Microsoft Word Altavilla Boucinha Peydro ep word version docx

t
0.00436
-0.0273
-0.0241
-0.0184
(0.0148)
(0.0210)
(0.0204)
(0.0202)
Slope
j,t
0.00150
0.00127
0.00128
0.00121
(0.00152)
(0.00153)
(0.00153)
(0.00153)
VIX
t
0.00220
0.000258
0.000360
0.00104
(0.00199)
(0.00207)
(0.00207)
(0.00214)
Real GDP growth
j,t
-0.00145
0.00717
0.00680
0.00384
(0.00962)
(0.0100)
(0.0100)
(0.0103)
Inflation
j,t
0.0330
0.0300
0.0302
0.0318
(0.0393)
(0.0387)
(0.0387)
(0.0388)
Expected real GDP growth
j,t
0.106***
0.105***
0.105***
0.106***
(0.0195)
(0.0193)
(0.0193)
(0.0194)
Expected inflation
j,t
0.0819
0.0669
0.0673
0.0759
(0.0621)
(0.0620)
(0.0620)
(0.0628)
Expected default frequency
j,t
-0.0587**
-0.0596**
-0.0596**
-0.0665***
(0.0259)
(0.0251)
(0.0251)
(0.0254)
NPL ratio
i,j,t-1
-0.00942**
-0.00802**
-0.00807**
-0.00936**
(0.00401)
(0.00405)
(0.00405)
(0.00390)
Regulatory capital ratio
i,j,t-1
0.00573
0.00872**
0.00859**
0.00770*
(0.00374)
(0.00378)
(0.00377)
(0.00395)
Cost-to-income ratio
i,j,t-1
-0.00324**
-0.00337**
-0.00336**
-0.00327**
(0.00157)
(0.00156)
(0.00156)
(0.00156)
Low-for-long (D
MRO≤1.5
)
t
-0.00695***
(0.00226)
Low-for-long (D
Eonia≤1.25
)
t
-0.00662***
(0.00222)
Low-for-long (Taylor rule)
j,t
-0.00495**
(0.00221)
Bank FE
Yes
Yes
Yes
Yes
Number of observations
2959
2959
2959
2959
R
2
0.602
0.604
0.604
0.603

---

20 
The negative impact on profitability obtained with the other two low-for-long measures (i.e. using 
a threshold of 1.25% for the EONIA and 1.5% for the MRO) in columns 2 and 3 is of similar 
magnitude. This evidence is akin to the mechanism described in Brunnermeier and Koby (2017) 
where the temporary positive effects on the repricing of the securities held by banks are over time 
increasingly offset by the negative impact on net interest margins. 
The estimated impact can, however, be substantially different when the endogenous reaction of 
the macro variables associated with the low interest rate environment is taken into account. This is 
illustrated by the yellow line in Figure 4: a 1pp increase in the expected GDP (associated with an 
increase in bank profitability of about 10 basis points) would shift the blue line outward thereby 
contributing to a significant delay in the materialisation of the negative consequences for bank 
profitability associated with a low-for-long environment. For the first five years the change in 
expected GDP more than offsets the negative impact on profitability linked to the low-for-long; it 
would then take about ten years (twice as long as in the previous case) to reduce the profitability of 
the median bank by 25%. Overall, the adverse impact of a protracted period of low rates on 
profitability is likely to be offset by the respective impact on loan loss provisions and 
intermediation volumes, a mechanism not envisaged in Brunnermeier and Koby (2017) and further 
explored in the next subsection.
Figure 4: Low-for-long, bank profitability and macroeconomic outlook 
Notes: the chart illustrates the results of the estimates in column 4 of Table 4. The 
solid blue line indicates the impact on ROA of being in low-for-long and is 
constructed at unchanged macroeconomic outlook. The yellow line shows the change 
in ROA when although being in low-for-long, the macro outlook improves.