Business tendency survey for manufacturing, mining and quarrying
Updated: 22 July 2021
Next update: 22 October 2021
About the statistics
The statistics provide current data on the business cycle for manufacturing, mining and quarrying by collecting business leaders’ assessments of the economic situation and the short-term outlook.
Local unit (establishment) : An enterprise or part of an enterprise that is located in one particular place and can be identified geographically.
Enterprise : The smallest combination of legal units that is an organisational unit producing goods or services and that benefits from a certain degree of autonomy in decision making.
Branch unit : Unit, which comprises all establishments within an enterprise belonging to the same 3-digit industry group (SIC2007).
Altinn: digital government system for reporting data via the Internet.
NACE : Standard for industrial classification used by EUROSTAT based on the UN's international standard for industrial classification, ISIC Rev. 3.
Unadjusted figures : Raw data figures with primary information from the respondent.
Seasonally-adjusted figures : Time series for which calendar and seasonal effects have been removed. X12-ARIMA is used to calculate these figures.
Trend series : Time series for which calendar and seasonal effects together with the irregular component have been removed. The trend in a time series reflects the long-term tendency that influences the series and has a fairly smooth and monotonic character. The trend series are calculated in connection with the decomposition of the time series for the unadjusted figures in X12-ARIMA.
Response distribution : Employment weighted shares in percentages for valid response alternatives for a single question. For questions like «total level of production» the response alternatives are greater , unchanged and smaller respectively. The response distribution may be expressed in the following way:
(1) G + U + S = 100
G = Percentage that has replied: Greater
U = Percentage that has replied: Unchanged
S = Percentage that has replied: Smaller
Net figures : Defined as the difference between the percentage shares for the extreme response alternatives. For questions like «total level of production» the extreme response alternatives are greater and smaller respectively. The Net figure, N, is defined as:
(2) N = G - S
The net figure, as an indicator of the development in the variable, is often assigned turning point characteristics. A net figure greater then zero indicates that the growth rate of the variable is positive. A positive net figure, but reduced from one quarter to the next, indicates that the growth rate is still positive but reduced. The opposite applies for a negative net figure.
Diffusion index : Defined as the estimated positive percentages (greater) plus half of the neutral answers (unchanged). For questions like «total level of production» the diffusion index, D, is compiled as:
(3) D = G + 0.5 x U
The diffusion index has a simple intuitive approach as it compiles the respondents answering greater seen in association with half the share of respondents answering unchanged. The simplicity lies in the fact that the indicator builds on the assumption that half of the respondents answering unchanged in practice have experienced a growth in the variable, while the other half have experienced a decline. The diffusion index has 100 as the maximum value when all active respondents choose the response alternative greater. The minimum value is equal to 0 when all choose smaller. The index normally fluctuates around 50, which is also the turning point value. Below are some interpretations for the diffusion index as described in the literature:
- If the D value is greater than 50 it indicates that the growth rate of the variable is positive, and the opposite for a value below 50.
- If the D value rises from a level below to a level above 50, the growth rate of the variable has turned from negative to positive.
- If the D value is greater than 50 and increasing it indicates that the growth rate is increasing, while a falling index from above 50 indicates a falling rate of growth, but still positive. The opposite is the case for an index below 50.
Industrial Confidence Indicator (ICI) : The ICI is calculated on the basis of the net figures from three questions in the Business Tendency Survey:
- Actual development in total stock of orders compared with the previous quarter (X)
- Expected development in the level of production in the forthcoming quarter compared with the present quarter (Y)
- Assessment of stock of own products intended for sale (Z)
The ICI is the arithmetic average of the net figures (Z with inverted sign). Further, the trend is identified by the seasonal adjustment of the ICI. The Norwegian ICI is harmonised with the ICI defined by EUROSTAT, and the composition is described in detail in Economic Paper number 151, see DG ECFIN (2001). The ICI is supposed to be a leading indicator for the production in manufacturing industries whereby increases in production expectations indicate directly increases in the forthcoming level of output, increases in the total stock of orders indicate an increased level of production due to the fulfilment of the received orders, and finally, increases in stocks indicate slow sales and reduced activity.
