Dynamics of Services Exports in India

Abstract

The article examines the dynamics in specialization pattern of comparative advantage in services exports from India for the time period 2004–2013, while drawing a distinction between the pre- and the post-crisis trends. The analysis is based on a modified revealed symmetric comparative advantage index. It also employs the Galtonian regression and the Markovian transition matrices to study the distribution and intra-distribution dynamics of export patterns. The results reveal that India has a strong comparative advantage in modern services mainly business services and computer and information services. The findings suggest a stable but broadening pattern of specialization of India’s services exports. Further, the transition probabilities reveal that the chances of remaining in or switching to a state of comparative disadvantage are more pronounced than being or moving to a state of comparative advantage. Overall services exports exhibit high persistence in both the pre- and post-crisis period pointing towards a low impact of crisis on India’s services exports.

JEL: C14, F14, L80

Keywords

Services trade revealed comparative advantage specialization dynamics transition probability mobility indices India

I. Introduction

Significant role that services have emerged to play in the global business scenario is evident from its fast-rising contribution to the integration of economies into international trading system. For instance, the services sector contributes around 70 per cent towards GDP in developed countries and around 50 per cent towards the GDP of developing countries. Consistent with the role played by services sector in the domestic economies, international trade in services has also witnessed unprecedented growth over the past decades (UNCTAD, 2014b). From the period 1995 to 2014, the world exports of services multiplied four times from US$1,179 billion to US$4,872 billion (WTO, 2015). In tandem with the growth in world services trade, India has also experienced a surge in its services exports, witnessing a huge leap from US$6.76 billion in 1995 to US$155.6 billion in 2014 (WTO, 2015). Realizing the intensifying role of services all around, the Uruguay Round of the WTO in 1995 broadened the scope of multilateral negotiations to include services under the General Agreement on Trade in Services (GATS) (Taneja, Mukherjee, Jayanetti, & Jayawardena, 2004).

With this increased involvement in the world markets, a critical issue that emerges is that of the specialization and dynamic shifts in comparative advantage (CA) pattern of services (Widodo, 2009). Following this, there has been a renewed interest in the trade dynamics of countries (Brasili, Epifani, & Helg, 2000; Hinloopen & Marrewijk, 2001; Proudman & Redding, 2000; Redding, 2002) as these often reflect profound structural changes in the whole economy of a particular country. Further, when it comes to analysing trade specialization CA still remains the fundamental theoretical argument underlying the trade patterns. Economic theory argues that, under free trade, countries specialize in and export those products in which they have a CA based upon their factor endowments and technology (De Benedicts & Tamberi, 2001, 2004), while also acknowledging that internal and external shocks may influence the production, technology or institutional systems, inducing a change in the pattern of CA (Fertő, 2007).

Consequently, due to the economic restructuring warranted by market-oriented reforms in the 1990s and subsequent competitive pressure, India was also expected to move towards specialization on the basis of its CA (Nath, Liu, & Tochkov, 2015). Although there exists substantial empirical literature that presents CA measures for different countries, but such studies focusing on services trade appear to be quite scanty. Further, most of the studies on trade patterns are based on the static economy assumptions, with few studies accounting for dynamic pattern of trade especially in the context of European countries. One interesting study by Brasili et al. (2000) compares dynamics in the export patterns of six industrialized and eight Asian emerging economies. The results suggest that though both the groups seem to have similar degree of specialization, yet the latter appear to be more specialized. The industrialized countries have a highly persistent trade pattern but both categories show a tendency towards reduced polarization and symmetric distribution of the specialization index. Similar results have also been reported for European Union and east European countries by Fertő (2007) and Fertő and Soós (2008) and Chiappini (2014) who described the evolution of trade and technology specialization for 11 European countries. These studies, although undertaken for different time periods, do not support the idea of self-enforcing mechanism as suggested by endogenous growth and trade literature. However, they found polarization taking place for Latvia, Estonia, Lithuania and Slovakia and fall in specialization for Czech Republic, Poland and Slovenia. Another study by He (2010) used the same approach to study changes in the dynamics of Chinese agricultural products during pre- and post-accession of China into the WTO, and found that trade patterns follow the trends suggested by the trade theories. Doanh (2011) studies the dynamics of specialization pattern of Vietnam’s commodity exports for the period 2001–2009 using the revealed comparative advantage (RCA) index and the transition matrices and reported that a diversifying pattern of exports exists. One interesting study that explored the evolution of trade specialization in service sector of India reported fairly stable distribution of CA indices during 1980–2006 which implied overall lack of dynamism in the trade-related services sector (Pailwar & Shah, 2009). Using bilateral trade data for 16 service categories, Nath et al. (2015) examined the patterns, evolution and determinants of CA in U.S. services trade with China and India during 1992–2010. The results indicated that although U.S. has CA in most of the services, India and China have gravitated towards gaining CA in the modern services. The former is for a different time period than the present and the latter study services trade particularly with the U.S. There are hardly any studies that deal exclusively with India’s services trends.

