Measuring the effects of mergers and acquisitions on the economic performance of real estate developers

Real estate developers in China are using mergers and acquisitions (M&As) to ensure their survival and competitiveness. However, no suitable method is yet available to assess whether such M&As provide enhanced value for those involved. Using a hybrid method of data envelopment analysis (DEA) and Malmquist total factor productivity (TFP) indices, this paper evaluates the short and medium term effects of M&As on acquirers’ economic performance with a set of 32 M&A cases occurring during 2000-2011 in China. The results of the analysis show that M&As generally have a positive effect on acquirers’ economic performance. Acquisitions on average experienced a steady growth in developer Malmquist TFP, a more progressive adoption of technology immediately after acquisition, a slight short-term decrease in technical efficiency after acquisition but followed by a marked increase in the longer term once the integration and synergy benefits were realised. However, there is no evidence to show whether developers achieved any short or long term scale efficiency improvements after M&A. The findings of this study provide useful insights on developer M&A performance from an efficiency and productivity perspective.


INTRODUCTION
Real estate is China's fastest-growing industry as its formation and development follows China's economic reform in the 1980's (Choi, 1998). Since 2004, the Chinese government has enacted a series of macro-economic regulatory policies to mitigate the risk of a real estate bubble. This has resulted in a wave of mergers and acquisitions (M&As) of developers in the industry, and both the number of deals and volume of developer M&As have increased sharply since 2006 (see Table 1), making the industry one of the most M&A active in China today. However, it is not yet clear whether developer M&As lead to better post-acquisition performance or not. Table <1> here

Please insert
The most commonly used methods to evaluate M&A performance in the business domain include event studies, cash flow analysis and market value frontiers (Franks and Harris, 1989;Healy et al., 1992;Sudarsanam et al., 1996;Mitchell et al., 2004).
However, as Antoniou & Zhao (2011) point out, these methods are unable to provide meaningful insights or usable information on the extent to which M&As create value.
Furthermore, in order to evaluate the M&A performance, a single factor (e.g. cash flow, value added) is not enough to compare the acquirers' performance before and after a M&A. Multiple inputs and outputs should be considered for a comprehensive evaluation.
Data Envelopment Analysis (DEA) is a powerful methodology for assessing the relative efficiencies of multi-input and multi-output production units. Established by Charnes et al (1978) based on the work of Farrell (1957), DEA has the advantage of not needing to select a particular functional form, make distributional assumptions or set the relative weights of variables. It has good statistical characteristics and is a very convenient method for detecting efficiency and productivity changes in individual organisations (Charnes et al, 1978;Cooper, et al, 2007aCooper, et al, , 2008b, making it very suitable for evaluating and comparing the performance of developer M&As.
However, DEA measures the efficiency and productivity of decision units for a specific period of time and does not allow any analysis of changes over time.
Meanwhile, the Malmquist total factor productivity (TFP) indices can help evaluate the total factor productivity change of a particular organisation over a fixed period, although it could be applied equally well in other areas (Caves et al., 1982;Färe et al 1994a,b;Cooper et al, 2007b;Kortelainen, 2008). Malmquist indices have several desirable features and properties: (1) there is no need to make behavioural assumptions, such as cost minimisation or profit maximisation, which makes them useful when the producer's objectives differ, or are unknown or are unachieved; (2) no need to provide price information, which makes the indices of practical use when either prices do not exist, are distorted or have little economic meaning; and (3) they can easily be calculated by the DEA methodology (Caves et al, 1982;Färe et al, 1998). All these issues make the DEA-based Malmquist TFP Index very suitable for evaluating M&A induced performance changes as a result of real estate acquisitions.
The purpose of this study is, therefore, to establish the extent to which developer M&As increase the acquirers' economic performance using the DEA-based Malmquist TFP Index. The next section develops and explains the DEA-Malmquist method for assessing developer M&A performance. This is followed by an illustration of the method and test on a set of 32 Chinese developer M&As between 2000-2011.
The results of the analysis are then presented prior to some concluding remarks.

