• Methods and Projects
• Standards and Classifications

• Standardised Methodological Report

Minimise complete Methodological Report Complete Methodological Report Queried fields

Statistics on Global Value Chains

• 1Contact
  • 1.1Contact organisation
    National Statistics Institute of Spain
    
  • 1.5Contact mail address

    Avenida de Manoteras 50-52 - 28050 Madrid

    
    
• 2Metadata update
  • 2.1Metadata last certified
    27/11/2025
  • 2.2Metadata last posted
    27/11/2025
  • 2.3Metadata last update
    27/11/2025
• 3Statistical presentation
  • 3.1Data description

    Global Value Chain Statistics (GVC) is an operation compiled under Implementing Regulation (EU) 2022/918 whose overall objective is to measure the phenomenon of global value chains in which Spanish companies are involved for the 2021-2023 reference period, where this concept encompasses the entire range of cross-border activities necessary to bring a product or service from its conception to end consumers, through the various stages of production and delivery.

    The specific objectives established are:

    1.    To ascertain the distribution of employment in each company according to the business functions it performs.

    2.    To ascertain how many companies are involved in global value chains, i.e. are buying and/or selling goods and/or services abroad.

    3.    To ascertain how many companies have been involved in international outsourcing operations during the reference period, and how many jobs are affected.

    4.    To ascertain the extent to which certain international events have impacted global value chains.

    The information needed to compile these statistics is collected in the Structural Business Statistics questionnaire, through a series of additional questions specific to certain legal units in the sample that meet the criteria for the target population of global value chains. The aim of this approach is to obtain the necessary information while reducing the statistical burden on respondents by generating a new statistical operation, without compromising the quality of the aggregate estimates.

  • 3.2Classification system
    • Clasificaciones utilizadas

      NACE Rev. 2 coding is used to code the activity of the enterprises.

      
      
      

      The rest of the classifications are described in Regulation (EU) 2022/918 or in the GVC Compliers' Manual.

  • 3.3Sector coverage

    The economic sectors referred to in these statistics are those listed in sections B to N of the NACE Rev. 2 classification. Another condition is that enterprises must have 50 or more employees or self-employed persons.

  • 3.4Statistical concepts and definitions
  • 3.5Statistical unit

    The basic statistical unit for this operation is the enterprise, understood as "the smallest combination of legal units that constitutes an organisational unit for the production of goods and services and enjoys a certain degree of decision-making autonomy, mainly in the use of the resources at its disposal. The enterprise may carry out one or more activities in one or more locations. An enterprise may correspond to a single legal unit" (definition from European Union Regulation 696/93).

    The increasing complexity of the internal operations of business groups today has led the European Statistical System (ESS) to seek a way to reflect the activity of these groups in official business statistics. Indeed, legal units belonging to business groups sometimes sell their products or provide their services exclusively or mainly within the group, without being market-oriented or having decision-making autonomy over the entire production process.

    For all these reasons, and in accordance with the European Statistical System (ESS), an “enterprise” can be:

    • Either an independent Legal Unit, which is not part of a business group, and therefore is assumed to have decision-making autonomy.
    • Or a business group formed for one or more Legal Units.
    • Or a subset of one or more Legal Units of a business group. 

    The basic idea is that if the legal units of a statistical enterprise serve exclusively or mainly other legal units of the same enterprise (for example, because they sell products under a vertically integrated production process or provide services as an auxiliary), these servile legal units must be combined with the others they support to form the true statistical unit ‘enterprise’.

    It should be noted that most enterprises are independent legal units. For this majority, the identity Enterprise = Legal Unit applies. However, legal units that are part of enterprise groups (3% of the total) may be composed of two or more legal units.

  • 3.6Statistical population

    The GVC statistics are aimed at all companies, corporations and individuals that are market producers, with 50 or more employees and self-employed persons in the the reference period and whose main activity is between NACE Rev. 2 sections B-N. Also, the only enterprises that will report data for variables 2 to 5 will be those that have traded goods or services with at least €100,000.

  • 3.7Reference area

    All statistical units located within the national territory (Spain) are subject to investigation.

  • 3.8Time coverage

    This is a structural statistical operation carried out every three years.

  • 3.9Base period

    This statistic does not use index numbers, therefore it is not applicable.