Therefore, the main objective of the article is to empirically analyse if there were dynamic shifts in the CA pattern of India’s services exports to the world. In this light, the study particularly addresses the following issues: (i) services wherein India enjoys CA, (ii) specialization pattern of India’s services exports over the period 2004–2013 and (iii) possible shift in CA with the economic crisis of 2008 as the reference point. It may be interesting to study if the dynamics of services have shifted after the crisis, as it had hit India’s major services trade partners, namely the U.S. and Europe.

The rest of the article is organized as follows. The following section describes the data sources and methodology used for measuring dynamic changes in the specialization pattern of India. The next section focuses on the empirical part of the article and presents the results arising out of the distribution and the intra-distribution dynamics of the specialization index. The summary and conclusions have been presented in the last section.

II. Methodology and Data Sources

The most widely accepted tool for measuring trade specialization is based on the concept of CA as propounded by Ricardo. Although the first index of CA was devised by Liesner (1958), it was further refined and popularized by Balassa (1965) in his seminal work for 184 manufactured product categories produced in thirty six developed and developing countries. The Balassa index equals:

B R C A_{j}^{i} = (X ​_{j}^{i} / X^{​ i}) / (X_{j} / X_{w})

(1)

where ‘X ’ represents exports, ‘j’ is the particular service, ‘i’ represents country and ‘w’ represents all the countries of the world as a group. The index measures the product share in total exports of a country to the share of world exports of the product in total exports from the world. However, as its values range from zero to infinity resulting in a skewed distribution, it violates the assumption of normality of the error term, thereby, producing biased regression results. Moreover, the analysis produces results that give much more weight to the values above one when compared to those below one. To deal with this issue of asymmetry, Dalum, Laursen and Villumsen (1998) proposed a more symmetric measure, which is a simple decreasing monotonic transformation of the Balassa index (Widodo, 2009). The Revealed Symmetric Comparative Advantage (RSCA) is formulated as follows:

R S C A = (B R C A - 1) / (B R C A + 1)

(2)

The index value ranges from –1 to +1 thus assigning equal weight to the changes in the index below unity and above unity. CA is revealed if the value of the index is above zero and comparative disadvantage (CD) is revealed for values below zero. Hence, this improvisation over the traditional index of RCA is used to achieve the first objective of the study.

Stability of Indices

Here the focus is on two types of stability as suggested by Hinloopen and Marrewijk (2001): (i) stability of distribution of the index from one period to the other and (ii) stability of the value of index of products (services in this case) from initial to the final period. Both these types are used to analyse the second and the third issues which this article attempts to address, that is, specialization patterns for the entire period under consideration, 2004–2013 and in the light of financial crisis of 2008.

To assess the first type of stability of national export specialization patterns, or alternatively to understand the progression pattern of specialization over time, we follow the approach suggested by Dalum et al. (1998). Here, stability is tested by using the Galtonian regression equation for each country following Hart (1976) who employed this for observing changes in the distribution of income over time. The regression equation is as follows:

R S C A_{j i}^{t 2} = α_{j} + β_{j} R S C A_{j i}^{t 1} + μ_{j i}

(3)

The superscripts t₁ and t₂ refer to the initial year and the final year, α and β are standard regression coefficients and µ is the residual term. The dependent variable, RSCA at time t₂ for sector ‘j’ in country ‘i’ is tested against RSCA at the initial time period. If β = 1, it corresponds to an unchanged pattern from t₁ to t₂. If β>1, the specialization pattern of a country strengthens and 0<β<1 indicates that sectors with initially low RSCAs register increase over time and those with high RSCAs witness decrease. The regression effect is represented by (1–β). However, as pointed out by Dalum et al. (1998), the evolution of degree of specialization cannot be interpreted from the β coefficients. Therefore, following Cantwell (1989), the changes in the degree of specialization can be analysed by using:

σ_{t 2}^{2} / σ_{t 1}^{2} = β_{j}^{2} / R_{j}^{2}

(4)

where R is the correlation coefficient from the regression and σ is the dependent variable’s variance. β = R implies an unchanged pattern of a given distribution, β >R entails an increase in the overall degree of specialization and β<R indicates a fall. (1–R) measures the mobility effect. This allows the analysis of the tendency towards polarization of a country’s trade specialization pattern (Brasili et al., 2000).

