High executive pay may not pay off : case study of the Dutch social housing sector

(1)

Thesis submitted in partial fulfilment of the requirements

for the MSc Economics degree

High executive pay may not pay off:

Case study of the Dutch social housing sector

Author:

Flora Oudeboon

UvA NetID: 6296432 flora.oudeboon@student.uva.nl

Supervisors:

Prof. Dr. MP Schinkel

F Fels¨

o

August 29, 2014

(2)

Abstract

This study evaluates the relationship between executive pay and the performance of Dutch Housing corporations. Executive pay of housing corporations is determined by the size of the corporation and region in which it is active. Possibly, a higher pay attracts more talented and/or more motivated executives, who can realize a better performance. This study firstly determines the relationship between pay and performance. Second, de-terminants of executive pay are identified. The results suggest that there is no significant relationship between pay and performance. Firm characteristics such as size, and the size of ones competitor do significantly influence executive pay. The results of the study may serve as an input for the discussion concerning the Dutch Top Income Standardization Act, focusing on the performance of Dutch housing corporations. This Act moderates executive pay of all senior public officials and quasi-autonomous non-governmental organizations such as regulatory authorities, health care providers, and social housing corporations.

(3)

Acknowledgements

This thesis has been submitted in partial fulfillment of the requirements for the MSc Economics degree of the University of Amsterdam. Between April and August 2014, I have been an intern at the Netherlands Bureau for Economic Analysis (CPB). Here, I got the opportunity to work on my thesis and draw on its resources.

I am very grateful that the CPB, jointly with the Ministry of Economic Affairs, has given me the opportunity to combine the completion of my studies and answer an important policy question. I hope I lived up to their expectations. In particular, I would like to thank

two individuals who taught me a lot. First, I am grateful for Fl´ora Fels¨o’s enthusiasm, her

extensive comments and her around the clock availability to talk about my research. Among many things, she has taught me how to focus on the big picture, and how to decide what specifications to use. She was a true mentor over the last five months. Second, Dr. Bjørn Volkerink helped me by providing thorough feedback. His valuable comments improved the structure of my thesis a lot, and clarified the results section.

Other people I would like to thank are Dr. Dreim¨uller (roAg), Mr. Alexander Luijten

(Aedes) prof. dr. ir. Priemus (TU Delft) who familiarized me with the empirical reality of social housing sector, and deepened my understanding of the labor market for executives. Finally, I am grateful to Prof. Dr. Maarten Pieter Schinkel, who supervised my thesis, provided constructive criticism and showed me vivid examples of how corporations operate and engage its tenants.

The analysis in Chapter 3 and 4 draws on data retrieved from the Ministry of the Interior and the Central Social Housing Fund for which I am appreciative. Particularly Mario Smient and Frans Vroonhoven contributed to my understanding of the data, and always replied very quickly when I asked for particular data.

(4)

Glossary

National umbrella organization of

housing corporations

Landelijke branchevereniging van

woningcorporaties

Aedes

Social Housing Management Decree Besluit Beheer Sociale Huisvesting BBSH

Authorized social housing corporations Decree

Besluit Toegelaten Instellingen

Volkshuisvesting

BTIV Ministry of the Interior & Kingdom

Relations

Ministerie van Binnenlandse Zaken en Koninkrijksrelaties.

BZK

Common Labour Agreement Collectieve Arbeidsovereenkomst CAO

Dutch Central Social Housing Fund Centraal Fonds voor de

Volkshuisvesting

CFV Ministry of Housing, Spatial Planning

and the Environment

Ministerie van Volkshuisvesting,

Ruimtelijke Ordening en Milieu

VROM

Association for Supervisory Board

members housing corporations

Vereniging Toezichthouders

Woningcorporaties

VTW

Top Income Standardization Act Wet Normering Topinkomens WNT

The Act on Disclosure of Publicly

Funded Remuneration of Senior

Officials

Wet Openbaarmaking uit publieke

middelen gefinancierde Topinkomens

(7)

