average earnings in all three cases).108 All these indicators are available for 21 OECD
countries over the period 1985-2003.
The threshold beyond which a change in the policy indicator is assumed to signal a
major policy reform is set as follows. For each of the five policy areas mentioned above, the
standard deviation of the annual change in the corresponding quantitative indicator is
calculated over all available observations (typically 399, corresponding to 21 countries over
the period 1985-2003). A major reform is then assumed to have been undertaken when the
overall change in the policy indicator exceeds two standard deviations. In practice, this
corresponds respectively to: a decline in the benefit replacement rate indicator larger than
6.5 percentage points; a decline in the tax wedge measure larger than 5 percentage points; a
decline in the summary index of employment protection legislation larger than 0.5 units (the
average value of the index over the sample considered is 2.1); a decline in the de-trended
indicator of product market regulation larger than 0.6 units (the average value of the de-
trended index over the sample considered is 5);109 a decline in the average implicit tax rate on
continued work larger than 6.5 percentage points.
We obtain one reform indicator for each policy area, which takes value 1 when
observations - i.e. pairs (country, year) - correspond to a major policy reform and 0 otherwise.
Two alternative ways can then be envisaged to exploit these five reform variables:
- One option is to merge them so as to obtain an aggregate policy reform indicator which
takes value 1 for those pairs (country, year) that correspond to at least one major policy
reform and 0 otherwise.
- Alternatively, they can be stacked up in order to expand dramatically the amount of
information available for the econometric estimates. One problem with this option is that it
rests implicitly on the assumption that structural reforms undertaken in a given pair
(country, year) in different fields are independent from one another. However, this can be
accounted for in the econometric estimates by controlling for the impact of reforms
undertaken in other fields.
Therefore, both aggregate and stacked up datasets are used alternatively in the Probit
regressions presented in Section 4.3 of the main text.
8. The main sources for these indicators are respectively: OECD, Benefits and Wages, various issues ; OECD, Taxing Wages,
various issues; OECD (2004a), Employment Outlook, Paris; Nicoletti, G. and S. Scarpetta (2003), “Regulation, productivity and growth”,
Economic Policy, No. 36; Conway P. and G. Nicoletti (2006), “Product market regulation in non-manufacturing sectors in OECD
countries: measurements and highlights”, OECD Economics Department Working Paper, forthcoming, OECD, Paris; Duval, R. (2003),
“Retirement Behaviour in OECD Countries: Impact of Old-age Pension Schemes and Other Social Transfer Programs” OECD Economic
Studies No. 37, 2003/2, OECD, Paris.
9. The rationale for de-trending the product market regulation indicator is to control for the general decline observed across the
OECD area over the sample period. In the absence of such a control, our methodology would identify a disproportionately large number of
reforms compared with other policy fields. Instead, here only declines in product market regulation that have gone substantially beyond the
general trend towards deregulation are considered as major policy reforms.
206