The second type of stability deals with the issue of persistence or mobility in the initial patterns of international specialization. This corresponds to the question of intra-distribution dynamics and involves an evaluation into the entire cross-section distribution of the index over time (Proudman & Redding, 2000; Redding, 2002). Following the technique pioneered by Proudman and Redding (2000), and subsequently used by Brasili et al. (2000), Hinloopen and Marrewijk (2001), Redding (2002), Fertő (2007), Fertő and Soós (2008), He (2010), Doanh (2011), Chiappini (2014) and Nath et al. (2015) among others. The study applies the Markovian transition probability matrices model to estimate the intra-distribution dynamics of the RSCA index over time.

In examining specialization of exports, the RSCA index distinguishes between the two states of CA and disadvantage. On this basis, the two states of CA and CD are defined for analysing the intra-distribution dynamic pattern of this specialization. The distribution of RSCA index across services is defined by Q(t), where t is the current time period. The distribution at time t+τ is then given by:

Q_{t + τ} = P * Q_{t}

(5)

where P is the Markov transition probability matrix and Q(t) as defined previously. The matrix is given by:

P = [\begin{matrix} p 11 & \dots & p 1 n \\ ⋮ & ⋱ & ⋮ \\ p n 1 & \dots & p n n \end{matrix}] p_{k l} \geq 0, \sum_{k = 1}^{n} p_{k l} = 1

(6)

where ‘n’ is the number of states (2 in this case, i.e., CA and CD) and p_kl with k, l = 1,…, n is the probability of transiting from initial state in time t to final state in time t+τ (Nath et al., 2015), that is, the number of transitions, which are 2-year, 4-year and 8-year transitions for the period spanning from 2004 to 2013 and for the pre- and post-crisis period, taking 2004, 2008 and 2009, 2013 as the initial and the final year, respectively, for the two time periods.

Mobility or persistence throughout the distribution can be interpreted from the off-diagonal and diagonal elements of the derived probability transition matrices. High values along the diagonal indicate high persistence, while higher values off-diagonal denote high mobility.

In order to further strengthen the above interpretation, appropriate mobility indices have been used to estimate the degree of mobility in the specialization pattern (Chiappini, 2014; Fertő & Soós, 2008; Proudman & Redding, 2000). These have been calculated for the transition over the entire period considered and the pre- and post-crisis period. This process helps to reduce the information regarding mobility of transition probabilities to a single statistic (Proudman & Redding, 2000; Redding, 2002). This study uses indices proposed by Shorrocks (1978). The first one among these, M₁, evaluates the trace of the probability matrix, thus directly capturing the relative magnitude of diagonal and offdiagonal terms. It equals the inverse of harmonic mean of the expected duration of remaining in a given state.

M_{1} = n - t r a c e (P) / n - 1

(7)

where n is the number of states and P is the transition probability matrix.

The second index evaluates the determinant (det) of the transition probability matrix.

M_{2} = 1 - | \det (P) |

(8)

For both indices, a higher value is indicative of greater mobility, with a value of zero indicating perfect immobility (Fertő, 2007; Fertő & Soós, 2008).

The mobility indices have also been estimated for all the time periods considered for stability analysis.

Data

The study has taken data from UNCTAD services trade statistics for the time period 2004–2013. It defines the services in conformity with the concepts and definition of the IMF Balance of Payments Manual (BPM 5). Here the services data have been compiled under three broad categories namely travel, transport and other services. The third component of other services is further divided into sub-categories to include communication, construction, insurance, computer and information services, financial services, other business services, royalties and licence fees, personal cultural and recreation services, and government services not included elsewhere (g.n.i.e.).

III. Empirical Results

India’s Comparative Advantage

Applying the above techniques to the data obtained from the above-stated source, India’s RSCA has been estimated which is presented in Table 1.

As is evident from Table 1, India appears to be enjoying strong CA in the domain of computer and information services and other business services. The higher CA seemed to be in computer and information services, particularly in IT and software-related services, owing to low cost of operations, high quality of products and services, and readily available low wage skilled English-proficient manpower (Eichengreen & Gupta, 2013; Mukhopadhyay, 2002; Ray, 1991; Sahoo, Dash & Mishra, 2015). Furthermore, a favourable time zone difference with North America and Europe helps Indian companies to achieve round the clock international operations and customer services, to cater to high global demands (Raipuria, 2001; RBI, 2010; Meyer, 2007; UNCTAD, 2012). An earlier study undertaken by Goswami, Gupta and Mattoo (2012) with the objective to analyse India’s services growth pattern and characteristics of its services exports also supports these results. Of the other sectors displaying CA, the advantage of communication services turned into disadvantage since 2010 onwards which can be attributed to the economic crisis of 2008 (Sahoo et al., 2015). Interestingly, prior to 2010, CA of communication services seemed to be higher than that for insurance. However, insurance continued to have an advantage till 2013. These results differ slightly from those reported by Burange, Chaddha and Kapoor (2010) and Raychaudhuri and De (2012) probably because of the difference in classification of data used by different data sources. India had registered a CD in the construction sector owing to its persistent decline compounded by recent slowdown in growth and business sentiments (Sahoo et al., 2015) and by the fact that majority of the construction companies are engaged in domestic market with very few engaged in exportoriented services (Mukherjee, 2001; Taneja et al., 2004). However, all the studies reported the strong CA of India in the realm of computer and information services sector.

