It turns out that according to this Reuters article, the unemployment rate data collected over the last few months is basically wrong and has been overestimated. The data I used (via FRED) showed that the Swedish unemployment over the last two months was edging up to 7.4%. Based on that figure, the recent increase would have meant that Sweden was heading for a recession, since the 3-month moving average was increasing by more than 0.5 percentage points over its previous 12 month low. However, now it turns out that this recent increase is fake. According to the Swedish Statistics office, the seasonally adjusted unemployment rate is only 6.8% while the non-seasonally adjusted figure is a little lower. While these new figures are still somewhat higher than what the unemployment rate was at the beginning of the year, the Sahm rule would not give us a recession warning based on this updated data. It is still a little unclear to me what precisely has happened over the last months and why the initial unemployment rate data was wrong to such an extent. But this error is obviously a huge problem and the agency responsible for collected the data must figure out as soon as possible what has happened since this data is also of crucial importance for domestic policy makers, including the Riksbank.
Maybe the key takeaway from all this is that even current Macro data can be of poor quality at times. So beware!