In a detailed interview with the News Agency Direkt, then Riksbank Deputy Governor Karolina Ekholm reveals how the Riksbank’s judgmental adjustments of the model forecasts have given the Riksbank’s inflation forecasts an upward bias. She also points out that I warned about this upward bias and explicitly entered reservations about the inflation forecast from december 2012. It was thus not a secret at the Riksbank that there were problems with the forecast.
This was Karolina Ekholm’s last interview as Deputy Governor. She has now resigned from the Riksbank and will be appointed State Secretary with the new Finance Minister, Magdalena Andersson.
This is an English translation of the interview (previously published in Swedish on Ekonomistas):
Ekholm (Riksbank): More confident with more model-driven inflation forecast
STOCKHOLM (Direct 2014-09-23) Since July, the Riksbank nowadays puts more weight on the results from forecast models than on judgment in its inflation forecasts. This does not necessarily mean that forecasts become more accurate, but it reduces the risk that the forecast errors tend to go in the same direction.
This says Deputy Governor Karolina Ekholm in a telephone interview with the News Agency Direkt.
“The general uncertainty is the way it is. One should not expect that the forecast immediately becomes more accurate because of this … but what’s important to me is that we do not make errors that are ‘biased’, errors that tend to be in the same direction all the time, and I feel more confident about that now,” she says.
The minutes from the Riksbank’s monetary policy meeting in September shows that Karolina Ekholm stated that forecast evaluations made by the Riksbank’s Monetary Policy Department showed that models proved to make “relatively good forecasts” of future inflation, and that the Riksbank therefore since July has put more weight on them than on judgment.
“This change may have eliminated a tendency towards biased estimates of future inflation that the forecast has exhibited for some time,” she said.
Karolina Ekholm notes that evaluations of previous inflation forecasts carried out by the Riksbank suggest that “judgment has not improved the predictions, but rather tended to make them poorer.” This could mean that the Riksbank would have pursued an easier monetary policy if less weight had been put on judgment in forecasting inflation in recent years.
“It is impossible to know, but at least it had made the trade-offs involved look a little different,” she says.
She notes that in a situation where inflation has been below target for some time, the entire Executive Board is intent on bringing it back to target “within a reasonable time frame.”
“If the forecasts then show that (inflation) will increase pretty soon, maybe you think that the need to stimulate the economy with more monetary policy is not so great,” says Karolina Ekholm.
According to Karolina Ekholm, you cannot say what it is that has made the judgments poor while the forecasting models have been better.
“But I think there has been a fairly widespread view, not just at the Riksbank, that the underlying inflationary pressure has been stronger than what in retrospect seems to have been the case.”
The Riksbank has received a lot of criticism in recent years for pursuing a monetary policy that is too tight, but given that inflation forecasts have shown that inflation will rise towards the target in about two years, the monetary policy decisions of the majority do not appear markedly tight compared with historical patterns.
“Yes, in any case they have not stood out as remarkable,” says Karolina Ekholm.
She points out that former Executive Board member Lars EO Svensson explicitly entered reservations against the inflation forecasts from December 2012 until he resigned in May 2013. His reservations meant that also Karolina Ekholm “started thinking” about the inflation forecasts, but she chose not to join in the dissent on the forecast.
“So he really had the view that the forecast was an overestimation. I myself dissented against the (repo rate) decisions, and maybe some components of the forecasts, but I did not think I had any reason to directly enter a reservation against the (inflation) forecast”, says Karolina Ekholm.
She points out, however, that the issue of judgments in the inflation forecast has been discussed regularly, and since the Riksbank’s forecast ability is continuously evaluated it is not possible to deliberately publish “strange forecasts – that is something that everyone sees immediately.”
“Now we have been fairly criticized for our forecasts. Because of this we have had reason to continuously evaluate them, and this is of course a result of that evaluation process; we have been able to conclude that several of the forecasting models we use actually have made fairly good predictions,” says Karolina Ekholm.
She does not mean to say that “judgments” have made the forecasts worse all the time, “but there has been such a tendency.”
According to Karolina Ekholm inflation forecasts are always based on judgment as well as models, and the Riksbank uses several forecasting models. The results from the models are weighed together and then judgment is added on top of this.
“But because we’ve made a fair amount of forecasting errors recently, which we are not the only ones to have made, we’ve had reason to look into where these mistakes enter the forecasts,” says Karolina Ekholm.
Per-Anders Degerman, The News Agency Direkt
That the inflation forecasts were biased upwards was not a secret at the Riksbank. Karolina Ekholm refers to my dissenting against the inflation forecast from December 2012 until I left the Riksbank in May 2013. At the December 2012 policy meeting, I said according to the minutes (p. 5-6):
Mr Svensson questioned the forecast in the main scenario that CPIF inflation would rise
quickly to 2 per cent. As shown in Figure 7, the forecasts from 2010 and onwards have
systematically overestimated CPIF inflation for 2012. The forecasts rise fairly quickly to 2
per cent, while actual outcomes show a downward trend. Why should the current
forecast, with a rapid rise to 2 per cent, be better than the previous forecasts, wondered
The figure 7 that I referred to was brought by me to the meeting. It is shown at the end of the minutes and looks like this:
My formal dissent at the December 2012 reads like this (the minutes, p. 26-27):
Deputy Governor Lars E.O. Svensson entered reservations against the Monetary Policy Update and the decision about the repo rate and repo-rate path in the Monetary Policy Update. He advocated lowering the repo rate to 0.75 per cent and then a repo-rate path that stays at 0.5 per cent from the second quarter of 2013 through the first quarter of 2014, and then rises to 1.5 per cent by the end of the forecast period. This was justified by his assessment that the Update’s forecasts of foreign policy rates further ahead, foreign growth and Swedish inflation are too high and that given these circumstances his repo- rate path implies a forecast for CPIF inflation that is closer to the inflation target and a forecast for unemployment that is closer to a long-run sustainable rate and therefore constitutes a better-balanced monetary policy. His assessment was that his lower repo- rate path would not noticeably affect any risks associated with household indebtedness, since monetary policy normally has very small short-run effects on household indebtedness and with low and stable inflation no long-run effects.