As a first step to examine whether the Business Survey indicator can be used as a measure of economic activity, the correlation between the indicator and GDP growth was calculated. Statistics Sweden's (SCB) GDP time series are revised when new or better information becomes available. The first publication of a GDP outcome at a given point in time therefore differs from what subsequent publications show.^[14]Since May 2020, Statistics Sweden has also published a monthly GDP indicator, which is an initial compilation of the economic statistics. Here, the initial publication is the publication of the regular quarterly national accounts and not the GDP indicator. Even though revised GDP data provide more accurate measures of actual economic developments, there are reasons to consider real-time data as well. The first reason is that the information in the Business Survey indicator is largely based on companies' expectations and therefore data should be used that are available when these expectations are formed. The second reason is that the indicator should provide reliable early signals about the economic condition and therefore there is value in exploring whether the signals are most consistent with unrevised or revised GDP data.

The correlation analysis uses revised and unrevised data for the quarterly change in GDP and the annual change in GDP. The correlation was first calculated for the period 2008 Q3 – 2019 Q4 and then for the entire available period, 2008 Q3 – 2023 Q2. The first calculation period excludes the pandemic years in order to evaluate how the indicator correlates with GDP without the shocks from the pandemic.^[15] During the pandemic, there were several factors that complicated the assessment of the economy and business activity. For example, companies’ production and sales opportunities were greatly affected by restrictions and the spread of infection. However, it is worth examining the information value of the indicator even during the pandemic and therefore the correlation has also been calculated until 2023 Q2.

The correlation coefficients, reported in Table 2, show that the indicator correlates strongly with the annual change in GDP but weaker with the quarterly change over the period 2008 Q3 – 2019 Q4. The correlation with the annual change in GDP is relatively high even when the period is extended to 2023 Q2. The correlation with quarterly GDP growth for the period 2008 Q3 – 2023 Q2 is much weaker, but not very different from the correlation of the other survey-based indicators with quarterly GDP growth for the same period (see Table 4 in the Appendix).

Overall, the indicator is judged to reflect GDP growth well, as illustrated in Figures 2 and 3. However, during the pandemic, there is considerable discrepancy between GDP and the indicator. GDP growth reached historically low levels, while the Business Survey indicator suggests only a cautious deterioration in the economy. This can be explained by the fact that the indicator consists of a number of forward-looking variables. For example, in the spring of 2020, a majority of the companies surveyed assessed that the economic situation would improve in six months, which softens the decline in the indicator. Compared to the financial crisis, whose impact on economic activity is reflected well in the indicator, the opposite is true. A majority of the companies surveyed at that time assessed that the economic situation would deteriorate in six months' time and this contributes to the decline in the indicator for that period. However, the objective of the Business Survey, and thus the indicator, is not to identify unpredictable crises (such as a pandemic) but to reflect the trend in economic developments. It is considered to do so even during the pandemic, as it shows, like the real economy, a sudden downturn followed by a rapid recovery.

Figure 2. The Business Survey indicator and quarterly GDP change Standardised units and quarterly change, percent

Figure 3. The Business Survey indicator and annual percentage change in GDP Standardised units (left axis) and annual percentage change (right axis)

Note: Coefficients are significant with p-values less than 0.001. The time series for GDP are calendar-adjusted and seasonally adjusted.

Source: The Riksbank.