Archives

  • 2018-07
  • 2018-10
  • 2018-11
  • 2019-04
  • 2019-05
  • 2019-06
  • 2019-07
  • 2019-08
  • 2019-09
  • 2019-10
  • 2019-11
  • 2019-12
  • 2020-01
  • 2020-02
  • 2020-03
  • 2020-04
  • 2020-05
  • 2020-06
  • 2020-07
  • 2020-08
  • 2020-09
  • 2020-10
  • 2020-11
  • 2020-12
  • 2021-01
  • 2021-02
  • 2021-03
  • 2021-04
  • 2021-05
  • 2021-06
  • 2021-07
  • 2021-08
  • 2021-09
  • 2021-10
  • 2021-11
  • 2021-12
  • 2022-01
  • 2022-02
  • 2022-03
  • 2022-04
  • 2022-05
  • 2022-06
  • 2022-07
  • 2022-08
  • 2022-09
  • 2022-10
  • 2022-11
  • 2022-12
  • 2023-01
  • 2023-02
  • 2023-03
  • 2023-04
  • 2023-05
  • 2023-06
  • 2023-07
  • 2023-08
  • 2023-09
  • 2023-10
  • 2023-11
  • 2023-12
  • 2024-01
  • 2024-02
  • 2024-03
  • 2024-04
  • 2024-05

    • The next step is to obtain the minimum value of

    2018-10-30

    The next step is to obtain the minimum value of the pseudo t-ratio for testing whether the autoregressive parameter (ρ) is 1 for all the break time combinations. The derivation of the asymptotic distribution of this statistic, Clemente et al. (1998) assume that 0<λ0<λ1 and λ2<1−λ0<1, implying that the test is not defined at the sample limits and it is necessary to choose some trimming value (λ0), and also λ1 and λ2 takes the values in the interval [(k+2)T, (T−1)/T]. The authors impose one additional restriction, λ2>λ1+1, in order to eliminate cases of consecutive time breaks. The unit root null sirtuin 1 to test for two structural breaks, when the shifts are better represented as addictive outliers (AO), is developed by a two-step procedure. The first one is to remove the deterministic part of the variable and estimate the following model: The second step is to test the hypothesis of ρ=1 and search for the minimal t-ratio in the following model:where are the residuals obtained from equation (15), DTB are dummy variables included in equation (16) to assure that converges to the distribution described in Eq. (4) of Clemente et al. (1998). The idea is to run a regression of the residuals on their lagged values, a number of lagged differences, and a set of dummy variables needed to make the distribution of the test statistic tractable. Table 3 summarizes the results and shows that there is rejection of the null, i.e., all series are stationary. The two breaks estimated are located at the end of 2002 (first break) and beginning of 2004 (second break). As for the Brazilian GDP, the second break is in 2010, which coincides with the international financial crisis, which began in 2008. In the case of inflation, the first break coincides with the inflationary impact of the confidence crisis, between 2002 and 2003.
    Determinants of yield spreads: model and results The measurement equation can be defined as follows:where Spread represents the yield spreads related to 1, 3, 6, 12, 48 and 60 months. It means that for each yield spread, there will be an estimation with 2 lags taking also into consideration the output gap, inflation (IPCA), and the interest rate on fixed and floating rate bonds. The state equation can be defined as:where ‘i’ represents the time-varying parameter to be estimated, which follow a random walk process. Table 4 reports the estimation results. First of all, we notice that both short-term and long-term yield spreads have a statistically significant relationship (in the second lag) with the Brazilian economic activity (measured by the output gap) and they increase as maturity increases. The same applies to the IPCA inflation, which is statistically significant with one lag. It means that changes in yield spreads can be explained by expected changes in economic activity, and the higher the gap between actual and potential products, the greater the demand for protection. The same reasoning applies to inflation: the more agents are inclined to believe that prices will go up, the higher the yield spread required. Nevertheless, our main aim is to examine how agents respond to changes in the profile of government securities along the yield curve, i.e., the behavior of agents to changes in public debt. When we look at the fixed rate bonds, we notice that there is no statistical significance at lag 0 (FixedRate) and lag 2 (FixedRate). On the other hand, at the first lag the results show that an increase in issuance of fixed rate bonds may also increase the yield spread, and the longer the maturity the greater the effect. The coefficient ranges from 0.00585 for the one-month spread up to 0.20596 for the 60-month spread. As for floating rate securities, the effects are more significant for the yield spread at the first lag (FloatRate), in which the coefficient ranges from −0.01031 to −0.21483. The negative sign is an indication that the effect is opposite to what happened with the fixed rate bonds, i.e., an increase in the issuance of floating rate bonds leads to a reduction of yield spread.