(5) ICI = (X + Y - Z)/3
The survey is classified according to the Standard Industrial Classification 2007 (SIC2007). This is a Norwegian adaptation of Eurostat’s industry classification, NACE Rev. 2. SIC2007 forms the basis for classifying units according to principal activity in the Central Register of Establishments and Enterprises. The use of common standards is essential in order to enable the comparison and analysis of statistical data at an international level and over time.
The survey is also classified according to EUROSTAT's end-use categories (Main Industrial Groupings, MIG). The end-use categories (MIGs) are based on the 3-digit level industrial groupings in SIC2007. Five end-use categories are included in the survey:
Consumer goods (E3+E4)
The following table summarises the most important industries included in the different end-use categories:
Main industries included
Wood and wood products, Paper and paper products, Basic chemicals, Rubber and plastics products, Non-metallic mineral products, Basic metals
Machinery and equipment, Building of ships, boats and oil platforms, Repair and installation of machinery
Manufacture of furniture
Food products, Printing and reproduction, Basic pharmaceuticals
Consumer goods (E3+E4)
Manufacture of furniture, Food products, Printing and reproduction, Basic pharmaceuticals
For a complete description of industries covered in each MIG, see Commission regulation (EC) No 656/2007 .
The objective of this classification is to provide an activity breakdown of NACE, which is more detailed. The classification of the different units is based on the application of the produced products. It should be noted that the MIGs are not comparable in size, in particular the consumer durables heading is smaller than the others.
Name: Business tendency survey for manufacturing, mining and quarrying
Topic: Energy and manufacturing
Division for Business Cycle Statistics
National level only
Published about one month after the end of the quarter
Non-revised and revised micro data are stored in accordance with Statistics Norway's guidelines for storing computer files.
The survey maps out business leaders’ evaluations of the economic situation and the outlook for a fixed set of indicators. Even if the survey does not give precise measures of economic variables, this kind of business survey helps to monitor the economic trend in present time and in the short-term outlook.
The survey was established in 1973 and put into operation on a regular basis from the first quarter of 1974. In 1995, a major review of the survey was conducted. The review was especially concerned with a modernisation of the bottleneck questions and related response alternatives i.e. factors delimiting the production activity. A second review of the survey was conducted in 2011. Some questions concerning delivery time and inventories were replaced by other ones concerning input factor prices, profitability and factors delimiting gross capital investments.
As from the first quarter 2009, all results will refer to SIC2007 (see paragraph 4.2). The historical series are recalculated according to this version of SIC, and results for the business tendency survey dating back to 1990 are available in the StatBank database. Historical series based on SIC2002 (see paragraph 6.1) also remain available for the period 1988 to 2008.
The survey is wholly financed by government appropriations.
The users of the business tendency survey are found within the financial sector, the macro economic analytic environment, media and public institutions (the Ministry of Finance and Norges Bank among others). The results are mainly used for monitoring the economic performance during a business cycle, for analyses and for predicting the short-term development.
The survey is one of many indicators that form a basis for monitoring the performance of the economy during a business cycle. A collection of business cycle indicators is available in the theme page Economic indicators and Key Economic Figures. The survey does not, however, have a direct connection with other statistics.
The population covers all branch units in mining and quarrying (SN05, SN07-08, SN09.9) and manufacturing (10-33), see Standard Industrial Classification 2007 (SIC2007) . The Central Register of Establishments and Enterprises (CRE) defines the population, and the branch unit is the unit for analysis in the survey. The branch unit comprises all establishments within an enterprise belonging to the same 3-digit industry group. See paragraph 4.1 for a complete definition of establishment, branch unit and enterprise.
The survey uses data collected by questionnaires from the units included in the sample, in addition to information from the CRE. The CRE is Statistics Norway's own register of all legal units and establishments in the private and public sectors in Norway. Employment data from the annual Manufacturing statistics are also used in the estimation process.
The gross sample includes about 800 units and represents about 3.5 per cent of the total population of branch units. The sample units cover about 40 per cent of the total level of employment for the industries covered by the survey. The sample includes all branch units with 300 employees or more (panel). The remaining units are drawn by methods based on stratification and optimal allocation with probability proportional to the size of the unit measured by the number of employees. The sample does not include establishments with fewer than 10 employees.
