To answer our main problem statement, we first take a closer look at the sets of questions and hypotheses we have been investigating. We will start off with the set of sub-questions related to investments.
The first sub-question was related to a potential preference shift and analysed in section 5.1.1. We found that the correlation between deposit rates and key policy rates has been high both historically and post NIRP, indicating that the transmission to the market has been strong. Further, we isolated an increase in households’ savings following NIRP
implementation. Focusing on corporate savings, we were not able to deduct a common trend, nor isolate a NIRP effect. For the investments, we found growth in both sectors.
However, we could not conclude that NIRP was the sole key driver behind the observed development. Overall, we were not able to find evidence of a preference shift due to the negative interest rate policy implementation and had to reject our hypothesis.
The second investment sub-question aimed at answering how the level of corporate
investments had been affected by NIRP through the discount rate. Due to a relatively strong correlation between the key policy rates and the discount rates, we concluded that the transmission was strong. The negative coefficients between the GFCF investments and the discount rates indicated the expected inverse relationship, but the correlations were weak.
To address whether the NIRP was the key driver behind the development, we compared the correlations for the NIRP economies to the one of the United Kingdom. As the coefficients were quite similar, and we did not manage to obtain reliable post-NIRP coefficient, we could not accept, nor reject the hypothesis that NIRP influenced the investment level through the discount rate.
In the third analysis, we investigated aggregated macroeconomic variables related to investments, to answer whether NIRP caused the development in the aggregate economy.
First, a test was conducted to check the correlation between GDP and investments, and the analysis established a significant relationship between the two variables. Further, we studied the development of GFCF investment and uncovered that there has been an investment growth in all economies following NIRP implementation. However, as the control economy
faced a similar development, we concluded that NIRP was not the isolated driver. The evidence rather suggested expansionary monetary policy to be the cause of the
development, and a sector-specific analysis further supported these findings. Summing up, we established aggregate investment growth but had to reject the hypothesis that NIRP was causing this development.
Lastly, we investigated NIRP effects with respect to overinvestments through a series of different analyses including capacity utilization, output gaps, GFCF to GDP ratio, M1 to GDP ratio and indicators of credit induced overinvestment bubbles. We found a few warning signs for the Swedish economy, but the overall results did not show any evidence of an overinvestment bubble building. We thereby had to reject the hypothesis.
Turning to the sub-questions related to liquidity-constrained households and firms, the first hypothesis concerned whether NIRP could contribute to an increase of zombie firms. We started the analysis by investigating the transmission effect of policy rates to the commercial lending rates, and the test results indicated a high degree of transmission. However, none of the lending rates had entered negative territory. Later, we investigated the development of the TFP index controlled for the business cycle, non-performing loans and corporate profits to corporate debt, to see if zombie presence was increasing on the aggregate level. Even though it is reasonable to assume that the rate cuts have facilitated some degree of zombie survival, the analyses did not provide any evidence of an increase in zombie presence, and we had to reject the hypothesis.
Focusing on the hypothesis related to potentially depressed market restructuring following NIRP implementation, we looked at the development in the average number of bankruptcies controlled for the business cycle. The output showed an inverse, healthy relationship for all economies, including the UK. There were some interesting deviations, however, they were not large enough to influence the conclusion. Thereby, we had to reject the hypothesis.
Finally, we investigated if NIRP implementation had caused rigidity in labour resource reallocation, and influenced labour utilization. Through the analyses, we were not able to isolate a NIRP effect, and in contradiction to the hypothesis, the evidence suggested an improvement in labour reallocation rigidity. In addition, the current trend in the NIRP
economies is an improvement with respect to labour utilization. Based on the findings, we had to reject our hypothesis.
After reviewing the sub-questions and hypothesis conclusions, we are now ready to draw a conclusion with respect to our main problem statement: Following NIRP implementation, what are the main hypotheses and concerns related to investments and
liquidity-constrained households and firms, and have these materialized? We conclude that there has been limited realization of the concerns related to NIRP implementation. We were only able to isolate a few NIRP effects and these uncovered effects of NIRP implementation
contradictory to what we expected to find. For example, increased savings and a higher degree of labour utilization. However, we cannot disregard the possibility that another research design, for instance, a sector analysis or an analysis including abnormally low, but still positive, key interest rates, could have resulted in another conclusion. In addition, the empirical findings in this thesis only uncover some of the short-term effects of NIRP
implementation. As time goes by and more data become available, we cannot rule out that one or several of these conclusions might change and that the long-term effect of NIRP implementation may differ largely from the short-term effects.