• 4Unit of measure
  • 4.1Unit of measure
    • Business functions: number of employees

    • GVC arrangements: number of enterprises.

    • International outsourcing: number of enterprisesand, number of jobs lost or created as a result of outsourcing, as applicable.

    • Motivations and barriers for international outsourcing: number of enterprises

    • Effects on GVCs: number of enterprises
• 5Reference period
  • 5.1Reference period

    Data referring to the period: Triennial 2021–2023, excluding variables 1 to 5 which refer only to 2023.

• 6Institutional mandate
  • 6.1Legal acts and other agreements

    The compilation and dissemination of the data are governed by the Statistical Law No. 12/1989 "Public Statistical Function" of May 9, 1989, and Law No. 4/1990 of June 29 on “National Budget of State for the year 1990" amended by Law No. 13/1996 "Fiscal, administrative and social measures" of December 30, 1996, makes compulsory all statistics included in the National Statistics Plan. The National Statistical Plan 2009-2012 was approved by the Royal Decree 1663/2008. It contains the statistics that must be developed in the four year period by the State General Administration's services or any other entity dependent on it. All statistics included in the National Statistics Plan are statistics for state purposes and are obligatory. The National Statistics Plan 2025-2028, approved by Royal Decree 1225/2024, of  3 December, is the Plan currently implemented. This statistical operation has governmental purposes, and it is included in the National Statistics Plan 2025-2028. (Statistics of the State Administration).

    
    

    The entry into force of the new Regulation (EU) 2019/2152 of the European Parliament and of the Council on European Business Statistics (EBS) and its Implementing Regulation (EU) 2020/1197 of the Commission of 30 July 2020 gave rise to a subsequent specific implementing act, Commission Implementing Regulation (EU) 2022/918 of 13 June 2022, which determined that information on the participation of enterprises in global value chains should be obtained and specified the aggregate information required.

  • 6.2Data sharing
    The exchanges of information needed to elaborate statistics between the INE and the rest of the State statistical offices (Ministerial Departments, independent bodies and administrative bodies depending on the State General Administration), or between these offices and the Autonomic statistical offices, are regulated in the LFEP (Law of the Public Statistic Function). This law also regulates the mechanisms of statistical coordination, and concludes cooperation agreements between the different offices when necessary.
    

    Various administrative records and available statistics are consulted:

    • Value Added Tax (VAT) and International Trade in Goods Statistics (ITGS), both provided by the Spanish Tax Agency, AEAT
    • International Trade in Services Survey (ITSS), compiled by the National Statistical Institute of Spain.
• 7Confidentiality
  • 7.1Confidentiality - policy

    The Statistical Law No. 12/1989 specifies that the INE cannot publish, or make otherwise available, individual data or statistics that would enable the identification of data for any individual person or entity. Regulation (EC) No 223/2009 on European statistics stipulates the need to establish common principles and guidelines ensuring the confidentiality of data used for the production of European statistics and the access to those confidential data with due account for technical developments and the requirements of users in a democratic society

    
    
  • 7.2Confidentiality - data treatment
    INE provides information on the protection of confidentiality at all stages of the statistical process: INE questionnaires for the operations in the national statistical plan include a legal clause protecting data under statistical confidentiality. Notices prior to data collection announcing a statistical operation notify respondents that data are subject to statistical confidentiality at all stages. For data processing, INE employees have available the INE data protection handbook, which specifies the steps that should be taken at each stage of processing to ensure reporting units' individual data are protected. The microdata files provided to users are anonymised.
    
• 8Release policy
  • 8.1Release calendar

    The advance release calendar that shows the precise release dates for the coming year is disseminated in the last quarter of each year.

    
    
  • 8.2Release calendar access

    The calendar is disseminated on the INEs Internet website (Publications Calendar)

    
    
  • 8.3User access

    The data are released simultaneously according to the advance release calendar to all interested parties by issuing the press release. At the same time, the data are posted on the INE's Internet website (www.ine.es/en) almost immediately after the press release is issued. Also some predefined tailor-made requests are sent to registered users. Some users could receive partial information under embargo as it is publicly described in the European Statistics Code of Practice

    
    
• 9Frequency of dissemination
  • 9.1Frequency of dissemination

    Triennal

• 10Accessibility and clarity
  • 10.1News release

    The results of the statistical operations are normally disseminated by using press releases that can be accessed via both the corresponding menu and the Press Releases Section in the web

    
    

    News release for GVC 2021-2023 will be published on INE's website on 27 November: https://ine.es/dyngs/INEbase/operacion.htm?c=Estadistica_C&cid=125473617711&menu=ultiDatos&idp=1254735576550

  • 10.2Publications

    The results of the statistics are published on the INE website (INEBase). Some results are included in publications such as the Anuario Estadistico [Statistical Yearbook], Cifras INE [INE Figures], etc. 