The survey is based on data collected by questionnaire. The questionnaires are returned electronically via Altinn. The questionnaire is released around the 10th in the last month of each quarter. The deadline for returning the questionnaire is the last day of the month. Units registered with an e-mail address in Altinn are notified by e-mail when the questionnaire is available on the Internet.
The largest of the establishments within the branch unit is used as the reporting unit. For practical reasons, some enterprises prefer to report from the head office. The person responsible for filling in the questionnaire should be the leader of the unit or a member of the management staff. Establishments that fail to return the questionnaire receive a reminder within a few days of the deadline. A new deadline of 12 days is set.
The questionnaires are downloaded from the Internet, and the data are automatically checked for duplicates. Questionnaires edited close to the release day, and faxes, are manually registered. When data from the questionnaires are loaded to the production database, they are controlled for logical errors, such as multiple response alternatives chosen where this is not valid. Revision on aggregated level is done by assessing the development over time, and unacceptable series lead to further revision of the data. Results and tendencies are also compared with other relevant quantitative statistics.
Employment weighted results (response distribution) are calculated for each question. The sample units are classified in different strata depending on the number of employees in the branch unit and in which industry they belong to (3-digit NACE). For each question by stratum a response distribution is estimated using employment data as weights. The response for each branch unit is given a weight equal to the number of employees. For aggregation to the industrial group level and totals, the stratum results carry a weight equal to the stratum population employment.
Time series sometimes contain significant seasonal variations that make it difficult to interpret the development from one period to another. To facilitate the interpretation of such time series, the figures are seasonally adjusted. For more information on seasonal adjustment, see Seasonal adjustment: general information .
In the Business Tendency Survey, seasonally adjusted figures and trend figures are calculated with X12-ARIMA for most of the questions. Only the trend figures are released, together with the unadjusted figures. For survey specific documentation of seasonal adjustment practices, see About sesaonal adjustment .
Confidential micro data : According to § 2-4 of the Statistics Act, collected data are subject to secrecy and must be kept or destroyed in a secure manner. Any use of the data must be in accordance with the rules set out by the Data Inspectorate.
Time series that are not to be published : The publication of data is subject to the provisions in § 2-6 of the Statistics Act. The main rule is that data should not be published if they can be traced back to the respondent, i.e. figures for which less than three respondents make up the foundation for a cell in the table, figures where one respondent represents more than 90 per cent of the total value or figures where two respondents represent at least 95 per cent of the total value. In the business tendency survey this is the case for divisions 05, 12 and 19 (SIC2007), and for a number of other more detailed levels of aggregation. As a main rule, all series that are not published are considered confidential.
Unpublished data : Revised data that are not published are subject to secrecy. This means that they are unavailable without specific approval.
As from the first quarter 2009, SIC2002 has been replaced by SIC 2007 (see paragraph 2.1). The historical series based on this new version of SIC have been recalculated back to 1990. For the years 1990 to 2003 a macro-approach is used for back-casting the time series. For the years 2004 - 2008 the survey is recalculated by a detailed re-working of individual data (micro-approaches). Users of the data must ensure they use results based on the same version of SIC when making comparisons over time. Historical series based on SIC2002 remain available in the StatBank database under Completed time series. However, as from the first quarter 2009, only series based on SIC2007 will be continued.
Measurement errors are caused by the questionnaire or the respondent’s internal systems for obtaining the data. Sources of measurement errors may be ambiguous guidelines or the respondent’s insufficient accounting systems. In the Business Tendency Survey, errors in reported answers may originate from misunderstandings of the definition of the main variables used in the survey. Unambiguous guidelines and definitions are therefore emphasised.
Processing errors can occur when Statistics Norway processes the data. Typical examples are misinterpretations of the answers on the questionnaires &– for example that a chosen response alternative is not registered. Paper questionnaires are optically read with automatic verification and transfer to an electronic medium. The current techniques for optical reading are of a high quality, and few errors are found in this phase of the production. The introduction of Altinn has also helped reduce such errors, as data from electronic questionnaires are loaded directly into the system. Questionnaires that are not verified by optical reading are processed manually. Thus there is room for human error, but because the proportion of manually registered questionnaires is very small this type of error seldom occurs.