Table 1.
Revealed Symmetric Comparative Advantage Index of India’s Services Exports

Sector/Year 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013

Transport –0.08 –0.08 –0.04 –0.08 –0.09 –0.02 –0.06 0.00 0.00 –0.03

Travel –0.04 –0.06 –0.07 –0.06 –0.11 –0.13 –0.09 –0.09 –0.09 –0.12

Communications 0.33 0.38 0.40 0.34 0.24 0.12 –0.07 –0.12 –0.14 –0.05

Construction 0.04 –0.32 –0.18 –0.23 –0.35 –0.38 –0.55 –0.43 –0.39 –0.26

Insurance 0.31 0.37 0.27 0.29 0.18 –0.04 0.00 0.19 0.10 0.05

Financial services –0.64 –0.31 –0.11 –0.10 –0.03 –0.09 0.06 0.00 –0.08 –0.06

Computer & information 0.85 0.86 0.86 0.85 0.85 0.82 0.82 0.80 0.80 0.79

Royalties & licence fees –0.93 –0.81 –0.95 –0.89 –0.92 –0.91 –0.95 –0.90 –0.89 –0.87

Other business services 0.39 0.44 0.47 0.45 0.44 0.30 0.31 0.27 0.35 0.34

Personal, cultural and recreational –0.67 –0.43 0.00 0.16 0.22 –0.03 –0.30 –0.38 0.00 0.19

Government services n.i.e. –0.21 –0.35 –0.50 –0.50 –0.47 –0.46 –0.44 –0.44 –0.50 –0.53

Sector/Year	2004	2005	2006	2007	2008	2009	2010	2011	2012	2013
Transport	–0.08	–0.08	–0.04	–0.08	–0.09	–0.02	–0.06	0.00	0.00	–0.03
Travel	–0.04	–0.06	–0.07	–0.06	–0.11	–0.13	–0.09	–0.09	–0.09	–0.12
Communications	0.33	0.38	0.40	0.34	0.24	0.12	–0.07	–0.12	–0.14	–0.05
Construction	0.04	–0.32	–0.18	–0.23	–0.35	–0.38	–0.55	–0.43	–0.39	–0.26
Insurance	0.31	0.37	0.27	0.29	0.18	–0.04	0.00	0.19	0.10	0.05
Financial services	–0.64	–0.31	–0.11	–0.10	–0.03	–0.09	0.06	0.00	–0.08	–0.06
Computer & information	0.85	0.86	0.86	0.85	0.85	0.82	0.82	0.80	0.80	0.79
Royalties & licence fees	–0.93	–0.81	–0.95	–0.89	–0.92	–0.91	–0.95	–0.90	–0.89	–0.87
Other business services	0.39	0.44	0.47	0.45	0.44	0.30	0.31	0.27	0.35	0.34
Personal, cultural and recreational	–0.67	–0.43	0.00	0.16	0.22	–0.03	–0.30	–0.38	0.00	0.19
Government services n.i.e.	–0.21	–0.35	–0.50	–0.50	–0.47	–0.46	–0.44	–0.44	–0.50	–0.53

Source: Author’s calculations based on UNCTAD statistics.

Distribution Dynamics

This section illustrates some stylized facts about the evolution of trade patterns. Table 2 reports some simple summary measures depicting a general picture of the distribution of RSCA index of India’s services for the pre- and post-crisis periods.

As, median of the index measures the overall level of specialization of a country (De Benedicts & Tamberi, 2004), extremely low median values suggest that the structure of exports is concentrated in sectors having CA. High variance would imply a narrow range of advantaged sectors or high degree of specialization and vice versa for a low variance value (Cantwell & Iammarino, 2001). Considering the standard deviation and maximum values, Table 2 reports a monotonic decrease in the degree of specialization between 2004 and 2013. Median values are slightly higher than the mean pointing towards a broad band of specialization instead of a narrow band of highly specialized sectors. Further, the shift of skewness from left to right indicates that there is an increase in the RSCA values of sectors with initial low RSCA and vice versa. Overall, the results indicate that some services have lost CA over the time.