Chapter 1 Introduction

Recently, executive pay has become a more regular topic of discussion. Cases of

self-enrichment, fraud and mismanagement sparked a critical debate about the functioning of and executive pay in the semi-public sector (Veenstra et al., 2013). For instance, the former

executive of housing corporation Vestia was responsible for a loss ofe2 billion due to

deriva-tive speculation of the organization. Next to his base pay of e500.000 in 2011, he received a

e3.5 million severance pay (NRC Handelsblad, 2014; Aedes, 2014a).

After a public outcry, government interventions have monitored and regulated executive pay of public officials (Ministry of the Interior and Kingdom Relations, 2011, 2014a). From 2006 onwards, semi-public institutions were stipulated to report its executive pay to the

Min-istry of Interior Affairs and Kingdom Relations (BZK) if it exceeded a minister’s pay1. In

January 2013, the Top Income Standardization Act (WNT) was instated to moderate

exec-utive pay: it stipulates that newly hired public officials are allowed to earn up to e187 340,

130 percent of executive pay2. The WNT applies to all government employees, educational

institutions, national broadcasters, and quasi-autonomous non-governmental organizations such as regulatory authorities, health care providers, and social housing corporations. The government has proposed a bill to lower this cap to 100 percent of a minister’s salary from January 2015 onwards (Ministry of the Interior and Kingdom Relations, 2014a).

The government presumes in the Explanatory Memorandum of the WNT that qualified and adept senior officials are still interested in becoming an executive, and current executives can be retained against a moderate pay (Ministry of the Interior and Kingdom Relations, 2011, 2014a). However, no studies have looked at the policy implications of moderating executive pay in the semi-public sector in the Netherlands. Hence, this thesis tests whether

1_{Wet openbaarmaking uit publieke middelen gefinancierde topinkomens, BWBR0019562.} 2_{Art. 2.3 Wet Normering Topinkomens, BWBR0032249.}

(8)

fluctuations in executive pay leads to variation in performance. Policy implications of the WNT act are assessed by looking at one industry in particular: the Dutch social housing sector. Social housing has been chosen, because data concerning both firm characteristics and managerial abilities is available. Moreover, variation in pay due to exogenous changes in

the common terms of employment help determine the effect of pay on performance3_.

Social housing in the Netherlands

Dutch social housing corporations are supported by the government to fulfill the public task of the provision of affordable and adequate housing of those in need (Conijn, 2005). At present, approximately 2.4 million dwellings are owned by housing corporations. In total, housing corporations own 31 percent of the Dutch housing stock, and represent the largest national social housing sector in Europe (Fleurke et al., 2009). The majority of its tenants have limited financial means to afford housing: the primary target group earns a disposable income of 1250 euros per month, of which 31% is spent on housing (Kromhout, 2013).

According to the most recent coalition agreement, the core activities of housing corpo-rations include the building, leasing and managing of social housing (VVD–PvdA, 2012). Its primary responsibilities are outlined by the Social Housing Management Decree (BBSH). These include: (1) provide adequate housing for people with a low income, (2) improve or sustain the quality social housing dwellings, (3) enlarge the liveability of the neighborhood, (4) offer real estate property to help out elderly and other individuals in need, (5) safeguard

the financial continuity of the corporation, and (6) engage tenants in one’s housing policy4_.

Performance and executive pay

Executives can be considered to be experience goods. Before you know what an executive will contribute to a firm, you commit to a price. Theoretically, an executive should receive a part of its expected marginal product (Gabaix and Landier, 2008). Empirically, scholars report that observable firm characteristics, such as size and performance, can partially explain the level and variation in executive pay (Gabaix and Landier, 2008; Graham et al., 2012).