  • 10.3On-line database

    INEbase is the system the INE uses to store statistical information on the Internet. It contains all the information the INE produces in electronic formats. The primary organisation of the information follows the theme-based classification of the Inventory of Statistical Operations of the State General Administration . The basic unit of INEbase is the statistical operation, defined as the set of activities that lead to obtaining statistical results on a determined sector or subject based on the individually collected data. Also included in the scope of this definition are synthesis preparation.

    
    

    Information on Global Value Chain Statistics can be found at the following link: https://ine.es/dyngs/INEbase/operacion.htm?c=Estadistica_C&cid=1254736177116&menu=resultados&idp=1254735576550

  • 10.4Micro-data access
    A lot of statistical operations disseminate public domain anonymized files, available free of charge for downloading in the INE website Microdata Section
    

    GVC microdata are not disseminated

  • 10.5Other

    Users may request specific customised information analyses, which are carried out while preserving data confidentiality at all times, through the User Support Area at the following link: User Request

  • 10.6Documentation on methodology

    The methodology can be found at: https://ine.es/dyngs/INEbase/operacion.htm?c=Estadistica_C&cid=1254736177116&menu=metodologia&idp=1254735576550 

  • 10.7Quality documentation
    Based on Regulations 2019/2152 and 2020/1197 of the European Parliament and of the Council, the European Commission (Eurostat) assesses the quality of the data transmitted and produces reports on the quality of European statistics. Eurostat produces an internal quality report containing quantitative and qualitative information.

     

    Sections 11 to 19 of this document constitute the user-oriented quality report for this operation.
• 11Quality management
  • 11.1Quality assurance

    Quality assurance framework for the INE statistics is based on the ESSCoP, the European Statistics Code of Practice made by EUROSTAT. The ESSCoP is made up of 16 principles, gathered in three areas: Institutional Environment, Processes and Products. Each principle is associated with some indicators which make possible to measure it. In order to evaluate quality, EUROSTAT provides different tools: the indicators mentioned above, Self-assessment based on the DESAP model, peer review, user satisfaction surveys and other proceedings for evaluation.

    
    

    Based on Regulations 2019/2152 and 2020/1197 of the European Parliament and of the Council, the European Commission assesses the quality of the data transmitted and produces reports on the quality of European statistics

    The National Statistical Institute of Spain ensures data quality in line with ESS Quality Assurance Framework (QAF).

    Several validation checks are involved in this statistical process from the beginning:

    • All the data sources used in the statistics are subject to integrity checks.
    • Once the integration of data from all sources has been completed, various checks are performed again, verifying whether the information is accurate.
  • 11.2Quality assessment

    The statistics are guided by the European Statistics Code of Practice and, in line with this, focus primarily on:

    • Evaluating and verifying the quality of primary data from the administrative records of the Tax Agency and the INE itself.
    • Guaranteeing statistical confidentiality.
    • Clearly describing the methodology and reviewing it annually.
    • Not adding statistical burden to reporting units.
    • Using definitions and standards that allow consistency with data from other European statistical institutes.

    Planned improvements on analysing the results in greater depth, focusing on the meaning and integrity of the variables in relation to each other and their totals.

• 12Relevance
  • 12.1User needs

    The primary users of this survey are the European Commission and the Spanish economic authorities with a view to the implementation and review of the General Agreement on Trade in Services (AGCS) and the Agreement on Trade-Related Aspects of Intellectual Property Rights (ADPIC):

    • Eurostat
    • Chambers of Commerce
    • Central and Regional Governments
    • Scholars
    • Researchers
    • Individuals
  • 12.2User satisfaction

    The INE has carried out general user satisfaction surveys in 2007, 2010, 2013,  2016 and 2019 and it plans to continue doing so every three years. The purpose of these surveys is to find out what users think about the quality of the information of the INE statistics and the extent to which their needs of information are covered. In addition, additional surveys are carried out in order to acknowledge better other fields such as dissemination of the information, quality of some publications...

