In recent years, the Chinese domestic petroleum market has been dealing with oversupply, resulting in decreasing industry profits. Oil companies have been facing fierce competition and unprecedented operational pressures. Increasingly, oil companies have turned to management accounting to increase operating profits and add enterprise value.
The China National Petroleum Corporation (CNPC) is one of the companies affected. Its subsidiary, the CNPC Hubei Sales Company (the HSC), which has increased its operating profits and enterprise value by harnessing extensible business reporting language (XBRL) and big data, offers a case study into how management accountants can utilise valuable data assets.
The HSC took advantage of XBRL in cross-platform and heterogeneous systems to share and integrate data at the enterprise level and to support decision-making in areas including daily operations, market prediction, asset evaluation, cost control, investment, and risk identification.
The CNPC HSC, a state-owned enterprise, is responsible mainly for building a sales network to sell CNPC refined oil products in Hubei province. The HSC also organises oil procurement, transportation, storage, and wholesale and retail transactions.
The HSC has 824 petrol stations and 13 oil depots in operation, and annual sales have increased from 6,300 tonnes to 3.51 million tonnes over 18 years, achieving the objective of occupying one-third market share by sales volume with around 20% of the petrol stations in the area.
To increase sales, the HSC has previously put a lot of resources into promotional marketing activities, the effects of which have been difficult to evaluate. For a long time, management accounting has been limited to traditional quantity analysis, forecasting, performance, and incentives. The HSC also has 21 additional and heterogeneous information systems with decentralised data storage and nonstandardised and inconsistent data, which had hindered further data integration.
To meet new decision-making information requirements, the HSC built a new management accounting system that could process big data (see the sidebar, "Factors Leading to Successful Implementation," below).
The HSC established a financial control system that has integrated financial accounting, fund settlement (the transfer of funds from buyer to seller and the transfer of an asset's title from seller to buyer), and operational budgeting and control. The financial shared services center (FSSC) is responsible for the system's data classification, processing, calculation, and quality control.
XBRL and big data solution
XBRL is business reporting markup language. The standards are developed and managed by XBRL International and are currently used primarily for the definition and exchange of business and financial information. It has been widely used for financial reporting of detailed ("tagged") aggregate disclosures from regulated companies to regulators. This traditional use of XBRL is well established in China as well as the rest of the world. Nearly 150 securities, tax, banking, insurance, statistics, and business regulators mandate its use in about 70 countries. Whilst more than 20 million companies use XBRL for this kind of financial, performance, risk, and compliance reporting to external authorities, the number of companies taking advantage of this sort of standardisation of data inside the enterprise has been extremely limited.
The idea behind the HSC's project is to implement use of XBRL for internal management reporting to structure and standardise heterogeneous data within enterprises — utilising big data to ensure businesses are equipped for the digital age.
The HSC launched the XBRL and big data project to explore the transition path for management accounting in an internet and cloud computing environment.
By using XBRL's technical standards, the HSC aimed to establish a unified data platform and a unified standard index system, and to build a digital enterprise with data-driven scientific management and intelligent decision-making with the involvement of innovators from across the business, connecting past, present, and future.
By using XBRL to tag data, the project breaks each type of business data free from specific software vendors and product restrictions, establishing a standardised, self-describing, private data cloud with a connected semantic information database.
By using XBRL and big-data-based management accounting information, the project also aimed to expand the information available to cover internal management, customer behaviour management, and responsibility centres management, digging deeper into the value of data.
The financial control system and the other 21 disparate information systems in use at the HSC could provide a large amount of management accounting information, but the data is scattered through all the systems in different formats with significant noise and error, meaning it needs cleaning to be converted into valuable data.
Therefore, it was necessary to set up an automatic interactive mechanism within the existing system.
The HSC set up a new shared data platform based on Hadoop/XBRL. The XBRL database is built using a relational and in-memory database. It is based on XBRL multidimensional data characteristics, with object-oriented data storage to complete the calculation and processing, and uses open source Hadoop architecture for distributed computing to solve the XBRL and big data storage challenges and operational efficiency issues.
