ORMAT
Parametric Decomposition of Total Factor Productivity Growth in the European Insurance Industry: Evidence from Life and Non-Life Companies
Dev Vencappa
Paul Fenn*
Stephen Diacon
*Centre for Risk and Insurance Studies
Nottingham University Business School
Jubilee Campus
Nottingham, NG8 1BB
United Kingdom
T +44 (0) 115 951 5254; F +44 (0) 115 846 6667;
Abstract
This paper estimates and decomposes productivity growth for a sample of European life and non-life insurance companies to gauge the relative importance of three major sources through which a firm’s productivity can change over time. We focus on a period where substantial deregulation took place in the European insurance industry, as well as significant shocks to the global capital markets. Using Stochastic Frontier Analysis, we estimate a production frontier from which we then proceed to construct three main components that make up productivity growth through a derivative-based measure along the lines of Bauer (1990a) and Kumbhakar & Lovell (2000) using the Baltagi & Griffin (1988) general index of technical change. We observe temporal variations in the rate of overall productivity growth for both life and non-life insurance, which are driven by patterns of technological progress and regress, together with consistent positive contributions from scale efficiency. In most years we find evidence of modest growth in technical efficiency, with some indication that this differs across EU member states.
JEL Classification: G21
Keywords: Insurance; Total factor productivity growth; Stochastic frontier analysis
Parametric Decomposition of Total Factor Productivity Growth in the European Insurance Industry: Evidence from Life and Non-Life Companies
1. Introduction
Productivity measurement is crucial to all sectors of the economy. The academic literature is burgeoning with an increasing number of studies that are developing new and more refined measures of performance as well as providing new evidence of changing performance in areas where hitherto, the non-availability and/or complexity of data made measurement difficult. Over the years, the focus of attention has moved towards the services sector, given that in most developed countries, the contribution of this sector to GDP has increased relative to traditional goods industries such as agriculture and manufacturing[1]. Despite this ever rising importance of services in developed nations’ economies, measured productivity growth in services industries has generally been very slow compared to manufacturing,. Such findings have attracted interest amongst academics, many of whom have been concerned with testing the validity of Baumol’s hypothesis of a “cost disease” in services.[2]
There have been several studies on productivity growth in the services sector, mostly at the industry and sectoral level. At the micro level, financial services industries such as banking and insurance have been studied quite extensively. For insurance, however, most of the latter studies have focused on measuring cost performance, with some attempting to estimate scale economies and some measure of technical progress (see Cummins and Weiss, 2000, for a good review). Very few studies have attempted to estimate productivity growth for the insurance industry. This paper therefore attempts to fill in the gap by providing micro-level evidence on productivity growth for European life and non-life insurance industries. We estimate productivity growth for the European insurance industry and decompose it into three component sources.
We carry out this exercise over a period where substantial deregulation took place in European financial services industries. Over the past 15 years or so, European insurers have been exposed to a changing economic landscape due to the deregulation brought about by the EU Directives on Insurance and Banking. The liberalisation of this sector resulted in an increasing number of M&A activities, especially towards the end of the 1990s, increased cross-border trade in insurance services and a change in the market structure of European insurance industry. It is believed that such consolidation and increased competition have forced firms to focus on different ways of increasing their performance. Some firms might have seen it advantageous to become more efficient in their scale of operations, while some others have seen technological progress or improvements in managerial efficiency as a way of surviving in the new economic landscape. All these factors combined would have clearly had an impact on the productivity growth of the insurance industry. Moreover, the end of the 1990’s and the beginning of the next decade witnessed two significant shocks to the global capital market (the stock market crash of 1999 and the terrorist attack on New York in 2001) and these shocks arguably imposed severe constraints on the availability of capital in a way which may well have affected productivity in the financial sectors including insurance. Clearly the combined influences of technical efficiency, scale and technical progress are important issues faced by the industry in Europe over the last decade or more. A decomposition of productivity growth will help to pin down the factors driving productivity growth over time.