REVIEW OF POST-ACQUISITION PERFORMANCE EVALUATION
Many studies have examined the stock returns of acquisitions to investigate the effect of M&A transactions on acquisitions. The event-study methodology, first proposed by Fama et al. (1969) is often used. This focuses on the long-term (e.g. one to five year) effect following an event (e.g., a takeover) and can provide key evidence concerning market efficiency (Brown and Warner, 1980;Fama, 1991). However, the event-study methodology has several shortcomings. Firstly, for a long-term event study it is more difficult to isolate the takeover effects from many other strategic and operational decisions or changes in the financial policy arising in the long term. Secondly, benchmark performance often suffers from measurement or statistical problems (Barber and Lyon, 1997). For example, according to Barber and Lyon (1997), cumulative abnormal returns (CAR), which is often used to investigate the effect of extraneous events on stock prices by calculating the sum of all the differences between the expected returns and the actual returns up to a given point in time, is a biased predictor for long-term event studies.
Two other main methods of assessing and calibrating post-event risk-adjusted performance have been adopted in the past to measure long-run abnormal stock returns: a characteristic-based matching approach and Jensen's alpha approach, which is also known as the calendar time portfolio approach (Fama, 1998;Eckbo, Masulis, and Norli, 2000;Mitchell and Stafford, 2000). However, despite extensive studies of these two types of long-term event study methods, there is still no clear preference (Kothari and Warner, 2005). Both have low power against economically interesting null hypotheses and neither is immune from misspecification (Jegadeesh and Karceski, 2004). Considering these power and specification problems, the challenge of refining long-term event methods remains (Kothari and Warner, 2005).

Selection of Performance Evaluation Indicators (Inputs and Outputs)
In order to select appropriate input and output indicators for performance evaluation, a close examination of the real estate industry in China is required. Real estate development is a capital intensive industry that demands huge financial commitments to cover the high price of land acquisition and substantial expenditure in the construction process. This implies that one of the first requirements of a developer is a strong financial capability. To reflect this, the equity ratio (input 1) was selected as it reflects the financing capability and capital structure of the organisation. Accordingly, the stockholders return ratio (output 1) is adopted as an output indicator for assessing the contribution of capital input.
Furthermore, the production process of the real estate industry involves a long period of time and massive capital investment. Although property project presales can produce some capital in advance, developers still need to access further finance through other channels for the large amount of capital needed to cover construction costs. Therefore, the quick ratio (output 2), which measures the ability of an organisation to use its liquid capital to immediately overcome its current liabilities, is also adopted as an output indicator.
As property is expensive, the customers' purchasing intentions are normally influenced by their expected income and attitude towards future economic prospects.
Developers, on the other hand, need to formulate appropriate operation strategies depending on the economic situation. To reflect the impact of this, inventory turnover (input 2) -representing the property selling condition and resources commitment -is used as an input indicator. Return on sales (output 3) is the corresponding output indicator.
In China's real estate industry, land is generally regarded as a core production material and long-term asset, and developers always experience fierce competition and need a substantial amount of capital commitment for its purchase. To raise the funds needed, developers generally use land as a mortgage tool for obtaining quick cash.
Additionally, developers in China use presales (such as a 20% to 30% down payment) to lower investment barriers for individuals and appeal to more consumers to buy properties. By using these methods, developers improve their cash flow and transfer the risks involved in money collection to financial institutions such as banks. To reflect this aspect of developer performance, the receivable turnover ratio (input 3) is adopted as an input indicator, with the cash flow ratio (output 4) as an output indicator.
Finally, due to the ferocity of competition, developers need to compete in price, quality, service, product delivery, etc. Specifically, requirements such as vast investment resources and long construction periods make the profitability of developers particularly vulnerable to fluctuations in the economic environment and the market. To reflect the effect of market competition on developer performance, the return of assets (output 5) is used as an output indicator.
In short, three input indicators (Stockholder Equity Ratio, Inventory Turnover and Receivable Turnover Ratio) and five output indicators (Return on Equity, Return on Sales, Quick Ratio, Cash Flow Ratio and Return on Assets) are used in the analysis. It should be noted that all the output indictors are related to economic performance, and it is not the intention to investigate social or environmental performance at this stage.