    On the INE website, in its section Methods and Projects / Quality and Code of Practice / INE quality management / User surveys are available surveys conducted to date.(Click  next link)

     

    
    

    The INE has carried out general user satisfaction surveys in 2007, 2010, 2013, 2016, 2019 and 2025 (not carried out yet). The purpose of these surveys is to find out what users think about the quality of the information of the INE statistics and the extent to which their needs of information are covered. In addition, additional surveys are carried out in order to acknowledge better other fields such as dissemination of the information, quality of some publications...

    On the INE website, in its section Methods and Projects / Quality and Code of Practice / INE quality management / User surveys are available surveys conducted to date (User Satisfaction Surveys)

    In the latest general user satisfaction survey of 2019, the assessment of the product quality dimensions (relevance, precision, timeliness, coherence and comparability) for the groups of statistics referring to the Industry and Services (in which this statiscal operation in framed) can be consulted.

    On the other hand, those responsible for this statistical operation are in permanent contact with the main users, in order to meet any specific need for information.

    Likewise, given the possibility that there are requests not attended due to their complexity, these are evaluated as well as any suggestions made by the main users. Most of these demands are satisfied.

  • 12.3Completeness

    This statistical operation covers all the variables required by Eurostat regulations and guidelines.

• 13Accuracy and reliability
  • 13.1Overall accuracy

    The statistics use data from INE surveys (SBS and ITSS), as well as statistics and administrative records from the AEAT (VAT and ITGS). These data are highly reliable as they are refined and definitive in the corresponding sources.

    On the other hand, the data is not exactly tailored to the statistics, so certain information needs to be imputed, which lowers the level of accuracy of the data.

  • 13.2Sampling error

    Not applicable, as the Central Business Register compiled by the INE is used as the census.

  • 13.3Non-sampling error

    Throughout the statistical process, non-sampling errors are monitored. The overcoverage rate and non-response rate per unit are calculated:

    Overcoverage = 2,4%
    Non-response rate = 3,6%

• 14Timeliness and punctuality
  • 14.1Timeliness

    Data and metadata were transmitted to Eurostat within the legal deadline of T+21 months. However, the INE had to do a second transmission of the data due to some inconsistencies detected in tables T2, T3, T6 and T7. This transmission was on 4 November.

  • 14.2Punctuality

    The survey data was collected according to the established deadlines. This data was cleaned, validated, and processed approximately one year prior to transmission.

    However, we were dependent on the aforementioned entities to send us the data. Therefore, managing the collection of some results was a bottleneck.

• 15Coherence and Comparability
  • 15.1Comparability - geographical

    The consistency provided by the data sent to Eurostat by member countries allows for accurate geographical comparability at international level.

  • 15.2Comparability - over time

    As this is a newly created statistic, there is no comparability over time with other series, since it has only been published once.

  • 15.3Coherence - cross domain

    The statistics use the DIRCE (Central Business Register) to construct the target population. This register is a national reference, so cross-referencing with data from INE statistics and external sources will be consistent.

  • 15.4Coherence - internal

    The results are consistent with each other, as an exhaustive imputation and cleansing process has been carried out in accordance with the nature of the statistical data as a whole.

    Some internal validation routines were carried out at the final stage. All the validation rule were checked with Python programming.

• 16Cost and burden
  • 16.1Cost and burden

    Although this statistic makes extensive use of existing sources, information from surveys is also necessary. Therefore, the budgetary appropriations required for its financing in the four-year period 2025-2028 are €212.500, as provided for in the INE budget. 

• 17Data revision
  • 17.1Data revision - policy

    The INE of Spain has a policy which regulates the basic aspects of statistical data revision, seeking to ensure process transparency and product quality. This policy is laid out in the document approved by the INE board of directors on 13 March of 2015, which is available on the INE website, in the section "Methods and projects/Quality and Code of Practice/INE’s Quality management/INE’s Revision policy" (link).

    This general policy sets the criteria that the different type of revisions should follow:  routine revision- it is the case of statistics whose production process includes regular revisions-; more extensive revision- when methodological or basic reference source changes take place-; and exceptional revision- for instance, when an error appears in a published statistic-.

