important aspects. First, whilst their papers are cross-sectional in design, we conduct
the first time-series analysis of privatised NOCs. Second, they have to restrict
themselves to the analysis of high-level data from a third party provider, we can
consider firm performance and efficiency in more granular detail based on primary
company sources. Third, our study is the only one to make use of panel data analysis
to control for time-invariant fixed firm effects.
There are typically two options to privatise a state-owned company: either a
private trade sale to an industrial or financial buyer, or a public share offering. This
analysis focuses on the latter because for trade sales there is rarely any comparable
pre-vs.-post disclosure available – SIPs are the only transactions for which changes
can practically be observed over time. But because the most important and politically
sensitive privatisations usually occur in the SIP format anyway, it is possible to argue
that a sample of SIPs represent a meaningful picture of oil privatisations in general.
Overall, in the period from 1977 up to and including 2004 a total of 41 privatised
companies have been identified based on previous studies, third party databases and a
detailed press search by country.7 Of these 41 companies, three companies were
acquired shortly after privatisation, in one instance the government only sold a very
minor stake relative to third party investors, in two cases the SIP constituted a
negligible stake listed on the domestic stock exchange (largely employee shares), and
in seven cases data could not be found or was not made available. For the remaining
28 NOCs from 20 countries extensive accounting and share price data was collected
7 All voucher privatisations and all Russian privatisations in oil and gas (whether voucher or not) have been excluded from the analysis, largely for concerns over the transparency of the privatisation process.