X hits on this document

PDF document

PRIVATISING NATIONAL OIL COMPANIES: ASSESSING THE IMPACT ON FIRM PERFORMANCE - page 15 / 30

67 views

0 shares

0 downloads

0 comments

15 / 30

Visual inspection of Figures 1(A) to 1(E) indicates that it is difficult to pinpoint a

single discrete change for most of our preferred performance measures. Most

measures start improving well before the privatisation takes place in the capital

markets. Figure 1(F), the development of the real terms oil price, confirms that

governments do not sell at the peak of the oil price cycle.

In order to confirm the visual inspection we estimate the following fixed-effects

panel data model (see Villalonga 2000):

Perfit = αi + β1 Postit + β2 Yearit + β3 [Post*Year]it + γ Oilit + µ CtrGrpt + εit , where:

  • ‘Perf’ is the relevant performance metric,

  • ‘Post’ is a dummy variable for the years post privatisation (i.e. years +1 to +3),

  • ‘Year’ is a discrete variable, ranging from 1 (for observations in year –3) to 7

(for observations in year +3),

  • ‘Post*Year’ is a slope dummy variable,

  • ‘Oil’ is a control variable for the oil price in real terms, and

  • ‘CtrGrp’ is the median performance of the control group (where available).

  • Unit fixed effects αi are significant and thus included in the specification.

In this model the coefficient of ‘Post’ captures differences in the (average)

performance levels before and after privatization, the coefficient of ‘Year’ indicates

the year-on-year performance trend, and ‘Post*Year’ evidences any changes in such

performance trends that take place after the privatisation transaction. A positive

coefficient in ‘Post*Year’ thus indicates that the performance trend further increases

after privatization (or decreases less, if the ‘Year’ coefficient is negative). The model

also includes two control variables, one for oil prices – the most important and most

volatile driver of corporate performance – and another for the performance of an

industry control group of firms which did not experience any changes in

15

Document info
Document views67
Page views67
Page last viewedFri Dec 02 23:12:20 UTC 2016
Pages30
Paragraphs1873
Words11467

Comments