Table 2.

Descriptive Statistics of RSCA Index

	2004	2008	2009	2013
Mean	–0.058	–0.004	–0.072	–0.050
Median	–0.037	–0.035	–0.039	–0.045
Std. dev.	0.532	0.475	0.439	0.434
Maximum	0.854	0.845	0.824	0.793
Skewness	–0.096	–0.176	0.141	0.013

Source: Author’s calculations based on RSCA index.

Since the analysis with the help of aggregate statistics such as standard deviations or means, may not be suffice for the reason that such statistics may hide intra-distribution movements (Quah, 1993), the intra-distribution dynamics has been analysed through regression. Also, normality of the residuals from regression has been tested by using the Jarque-Bera test which does not reject the null hypothesis of normality. The results of regressing Equation (3), that is, RSCA of initial on final years of for the two time periods and their subsequent scatter graphs are presented in this section.

The results in Table 3 suggest that trends in trade specialization pattern have remained same for the pre- and the post-crisis periods indicating that the crisis did not alter the specialization pattern to a great extent. The β value remained less than unity at significant levels (1% and 5%) for both time periods as well as for the entire period under consideration. It implied that β de-specialization, that is, sectors with initially low RSCA values improved their positions, while sectors with high RSCA values slided back. This is true in our case as the RSCA values of disadvantaged sectors like transport, travel, financial services etc. have recorded improvement, although they still remained at comparative disadvantaged level. On the other hand, computer and information services and other business services which have been enjoying CA noted some deceleration. These findings are also supported by an earlier study (Nath et al., 2015) wherein the evolution and pattern of services trade of U.S. vis-a-vis India and China were analysed by using kernel density functions and Markov transition matrices. The regression effect (1–β) is found to be small, for all the time periods, suggesting sticky specialization pattern of India’s services trade. The β value is significantly higher than 0.5, displaying a relatively more persistent pattern in the post-crisis period than the earlier one. This has been accompanied by low mobility effects as suggested by 1-R value, lower in the second period, indicating little change in the relative positions of the sectors. It certainly suggests a more stable pattern of specialization in the post-crisis period. A low mobility in relation to a decreased dispersion also indicates no shift in the pattern of specialization. In both pre- and the post-crisis period, India remained specialized in the computer and information services and other business services sectors, although with some loss of CA when compared to their pre-crisis figures. The ranking of the sectors has not witnessed any change during the study period. This is confirmed by the extent of specialization as measured by the estimated variance of the RSCA index, using Equation (4).

Table 3.

RSCA Regression Results in regard of India’s Export Specialization Patterns

Time Period	A	β	R	β/R
2004–2008	0.0341	0.6524*	0.7317	0.8916
2009–2013	0.0194	0.9624*	0.9741	0.9879
2004–2013	–0.0173	0.5654**	0.6940	0.8146

Source: Authors’ calculation.

Notes: * and ** Significantly different from zero at 1% and 5% levels, respectively.

The β/R ratios suggest a change towards more broad trends in specialization implying a fall in the degree of specialization. This denotes that the regression effect which suggests a reduction in the degree of specialization due to proportional move of sectors towards the average outweighs the mobility effect, leading to the value of β/R to be less than 1. Further, this broadening out effect was found to be stronger in the pre-crisis period than the post one, with a value of 0.89 and 0.98 in the first and second periods, respectively, indicating stronger despecialization in the first period. It is further highlighted by the scatter diagrams which display changes in trade patterns between the pre- and the post-crisis period. Figure 1(a) displays the scatter diagram of the RSCA index for the pre-crisis period, where the x-axis represents values for 2004 and y-axis gives the corresponding RSCA values for 2008. Similarly, Figure 1(b) corresponds to the RSCA index values for the post-crisis period, that is, from 2009 to 2013. Here x-axis represents values of the index for the year 2009 and y-axis denotes the index values for the year 2013. The two vertical and horizontal lines, demarcated at RSCA = 0, split the area into four quadrants and separate the sectors with RCA and RCD. The upper-left and the lower-right quadrants highlight sectors which have moved from CD to CA and from CA to CD, respectively. The upper-right quadrant represents sectors that displayed unchanged relative positions in terms of CA s, while the lower-left quadrant displayed unchanged relative position in terms of CD during the study periods (He, 2010).

Figure 1(a).

Pre-crisis Period

Figure 1(b).

Source: Author’s plot based on estimated RSCA values calculated in Table 1.

