade provided bythe United Nations Conference on Trade and Development (UNC-TAD). The total number of exporters in the sample is 88. Based onStandard Industrial Trade Classification (SITC 4) data in Comtrade,about 62% of the trade between 1988 and 2005 were in differenti-ated products (Table 1).

We convert each reform measure into 4 binary indicator vari-ables: for each measure we identify if a country is ‘‘fully repressedor not’’, ‘‘partially repressed or not’’, ‘‘largely liberalized or not’’,and ‘‘fully liberalized or not’’. The sample is summarized in Table 2by each aspect of reform and in terms of its different intensity cat-egories.

3. Empirical framework and results

3.1. Fixed effect regressions

We estimate the following equation to estimate the diverseimpact of financial reforms on exports of differentiated products:

ln(X sjt) = αk +

3m=1

βmk

Rmjkt ∗ PDs

+ θj ∗ t + θj ∗ s + ξ s

jt , (1)

where, X sjt is country j’s exports in industry s at time t, Rm

jkt dummydenotes the k-th reformof intensitym in country j at t (for example,partially repressed equity markets at time t in country j), PDsdummy equals 1 if sector s is differentiated and equals 0 when it isnon-differentiated/homogeneous. The error term is denoted by ξ .The omitted category for each reform is the state of full repression.

The specification in Eq. (1) incorporates two kinds of fixed ef-fects: (a) exporter–time fixed effect, i.e., a separate fixed effect forevery exporter in each time period (θj∗t), and (b) exporter–industryfixed effect, i.e., a separate fixed effect for each industry in each ex-porting country (θj ∗ s). In a regression for a specific aspect of re-form (e.g., directed credit), the exporter–time fixed effects accountfor not only all the country specific factors (observed factors suchas GDP and unobservable factors such as institutional quality) butalso the states of other reforms in the country. Additionally, theyaccount for other channels affecting trade in differentiated prod-ucts that have been identified in the literature (e.g., country specificdiaspora, ethnic networks, etc.). With the exporter–industry fixedeffects, country–industry attributes such as reputation (for exam-ple, automobiles from Germany) are accounted for. The specifica-tion in Eq. (1), therefore, limits the possibility of omitted variablebiases on several fronts.

In each of the seven regressions, our coefficients of interest aretheβm

k ’s.We expectβmk > 0∀k,m, i.e., reformswith varying inten-

sities to increase exports of differentiated products relative to ho-mogeneous products. Finally,we expect greater effects fromhigherintensity of reforms unless diminishing returns from reforms be-come a factor.

A. Munasib, D. Roy / Economics Letters 123 (2014) 37–41 39

Table 2Intensities of financial reforms (descriptive statistics).

Directed credit Interest ratecontrols

Entry barriers Bankingsupervision

Privatization Internationalcapital

Securitiesmarket

Mean s.d. Mean s.d. Mean s.d. Mean s.d. Mean s.d. Mean s.d. Mean s.d.

Intensity measure 2.18 0.96 2.67 0.77 2.40 0.90 1.43 0.99 1.68 1.16 2.30 0.93 2.19 0.88

Dummy variables created from the intensity measures

Fully repressed 0.05 0.22 0.04 0.19 0.14 0.35 0.20 0.40 0.22 0.41 0.05 0.22 0.03 0.17Partially repressed 0.24 0.43 0.07 0.25 0.14 0.35 0.34 0.47 0.22 0.41 0.17 0.38 0.23 0.42Partially liberalized 0.20 0.40 0.08 0.28 0.17 0.38 0.29 0.46 0.22 0.41 0.20 0.40 0.27 0.44Fully liberalized 0.52 0.50 0.81 0.39 0.64 0.48 0.17 0.37 0.34 0.47 0.58 0.49 0.47 0.50

Note: Reforms intensity measures {0 = fully repressed, 1 = partially repressed, 2 = partially liberalized, 3 = fully liberalized}.

Table 3Regression of log of SITC 4-digit industry exports on the interaction of financialreform and product differentiation.

Financial reform index * (product differentiation) 0.6944***

(0.135)Exporter–time FE YesExporter–industry FE Yes

N 753166Adjusted R-squared 0.882MSE 1.2567

Exporter–time: F(1362, 693 446) 36.82Exporter–industry: F(58 356, 693 446) 50.881Both: F(59 719, 693 446) 93.445

Notes: (a) 1362 exporter–time fixed effects and 58356 exporter–industry fixedeffects. (b) Bootstrapping standard errors, clustered over exporters. (c) All F-statistics significant at 1%.∗ p < 0.1.∗∗ p < 0.05.*** p < 0.01.

While we focus individually on each of the seven aspects of re-forms, to begin with, we use the ‘‘Financial Reform Index’’ (or FRI),a continuous variable∈ [0, 1], that combines the different reformsinto one (Table 3). Subsequently, Table 4 presents the results withspecific reforms measured with their respective intensities.