The objective of this paper is therefore to measure and decompose productivity growth in European Insurance over a period where regulatory change and financial turbulence has been taking place. We provide direct evidence of productivity growth at the micro level for the insurance industry and are thus able to make international comparisons across European economies. We use Stochastic Frontier Analysis (henceforth SFA) to estimate a Fourier Flexible production frontier. From this frontier, we use a primal approach to construct estimates of productivity growth from its component sources, hence following the bottom-up approach advocated by Balk (2001). This paper is structured as follows.
Section 2 reviews the changing landscape of the European insurance industry during the period of our analysis. Section 3 reviews the techniques used to measure and decompose productivity growth. Section 4 reviews the literature on productivity measurement in the services sector. Section 5 discusses data and the definition of output and input variables, and this is followed by a section explaining the estimation methodology pursued here. Section 7 reports the results and a concluding section follows.
2. The Changing Economic Landscape of the European Insurance Industry.
Over the past 15 years the European Union implemented a series of banking and insurance directives with a view to deregulate the financial services sector and to establish a single European market in financial services. It is believed that for the first time, true price and product competition in both life and non-life insurance were introduced in European retail insurance markets, particularly through the implementation of the “single passport”, whereby, from 1994, an insurer could do business in all EU countries provided that it is licensed in one EU country. Such deregulations resulted in an increased amount of cross-border trade in insurance services as well as an unprecedented wave of mergers and acquisitions (M&As) that had been taking place in the European financial sector. Cummins and Weiss (2004) report that from 1990-2002 there were 2,595 M&As involving European insurers of which 1,669 resulted in a change in control. It is also believed that with liberalisation, existing firms face cross-border competition through the entry of foreign firms determined to cut into their market share.
Table 1 shows developments in the structure of the European insurance sector. The table is based on data from the European Insurance and Reinsurance Federation (CEA), which provides yearly reports on the insurance sector of most of the European countries. Included in these reports is a 5-firm, 10-firm and 15-firm concentration ratio for each of the countries covered and for both the life and the non-life insurance market. The overall picture that emerges from table 2 is a substantial increase in 5-firm concentration ratios as a result of the merger and acquisition activity in most European markets between 1992 and 2004. In general, concentration appears to be inversely related to the size of the domestic market. Concentration in the non-life insurance sectors increased between 1992 and 2004 for almost every country (except notably Switzerland). In the life insurance sector, the leading markets also saw an increase in concentration over the twelve years to 2004.
Table 1: Five-firm concentration ratios in life and non-life business[3]
Non-Life Insurance Companies / Life Insurance Companies1992 / 2000 / 2002 / 2004 / 1992 / 2000 / 2002 / 2004
Austria / 54.5 / 55.4 / 55.1 / 63.8 / 51.1 / 50.4 / 47.4 / 52.6
Belgium / 34.9 / 55.6 / 59.1 / 60.1 / 54.6 / 72.9 / 73.3 / 73.5
Germany / 23.5 / 27.6 / 28.3 / 31.5 / 30.8 / 32.4
Denmark / 60.3 / 72.2 / 68.5 / 69.9 / 77.0 / 54.8 / 57.6 / 55.3
Spain / 13.8 / 27.8 / 37.1 / 38.7 / 45.6 / 37.5 / 44.7 / 41.5
Finland / 82.7 / 88.5 / 91.5 / 91.3 / 98.7 / 99.7 / 89.8 / 84.9
France / 40.7 / 53.4 / 55.5 / 52.4 / 46.4 / 54.4 / 57.1 / 55.6
Ireland / 47.4 / 71.6 / 71.1 / 67.0 / 62.6 / 71.9 / 69.1 / 69.0
Italy / 33.8 / 35.1 / 65.8 / 69.3 / 53.9 / 27.2 / 52.9 / 62.8
Luxembourg / 84.5 / 54.6
Netherlands / 40.4 / 40.4 / 42.8 / 57.5 / 57.7 / 60.8 / 74.1
Portugal / 54.7 / 49.5 / 70.0 / 70.1 / 49.4 / 74.6 / 81.0 / 78.1
Sweden / 89.0 / 84.9 / 87.9 / 90.8 / 70.3 / 74.3 / 74.1 / 70.3
Switzerland / 61.0 / 61.5 / 56.3 / 56.0 / 71.1 / 80.5 / 81.5 / 78.3
U.K. / 28.6 / 35.7 / 49.9 / 53.2 / 29.2 / 48.9 / 46.1 / 50.0
Source: CEA European Insurance and Reinsurance Federation