The new shared data platform integrates the enterprise internal systems with external correlated data, cleaning and transforming so far 1.8 billion trading data points. Internal systems included the financial system, main business system, investment assets system, and materials purchasing system. The externally correlated data was extracted from communications data, tax data, weather data, and other sources. Data extracted from internal and external systems is tagged using business model, index model, element model, or analytical model concepts within a model layer, and then waits to be called by the service layer. Associating highly granular (including transaction-level) data observations with these detailed — and connected — definitions, in an enterprise standardised and systems agnostic format, is key to the design of the system.
The platform implements data connectivity in a two-way flow to achieve the data analysis of user-driven cloud services. At present, it can provide a calculation engine, standard services, an index service, reporting services, analysis services, budget control, parallel computing, and cache services, which can be accessed by a statement reporting program, statistical query program, mobile application, or data mining tools.
Unified data standard system
To achieve automatic data interaction in the new unified data shared platform, it was essential to standardise the data.
The HSC standardised its internal data and its external correlated data in accordance with the XBRL standards technology and created its own practice methodology to standardise data using XBRL.
In this standardised process, data was collected and sorted, real elements were identified and tagged, dimensions and indicators were identified, and the directory was structured. Indicators are used to illustrate the concept of overall quantitative characteristics. Dimensions are used to specify the analysis attributes or characteristics for combining or subdividing indicators' values. At that point the XBRL index system could finally be determined.
More than 2,000 metrics and 23 dimension items were defined from more than 15,000 report items in 341 internal reports, making the underlying data fine-grained, consistent, and flexible.
Easy to use
The XBRL taxonomy for this system uses a language that can be understood by business users easily. By using the same business language as typical human-computer interactions, it shields the financial and business staff from the technical complexity of both XBRL and the underlying IT systems. The interface of the XBRL index system is concise, and through the Excel plug-in and function, nontechnical users can quickly select a target, define custom analytics, and update the data without having to understand technical details such as XBRL syntax or underlying SQL repository structures.
At the same time, the system uses methods of web-based graphical analysis so the company's upper-management users can utilise big data analysis and forecast results intuitively, making it easy to understand comprehensive figures that assist with management decision-making.
Value in data analysis
The HSC has built a multidimensional data cube that is tagged using XBRL. The data cube is a model that can be used to evaluate the aggregated data from a variety of viewpoints and analyse the data separately to discover the underlying cause of business-transactions-related operations and their results.
After tagging the transaction-level data with XBRL concepts, the HSC could quickly find certain petrol stations' discount exceptions, abnormal price periods, the cost of each discount promotion activity, and so on. It could penetrate data freely through various organisation layers, trading hours, sales, product categories, promotional types, and any other dimensions that have been defined, to access the value of transaction-level data from the system. In addition to providing enhanced data mining that yields significantly more timely and effective analysis for decision-making, the integration of data standards across professional lines ensures more accurate and more compliant data.
Value in risk management
The HSC designed an early-warning risk mechanism by using XBRL technology and a machine learning model that could help identify hidden risks, pre-detect potential risks in station-level sales, and help control risk effectively. The model works by combining massive amounts of transaction data from the station-level system and fuel system, with business experience and knowledge of past problems.
It provides real-time tracking to send risk warnings, including of the risk of fuel card fraud and risk of transport-related oil loss, as well as the risks to inventory from the risk of oil losses, including from abnormal pump behaviours. All these risks can be differentiated for every single station, single fuel nozzle, single fuel tank, and each deal, and could support responsible units to deal with the related risk rapidly. For example, on 6 April 2016, the system notified a petrol station branch that No. 4 tank 93# had a petrol loss of 107 litres, a suspected spill. The branch immediately took measures to deal with disposal and repair, effectively reducing the risk of loss.
Value in budget management
By combining XBRL with big data, the HSC is making full use of internal and external data sources. It uses dynamic management models to balance quantity and efficiency within a single station's customised forecast model.
For example, in the single-station fuel nozzle prediction process, a range of business data was used to determine clusters of similar types of petrol stations. Data gathered included the location of the station, station traffic flow, station incoming rate, ratio of petrol-to-diesel sales, customer group characteristics, and other elements, all of which were then analysed using cluster analysis and machine learning methods.