In Table 3, we find strong evidence that countries with finan-cial reforms export more from industries producing differentiatedproducts. The coefficient of interaction of FRI with product differ-entiation indicator, βm

k , is significant at 1% level. The coefficientrepresents an increase in exports of differentiated products byabout 18% of one standard deviation of log exports in our sam-ple. These findings support the idea that financial reforms allevi-ate credit constraints that otherwise would restrict exports andthat this pathway is comparatively consequential for differentiatedproducts.

Table 4 presents estimates for each of the seven reforms withvarying intensities. For banking supervision, privatization of banks,international capital and securities markets, there are significanteffects on exports of differentiated products for any intensity ofreform. Also, for each of these reforms, exports rise monotonicallywith the intensity of reform. At the highest intensity, the interac-tion between securities market reform and product differentiationaffects exports by about 8.4% of one standard deviation of log ex-ports. The corresponding figures for banking supervision, privati-zation and international capital markets reforms are 4.9%, 5.6% and5.3% of one standard deviation of log exports, respectively.

In contrast, the relaxation of interest rate controls affectsexports of differentiated products only when the intensity levelis the highest. In the case of directed credit and entry barriers,only the startup reforms do not show any effect; effects appearas reforms intensify. At the highest intensity, the magnitudes arecomparable to other reforms except for securities market reformsthat have much bigger effects.

These results could follow from the access to credit and types oflenders that can vary with product characteristics. Giannetti et al.

(2011) discuss the ease for sellers of differentiated products in get-ting trade credit. In the case trade credit substitutes more readilywith bank credits then non-banking reforms such as in securitiesmarkets could be comparatively important for differentiated prod-ucts.

3.2. Event history regressions

An alternative approach to assess the impact of financial re-forms on trade in differentiated products is an event study set-ting similar to Trefler (2004) and Manova (2008). This approachuses reform dummies to pin the event of reforms. The event studyestimation equation for reform k (the event) that changed from(fully/partially) repressed to (partially/fully) liberalized state attime t is given as:

X̄js,t+w − X̄js,t−w = πk + φk PDs + θj ∗ t + ςjst , (2)

where, X̄js,t+w − X̄js,t−w denote the change in w-year average ofexports in industry s post and pre reform. Since, these differencescan be negative as well as zero, the dependent variable is not spec-ified in logs. Eq. (2), includes exporter–time fixed effects θj ∗ tand, therefore, does not include changes in exporter specific fac-tors such as GDP. Since differencing is done for each country atthe industry level, the formulation controls for observed and un-observed exporter–industry characteristics. As the change in ex-ports is assessed around an event, only observations that includethe reform of the particular type are retained. Since w denotes theduration over which it is assumed that reforms affect exports ofdifferentiated products, we allow for 5 different timewindows, i.e.,w = 1, 2, . . . , 5 in Eq. (2).

Table 5 presents the results of the event study regressions.After each reform, the exports of differentiated products increaserelative to the homogeneous products. This approach helps pindown the time it takes for reforms to influence exports. Securitiesmarket reforms are associated with significant changes in exportsof differentiated products at the minimum over a 4-yearswindow. On the other hand, directed credit, interest rate controls,entry barriers, and banking supervision reforms tend to affectdifferentiated product exports immediately. As for reforms relatedto privatization of banks and international capital markets, outestimates show effectiveness with a 2-year window and beyond.

Further, out of the seven reforms, in five cases there is strength-ening of the effects with time. The two exceptions are interest ratecontrols and international capitalmarkets: for the former the effectdoes not strengthen while for the latter it decays with time.

3.3. Robustness checks

Although our specifications in Eqs. (1) and (2) limit the possibil-ity of omitted variable bias, we carry out the following robustnesschecks to validate our results. First, Rauch (1999) classification isdone in twoways: conservative classification and liberal classifica-tion. Since our main analysis uses the conservative definition, we

40 A. Munasib, D. Roy / Economics Letters 123 (2014) 37–41

Table 4Regressions of log of SITC 4-digit industry exports on interactions of product differentiation and financial reform intensities.

Directedcredit

Interest ratecontrol

Entrybarrier

Bankingsupervision

Privatization Internationalcapital

Securities market

Partially repressed * (product differentiation) 0.0604 0.1740 0.0884 0.141*** 0.1087** 0.1573** 0.2407***

(0.0664) (0.1259) (0.0993) (0.0524) (0.0539) (0.0723) (0.0524)Partially liberalized * (product differentiation) 0.2028*** 0.1502 0.2025** 0.2293*** 0.2143*** 0.2388*** 0.3881***

(0.0700) (0.1555) (0.0936) (0.0682) (0.0609) (0.0683) (0.0731)Fully liberalized * (product differentiation) 0.3332*** 0.2778** 0.3281*** 0.2938*** 0.3363*** 0.3116*** 0.4905***

(0.0806) (0.1172) (0.0989) (0.0830) (0.0688) (0.0751) (0.0981)

Exporter–time FE Yes Yes Yes Yes Yes Yes YesExporter–industry FE Yes Yes Yes Yes Yes Yes Yes

N 753166 753166 753166 753166 753166 753166 753166Adjusted R-squared 0.8819 0.8819 0.8819 0.8819 0.8819 0.8819 0.8819MSE 1.257 1.2573 1.257 1.257 1.257 1.2571 1.2571

Exporter–time: F(1362, 693 444) 43.3710 45.3480 41.5800 37.7220 44.4650 42.8420 41.7940Exporter–industry: F(58 356, 693 444) 50.9150 50.9230 50.9130 50.8600 50.9570 50.7500 50.5870Both: F(59 719, 693 444) 93.8010 93.2060 94.7910 89.4550 94.3470 88.9540 81.3160

Notes: (a) Each regression has 1362 exporter–time fixed effects and 58356 exporter–industry fixed effects. (b) Bootstrapping standard errors reported. (c) Standard errorsclustered over exporters. (d) All F-statistics significant at 1%.∗ p < 0.1.** p < 0.05.*** p < 0.01.

