PART III QUANTITATIVE ANALYSIS
3.2 Review of empirical findings from previous international studies of buyouts
39
The above reviewed Nordic studies on changes in operational portfolio company performance due to PE-ownership are described in details as these are evaluated as the most important, cited, and valid studies. Other academics researching the impact of PE ownership on Nordic portfolio company operational performance, but which studies will not be reviewed in detailed, are Jääskeläinen (2011), Viitala (2012), and Rem and Köhn (2016) who document a positive impact of PE ownership measured in terms of growth, and Christensen and Andersen (2009) and Lund-Nielsen (2010), who document a positive impact of PE ownership measured on profitability only one year after entry and during the financial crisis, respectively, however without statistical significance.
3.2 Review of empirical findings from previous international studies of
40
The scientific method applied in this study is an event study with the use of Wilcoxon test statistics. Three cash flow variables are tested, i.e. i) operating income (before depreciation), ii) capital expenditures, and iii) net (operating) cash flow. The event window used measures the percentage differences on changes in the cash flow variables in the first three years after exit (+1, +2, +3) compared to the last fiscal year before exit (-1). To control for industry, Kaplan (1989) has collected data for companies in related industries as the 48 buyouts through the database of Standard and Poor’s Compustat Industrial Research.
Testing the 48 companies on the three cash flow variables, Kaplan (1989) finds i) that operating income, measured net of industry changes is unchanged in the first two years after exit, but 24% higher in the third year, and ii) the median net cash flow (difference between operating income and capital expenditure), net of industry changes, is 22%, 43%, and 81%
higher the first three years after exit, respectively, compared with the last pre buyout year. It was further found that the explanation for these operating changes is to the largest extend supported by hypothesis 3 that is reduced agency costs (Kaplan S. , 1989).
Main differences between the scientific setup of the study of Kaplan (1989) and the study of this thesis is i) the dataset and period of testing, ii) performance measurements, and iii) the event window.
Lichtenberg, F., Siegel, D., 1990. The Effects of Leveraged Buyouts on Productivity and Related Aspects of Firm Behaviour. Journal of Financial Economics, Vol 27, No. 1, 165-194.
[Overall conclusion from study: Plants involved in LBOs had significantly higher rates of total factor productivity (TFP) growth than other plants in the same industry.]
In this paper, the effects of the specific type of ownership change i.e. the leveraged buyout (LBO) on total factor productivity and related aspects of firm behavior, was investigated (Lichtenberg & Siegel, 1990). The authors document a growth in LBO activity, which was the rationale for conducting this paper in order to analyze the relationship between LBOs and total factor productivity - output per unit of total input.
In this scientific paper, the author’s tests for the above-mentioned using large longitudinal establishment and firm-level U.S Census Bureau data sets linked to a list of 1100 American LBOs between 1981-1986 compiled from public data sources (Lichtenberg & Siegel, 1990).
In order to test for differences in productivity i.e. TFP between plants involved in LBOs in 1981-1986, the measure TFP is derived as a raw or studentized residual from a local first order logarithmic approximation Cobb-Douglas production function:
41
ln 𝑉𝑄𝑖𝑗𝑡 = 𝛽0𝑗𝑡+ 𝛽𝐿𝑗𝑡ln 𝐿𝑖𝑗𝑡+ 𝛽𝐾𝑗𝑡ln 𝐾𝑖𝑗𝑡 + 𝛽𝑀𝑗𝑡ln 𝑉𝑀𝑖𝑗𝑡+ 𝑢𝑖𝑗𝑡 (4)
Where VQ denotes the value of production; L denotes labor input; K denotes capital input);
VM denotes the value of materials consumed (materials purchased adjusted for changes in raw-materials inventories); u is a disturbance term; and the subscript ijt refers to establishment i in 4-digit SIC industry j in year t (Lichtenberg & Siegel, 1990).
Above production function is estimated separately by industry and year, hence, in order to assess productivity differences across industries, studentized residuals were examined:
𝑆̂𝑗𝑡2 = 1
𝑁𝑗𝑡− 4∑ 𝑒𝑖𝑗𝑡2
𝑖
(5)
With this residual, a given value of a raw residual 𝑒𝑖𝑗𝑡 represented a larger relative departure from mean productivity in some industries and years (those with "low" 𝑆̂𝑗𝑡2) than in others.
Hence, Lichtenberg and Siegel (1990) were able to scale the raw residuals by the corresponding estimated standard error. For instance, an observation with studentized residual 𝑒𝑖𝑗𝑡/𝑆̂𝑗𝑡 equal to 0.5 has productivity half of a standard deviation above average.
