Depressed market restructuring

low in a historical perspective, indicating that the effect we see is more of a recovery tendency following the Great Financial Crisis. The historical development is presented in appendix 23.

Summing up, the evidence may be interpreted in multiple ways, supporting several arguments. The findings are thereby inconclusive. Still, it is interesting that the United Kingdom has faced a more significant ratio growth than the other economies. This might imply that NIRP is slowing the growth of the ratio. However, this is only speculation. Further, the findings do not support growth in the number of zombie firms. At the same time, the dampening development in the NIRP economies may suggest that there are zombies influencing the results.

5.2.1.5 Conclusion

There is little empirical evidence of zombie growth following NIRP implementation.

However, we have found some economy specific indicators that could indicate an increase, but these findings are inconclusive. We thereby must reject our initial hypothesis. However, one can always argue that it is unreasonable that NIRP implementation and transmission to the lending rates have not facilitated survival of otherwise non-viable firms. Still, these effects may not have been captured by the analyses. The analysis focus has been on the aggregate economy, and this may not be the best approach to uncover expansion of zombie presence. Sector or industry-level analyses could have provided more specific details and reliable results. It is reasonable to expect that lower lending rates would result, to some degree, in increased survival of non-viable firms, and that the lack of empirical evidence in these analyses, therefore, might not be evidence of zero growth, but rather confirm that zombie presence is difficult to uncover on an aggregate economy level.

To investigate whether the fear has materialized, we will focus on the development of bankruptcies together with the business cycle, to account for the cycle effect on

bankruptcies. In general, one would expect an inverse relationship between the business cycle and the number of bankruptcies. However, the artificially low interest rates could facilitate the survival of otherwise insolvent firms. Please note that we do not have a

complete set of bankruptcy figures for the Euro Area, and the findings will only be indicative.

Figure 5.2.2.1 Average annual number of bankruptcies versus the cycle

A. Denmark B. Sweden

C. Switzerland D. Euro Area

E. United Kingdom

Last observation: Average annual bankruptcies: Last observation end of year 2015 for Switzerland, end of year 2016 for all other economies. Ratio to trend GDP: Last observation September 2016 for all economies. Source: Bankruptcy data

downloaded from Trading economics; Euro Area figures are self-calculated based on data for different Euro Area economies;

The uncovered general tendency is a common inverse relationship between the business cycle and bankruptcy rate. Despite the general common tendency, there are a few

deviations worth taking note of. In the initial period following NIRP, Denmark experienced a reduction both in the ratio to trend GDP and bankruptcies. This could be a sign of an initial NIRP effect. However, only Switzerland experience a similar co-movement of the variables in the first period, but here we see an increase in bankruptcies and an upswing in the business cycle. The effect in Switzerland is thereby the opposite of the expected NIRP effect.

However, it is important to note that the Swiss National Bank implemented NIRP in December 2014, and the data available may thereby not capture the actual NIRP

response. Further, the implementation of NIRP in Sweden was also late 2014, so the initial NIRP effect in that economy could be argued to be the development presented in the 2015-2016 timeframe. These findings suggest a NIRP effect with a reduced co-movement of both variables. However, there is no co-movement effect uncovered in the Euro Area.

Nevertheless, the large scale of the Euro Area and the lack of data could be a disturbing factor. Still, the data is inconclusive, and we are not able to conclude that there is an initial NIRP effect.

A further deviation of interest is the deviation uncovered towards the end of the period. In Denmark, we find a significant increase in bankruptcies, while there are no significant fluctuations in the business cycle. In Sweden, the two variables both decline. In the Euro Area, the two variables move in each their direction, as expected, but there is a significant reduction in bankruptcies, while the business cycle growth is only moderate. The findings in the Euro Area and Sweden could be argued to be a NIRP effect on bankruptcies. However, there is contradicting evidence in Denmark. In Switzerland, we lack corresponding data. NIRP implementation was made at different points in time, and this timing difference makes it hard to uncover periodic NIRP effects. Highlighting deviations is, therefore, more a statement of anomalies rather than a tool to uncover periodic NIRP tendencies. The different findings in the economies do not support a common NIRP tendency but could suggest economy-specific NIRP effects.

However, there is an important factor to be noted related to the end of period deviation in Denmark. The development we find here may be a long-term effect of NIRP implementation:

The sharp increase in bankruptcies, despite the flat business cycle, could be a sign that the NIRP effect keeping zombie firms alive has come to an end. Even though we have not found any conclusive evidence of a NIRP effect on bankruptcy rates, an effect might still be

present. If NIRP was not implemented in the economies, the bankruptcy rates could

potentially have been significantly higher. The NIRP would thereby have a calming effect on the bankruptcy rates, even though we were not able to prove this in our analysis. The end-of-period development in Denmark could support this theory. However, at this point the data available is insufficient, so the discussion is only speculative. Based on our findings we must reject our hypothesis, but as more time goes by and more data become available this conclusion might change.