Table 5Cumulative impact of each financial reform on SITC 4-digit industry exports (event studies regressions).

Directedcredit

Interest ratecontrol

Entrybarrier

Bankingsupervision

Privatization Internationalcapital

Securities market

1-year window

Variety 22.85*** 14.97** 27.46** 22.51*** 8.588 0.796 1.530(8.144) (6.942) (11.11) (8.216) (5.240) (4.076) (2.850)

Observations 20,457 21,732 22,052 28,301 15,024 27,523 19,863MSE 207.7 249.8 320.2 311.3 137.5 182.8 140.0

2-year window

Variety 29.66*** 19.61** 33.99** 26.53*** 14.68** 6.639* 2.190(10.90) (9.674) (13.42) (9.104) (6.666) (3.876) (3.913)

Observations 17,344 17,566 20,139 25,961 12,763 21,568 17,163MSE 262.5 302.7 370.5 324.9 156.8 142.5 171.1

3-year window

Variety 38.55** 21.29* 47.94*** 37.26*** 19.70** 9.535** 5.266(15.40) (11.00) (16.94) (11.47) (9.029) (4.276) (3.854)

Observations 14,756 14,397 17,212 20,771 11,028 17,849 13,621MSE 317.3 338.7 415.0 372.0 190.6 171.5 180.0

4-year window

Variety 47.13** 19.72 59.45*** 42.08*** 25.57* 13.00** 12.41**

(21.21) (13.61) (19.66) (13.71) (14.46) (5.204) (5.066)Observations 12,342 12,200 15,150 18,882 7610 14,258 10,462MSE 365.3 333.0 471.7 388.6 248.7 208.3 224.6

5-year window

Variety 61.25** 25.05 67.19*** 54.39*** 36.53* 10.41** 16.94**

(27.96) (18.17) (23.13) (18.25) (18.74) (5.002) (6.869)Observations 11,301 10,404 13,097 15,529 4976 10,247 9108MSE 436.2 377.9 519.9 458.6 235.3 245.9 269.3

Notes: (a) An ‘event’ in these event studies regressions is defined by the Reform dummy: 0 = fully/partially repressed, 1 = partially/fully liberalized. (b) Each regressionincludes exporter–time fixed effects. (c) Standard errors clustered over exporters.

* p < 0.1.** p < 0.05.*** p < 0.01.

repeat the analyses using the liberal classificationmeasure. The re-sults are robust to using this alternative classification of products.

Second, China has low levels of financial reformsbut a very largeshare in international trade. We, therefore, carry out the aboveanalyses with a sample excluding China. Again, the results arerobust to this sample change.

4. Concluding remarks

Using robust empirical techniques we show that financialreforms disproportionately increase the exports of differentiatedproducts, the main drivers of exports intensification. The resultssuggest that in the trade of differentiated products the concept ofcomparative advantage needs to be expanded to include credit as

a factor. Policies aimed at promoting exports and gaining share inthe world market, therefore, need to be combined with financialsector reforms, especially in countries where the level of financialdevelopment is low.

References

Abiad, A., Detragiache, E., Tressel, T., 2010. A new database of financial reforms. IMFStaff Pap. 57 (2), 281–302.

Artopoulos, A., Friel, D., Hallak, J.C., 2011. Lifting the domestic veil: the challenges ofexporting differentiated goods across the development divide. National Bureauof Economic Research (NBER). Working Paper Number 16947. April.

Becker, B., Chen, J., Greenberg, D., 2013. Financial development, fixed costs, andinternational trade. Rev. Corp. Finance Stud. 2 (1), 1–28.

Fink, C., Mattoo, A., Neagu, C.I., 2005. Assessing the impact of communication costson international trade. J. Int. Econ. 67 (2), 428–445. Elsevier.

A. Munasib, D. Roy / Economics Letters 123 (2014) 37–41 41

Giannetti, M., Burkart, M., Ellingsen, T., 2011. What you sell is what you lend?Explaining trade credit contracts. Rev. Financ. Stud. 24 (4), 1261–1298.

Manova, Kalina, 2008. Credit constraints, equity market liberalizations andinternational trade. J. Int. Econ. 76 (1), 33–47. Elsevier.

Rauch, J.E., 1999. Networks versus markets in international trade. J. Int. Econ. 48,7–35.

Trefler, D., 2004. The long and short of the Canada–US free trade agreement. Amer.Econ. Rev. 94, 870–895.