With above assumptions in place, the difference in the growth in TFP during 1981-86 between plants involved in LBOs during that period and other plants, conditional on the level of productivity in 1981, was estimated using a regression model:
𝑌𝑖𝑗86 = 𝛾0+ 𝛾1𝑋𝑖𝑗81−86+ 𝛾2𝑌𝑖𝑗81+ 𝜀𝑖𝑗86 (6) Where Y denotes either the raw residual or the studentized residual, and X denotes either an LBO dummy (= 1 if the establishment was involved in an LBO during 1981-86, = 0 otherwise) or an MBO dummy (defined similarly) (Lichtenberg & Siegel, 1990).
The overall finding from the regression suggests that plants involved in LBOs during 1981-1986 had significantly (about 14%) higher rates of productivity growth over that five-year span than other plants in the same industry. The authors points out two reasons for the relative productivity increase, that is i) increased intensity of effort by labor, and ii) reduction in the proportion of resources misallocated to inefficient activities (Lichtenberg & Siegel, 1990).
Key differences between the scientific methods applied in the study of Lichtenberg and Siegel (1990) and the study of this thesis is i) the dataset and period of testing, ii) the econometric methods applied, iii) the type of used peer group when testing relatively, and iv) the event window.
42
Davis, S.; Haltiwanger, K.; Handley, K.; Jarmin, R.; Lerner, J.; Miranda, J., 2014. Private Equity, Jobs, and Productivity, American Economic Review, Vol 104, No. 12, 3956-3990.
[Overall conclusion from study: Private equity buyouts lead to greater job loss at establishments operated by target firms as of the buyout year.]
How does PE-buyouts affect the creation/destruction of jobs? This hypothesis is tested in this scientific paper of Davis et al. (2014). They construct and analyze a huge dataset that covers US buyouts from 1980-2005, in which 3,200 target firms and their 150,000 establishments are tracked before and after acquisition, comparing to controls defined by industry, size, age, and prior growth (Davis, et al., 2014).
The statistical approach used to estimate the effects of buyouts on employment outcomes is attributed to i) an event study of nonparametric comparisons that control for the cross-product of categorical variables, ii) semi-parametric regressions that include additional controls, and where employment growth rate differences between targets and controls in the buyout year and following years are estimated, and iii) propensity score methods (Davis, et al., 2014).
The measurement of the employment growth rate used as input in above described statistical methods, the authors define these measurement as:
𝑔𝑖𝑡 =𝐸𝑖𝑡− 𝐸𝑖𝑡−1 𝑋𝑖𝑡
(7) Where 𝐸𝑖𝑡 is the employment (number of workers on the payroll in the pay period) at establishment or firm i in year t, and 𝑋𝑖𝑡 = 0.5 ∗ (𝐸𝑖𝑡− 𝐸𝑖𝑡−1).
The event window applied around transactions is such that employment outcomes are considered five years on either side of a private equity transaction (Davis, et al., 2014).
Running above mentioned tests, Davis et al. (2014) yields three main findings: First, employment shrinks more rapidly, on average, at target establishments than at controls after private equity buyouts. The average cumulative difference in favor of controls is about 3% of initial employment over two years and 6% over five years. Second, the larger post-buyout employment losses at target establishments entirely reflect higher rates of job destruction at shrinking and exiting establishments. In fact, targets exhibit greater post-buyout creation of new jobs at expanding establishments. Adding controls, cf. the regression analysis, for pre-buyout growth history shrinks the estimated employment responses to private equity pre-buyouts but does not change the overall pattern. Third, earnings per worker at continuing target
43
establishments fall by an average of 2.4% relative to controls over two years post buyout (Davis, et al., 2014).
Further, the study concludes that private equity buyouts involve much larger effects on the gross creation and destruction of jobs. The job reallocation rate at target firms exceeds that of controls by 14 percentage points over two years post buyout. About 45% of the extra job reallocation reflects a more rapid pace of organic employment adjustments, and the rest reflects acquisitions and divestitures. Hence, these novel findings provide evidence that private equity buyouts catalyze the creative destruction process as measured by gross job flows and the purchase and sale of business establishments (Davis, et al., 2014).
In addition, the study of Davis et al (2014) complements the study of Lichtenberg and Siegel (1990), as it was further found that target firms more aggressively close plants with low TFP, and they more aggressively open new plants with high TFP, relative to controls. In other words, target firms direct job reallocation activity on the plant entry and exit margins in ways that raise TFP (Davis, et al., 2014).
The dataset and period of testing, the statistical methods applied, the used peer group category, and the event window, all differentiates from the study of Davis et al. (2014) and the study of this thesis.
Other international studies worth mentioned, but which will not be reviewed in details, that document a positive impact of PE-ownership on operational value creation in terms of growth, operating profitability, and productivity, within the portfolio companies, are Muscarella and Vetsuypens (1990); Smith (1990); Wright et al. (1992); Wright et al. (1997); Harris et al.
(2005); Cao and Lerner (2006); Cressy et al. (2007); and Guo et al. (2007). In addition, Ravenscraft and Scherer (1987), and Desbrières and Schatt (2002) document a negative impact of PE-ownership on operational value creation within the portfolio companies.