STRI for Computer and Related Services Molly Lesher and Hildegunn Kyvik Nordås OECD Trade and Agriculture OECD Services Experts Meeting, Paris, 2 July 2009
Overview Regulations included in the index Classification of regulations Methodology The indices Robustness Relevance Conclusions This STRI focuses on barriers to computer and related services – a sub-sector of business services. The business services sector was selected as one of the pilot sectors because it has been one of the most dynamic services sectors over the past decade or so. Business services are, however, a large and diversified category within which highly regulated services (e.g. legal services) are combined with competitive services such as computer services. It was therefore decided to develop separate STRIs for professional and computer services. Computer and related services account for a relatively small share of services trade, but it has been growing faster than average and the share is consequently increasing. To introduce this STRI, we will keep to the following outline. First I will discuss regulations included in the index, the classification of regulations, and briefly recap the methodology. Then Hildegunn will present the indices for this sector, discuss the robustness of the results and their relevance, and offer some concluding remarks.
Computer services and trade restrictions Lightly regulated sector Most restrictions are horizontal, not sector- specific Skilled labour-intensive sector Selection criteria: Mentioned in the GATS and/or RTAs Expert judgement Computer and related services is a lightly regulated sector, and is subject to few sector-specific barriers. As a result, the STRI is composed of the least number of regulatory measures (28) among the pilot sectors. In addition, the STRI for this sector mainly consists of horizontal trade and regulatory measures that are found to impact trade and foreign investment in the sector. It is also a very skilled labour-intensive sector. After careful analysis of the regulatory regimes of OECD countries and input from a wide range of industry experts, we selected regulations based on several criteria: (1) whether they were mentioned in the GATS (2) whether they were mentioned in recent RTAs signed by OECD countries, and (3) expert judgement, particularly from the 2008 OECD Expert Meeting on Business Services. RTAs analysed: US-Australia FTA (2005), NAFTA (1994), Japan-Mexico FTA (2005), EFTA-Mexico (2001), EC- Mexico (2000).
Restrictions included in the STRI Categories Sector specific Source Restrictions on foreign ownership Yes/Horizontal OECD TAD survey; PMR Restrictions on movement of people Yes OECD TAD survey Discriminatory measures Barriers to competition Horizontal PMR Regulatory transparency and licensing systems There are 5 categories of restrictions included in the STRI. These restrictions are derived from a sector-specific questionnaire that was given to countries as part of a special survey as well as horizontal measures that come from the OECD PMR survey. (1) Restrictions on foreign ownership (ex: foreign direct equity limits) (2) Restrictions on the movement of people (ex: restricting the number of foreign professionals via quotas) (3) Discriminatory measures (ex: restrictions on foreign participation in public procurement) (4) Barriers to competition (ex. providing foreign firms with redress when they believe business practices restrict competition) and (5) Regulatory transparency and licensing systems (ex. the existence of one-stop-shops for obtaining information and notifications and licenses). A total of 28 measures are included in the index, and more details about these measures can be found in the paper. NB: Subsidies and taxes horizontal; public procurement sector specific
Classification of restrictions Number of Categories Category type 5 GATS 2 (NT+MA and DR/Other) By mode 3 Discriminatory or not 2 Entry or on-going operations The measures were classified under different typologies. Classifying barriers and regulations in computer services under different typologies can increase the usefulness of the regulatory profile and STRI by highlighting different dimensions of the data specifically for negotiators and regulators. The first classification is by the five broad categories of measures indicated in the last slide that groups by topic the 28 measures. The second classification is by the GATS (that is, those measures that are scheduled and those that are not scheduled). This classification groups market access and national treatment measures in one category – they are difficult to distinguish in practice (though less so for scheduling purposes) – and domestic regulation and other regulations in another category. This grouping also allows a distinction to be made between restrictions subject to scheduling under the GATS – and consequently to negotiations for their removal – and other largely domestic regulatory measures that do not need to be scheduled. The next classification is a discriminatory and non-discriminatory classification which provides additional information from the previous classification because NT and MA are combined. Non-discriminatory measures affect total demand whereas discriminatory ones typically only have a bearing on import demand. A classification by the GATS modes of supply is also used. Modes 1 and 2 are combined, whereas Modes 3 and 4 are treated as separate categories. By highlighting which modes are most restrictive, negotiators can better tailor their requests and offers in the context of services trade talks. Finally, a classification that divides the measures into those that affect entry or on-going operations is employed. Establishment restrictions can generally be regarded as impediments to the movement of capital, while those applying to firms’ operations constrain service provision after establishment. These distinctions are important because diverse types of barriers may have different economic costs, and they can be useful in helping regulators and industry analysts identify priority areas for reform.