On the one hand, pay is determined by supply and demand in a perfectly competitive market according to economic theory (Varian and Norton, 1992). Moderating executive pay bends part of the supply curve and could hence create a shortage of executives. On the other hand, it may be difficult to justify an exorbitant executive pay level, if your tenants have 3_{Guidelines on executive pay were revised in 2000, 2004 and 2010 respectively (Peters Committee, 2000;}

Izeboud et al., 2004; VTW and Het Directeurencontact, 2010).

(9)

limited financial means to pay their rent (NRC Handelsblad, 18 November 2013).

Even though the government has argued that executive pay should be at a more “socially acceptable, balanced, responsible level” (Ministry of the Interior and Kingdom Relations, 2014a), higher executive pay may actually pay off. In other words, a higher executive pay possibly attracts a more talented executive, who can realize a better performance. Conse-quently the better performance of housing corporations benefits the general public, because the adequate provision of social housing is a public interest (VVD–PvdA, 2012).

To the best of my knowledge, no research considers the relationship between pay and performance of the Dutch (semi-)public sector. Yet, understanding if and how executive pay influences performance is key to evaluate the policy implications of the WNT. The evaluation of the WNT Act depends strongly on the understanding of the determinants of executive pay and performance individually. A better apprehension of sector-specific determinants of pay yield insights into the policy implications of its regulated market for executives.

The implications of the WNT Act are analyzed by use of two hypotheses. The first hypothesis concerns the relation between executive pay and performance. It is expected that pay significantly influences performance. Second, the effect of managerial abilities and firm characteristics on executive pay are discussed.

Outline

This thesis is structured as follows: first, the organizational structure of the social housing sector is presented in Chapter 2. Besides the history and market structure, the responsibilities of and common terms of employment in the social housing sector are discussed.

A cost model is used in Chapter 3 to define performance of housing corporations. The model estimates the cost efficiency of housing corporations, that is used as a proxy for per-formance. The dependent variable is total costs, and explanatory variables include outputs, controls and executive pay. Alternative specifications, that check the robustness of the re-sults, include various measures of costs, and the executive’s start date. The results suggest that there is no positive effect of executive pay on cost efficiency.

Chapter 4 tests the prediction derived from the literature that firm characteristics and managerial attributes can partially explain variation in executive pay (Graham et al., 2012). In order to do so, the ordinary least squares method is used to determine the effect of these characteristics on executive pay. The results suggest that the most important determinants of executive pay is the size of the corporation itself.

(10)

Chapter 2 Social housing in the Netherlands

2.1 Historical Overview

Right to housing

During the industrial revolution, a serious shortage of housing in cities was caused by the accelerating growth of people living in urban areas. From 1850 onwards, private initiatives, such as the Vereeniging ten behoeve der Arbeidersklasse te Amsterdam, arose to provide housing for laborers. Often, the aim of the investors of these initiatives was to increase the number and quality of living accommodations of laborers (van der Schaar, 2009).

The Housing Act of 1902 legally introduced the shared national responsibility to serve those inhabitants who are unable to secure adequate housing by themselves. The objective of the Housing Act was to put an end to unhealthy basement tenements and bad speculative building. In order to prevent politicians lowering rents to win votes and consequently endan-gering the financial continuity of the organization, private actors were preferred to provide social housing (van der Schaar, 2009). The Act allowed housing associations and coopera-tives to receive state aid, if their primary activity concerned social housing. The selected undertakings were granted loans, tax reliefs and guarantees.

The Housing Act enabled the Crown to formally approve and accept housing associations

(Fleurke et al., 2009). If housing corporations wanted to be eligible for state aid, they

were expected to operate in the social housing sector and all of its profits were supposed to find their way back to social housing. Municipalities mainly supervised the social housing sector, and determined the success of the development of social housing in the following years (Fleurke et al., 2009).