After data has been loaded into the production database, in the revision process, there will be occurrences of multiple response alternatives where this is not valid. When this occurs the answer must be evaluated and a response alternative must be chosen as the valid one. These corrections are based on assumed logical coherences with other questions in the same questionnaire or by the use of response alternatives from previous questionnaires for the same respondent. This assessment can cause processing errors when the response alternative registered is not what the respondent had in mind. However, the number of occurrences of multiple response alternatives is very small, and this kind of error does not exist in the electronic questionnaire. In this questionnaire it is not possible to choose more than the valid number of response alternatives.
Errors of non-response refer to errors that either occur due to unit non-response or item non-response. Unit non-response occurs when the respondent has not returned the questionnaire, while item non-response occurs when at least one of the questions in the questionnaire is not answered.
Unit non-response for the survey is around 7 per cent when final production file is ready. Critical units, i.e. units that have a considerable impact on the results at a detailed level aggregation (2-digit NACE), are contacted by telephone. Calculations of the effect of missing units have been carried out (see Documents 2004/3), and no systematic skewness has been uncovered. Unit non-response is considered to be neutral, and is covered indirectly in the aggregation when inflating to population level.
Item non-response (single questions not answered in the questionnaire) is coded automatically as response alternative non-response and is not normally imputed.
Sampling errors refer to uncertainty that occurs when figures are produced based on a sample survey as opposed to a full count. The sample variance equals the expected deviation between a sample survey and a full count. In the business bendency survey the sample represents 3.5 per cent of the branch units in the population and covers about 40 per cent of the population’s total employment rate. In order to ensure a high degree of relevance at the lowest cost possible, great effort is put into including all large units in the population in the sample. The effect of sampling errors occurring in the population estimate is calculated, and the results are published in Documents 2004/10 .
Units in the sample that close down can be a source of skewness if the proportion of units closing down in the sample deviates from the population. The business tendency survey is mainly based on a fixed sample (panel). Periodic updates of the sample ensure that the structure of the sample is in accordance with the population.
Coverage errors refer to errors in registers that define the population. As a result, units may be incorrectly included in or excluded from the population. Other problems are related to delays in the update of the registers and units that are incorrectly classified. From experience a limited share of the population units are incorrectly classified. This is usually due to misleading or insufficient information at a certain time. Calculations of the size and significance of such errors have not been carried out. However, such errors are not considered to be greater than for other quantitative short-term statistics. Industry classification for the sample units is revised annually in the first quarter to ensure correct classification in accordance with the CRE.
What is seasonal adjustment?
Monthly and quarterly time series are often characterised by considerable seasonal variations, which might complicate their interpretation. Such time series are therefore subjected to a process of seasonal adjustment in order to remove the effects of these seasonal fluctuations. Once data have been adjusted for seasonal effects by X-12-ARIMA or some other seasonal adjustment tool, a clearer picture of the time series emerges.
For more information on seasonal adjustment: metadata on methods: seasonal adjustment
Why do we seasonally adjust the business tendency survey?
The business tendency survey is part of a system of short term statistics to monitor the economy. The primary goal of the survey is to provide current data on the development in the business cycle for manufacturing, mining and quarrying. The survey does not give precise measures of economic variables, but it still provides useful information on the current situation and the short-term outlook.
The level of activity within manufacturing, mining and quarrying will vary throughout the year because of public holidays etc. Some industries also experience fluctuations due to a change of seasons. An example is the demand for and production of certain food products which depend on whether it is summer or winter. This kind of effects will influence the reported data for a number of indicators in the business tendency survey and make it difficult to compare the results from quarter to quarter.
The business tendency survey is subjected to a process of seasonal adjustment in order to remove the effects of seasonal fluctuations. In this way we are able to analyse the underlying development in the business cycle. It is mainly the smoothed seasonally adjusted time series (trend) that are released and analysed.