An indication of the importance of mergers and acquisitions within the European insurance industry can be obtained from Figure 1, using data extracted from the Bureau Van Dijk Zephry M&A database. The figure shows that the vast bulk of M&A activity in Europe has occurred in the most open of Europe’s insurance markets (the United Kingdom), and that M&A activity in several countries have increased sharply since 2003. The mergers and acquisitions figures reflect the cross-border consolidation in the European insurance industry which itself is a response to the globalisation of financial services. Cummins & Weiss (2004) provide detailed figures which indicate that the majority (70% by number) of takeover targets in the European insurance market in the period 1990 to 2002 were life insurance firms – the majority of which were taken over by other life companies[4].
Although M&As have the potential to increase the market power of consolidated groups, they may also be the primary process whereby firms can harness the efficiency gains that come with economies of scale. Mergers and acquisitions are also an important instrument in the market for corporate control, whereby the managers of X-inefficient firms are replaced. The mere threat of a potential merger or acquisition may also produce powerful incentives on management to maintain efficient operations. Thus the number of mergers and acquisitions can also proxy for the degree of liberalisation and competitiveness in the relevant market.
Figure 1: Mergers and Acquisitions in European Insurance Markets 1998-2005
3. Productivity Growth Measurement & Decomposition
Measuring TFP growth is not a straightforward exercise, as measurement is undermined by a number of conceptual and empirical issues none of which has been satisfactorily resolved in the literature. Broadly speaking, the literature to date has followed two main approaches to productivity measurement – those based on the estimation of a technological frontier showing what is feasible for best-practice firms, and those based on some form of averaging process reflecting what has been achieved by representative firms in the industry. Within the latter, non-frontier approach, traditional measures of TFP growth include the index number approach (which also encompasses the growth accounting methodology) and the econometric production (or cost) function approach. The frontier approach is more recent and within these, parametric (such as Stochastic Frontier Analysis) and non-parametric (such as Data Envelopment Analysis) estimation techniques have been used extensively. With this approach, changes in the estimated frontiers over time are used to capture overall productivity growth.
While overall productivity growth figures obtained using the above methods are meaningful on their own, it is important to understand the various sources through which such growth arises. Thus a decomposition of TFP growth is necessary to identify these sources. The literature on measuring the sources of productivity change can essentially be summarised under two approaches: top-down and bottom-up. Under the former, a measure of TFP growth is obtained and an interpretation of the measure is required. For instance, do the estimates represent pure technical change or do they also capture efficiency change? Under this approach, there is the possibility that some of the TFP growth cannot be adequately accounted for and interpretation of the results may become difficult. Under the bottom-up approach, all possible sources of productivity growth are first identified and then estimated in the best possible way. These estimates are then appropriately combined to construct a measure of TFP growth.
Balk (2001) advocates this bottom-up approach and discusses four sources of productivity growth. The first source, which is widely recognised, given its early association with TFP growth by Solow (1957) is technical change, which arises through a shift in the production technology. The second source is efficiency change, which arises as a result of the firm being able to use its inputs more efficiently to produce its output, given the existing technology. The third source is scale efficiency change, whereby a firm is able to produce at levels of operations closer to the technologically optimum scale of production. The fourth source is the output mix effect, which captures the effect of the composition of the output mix on scale efficiency.