These graphs provide some important insights. Figure 1(a) shows that most of the points are located in the lower-left and the upper-right quadrants indicating that most of India’s services sectors did not change their relative positions in terms of CA s (disadvantages) in the pre-crisis period. The upper-left and lower right quadrants contain one sector each, namely construction services moved from an advantageous position to a disadvantaged one, while personal, cultural and recreational services sector showed a shift from CD position to CA. The positive slope of the regression line implies that average reversal or strengthening of initial specialization was absent. In Figure 1(b), most of the sectors are positioned lower-left quadrant demonstrating that majority of the sectors had CD and their relative positions did not alter in the post-crisis period. Computer and information services and other business services were the only two sectors which did not witness change in their relative positions throughout the entire period, despite the crisis, although the value of their indices fell probably due to decline in the business services orders from the US corporations and rise in the internal wage rates (Economic Survey 2012–2013, GOI). Only communication services appeared to have moved from CA to disadvantage, while insurance and personal, cultural and recreational services improved their status from CD to advantage during the post-crisis period. The results, therefore, appear to suggest that post-crisis period was more stable. The sectors which maintained their CD all through the study period comprised traditional services, that is, transport and travel services and modern services which included financial services, royalties and licence fee, and the government services not included elsewhere (g.n.i.e.).

Intra-distribution Dynamics

The regression estimates, as displayed in Table 3, only provide information on the conditional average of the distribution. It, therefore, requires use of other supplementary tools such as Markov transition probability matrices, for obtaining a comprehensive picture of the pattern of specialization. It may reveal significant information regarding the extent of persistence or mobility in the trade pattern of services. As shown in Table 4, the matrices exhibit the probability of switching from one to another state in between the beginning year and the end year. Since here, we are taking the crisis year, that is, 2008 as the dividing point in time, the initial and the final years for the two time periods are 2004, 2008 and 2009, 2013. The results reveal a strong persistence in the trade index during both the time periods, as depicted by the high diagonal entries. During the period 2004–2008, the probability that sectors may stay in the same state in the final year, as the initial one, varies from 80 per cent to 84 per cent. In the second period as well, the probability of remaining in the same state are high: ranging from 66 per cent to 75 per cent. However, when compared to the first period, the persistence for the latter is less evident. As for the off-diagonal elements, representing the extent of mobility, the probabilities are low. It is slightly higher in the second time period ranging from 25 per cent to 34 per cent, while those in the first period lie within 16 per cent to 20 per cent. An interesting observation that can be made regarding the off-diagonal probabilities is that the probability of switching from an initial state of CA to a disadvantaged one is higher in both the time periods relative to the probability of transiting from the of comparative disadvantage to the advantaged one. For instance, shifting from a state of CA to CD is 20 per cent and 34 per cent in the first and second periods, respectively, while the corresponding probabilities of shifting from CD to CA are only 16 per cent to 25 per cent. These results tell a slightly different story of specialization trends from that of the previous analysis. However, the overall pattern is still very stable for both the time periods, but more stable in the pre-crisis period.

Dynamics of trade specialization pattern have been presented in Table 4, by estimating the probabilities of maintaining, losing or gaining CA during the transition periods for 2,4 and 8 years. The probability that India maintains CA in services varies from 66 per cent to 76 per cent while maintaining CD varies from 84 per cent to 87 per cent. Robustness of this high level of persistence is maintained across different time transitions. For instance, the likelihood of India maintaining its CD is 87.1 per cent, 85 per cent and 84.6 per cent over 2-year, 4-year and 8-year period, respectively. Correspondingly, there is only a 12–15 per cent chance that services sector will make a move from CD position to CA one, over these time horizons. Again, for all time transitions, moving from CD to CD is more likely than moving from CA to CA state. Similarly, making transitions from CA to CD is more probable than shifting from CD to a CA state.

Table 4.

Transition Matrices for India’s Services RSCA

Pre-crisis	CDA	CA
CDA	83.3	16.7
CA	20	80
Post-crisis
CDA	75	25
CA	33.3	66.7
2-year transitions
CDA	87.1	12.9
CA	23.6	76.4
4-year transitions
CDA	85	15
CA	30.8	69.2
8-year transitions
CDA	84.6	15.4
CA	33.3	66.7

Source: Author’s estimation using the RSCA measures calculated from UNCTAD.

Note: CDA—comparative disadvantage, CA—comparative advantage. Each number represents the probability (in %) of switching from one state to another from an initial period to the final one over different time periods.