(11)

After the second World War, the government took the lead to provide more (adequate) housing for the poor. In order to reduce the housing shortage caused by the war, subsidies were provided on a large scale for those corporations willing to invest, renovate, or demolish and rebuild designated local neighborhoods (Fleurke et al., 2009). The execution and plan-ning of these areas was agreed upon with the local corporation, who was supervised by the municipality.

Besluit Beheer Sociale Huursector

In 1977, the revision of the Housing Act and its implementing order, the Besluit Toegelaten Instellingen Volkshuisvesting (BTIV), stipulated more responsibilities for housing corpora-tions. Although the scope of social housing was not defined by the revision, corporations were now obliged to give low-income households priority when letting inexpensive homes (van der Schaar, 2009). The implementing order was replaced by the BBSH Decree in 1993, where

social housing was defined1_{. The Decree provided a demarcation of permitted activities and}

the core tasks of housing corporations, and provided a legal framework for the governance structure of the social housing sector (Fleurke et al., 2009).

BBSH specified the terms of reference and duties of housing corporations. The duties of housing corporations included the adequate provision of quality housing for households with a low income, the engagement of tenants in the organizations’ policies and the safeguard of its financial continuity2_.

The grossing and balancing operation

Up until 1995, housing cooperatives received subsidies intended to cover part of operational

costs and ultimately reduce housing expenses for its tenants3_{. The government redeemed}

its future commitment to the corporations by a one-off transfer of all real estate, writing off all outstanding loans and abolishing the supply-side subsidies in 1995 (Vandevyvere and Zenth¨ofer, 2012). This is called the ‘grossing and balancing operation’. Housing policy shifted from being determined by the government, to governance (Scanlon and Whitehead, 2008).

Housing corporations were strongly encouraged by (local) governments in the late 1990s and early 2000s to claim a wider role (Fleurke et al., 2009). In 1997, the BBSH was adapted such that the liveability of neighborhoods became a core responsibility for housing corpora-tions. The provision of homes for elderly, disabled and others with special needs was added

1_{Chapter III, Art 12b, Besluit Beheer Sociale Huursector, BWBR0005686.} 2_{Chapter III, Art. 11-25, Besluit Beheer Sociale Huursector, BWBR0005686.}

(12)

Figure 2.1: The external corporate governance structure of the social housing sector

Adoption of Management of Social Housing Decree, indicating the demarcation of permitted activities and core tasks of housing corporations.

Providing housing to those in need of care becomes a task of corporations, as part of the Decree. Duties Housing Corporations Supervisory body Responsible public official Government L. Spies Rutte-I Kok-I Balkenende-I Balkenende-III E. Heerma J. Remkes S. Dekker E. Vogelaar E. van Middelkoop Rutte-II

E. van der Laan Lubbers-III Kok-II Balkenende-II Balkenende-IV D. Tommel H. Kamp P. Winsemius P.H. Donner S. Blok 1993 1994 1995 1996 1997 1998 1999 2000 2001 2003 2004 2002 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014

Corporations are supervised by municipality.

Inspectorate for housing becomes responsible for supervision corporations.

Financial monitoring becomes responsibility CFV.

Law amendment to to improve supervision.

Inspectorate for housing becomes part of VROM inspectorate.

Inspectorate for housing becomes part of the Human Environment and Transport Inspectorate.

CFV starts to dispose annual evaluations of corporations, on request of the minister.

Source: Algemene Rekenkamer (2014).

Improving the liveability of neighborhood(s) a task of corporations, as part of the Decree.

(13)

to the Decree in 2002. In principle, housing corporations expanded their focus to urban de-velopment, the quality of the neighborhood and the well-being of its tenants throughout the late 1990s and early 2000s (Scanlon and Whitehead, 2008). Because housing corporations often own(ed) the majority of the housing stock in urban renewal areas, they took a leading role to improve the quality of the neighborhood (ibid).

