Inflation is bipolar, according to a new Bank of International Settlements (BIS) study, and prone to feedback loops not captured in standard economic models.
The BIS report proposes a dual-personality view where inflation exists in either as low and steady or in a high, unstable form, requiring different monetary medicine for each state.
But happy-go-lucky low inflation can quickly switch into its wild, unpredictable alter-ego through self-reinforcing price spirals.
Agustín Carstens, BIS general manager, says the report highlights “how the behaviour of inflation is quite different at low and high levels”.
“Critically, the self-stabilising properties of the inflation rate when it settles at a low level vanish once it moves to a higher one. And the focus on transitions highlights how the risk of higher inflation becoming entrenched should not be underestimated,” Carstens says.
“In other words, just as water transitions across states – frozen, liquid and gas – inflation changes as its temperature rises.”
The BIS findings suggest central banks must alter monetary policy settings according to the hot or cold inflationary environments.
“What may be less appreciated is that the different behaviour of inflation in high- and low-inflation regimes has first-order implications for the conduct of monetary policy,” the report says. “Those implications have relatively limited consequences for the conduct of policy in a high-inflation regime: in those circumstances, there is little choice but to tighten policy in order to bring inflation down.”
Conversely, authorities can be more tolerant of price level deviations during low-inflation regimes where policy decisions hold relatively little sway.
Of course, the tricky part is differentiating the beginning of an out-of-control price spiral from benign fluctuations during a low-inflation state.
“The challenge of identifying transitions in real time is particularly tough because it is precisely around those periods that standard models perform more poorly in predicting inflation,” the BIS study says. “For one, many of these models take for granted that inflation is mean-reverting.”
Furthermore, most economic analysis fails to account for the “self-reinforcing dynamics” identified by BIS.
“And, exacerbating the problem, those parameters may well have been estimated on a sample when inflation was mostly low and stable, as is the case for many estimates available today,” the report says. “Put differently, the models are least valuable when needed most.”
The BIS two-faced proposal is an early-stage effort, the paper says, with more statistical work required to understand inflation: both kinds.