To illustrate this, the chance of shifting from CD to CA is only 12.9 per cent, 15 per cent and 15.4 per cent over 2-, 4- and 8-year periods, respectively. In contrast, probability of reversal is more likely and lies in the range of 23–33 per cent, depending on the length of different transition horizons. This implies that the chances of remaining in or shifting to a comparative disadvantaged position is more likely in case of Indian services trade, unless the economic and technical efficiencies are constantly upgraded. In order to further verify the results, as displayed in Table 4, the mobility indices for the pre and post-crisis period and for different transition lengths have also been calculated and highlighted in Table 5.

Table 5.

Mobility Index

Period	M1
Pre-crisis	0.367
Post-crisis	0.583
2-year transitions	0.365
4-year transitions	0.458
8-year transitions	0.487

Source: Estimated from transition matrices shown in Table 4.

As for the mobility indices, both M₁ and M₂ revealed the same value therefore only one of them is presented here. The values, as displayed in Table 5, indicate that the mobility of trade pattern is low for both time periods, but lower for the pre-crisis period, indicating towards a higher level of stability in this phase. Overall, the trade specialization pattern of India’s services exports has been more or less similar for the entire 10-year period. Interestingly, there has hardly been any evidence of an increase in the overall degree of specialization of a given country (Redding, 2002). Nevertheless, it also implies that India’s services trade has shown resilience towards the effect of the crisis in terms of lower magnitude of decline, less synchronicity spread across countries and early recovery from the crisis (UNCTAD, 2014b).

IV. Summary and Conclusion

This article attempted to analyse dynamics in the specialization pattern of the Indian services exports using appropriate analytical tools. Firstly, the CA structure of its services exports was studied using revealed symmetric CA. The analysis revealed that India has high CA in modern services like other business services, and computer and information services. The transport and travel sectors demonstrated as disadvantaged for the time period under study. Further, the regression-based stability analysis indicated towards broad trend of de-specialization in the trade pattern, although the stickiness of pattern has been more pronounced during the post-crisis period. The transition probability analysis suggests that the probability of sectors maintaining same position as the initial one was marginally better for the pre-crisis period. Overall, India seemed to have experienced a very stable pattern of specialization which appears to be in line with the economic theory that suggests that these patterns do not change very quickly. Another interesting observation that emerges from the study is that while there is a high chance of shifting from an initial state of CA to that of disadvantage, vice versa is not true, probably because of slow progress of economic and technical efficiency that does not help in turning disadvantage to advantage. On the other hand, the relatively slower rise of economic and technical progress may transform advantage into disadvantage. Therefore, while Indian policy makers need to adopt measures aimed at further strengthening the sectors which are India’s forte through significant and faster enhancement in the economic and technical efficiencies, the trend towards stable but broadening of specialization pattern indicate the need to facilitate diversification of the export structure and its destinations, which may subsequently reduce the risk of vulnerability arising out of a relatively smaller basket of exportable and concentrated export destinations. The indications are that India needs to look beyond the software sector and traditional export destinations for further diversification.

References

Balassa

(1965). Trade liberalisation and ‘Revealed’ Comparative Advantage. Manchester School, 33(2), 99–123.

Brasili

Epifani

& Helg

(2000). On the dynamics of trade patterns. De Economist, 148(2), 233–258.

Burange

L. G.

Chaddha

S. J.

& Kapoor

(2010). India’s trade in services. Indian Economic Journal, 58(2), 44–62.

Cantwell

(1989). Technological innovation and multinational corporations. Oxford: Blackwell.

Cantwell

& Iammarino

(2001). EU regions and multinational corporations: Change, stability and strengthening of technological comparative advantages. Industrial and Corporate Change, 10(4), 1007–1037.

Chiappini

(2014). Persistence vs. mobility in industrial and technological specialisations: Evidence from 11 Euro area countries. Journal of Evolutionary Economics, 24(1), 159–187.

Dalum

Laursen

& Villumsen

(1998). Structural change in OECD export specialisation patterns: de-specialisation and ‘stickiness’. International Review of Applied Economics, 12(3), 423–443.

De Benedictis

& Tamberi

(2001). A note on the Balassa index of revealed comparative advantage. Available at SSRN 289602, 1–37.

De Benedictis

& Tamberi

(2004). Overall specialization empirics: techniques and applications. Open Economies Review, 15(4), 323–346.

10.

Doanh

Khanh N.

(2011). Patterns and dynamics of Vietnam’s revealed comparative advantage and export specialization. Journal of Economics and Development, 13(1), 19–37.

11.

Eichengreen

& Gupta

(2013). Exports of services: Indian experience in perspective. Indian Growth and Development Review, 6(1), 35–60.

12.

Fertő

(2007). The dynamics of trade in central and eastern European Countries. Managing Global Transitions, 5(1), 5–23.