DEA Efficiency Estimation
DEA is a modern frontier analysis method for efficiency estimation, comprising technical efficiency, pure technical, allocative, scale, cost and revenue efficiency (Cooper et al. 2007b). The efficient value range is from 0 to 1, where 1 is regarded as the most efficient. In this paper, technical efficiency, pure technical efficiency and scale efficiency are used to measure acquirer efficiency. Technical efficiency is measured by using an input-oriented model (Shepherd, 1970). Assume Decision Making Unit (DMU) i uses M inputs x to generate N outputs y in period t. The production technology of period t can be modelled by an input function. For any y ϵℝ , V y denotes the subset of all input vectors x ϵℝ which yield at least y , using a production technology characterised by returns to scale of type r, where r = c = constant returns to scale (CRS), r = v = variable returns to scale (VRS), and r = n = non-increasing returns to scale (NIRS). The input-oriented distance function is D x , y = sup θ : , y ϵ V y = inf θ : θ x , y where x , y is the input and output vector for DMU i during period s.
Technical efficiency TE x , y is thus defined as TE x , y = 1 D x , y ⁄ . CRS technical efficiency is measured for each DMU by solving a linear programming problem D x , y = TE x , y = min θ , Subject to: Y λ ≥ y , where X is a × input matrix and Y an × output matrix for all DMUs, λ is an × 1 intensity vector, and = the number of DMUs in the sample (i = 1,2,. . . , ). This estimation (with the λ constrained to be non-negative) generates a CRS frontier.
Technical efficiency can be divided into pure technical efficiency TE x , y (technical efficiency relative to a VRS frontier) and scale efficiency SE x , y , as TE x , y = TE x , y SE x , y . These are separated by solving equation (2) with the additional constraint: ∑ λ = 1 for a VRS frontier, and with the constraint

DEA-based Malmquist analysis of productivity
The Malmquist index approach is adopted to measure the total factor productivity (TFP) change of DMUs over time. The description below draws primarily upon the work of Fare et al (1994aFare et al ( , 1998 and recaps some of the discussion from Coelli et al (2005). The Malmquist TFP change index (output-orientated) between period s (the start period) and period t is given by (Caves et al, 1982) The distance function D x , y = inf ∅: x , y ∅ ⁄ ∈ S is defined as the reciprocal of the "maximum" proportional expansion of the output vector y s in given inputs x s . Similarly, the distance function D x , y = inf ∅: x , y /∅ ∈ S represents the distance from period t to the period s technology. A value of M 0 larger than one means that the TFP grows from period s to period t, otherwise a decline in TFP is indicated.
In equation (4), the ratio outside the square brackets is actually the efficiency change (EC), which evaluates the change in the output-oriented measure of Farrell technical efficiency between periods s and t: The remaining part of equation (4)   The SEC is actually the geometric mean of two SEC measures relative to period t and s technology respectively.
The Malmquist TFP index (distance measures) in equation (3) can be calculated by using a DEA-like linear programming methodology (Fare et al 1994a). For the organisation i-th, four distance functions need to be calculated to measure the TFP change between two periods. These four distances can be obtained by the four linear programming problems (equations 9-12): and D x , y = max ∅, ∅ , where θ is a scalar and λ is a I×l vector of constants. The value of θ is the efficiency score for the i-th organisation.  Coelli (1996).

Estimation Windows
To compare the value of the Malmquist TFP index before and after acquisition, three estimation windows are established that include four time points such as one year prior to acquisition (t-1), acquisition announcement (t+0), one year after acquisition (t+1) and three years after acquisition (t+3). The two windows: window (1) from (t-1) to (t+1); and window (2) from (t-1) to (t+3) represent the short-term and relatively long-term windows of the M&As respectively.