Time series that are seasonally adjusted
The business tendency survey publishes 220 seasonally adjusted time series which covers a wide range of indicators on the development within manufacturing, mining and quarrying and EUROSTAT's end-use categories (Main Industrial Groupings, MIG).
Pre-treatment is an adjustment for variations caused by calendar effects and outliers.
No pre-treatment is performed.
Calendar adjustment involves adjusting for the effects of working days/trading days and for moving holidays. Working days/trading days are adjustment for both the number of working days/trading days and for that the composition of days can vary from one month to another.
No calendar adjustment is performed.
Methods for trading/working day adjustment
No adjustment for trading/working day is performed.
Correction for moving holidays
No correction for moving holidays is performed.
National and EU/euro area calendars
Calendar adjustment is not required.
Treatment of outliers
Outliers, or extreme values, are abnormal values of the series.
No pre-treatment of extreme values.
Pre-treatment requires choosing an ARIMA model, as well as deciding whether the data should be log transformed or not.
Model selection is performed manually based on statistical tests.
Comments: (0,1,1) (0,1,1) or the "Airline model" is selected manually for all the time series.
The composition scheme specifies how the various components - basically trend-cycle, seasonal and irregular - combine to form the original series. The most frequently used decomposition schemes are the multiplicative, additive or log additive.
Manual decomposition scheme selection after graphical inspection of the series.
Comment: Log additive method is in use for some of the time series.
Choice of seasonal adjustment approach
Consistency between raw and seasonally adjusted data
In some series, consistency between raw and seasonally adjusted series is imposed.
Do not apply any constraint.
Consistency between aggregate/definition of seasonally adjusted data
In some series, consistency between seasonally adjusted totals and the aggregate is imposed. For some series there is also a special relationship between the different series, e.g. GDP which equals production minus intermediate consumption.
Do not apply any constraint.
Direct versus indirect approach
Direct seasonal adjustment is performed if all series, including aggregates, are seasonally adjusted on an individual basis. Indirect seasonal adjustment is performed if the seasonally adjusted estimate for a time series is derived by combining the estimates for two or more directly adjusted series.
The unadjusted data are aggregated, and direct seasonal adjustment is performed on aggregates and components using the same approach and software. Any discrepancies across the aggregation structure are not removed.
Horizon for estimating the model and the correction factors
When performing seasonal adjustment on time series, it is possible to choose the number of observations to be used when estimating the model and the correction factors. Correction factors are the factors used in the pre-treatment and seasonal adjustment of the time series.
The whole time series is used to estimate the model and the correction factors.
General revision policy
Seasonally adjusted data may change due to a revision of the unadjusted (raw) data or the addition of new data. Such changes are called revisions, and there are several ways to deal with the problem of revisions when publishing the seasonally adjusted statistics.
Seasonally adjusted data are revised in accordance with a well-defined and publicly available revision policy and release calendar.
Concurrent versus current adjustment
The model, filters, outliers and regression parameters are re-identified and re-estimated continuously as new or revised data become available.
Horizon for published revisions
The revision period for the seasonally adjusted results is limited to 3-4 years prior to the revision period of the unadjusted data, while older data are frozen.
Comment: The revision period for the seasonally adjusted figures is 4 years when new data are added. The whole time series may be revised when implementing new or improved methods.
Evaluation of seasonally adjustment data
Continuous/periodical evaluation using standard measures proposed by different seasonal adjustment tools.
Quality measures for seasonal adjustment
For most of the series, a selected set of diagnostics and graphical facilities for bulk treatment of data is used.
A table containing selected quality indicators for the seasonal adjustment is available here.
For more information on the quality indicators in the table see: metadata on methods: seasonal adjustment
Seasonal adjustment of short time series
All series are sufficiently long to perform an optimal seasonal adjustment.
Treatment of problematic series
All problematic series are treated in a special way.
Unadjusted data, seasonally adjusted data and smoothed seasonally adjusted data are available.
Comments: Only unadjusted data are released for lower aggregates within manufacturing.
In addition to raw data, at least one of the following series is released: Calendar adjusted, seasonally adjusted, smoothed seasonally adjusted (trend).