The companies' responses to eleven of the Survey's central questions form the basis of the Riksbank's new indicator of economic activity. The questions concern companies' perception of the economic situation, new orders, production/sales, employment, access to external financing, profitability and investment plans (see Table 1). The method chosen to summarise companies' responses to these questions is principal component analysis. By identifying dominant patterns and relationships between a set of variables, principal component analysis can filter out noise and extract common movements from a dataset.^[7] For a more detailed description of principal component analysis, see, for example, Jolliffe (2002). The method is considered to be an efficient way of summarising information and is particularly advantageous when the dataset is based on survey data as changes in individual series cannot always be assumed to be relevant to, in this case, the overall perception of the companies.^[8] Stock & Watson (2006). The analysis generates a number of underlying variables - principal components - where the first principal component explains most of the variation in the dataset. In this way, the method can reduce the problem of several of the included series being highly correlated with each other (multicollinearity) while preserving as much of the variation in the data as possible. The first principle component constitutes the Business Survey indicator^[9] The Business Survey indicator explains 57 per cent of the variation in the dataset. This means that the information extracted and summarised in the indicator retains 57% of the original variation between the variables included in the indicator. and it is therefore advantageous to use the indicator in different types of analyses instead of the individual questions.

The Business Survey is conducted three times a year (February, May and September) and therefore the series have first been interpolated to quarterly data (for quarter four).^[10] Until 2016, the interviews for quarter one were conducted in December and January. Between 2017 and 2019, the interviews were conducted in February, May and October/November. For these years, the third quarter has been interpolated. The series are weighted by company size and are seasonally adjusted in the sense that respondents are asked to provide seasonally adjusted responses. The Business Survey indicator is then calculated using principal component analysis, from 2008 quarter 3 to the latest survey period.^[11] In the early years, questions were added/removed. From the third quarter of 2008, series of the questions included in the Business Survey indicator are available. The Business Survey indicator has been standardised to a mean of zero and a standard deviation of one, which means that a negative value indicates that the situation is worse than normal and a positive value that it is better than normal.