*4.4. Test and Diagnostics*

This section provides a discussion of the results of the tests and diagnostics to assess the quality of the results presented in the previous section. The results of the poolability test are shown in Table 5, while Table 6 displays the diagnostics, and more specifically, the distribution of standard errors between idiosyncratic and time-invariant terms.


#### **Table 5.** Misspecification tests.

Source: Authors' calculation. \*\*\* denotes significance at the 1% level.



Source: Authors' calculation.

Table 5 presents the results of the poolability test, related to model misspecification, and allows us to decide between the pooled OLS estimates and the FE/RE estimates. The test results suggest that results from RE are more reliable in the case of Asia and the Pacific and North America. In the case of Europe, however, the test seems to favor pooled OLS, even though the model showed a lower R-squared overall. Overall, the results of the misspecification tests confirm the validity of our results.

Finally, we introduce empirical estimates of *i*,*<sup>t</sup>* and *ui*,*t*, time-varying and idiosyncratic error terms, in Table 6. As one would expect, the size of the idiosyncratic error term is rather large in all models. It is worth noting that, for European bonds, the majority of the unobserved terms are captured by time-varying factors, meaning that European bonds are quite homogenous in terms of risks. This was already observed by the meanvariance analysis of European bonds. As for the region of interest in this study, it appears that variance due to heterogeneity across bonds accounts for 36% of unobserved factors determining performance, thereby confirming the high risks associated with Asian green bonds. Indeed, if the performance of Asian bonds has such high variation, they are naturally considered less reliable by investors in general.

To prove this last point, we also provide ratings of Asian green bonds by major agencies, namely S&P and Moody's. Due to the lack of data availability, we only provide ratings for 48 bonds, all issued in Asia. The distribution of ratings is shown in Figure 5. This figure shows the heterogeneity of Asian green bonds, as no rating category dominates the sample. That being said, one can also see that the majority of bond ratings are mid-tier (A+, A, and A– for S&P, and A1, A3, Baa1, and Baa2 for Moody's). Furthermore, 5% to 8% of Asian green bonds are below BBB and Baa3, and hence considered risky investments, which is a relatively high percentage. Of course, agencies are not infallible, but Asian green bonds could be seen as relatively risky investments, based on their ratings.

**Figure 5.** Probability of default of Asian green bonds—ratings distribution. Source: Authors' compilation. (Aa3, Aa2, A1, A3, Baa1, Baa2, Baa3, <Baa3) And (AA, AA-, A+, A, A-, BBB+, BBB-, <BBB-) are the credit ranges by the two credit rating agencies.