Summary statistics
The financial operational indicators of acquirers and target developers are summarised in Table 3, which provides a general background and context of the sample cases' performances. This shows that acquirers are 2.77, 2.35, 3.51 and 2.25 times larger than the target developers on total assets, debt, cash and market value respectively. In contrast, the acquirers have relatively smaller financial leverage, Tobin's Q, Cash & Growth, return on equity, and return on assets. This first indicates that the management efficiency and profitability of the target developers is much higher than that of the acquirers in the sample. Second, it clearly shows that that the M&As generally occur between acquirers with a large business scale and target developers with high managerial efficiency and profitability. Table <2> here

Please insert
The information relating to the input and output indicators is summarised in Table 4.
For the input indicators, the equity ratio increased slightly from 0.36 in t-1 to 0.38 in t+3, indicating an increased capital commitment from acquirers after acquisition. The acquirers' average inventory turnover decreased significantly from 0.94 in period t-1 to 0.44 in period t+3. The receivables turnover ratio decreased dramatically from 227 to 117 from t-1 to t+1, but then increased slightly to 241 in t+3. In terms of output indicators, both ROE and ROS increased significantly from the pre-acquisition to post-acquisition phase. The acquirer's average quick ratio experienced little fluctuation from t-1 to t+3. Conversely, the cash flow ratio was highly volatile during the same period. Finally, there was a substantial rise in the acquirers' average return on assets from t-1 to t+3, increasing to a mean 0.06 in the short term and 0.04 in the longer term. All in all, the selected input and output indicators mainly show a trend of better financial performance of the sample cases after the M&As with five out of eight indicators recording an increased value. Of the other three indicators, one has only a slight decrease from 0.69 at t-1 to 0.68 at t+1, with the remaining two in an obvious relative decline. Table <3> here

Please insert
The efficiency of acquirers measured by the DEA method is illustrated in Table 5.
Their average technical efficiency decreased significantly from 0.93 in t-1 to 0.74 in t+3, with the lowest being 0.72 in t+1. This means that technical efficiency declined sharply in the short-term after acquisition, but recovered slightly since the realisation of the synergy and integration benefits of M&A in longer term. Similarly, pure technical efficiency sharply declined from 0.94 to 0.81 during t-1 to t+1, but rose to 0.86 in t+3. In contrast, scale efficiency experienced a continual decline in both the short and long-term, dropping gradually from 0.98 to 0.85 -implying that no economies of scale were achieved. Table <4> here

Productivity measuring result analysis
The Malmquist TFP index of each acquirer is provided in (TEC=PEC*SEC=1.12*1.17) indicates that its pure technical efficiency and scale efficiency had both also made some progress.

Please insert Table <5> here
It is noticeable that the mean Malmquist TFP index is highest (1.26) in the M&A year (t+0), and then decreased significantly to 0.81 in the following year (t+1). In the longer term after acquisition (t+3), the acquirers' mean TFP improved slightly to 1.02.
The reasons for these changes can be identified from the TEC and TC perspectives: The acquirers' technology had the largest upgrade in t+0, with a slight increase at t+1 and followed by a slight decrease by t+3. In contrast, the mean technical efficiency of the acquirers markedly declined in both t+0 (0.96) and t+1 (0.79), but rose in t+3 (1.03). This clearly suggests, therefore, that the M&As improved acquirers' technology in the short-term was due to a greater commitment of resources, but the growth began to diminish in the longer term as the effect of the M&As on resource investment weakened. Meanwhile, technical efficiency decreased in t+0 and t+1 because of organisational transition in the short-term while, in the longer term, technical efficiency increased as the synergy and integration benefits were realised.
Similarly, PEC can be interpreted in the same way as TEC. Surprisingly, neither short nor long term increases in scale efficiencies were realised after the M&As, which is consistent with previous scale efficiency analysis results. The reason for this is probably due to the real estate industry's unique characteristics of localisation and unmovable products .   Note: Tobin's Q is an indicator generally used to measure the management efficiency of organisations (Lang et al, 1989, Servaes, 1991, Chung and Pruitt, 1994. Tobin's Q = (the organisation's market value + liquidation value of preferred stock DEBT)/ Total assets; If T's Q less than one, an inefficient management is indicated. Cash & Growth is a useful indicator to identify any agent problems (Jensen, 1986 andLang et al, 1991). High levels of cash flow, but low growth opportunities imply the presence of an agency problem.