The social housing sector itself highlighted its corporate social responsibility in this period: the corporations were seen as independent enterprises with a social mission (Fleurke et al., 2009). In fact, they considered themselves to be “policymakers, implementers and social engineers”; all to improve the provision of housing, the neighborhoods, and the tenants’ well being (Scanlon and Whitehead, 2008). Consequently, corporations engaged themselves in activities outside of its traditional duties like the maintenance and refurbishment of homes. Governance post-1995

The grossing and balancing operation marked the start of more governance than government in the social housing sector (Scanlon and Whitehead, 2008). The years following the operation can be characterized by frequent changes in the responsible public official, as well as in policy. Indicative is the number of Ministers and Secretaries of State that have been responsible for social housing: out of the 12 responsible public officials since the grossing and balancing operation, 7 have been in charge since 2007 (See Figure 2.1). As a result, policy changes were implemented frequently and social housing goals were regularly modified: on average, four amendments were made annually to the Housing Act, including its BBSH Decree, in the period 1993-2013 (Algemene Rekenkamer, 2014).

In general, a distinction can be made between external and internal corporate governance. The Central Social Housing Fund (CFV) and the housing inspectorate are responsible for the external corporate governance. Its structure is explained by the timeline in Figure 2.1. CFV audits the financial performance of housing corporations, and can financially support

insolvent firms by requiring all housing corporations to contribute4. The housing inspectorate

is the government’s supervisory body that monitors the performance of housing corporations (Algemene Rekenkamer, 2014). The Supervisory Board and the corporation’s independent auditor are responsible for the corporation’s internal corporate governance.

(14)

2.2 The present market structure

A Segmented Housing Market

One of the prime characteristics of the present social housing sector is its distinct character within the overall housing market. More than seventy per cent of all rental dwellings are owned by housing corporations; 31% of the total housing stock. Figure 2.2 illustrates the division of the housing between the different types of owners. Because of the strong seg-mentation between the social rental sector and the other forms of housing, a “dual market” arises (Kemeny in Fleurke et al., 2009, 57). Below, a short description of the present market structure will be given to highlight the social housing sector’s distinct character.

Figure 2.2: The Dutch housing stock (2012)

The social housing sector is subject to artificially low, regulated, rents, and strict

reg-ulation. These regulations require 90% of newly vacant homes with a rent of e699.48 or

less to be allocated to households with an income less than e34,678 (price level 2014)5

(Ministry of the Interior and Kingdom Relations, 2014b). Because of the low, regulated, rental prices in social housing, most corporations let its social dwellings below its operational costs (Fleurke et al., 2009). In order to be able to invest or maintain the real estate, hous-ing corporations work in a protected environment. Here, the government still directly and indirectly supports all social housing corporations:

Direct support stems from favorable borrowing conditions: the government provides ac-cess to the capital market and is ultimately the guarantor of the corporation’s outstanding loans. Its favorable borrowing terms can officially solely be used to invest in social housing:

cross subsidizing is not allowed6 (van der Schaar, 2009). Because many tenants in social

housing dwellings receive household-specific subsidies, designed to lower housing expenses, corporations receive indirect support from the government.

5_{Chapter III, Art. 13, Besluit Beheer Sociale-Huurstector, BWBR0005686.} 6_{European Commission (2009). State aid No E 2/2005 and N 642/2009.}

(15)

In contrast, the private rental sector is free to set its (usually higher) rents and faces less regulatory constraints, yet does not receive financial benefits. Households of owner-occupied homes face less regulatory constraints too, but are supported by the government’s mortgage deduction plans.

Competition

Because housing is spatially fixed, competition between different actors is more complicated. Below, two differences with perfectly competitive markets are mentioned.

First, in a competitive rental market, social housing corporations and private investors compete, assuming that consumers will take out the better price/quality bundle. Yet,

house-holds earning less than e34,678 (threshold eligibility social housing dwellings) are unlikely

to find unsubsidized, affordable housing (Fleurke et al., 2009). As a result, these households are subjected to take out a rental dwelling from housing corporations. Especially in urban environments, waiting lists create a stable demand for low-rent dwellings from corporations. This indicates either a dearth of low-rent dwellings or a too low price point. In Amsterdam, for instance, waiting lists for social housing vary between 4 to 11 years (van Zoelen, 2012).

