IV. ANÁLISIS DE RESULTADOS
4.5. Resultados de pruebas de las baterías
While the focus of this chapter is the dynamics of the EFP, it is interesting to discuss to which extent the model’s predictions for other variables are consistent with theoretical predictions. The following figures show the impulse responses of the endogenous variables to the identified structural shocks, specified as in table 2.1. In particular, I will comment on the estimated impulse response functions of remaining unrestricted variables in relation to their theoretical counterparts, summarized in tables in appendix A. As already explained, most SVARs with sign restrictions generate wide credible sets and, therefore, the following impulse response functions will be interpreted with the help of median estimates.
Conditional on an adverse supply shock, depicted in figure 2.2, the EFP (measured by the GZ credit spread) increases. Real GDP and prices (together with the interest rate) move into opposite directions. Stock prices and credit fall and recover only slowly, which is in line with the implications from most of financial DSGE models considered in table A.1 and A.2 in appendix A.
Now I will turn to the analysis of an unexpected increase in the interest rate. The adverse one standard-deviation monetary policy shock causes a decrease in real GDP and the GDP deflator by identifying assumption. Restrictive monetary policy is effective only in the first two periods. The interest rate rises by 10 basis points initially and falls by the same amount in the following periods. This result is comparable with the median impulse response of the nominal interest rate in the analysis by Uhlig (2005), whose sign restriction on the nominal rate has a comparable duration. Contrary to Uhlig (2005), my model reports an unambiguous decline in real GDP as a result of the imposed sign restriction, which is derived from NK models.
It is worth noting that the structural model generates an increase in stock prices and the credit volume, which stands in contrast to the predictions from theoretical models outlined in table A.1 and A.2 in appendix A. Gal´ı and Gambetti (2015) report a similar result on stock prices in the context of a time-varying VAR and, therefore, argue that their result provides empirical evidence against the
Figure 2.2: Adverse aggregate supply shock 1 5 10 15 20 -0.8 -0.6 -0.4 -0.2 0 GDP 1 5 10 15 20 0 0.2 0.4 0.6 GDP deflator 1 5 10 15 20 -0.2 0 0.2 0.4 0.6 0.8 Interest rate 1 5 10 15 20 -1.5 -1 -0.5 0 0.5 Credit 1 5 10 15 20 -0.1 0 0.1 0.2 0.3 EFP 1 5 10 15 20 -6 -4 -2 0 2 Stock prices
Notes: The bold lines denote the median of the impulse responses, which are estimated from a Bayesian VAR with 1000 draws. The bounds are the 16th and 84th percentiles. The impulse responses are related to an adverse one standard deviation aggregate supply shock. GDP, GDP deflator, credit and stock prices are expressed in percentage deviations, whereas the EFP and nominal interest rate are reported in percentage points. The horizontal axis is in quarters. The time period is 1973Q1-2010Q3.
conventional wisdom that contractionary monetary policy shocks have a negative effect on asset prices. Additionally, the positive development of the credit volume appears to be a result of a slow readjustment of the market. It takes time for the new credit terms to be effective in financial contracts.
The GZ credit spread increases much more than the initial rise in the interest rate. This can reflect the degree of financial frictions in the corporate sector that intensifies the propagation of the shock. The countercyclical movement of the EFP, as already argued, is the feature of numerous financial DSGE models.
The financial shock is depicted in figure 2.4. My results indicate that an unexpected rise in the EFP (by 20 basis points) leads to a prolonged contraction of output and a significant easing of the monetary policy (by 50 basis points). These results are in line with the results established in the literature: Financial shocks have a persistent impact on the real economy (see Gilchrist and Zakrajˇsek, 2012; Meeks, 2012). Median estimates on prices speak clearly in favor of the deflationary nature of financial shocks (see, e.g., Gilchrist and Zakrajˇsek, 2012; Furlanetto et al., 2014). The magnitude of the increase in the EFP should not be surprising given that the measure of the premium is based on the broad coverage of firms with different credit standings. For a comparison, Meeks (2012) reports that his estimated one standard- deviation credit spread shock generates initially a 50-basis-point increase in the high yield spread, which is a difference between the speculative-grade corporate
Figure 2.3: Adverse monetary policy shock 1 5 10 15 20 -0.6 -0.4 -0.2 0 0.2 GDP 1 5 10 15 20 -0.6 -0.4 -0.2 0 GDP deflator 1 5 10 15 20 -0.4 -0.2 0 0.2 Interest rate 1 5 10 15 20 -0.5 0 0.5 1 Credit 1 5 10 15 20 -0.1 0 0.1 0.2 0.3 EFP 1 5 10 15 20 -2 0 2 4 Stock prices
Notes: The bold lines denote the median of the impulse responses, which are estimated from a Bayesian VAR with 1000 draws. The bounds are the 16th and 84th percentiles. The impulse responses are related to an adverse one standard deviation monetary policy shock using identification scheme in table 2.1. GDP, GDP deflator, credit and stock prices are expressed in percentage deviations, whereas the EFP and nominal in- terest rate are reported in percentage points. The horizontal axis is in quarters. The time period is 1973Q1-2010Q3.
Figure 2.4: Adverse financial shock
1 5 10 15 20 -1 -0.5 0 0.5 GDP 1 5 10 15 20 -0.6 -0.4 -0.2 0 0.2 GDP deflator 1 5 10 15 20 -1 -0.5 0 0.5 Interest rate 1 5 10 15 20 -2 -1.5 -1 -0.5 0 Credit 1 5 10 15 20 -0.1 0 0.1 0.2 0.3 EFP 1 5 10 15 20 -6 -4 -2 0 2 Stock prices
Notes: The bold lines denote the median of the impulse responses, which are estimated from a Bayesian VAR with 1000 draws. The bounds are the 16th and 84th percentiles. The impulse responses are related to an adverse one standard deviation financial shock. GDP, GDP deflator, credit and stock prices are expressed in percentage deviations, whereas the EFP and nominal interest rate are reported in percentage points. The horizontal axis is in quarters. The time period is 1973Q1-2010Q3.
bond yield and the closely matched government bond yield. Unlike my measure for the premium, the high yield spread measure is related to firms subject to the high risk of default.