Table 2.
Probit estimates of the determinants of structural reforms over 1985-20031
|
(1) |
(2) |
(3) |
(4) | |
|
Econometric method: |
Probit |
Random-effects probit |
Probit |
Random-effects probit |
|
Dataset: |
Aggregate |
Aggregate |
Stacked up |
Stacked up |
|
Dependent variable: |
Binary reform index |
Binary reform index |
Binary reform index |
Binary reform index |
|
Unemployment (-3) |
0.06 (m.p: 0.02) |
0.10 [1.97]** |
0.03 (m.p: 0.01) |
0.08 [3.69]*** |
|
Crisis (-1) |
0.95 (m.p: 0.36) |
1.19 |
0.38 (m.p: 0.09) |
0.71 [3.29]*** |
|
Small country |
0.62 (m.p: 0.23) |
0.64 [1.97]** |
0.38 (m.p: 0.07) |
0.26 [1.31] |
|
Cyclically-adjusted fiscal surplus (-1) |
0.12 (m.p: 0.05) |
0.15 [3.69]*** |
0.07 (m.p: 0.01) |
0.11 [4.97]*** |
|
D(Cyclically-adjusted fiscal surplus (-1)) |
-0.11 (m.p: -0.04) |
-0.13 [2.01]** |
-0.04 (m.p: -0.01) [1.30] |
-0.08 [1.95]* |
|
Other reforms (-1) |
0.2 (m.p: 0.04) |
0.27 | ||
|
Observations__________________________ |
_________333________ |
_________333_________ |
________1655________ |
_________1655________ |
1. m.p: marginal probability.
Absolute value of robust z-statistics in parentheses, * significant at 10% level, ** significant at 5%, *** significant at 1%.
Source: Author's calculations.
Standard probit estimates such as those presented in column 1 of Table 2 purposefully
ignore the panel structure of the data. As a robustness check, in column 2 the equation is re-
estimated using a panel data, random-effects probit framework in which countries are considered
as different individuals. The results are virtually unaffected, except the unemployment variable
and the small country dummy which are now significant only at the 5% (instead of 1%)
confidence level.
Estimates on stacked up data (columns 3 and 4) use the same specifications97 as those on
aggregate data, except that they also control for the (lagged) effect of reforms made in other fields
on the propensity to implement a reform in a given field.98 As already mentioned, this additional
control variable aims to capture positive spillovers between structural reforms in different areas.
The results are fairly comparable to those obtained on aggregate data, except that the fiscal
adjustment variable is not statistically significant at conventional levels in the standard probit
regression (column 3). In line with IMF (2004), reforms made in other fields are found to be
97.In principle, the use of stacked-up data also makes it possible to undertake the analysis of the determinants of structural reforms at the level of
each policy area. However, the scarcity of reforms in each field considered individually suggests - and estimates confirmed - that this is likely to
bring few robust findings.
98.For each of the five policy fields considered, this variable takes values comprised between 0 and 4 depending on the number of reforms carried
out in the other four fields.
195
More intriguing information
1. The name is absent2. Endogenous Determination of FDI Growth and Economic Growth:The OECD Case
3. IMMIGRATION AND AGRICULTURAL LABOR POLICIES
4. Announcement effects of convertible bond loans versus warrant-bond loans: An empirical analysis for the Dutch market
5. The Clustering of Financial Services in London*
6. Text of a letter
7. Structural Influences on Participation Rates: A Canada-U.S. Comparison
8. Evolutionary Clustering in Indonesian Ethnic Textile Motifs
9. REVITALIZING FAMILY FARM AGRICULTURE
10. Population ageing, taxation, pensions and health costs, CHERE Working Paper 2007/10