Second, most competitive markets are characterized by free entry and exit, such that only the most efficient firms are able to stay in the market (Pepall et al., 2005). Yet, (potential) entry into the social housing sector is limited, because up-front investments are high and approval is necessary to enter the market (Tegel et al., 2006; Fleurke et al., 2009). In practice, few entrants enter the social housing sector and only one corporation has left the market since 1990. In the same period, 21 insolvent corporations, however, have received financial support to reorganize its organization (Tegel et al., 2006).

After housing corporations gained independence, a wave of mergers did change the mar-ket structure significantly: in less than 20 years, the number of corporations halved (BZK Data). The total number of housing corporations decreased from 824 in 1990 to 381 in 2012 (Algemene Rekenkamer, 2014). In 2010, more than forty percent of the corporations had been involved in a merger between 1997 and 2010 (van den Berge et al., 2013).

In conclusion, corporations mainly operate in a market where sales are secured and hardly have to bid for the tenants’s favor (Fleurke et al., 2009). Moreover, entry and exit is severly limited.

(16)

2.3 Pay and performance in the social housing sector

2.3.1 The fulfillment of the corporation’s responsibilities

Corporations are free to prioritize the duties described in the BBSH according to one’s own insights (Koning and Leuvensteijn, 2010). In general two main goals can be identified: (1) the provision of adequate housing for its target group, and (2) minimize the negative external effects of the real estate and its tenants. These negative externalities include the quality of the housing stock and the liveability of the neighborhood. The firm’s financial continuity and engaging tenants can be considered to be prerequisites to fulfill the other goals (Veenstra et al., 2013).

The allotment of newly vacant dwellings illustrates whether the target group is in practice the prime group of tenants in the social housing sector. In Figure 2.3a, the adequateness of the allotments by year is presented in percentages, where a distinction is made between tenants below and above 65 years. On average, 73 % of all the dwellings are adequately allotted. These homes are provided either to households with a low income who pay little for their home, or to those with an income above the mean-tested income who pay for high-rent dwellings. More than 16 % of all dwellings are allotted to low-income tenants paying a too high rent. The rest of the dwellings are low-rent dwellings, and are provided to tenants with a too high income. Interestingly, the proportion of tenants paying too little rent has quadruppled from 5.5 % in 2007 to 20.6 % in 2012.

Figure 2.3: The provision of affordable and adequate housing

(a) The adequateness of the corporation its al-lotments.

(b) Dwellings provided to special attention groups.

(17)

Figure 2.3b illustrates the number of homes specifically intended to serve the households who receive special attention in the BBSH: the elderly, disabled and others with special needs. The number of single level homes is not taken into consideration in the Figure, because these may not be specifically intended for these groups. Overall, the number of homes specifically intended to house elderly has increased, whereas the other homes remain stable. The share of dwellings specifically designed for these groups remains around 23% (BZK Data).

Minimization of negative external effects

In 1997, the improvement the liveability of neighborhoods was added to the BBSH as a distinct responsibility of housing corporations. The Leefbaarometer provides a bi-annual index of the liveability of municipalities, based on the demographics of its residents, the quality of the public space, the safety of the area, its amenities and the quality of the housing stock. According to this index, visible improvements have been made across all fields over the last 6 years. In Table 2.1, aggregate scores by year for all of the before mentioned fields can be found. In general, a higher score indicates better performance in this field. For instance, a higher score of one’s housing stock illustrates that the quality improved over the last year(s).

