Slack and Cyclically Sensitive Inflation by Stock and Watson

Transcription

1 Slack and Cyclically Sensitive Inflation by Stock and Watson Discussion Lucrezia Reichlin ECB Annual Forum Sintra 8 th 2 th June, 28

2 What happened to the Phillips curve? Flattening? Disappearing? Or simply lost in noisy data? ØThe answer to this question has important policy implications ØDifficult empirical problem since an inflation component reflecting the slack in the economy may be dominated by persistent trends influenced by monetary policy and by noise due to shocks such as energy

3 My discussion the message Exploiting the heterogeneity of PCE components is a smart way to identify the Phillips curve component of inflation and construct an indicator of cyclical pressures on inflation Agree with the authors empirical finding that the Phillips curve is not dead However some problems : The CSI gives excessive weight to housing and energy (indirect effect via food) not clearly reflecting business cycle sensitivity This may make it a poor predictor/leading indicator of cyclical inflation Suggest an alternative model in the same spirit as a benchmark

4 My discussion outline A. A close look at CSI B. An alternative decomposition which also captures the cyclically sensitive component of inflation but may better suited for forecasting C. Where are we today?

5 A. SOME FEATURES OF THE CSI

6 CSI very strongly correlated with housing Housing and CSI NBER Core PCE housing cycle CSI cycle

7 Not surprising! Housing exc gas & electricity weighted 63% in CSI and 23% in PCE

8 Indeed housing is very correlated with the CAI Housing and the CAI

9 PROBLEM : why is the CAI so volatile? Output Gaps and CAI 4 CBO Output Gap CAI

10 PROBLEM 2: The CSI is persistently higher than PCE CSI and PCE in level 6 CSI, YoY (%) PCE, YoY (%) Core PCE, YoY (%)

11 Questions on the CSI. How should we interpret the higher level of the CSI with respect to core PCE? 2. Does this reflect the asset price persistent component rather than business cycle? 3. Why have cyclical pressures identified by the CSI since 22 not showed up in headline inflation up to now? 4. But if they don t, should we care about the signal?

12 B. READING THE CSI THROUGH THE LENS OF AN ALTERNATIVE METHOD

13 Alternative approach: trend-cycle model

14 Historical decomposition of the cycles - US Unemployment rate Real GDP Mar-85 Mar-9 Mar-95 Mar- Mar-5 Mar- Mar-5 Mar-2 Mar-85 Mar-9 Mar-95 Mar- Mar-5 Mar- Mar-5 Mar-2 Energy price cycle Phillips curve Idiosyncratic cycle Total cycle 6 Oil price 4 CPI inflation Mar-85 Mar-9 Mar-95 Mar- Mar-5 Mar- Mar-5 Mar-2 SPF: expected inflation Mar-85 Mar-9 Mar-95 Mar- Mar-5 Mar- Mar-5 Mar-2-6 Mar-85 Mar-9 Mar-95 Mar- Mar-5 Mar- Mar-5 Mar-2 UoM: expected inflation Mar-85 Mar-9 Mar-95 Mar- Mar-5 Mar- Mar-5 Mar-2

15 Zoom CPI inflation %, y-o-y Energy price cycle Phillips curve Idiosyncratic cycle Total cycle Mar-8 Mar-9 Mar- Mar- Mar-2 Mar-3 Mar-4 Mar-5 Mar-6 Mar-7 Mar-8 Mar-9 Mar-2

16 Historical decomposition Euro area HICP 2 HICP infation Energy price cycle Phillips curve Idiosyncratic cycle Total cycle Percent

17 Output gap comparison- the US Output Gaps and CAI 4 NBER recessions TC Output Gap CBO Output Gap Output Gap CBO Output Gap CAI CAI

18 Phillips curve and CSI in level: comparison Housing, CSI, and the Phillips curve 8. Core PCE Core housing, PCE YoY housing, % YoY (%) Phillips curve + T+PC inflation trend CSI, YoY (%) CSI, YoY %

19 Compare food (second most important component in CSI) and energy component from trend-cycle model Food Prices and the Energy Component Food & beverages Food & beverages TC energy cycle + inflation trend T+EP

20 Now the Euro area No housing cycle closer correlation between CSI and TC-PC Housing, CSI, and the Phillips Curve CEPR recessions HICP housing, YoY % TC Phillips curve + inflation trend CSI, YoY % CEPR HICP housing, YoY (%) T+PC CSI, YoY (%)

21 C. DOES IT MATTER?

22 Where are we today? The US The CSI: positive cyclical pressure from inflation more or less constant since 22 The trend-cycle model: upward pressure from 25 has now peaked The Euro Area The CSI: neutral pressure The trend-cycle model: upward pressure since 27 Implication for forecasting?

23 Quarterly forecast - US

24 Quarterly forecast the EA

25 Conclusions Some doubts about the model specification However agree with Stock and Watson s fundamental conclusion: there is a cyclical component of inflation hidden in the data This indicates that the Phillips curve is alive and well but needs to be found

26 Finding the Phillips curve is like finding a needle in a haystack but it is hidden there somewhere!