13.

Fertő

& Soós

K. A.

(2008). Trade specialization in the European Union and in post-communist European countries. Eastern European Economics, 46(3), 5–28.

14.

Goswami

A. G.

Gupta

& Mattoo

(2012). A cross-country analysis of services exports: Lessons from India. In Goswami

A. G.

Mattoo

& Sáez

(Eds.), Exporting services: A developing country perspective (pp. 81–120). Washington, DC: World Bank.

15.

Hart

P. E.

(1976). The dynamics of earnings, 1963–1973. Economic Journal, 86(343), 551–565.

16.

S. Q.

(2010). Dynamics of Chinese agricultural trade patterns: Pre- and post-WTO periods. Chinese Economy, 43(1), 5–25.

17.

Hinloopen

& Van Marrewijk

(2001). On the empirical distribution of the Balassa index. Weltwirtschaftliches Archiv, 137(1), 1–35.

18.

Liesner

H. H.

(1958). The European common market and British industry. Economic Journal, 68(270), 302–316.

19.

Meyer

Thomas

(2007). India’s specialization in IT exports: Offshoring can’t defy gravity (Research Notes 27). Deutsche Bank Research Working paper series. Retrieved January 6, 2016, from http://www.dbresearch.com/PROD/DBR_INTERNET_EN-PROD/PROD0000000000217530/India’s+specialisation+in+IT+exports%3A+Offshoring+c.pdf.

20.

Mukherjee

(2001). Trade in construction and consultancy services: India and the GATS (ICRIER Working Paper no. 75). Delhi: Indian Council for Research on International Economic Relations.

21.

Mukhopadhyay

(2002). Globalisation and Indian services sector. Economic and Political Weekly, 37(40), 4097–4098.

22.

Nath

H. K.

Liu

& Tochkov

(2015). Comparative advantages in US bilateral services trade with China and India. Journal of Asian Economics, 38(June), 79–92.

23.

Pailwar

V. K.

& Shah

N. R.

(2009). Revealed comparative advantages for India in services trade. International Journal of Trade and Global Markets, 2(2), 109–127.

24.

Proudman

& Redding

(2000). Evolving patterns of international trade. Review of International Economics, 8(3), 373–396.

25.

Quah

(1993). Empirical cross-section dynamics in economic growth. European Economic Review, 37(2), 426–434.

26.

Raipuria

(2001). Service exports: ‘Knowledge Bowl’ yet to yield major gains. Economic and Political Weekly, 36(35), 3351–3352.

27.

Ray

(1991). Trade in services for developing countries in 1990s: Problems and prospects. Economic and Political Weekly, 26(9/10), 519–527.

28.

Raychaudhuri

and De

(2012). International trade in services in India: Implications for growth and inequality in a globalising world. New Delhi: Oxford University Press.

29.

RBI (2010). Invisibles in India’s balance of payments: An analysis of trade in services, remittances and income. Monthly Bulletin, Department of Economic Analysis and Policy, Reserve Bank of India, Mumbai.

30.

Redding

Stephen

(2002). Specialization dynamics. Journal of International Economics, 58(2), 299–293.

31.

Sahoo

Dash

R. K.

& Mishra

P. P.

(2015). Determinants of India’s service exports. In Agarwal

(Ed.), Reviving growth in India (pp. 107–140). Delhi: Cambridge University Press.

32.

Shorrocks

A. F.

(1978). The measurement of mobility. Econometrica: Journal of the Econometric Society, 46(5), 1013–1024.

33.

Taneja

Mukherjee

Jayanetti

& Jayawardena

(2004). Indo-Sri Lanka trade in services: FTA II and beyond (ICRIER Working Paper No.145). Delhi: Indian Council for Research on International Economic Relations.

34.

UNCTAD (2012). Implications of The global economic crisis on India’s services sector (UNCTAD/DITC/TNCD/2009/22). Geneva: United Nations Conference on Trade and Development, United Nations.

35.

UNCTAD (2014a). Services policy reviews: A detailed methodology for reviewing policy, regulatory and institutional frameworks for services (UNCTAD/DITC/TNDC/2013/3). Geneva: United Nations Conference on Trade and Development, United Nations.

36.

UNCTAD (2014b). Services: New frontier for sustainable development. Exploiting the potential of trade in services for development 2. (UNCTAD/DITC/TNCD/2013/4). Geneva: United Nations Conference on Trade and Development, United Nations.

37.

Widodo

(2009). Dynamic comparative advantages in the ASEAN+ 3. Journal of Economic Integration, 24(3), 505–529.

38.

WTO (2015). International trade statistics. Geneva: World Trade Organization.