Methodology - scoring 81% of the measures are binary Scores are binary 0 indicates no restrictions; 1 indicates restriction The STRI is derived by aggregating regulations that are potentially trade restricting in computer and related services into a composite measure of restrictiveness. The construction of the index involves decisions concerning three main issues: scoring, weighting and aggregation. While the methodology for constructing the STRI was presented in detail this morning, I will just recap the highlights here for the benefit of those that could not attend this morning’s session. The first consideration in the construction of the STRI is scoring. Scoring relates to how regulatory measures are recorded. Since most of the data we collected on regulatory measures was binary to begin with (i.e. yes/no questions), all of the non-binary variables were transformed into a binary format. A “0” indicates no restriction and a “1” indicates the presence of a restriction. This method ensures that all variables are measured on the same scale.
Methodology - weighting Starting point: the 5 categories of regulation Each category (j) is weighted according to expert judgement : Equal weights are used under each heading (and sum to 1 within each category) Each weight is carried over from one classification to the next: Weighting is the second consideration in the construction of the STRI. Weighting captures the relative importance of impediments in terms of trade restrictiveness (the higher the weight the more restrictive a category of measure is considered relative to other categories). The starting point is the 5 categories of regulation. Each category heading is assigned a weight (according to expert judgement), and the sum of the weights of the 5 categories sums to 1. [j = category and k = measure] Likewise, each individual measure under each category is assigned a weight (according to equal weights), which sum to 1 within each category. Each weight is carried over from one classification to the next which ensures that no matter how measures are classified, a country’s overall STRI score remains constant.
Methodology: aggregation The next step involves aggregation of the five categories of regulations into the overall STRI, again using weights reflecting their relative restrictiveness and adding up to one across categories. As you can see from this equation, each country’s STRI is basically the weighted average of the scores, with k representing the measures, s representing the scores, and w representing the weights. A country’s STRI score ranges from 0 (no trade-related restrictions) to 1 (a completely restricted sector as far as trade is concerned). An advantage of this method is that each weight is fixed, and so whatever classification scheme is used to construct the STRI, the overall STRI score will remain constant. In addition, the final index is not influenced by the number of measures within each category of regulations. Now Hildegunn will discuss the results, the robustness of the results and the relevance of the STRI, and offer some conclusions.
Design of the STRI Equal weights Expert judgment within categories Restrictions on foreign ownership and market entry Individual measures Restrictions on movement of people Discriminatory measures Barriers to competition Regulatory transparency etc Equal weights within categories This next figure sums up the design of the STRI. You can see the 5 categories of regulations at the top which carry a weight derived from expert judgement. The individual measures are included under each category and carry an equal weight within the category. The “ranks” at the bottom show how much each category contributes to the overall STRI score for each country: Restrictions on foreign ownership (40%); Restrictions on the movement of people (appx. 22%); Discriminatory measures (appx. 15%); Barriers to competition (appx. 12%); Regulatory transparency and licensing (10%). Rank 1 40% Rank 2 22.5% Rank 3 15% Rank 4 12.5% Rank 5 10% Expert judgment Categories of measures are ranked from the most to the least trade restrictive
Results Mention that results changed because of database error. The highest possible score – the most restrictive is one. The most restrictive country has a weighted average of 30% of all possible restrivtions.
STRI by category type Sector specific ownership etc. restrictions are in all cases that board of directors must be residents. In addition there are horizontal restrictions on foreign M&A. Movement of people
STRI by GATS
STRI by mode
Robustness Rank correlation equal weights 0.90 Rank correlation PCA weights 0.49 Robust to the most common alternative weighting scheme (equal weights) But not robust to all weighting schemes Reason: most restrictions are in one or two categories
Relevance – STRI performance in gravity regressions Mode Correlation Significance Imports Negative No FDI 10% FATS 1% There are no regulatory measures that restrict mode ½ in the index. Impact on trade therefore comes through complementarity between modes.
Conclusions Sector-specific trade restrictions in computer services mainly relates to the movement of people either under mode 4 or mode 3 Other regulations that affect trade in computer services are horizontal transparency issues and barriers to competition The STRIs are somewhat sensitive to the weighting scheme – which means that it is important to choose the “right” weighting scheme The STRI appears to measure what it is supposed to measure – a negative and significant relation to FDI and FATS
OECD Trade and Agriculture www.oecd.org/trade/stri Contacts Molly.Lesher@oecd.org Hildegunn.Nordas@oecd.org