Then a multiple linear regression method was used to study the relationship between volume and price changes to determine the price sensitivity of different types and clusters of stations. Finally, the time series method was used to predict the daily sales volume, monthly sales volume, and annual sales volume.
Since the new budget model has been in operation, the difference between predicted and actual volume has been 3.12%, 1.67 percentage points lower than the difference between predicted volume using traditional methods and actual volume in 2016. By using this model, 37 petrol stations with stagnant sales exceeded their target of more than 1,874 tonnes in the first half of the year, an increase of about 65 tonnes per station over the same period in the previous year. This was accomplished by clustering stations by location, station traffic flow, station incoming rate, ratio of petrol to diesel in sales, customer group characteristics, and other elements. A multiple linear regression is used to study the relationship between volume and price changes to determine the price sensitivity of different types of sites. Finally, the time series method is used to predict sales volume on a daily, monthly, and annual basis. Based on all these analyses, potential critical factors for increasing sales are identified.
Value in customer management
Through the project, the HSC studied customer transaction data, dividing customers into high-value, low-value, stable, important potential, retained, etc., and analysing them from multiple dimensions such as their trading frequency, trading, purchase amount, purchase price difference, and stability. They then regularly updated their marketing and sales departments with dynamic customer information so staff could adjust product structure and marketing strategies and deliver promotions effectively.
A worthwhile endeavour
It's clear that the XBRL-powered big data analysis system has provided numerous benefits to the HSC. Improvements have been made in operations as well as management of risks, budgets, and customer focus. In this highly competitive industry — and in any industry — data-driven management can increase revenue and profits while providing indisputable evidence that it's delivering the value management seeks.
Factors leading to successful implementation
Implementing a big-data-led transformation using extensible business reporting language (XBRL) was achieved with executive leadership and a cross-functional approach at the China National Petroleum Corporation's Hubei Sales Company (the HSC). The following factors led to the project's success.
Government support and guidance
Since 2010, the accounting department of the Ministry of Finance has actively supported XBRL implementation in China. It has introduced the XBRL specification and has developed, published, and maintained a taxonomy for China Accounting Standards, as well as extension taxonomies for specific industries and specific regulatory requirements.
Top management emphasis
The HSC's top manager emphasised the importance of the project, helping to smooth implementation and making it easier to get the human, material, financial, and other needed resources.
Project team cooperation
The HSC's XBRL project team was led by the company's CEO and CFO. The financial department and information department were jointly responsible for the project. The project team cooperated very closely across the core project team and across internal departmental "silos".
Manager-driven implementation plan
The project plan was clarified and appropriate and was supported by different-level managers. The HSC began with the most difficult challenge faced by middle-level or tactical-level managers (the core business line leaders). After achieving significant, recognised results, the middle-level managers helped to push the potential for related achievements to top managers (strategy-level manager, chief-level manager). After receiving recognition from top managers, the project had a powerful driving force and enough resources to improve and maximise the project's potential benefits.
Meeting internally driven needs
The XBRL project was involved in almost all elements of the business. This helped the business department grasp the operational and management trends accurately, locate the risk points quickly, embed risk prevention and control procedures in all business processes, and support compliant and efficient operation. All these were internally driven needs of the enterprise and were highly recognised by all business units.
The cost of the project was less than 10 million RMB ($1.5 million) and was controlled to be lower than for an ordinary information system project, which usually costs almost 30 million RMB ($4.5 million) or even more. This increased the total profits of the HSC, which further increased support amongst the top managers.
Data Analysis Fundamentals Certificate, cpd.cimaglobal.com
Analytics and Big Data for Accountants, aicpastore.com
Yanchao Rao, Ph.D., is an associate professor at Shanghai University of Finance and Economics; Bing Leng is director of a division in the Accounting Regulatory Department of the Ministry of Finance of the People's Republic of China; Xiuping Mu, MBA, is CFO of China Petroleum Engineering Co. LTD; Chao Li is a partner at PwC Consulting in Beijing; Xiaofeng Hu is the director of the XBRL Business Division at Pansoft; and John Turner is the CEO of XBRL International. To comment on this article or to suggest an idea for another article, contact Ken Tysiac, FM magazine's editorial director, at Kenneth.Tysiac@aicpa-cima.com.