    
    

    The data are final once published. If errors are detected and the data need to be modified, an explanatory note will be added alongside the information, thus warning that the results have been altered. Where possible, users are directly informed of these errors.

  • 17.2Data revision - practice

    The data is published when it is final. It may only be modified if an error is detected in the data sources.

• 18Statistical processing
  • 18.1Source data

    The INE obtains most of the information needed to produce these statistics from the module attached to the Structural Business Statistics surveys. However, as mentioned above, this information is based on the DIRCE for the construction of the target population, the Spanish Tax Administration Agency (AEAT) for information relating to foreign trade operations, and the International Trade in Services Survey (ECIS) produced by the INE itself.

    The EEE is produced annually. However, in the case of these statistics, many of the questions referred to the three-year period 2021-2023.

    DIRCE is updated once a year using administrative sources, mainly tax and social security data, and information from INE statistical operations. It is an integrated information system at various levels, ranging from lowest to highest: establishment, Legal Unit, Enterprise and Business Group. For each of these levels, DIRCE contains information on the main economic activity, the number of employees and turnover, as well as identification and location data necessary for the correct collection of information. The DIRCE used as a population framework refers to the year prior to the reference period, although two main quantitative variables, number of employees and turnover, are updated to the year of study. Since the 2018 reference year, the DIRCE has included the new EE level, which is equal to the ULE in the case of independent ULEs, or to a set of ULEs in a business group, or to the entire business group, as determined by profiling techniques.

    As for the other sources, this was annual information for the 2023 reference year, not for the entire three-year period.

  • 18.2Frequency of data collection

    For each reference year t, data collection is organised through the direct collection of Structural Business Statistics questionnaires with the attached global value chains module, aimed at selected legal units (field collection, under the IRIA system). This takes place from April to September of year t+1.

  • 18.3Data collection

    Data collection using questionnaires and modules through the IRIA system is carried out by the INE's Collection Units. Almost all questionnaires are collected online.

    The Collection Units are responsible for managing the collection, recording and cleaning of the modules, as well as answering telephone enquiries from respondents. Telephone contact is also made with companies in cases where no response is received within the established deadline or where the response is considered insufficient or inconsistent.

    In order to monitor fieldwork, the different situations that may arise during the collection of information are taken into account. A company will be considered to have been effectively surveyed if its main activity is included in the population under study, the completed module has been obtained, and the data verify the established completeness and consistency controls.

    In addition, a series of incidents may arise during the information collection process that prevent the module from being obtained. Rigorous handling of these incidents is of great importance, as their analysis allows the survey framework to be updated and has an impact on the processing of the information.

    The incidents taken into account are:

    • Permanent closure or cessation of activity: the Legal Unit has permanently ceased its activity, a situation that is justified by an official document certifying this.
    • Temporarily closed or inactive: the Legal Unit remains closed during the information collection period and no informant can be located, or it has no activity during the year.
    • Incorrectly included: the Legal Unit's main activity is outside the scope of the survey.
    • Outside scope: other characteristics of the unit, other than its main activity, place it outside the scope of the survey.
    • Duplicated: the Legal Unit appears in the directory more than once.
    • Unlocatable, Negative and No response. These situations, which are in the minority at the end of the collection process, are dealt with specially by the Collection units in order to minimise non-response.
  • 18.4Data validation

    During the data collection phase, an initial process of data cleansing and coding is carried out. Both the electronic questionnaires completed by respondents online and the application used by the INE's Data Collection Units to manage, record and cleanse the data collected have error detection systems programmed to validate the data as it is entered by the user. A distinction is made between serious errors (which must be corrected) and second-level anomalies (which, once confirmed, must be justified). In addition, during data collection and cleaning, measures are also taken to reduce non-response.

    The records recorded by the Collection Units are used to create and feed, at least every two weeks, the complete recording files on which the subsequent phases of joint information processing are carried out. These files are processed at Central Services, where a new information coverage check is carried out to ensure the completeness of the recorded data, detect duplicates and coverage errors, and, at the same time, assess the quality of the variables collected.

    As for DIRCE, ITSS, ITGS and VAT, these are highly reliable and definitive data, and therefore their validation is not the responsibility of the GVC.

  • 18.5Data compilation

    As the collection of Legal Units progresses and complete recording files are created, the data undergoes additional micro-cleaning checks at INE, which are selectively focused on detecting and cleaning up errors and inconsistencies in the variables of each record, as well as cleaning up and imputing content errors. Depending on the characteristics of each type of error, automatic imputation procedures are used in certain cases. Likewise, systematic errors detected in previous studies and analyses of the recorded data are corrected.

    There may be several reasons why a response needs to be imputed:

    • Total or partial lack of response: due to lack of information from respondents or the application of prior filters.
    • Units not surveyed intentionally
    • Directory updates: the sample is obtained from a population frame based on the year prior to the last year of the three-year reference period, as the sample for the last year is not available at the time the EEE sample is drawn.
    • Not all questions included in the ECVG module are addressed to all reporting units, and a series of prior filters are applied to adjust the volume of units that will receive the ECVG module to the work capacity that can be assumed by the INE's collection units, selecting the most important ones in terms of size by number of employees or turnover, as appropriate.

    Imputation techniques are therefore applied, varying according to the section:

    Variable 1: Employment by business function
    Information from the Structural Business Statistics and the Commercial Register on employment and activity CNAE-2009 is used.

    Subsequently, the conversion table from CNAE-2009 to Business Functions provided by Eurostat is applied.

    Variables 2-5: International transactions in global value chains
    The imputation data for goods is obtained through a KNN (K Nearest Neighbours) imputation model developed in Python, taking as donors the units that have responded to the module in a valid manner. The proximity variables chosen to determine the nearest neighbour were the autonomous community, the sector of activity, the number of employees, turnover, whether or not it is a group, and if so, what type of group it is, whether it is a head, the country of that head, and continuous variables from VAT.

    The model proposes k=1 neighbours and a weighting method based on Minkowski distance. An overfitting system is desirable, as we want to specify the characteristics of the unit to be imputed as accurately as possible. This is why k=1 has been selected.

    On the other hand, the imputation of services is obtained directly from the International Trade in Services Statistics produced by the National Institute of Statistics.

    Variables 6-8: International outsourcing
    In general, the responses provided by the surveyed unit acting as the HEAD of the group apply to all units within the group. It is difficult to determine whether each unit actually outsources activities, as the public information available in the Consolidated Annual Accounts does not provide that level of detail.

    Average outsourcing percentages are calculated and applied to the number of employees and self-employed workers in each Legal Unit (LeU).

    It is assumed that small businesses do not outsource abroad.

    Variable 9: Motivations and barriers to outsourcing business functions
    Within the group, priority is given to the response that assigns the greatest importance to each item, and this response is transferred to the allocated units whenever possible.

    For all other units, a null response (Not applicable / Don't know) is assigned.

    Variable 10: Impact of recent events on economic globalisation
    A combination of methods is applied: direct imputation when data from units within the same group are available, and KNN imputation when no information can be obtained.

    The imputation carried out did not consist of a correction of the values already obtained with the survey attached to the EEA, but rather an extension of the survey to obtain data for the entire target population. The results by section are as follows:

    The imputation rate (IMP) was calculated with these results:

    Variable 1: IMP=36.59%

    Variable 2-3: IMP=1.64%

    Variable 4-5: IMP=100%

    Variable 6-8: IMP=0.61%

    Variable 9: IMP=0.61%

    Variables 10: IMP=3.77%

    It should be noted that many results were obtained through consistency between the unit and the response. In other words, if a unit had to be imputed but, due to its intrinsic characteristics, did not have to answer a specific question, it was directly imputed with ‘Not applicable’. These cases have not been included when calculating IMP.

    In order to obtain a single record for each Statistical Enterprise, the responses of the Legal Units belonging to the same enterprise are aggregated following the delimitation established by the Profiling methodology.

    The approach to combining responses depends on the type of variable:

    - For employment figures, the values are aggregated.

    - For the number of companies, the response with the highest declared intensity is selected.

    No weighting factors are needed for this statistic, as all units in the target population are included, with data collected or imputed.

    Finally, a complete data set is obtained, with all companies in the target population in the rows and all required variables in the columns, which serves as the basis for producing the final tables and series.

  • 18.6Adjustment

    No adjustments were applied. Neither outlier correction nor seasonal adjustment.

• 19Comment
  • 19.1Comment

    No additional comments were identified at this stage. We will discuss the future improvements.