Table 2.1: Liveability of municipalities of housing corporations

Demographics Social Public Safety Amenities Housing

Demographics space stock

2008 10.809 2.260 -2.361 10.955 -2.842 3.799

2010 10.381 3.583 5.262 11.106 -4.293 4.442

2012 12.115 0.577 5.217 13.695 -4.704 4.420

Source: www.leefbarometer.nl. Its values are based on a scale ranging from -50 to 50.

Promote cost-efficient operational management

The firm’s financial continuity and the extent to which it engages tenants can be considered to be prerequisites to accomplish its social housing goals. In practice, tenants are rarely not engaged as it is stipulated by law to engage them, yet the extent to which this happens varies (Koning and Leuvensteijn, 2010).

Safeguarding the financial continuity of the social enterprise can be promoted by cost-efficient operational management. Immediately after the grossing and balancing act, the overhead expenses increased significantly. Multiple annual reports of CFV, as well as a specialized report underline this trend (CFV, 2005). Today, overhead costs continue to rise.

(18)

Between 2008 and 2012, revenues increased with more than 9 million. Wages increased with 0.5 million euros to 3.6 million euros in 2012 (BZK Data).

The average balance of a housing corporation is shown in Table 2.2. The average balance sheet comprises 33 million euros of revenues, and 26 million euros. On average, revenues per

dwelling were e6,518, and expenses e4,921 in 2012.

Table 2.2: Average balance of a housing corporation 2008-2012

Income Expenses

Rents 27 285 Depreciation 7 401

Reimbursements 1 482 Ground lease 60

Government funding 429 Labor costs 3 993

Sales 3 120 Maintenance 7 020

Capitalization 353 Miscellaneous 6 608

Miscellaneous 830 Industry levy 1 041

Total 33 499 26 123

Source: BZK Data. Numbers are in thousands of euros. Only operational revenues and expenses are accounted for.

2.3.2 Executive pay policies over time

After the grossing and balancing operation, firms enjoyed more freedom and operated as (social) enterprises with a public task. As a result of the newly acquired freedoms, most housing corporations quickly became more professional organizations. Up until the transfer of funds, executive pay was still arranged according to the highest scale(s) of the common labor agreement. A job weight system was used where scales determined the pay levels of all jobs in the housing sector, according to seven types of responsibilities. Although directors could be exempted from this method and negotiate individualized terms of employment and pay levels, the highest base scale of the agreement was primarily used to determine executive compensation.

The Peters-Committee

Two proceedings led to a call for general terms of employment for executives in the social

housing sector in the late 1990s. As a result of these two developments Aedes7 asked Peters

et al. to draft a counsel that could guide employers to determine the terms of employment of the managing director. This Committee consisted out J.F.M. Peters, chair of the national 7_{Aedes is the national umbrella organization of the Dutch housing corporations. Approximately 90 percent}

(19)

corporate governance committee and former chief executive Aegon, seven chairs of supervisory boards and seven chairs of boards of directors of housing corporations. They were advised by compensation consultant Hay Group (Peters Committee, 2000; Aedes, 2003).

First, the updated sector Collective Labor Agreement (CLA) of 1999 stipulated that its clauses were not applicable to the statutory managing director any more. Executives were exempted, because (s)he fills the position of employer, has a different legal position and the pay scales in the agreement were not considered to fit the job profile of the chief executive.

Second, changes in the sector led to a more dynamic corporate environment and job profile for executives. Examples are the increase in scale, professionalization of the organization and the newly introduced market mechanisms after 1995 (Fleurke et al., 2009). A new corporate structure for housing corporations was proposed by the Glasz committee in 1998 to promote transparency: housing corporations should be governed by use of a two-tier board (Glasz et al., 1998). In short, a distinction between a non-executive and executive board was made. The Peters-Committee used this corporate structure as a point of departure when drafting the guidelines.

The main recommendation of the Peters Committee stated that the statutory managing director should earn a wage in line with the prevailing market wage (Peters Committee, 2000). The report specified different job profiles for different types of corporations. The Peters-Committee used peer groups consisting out of firms from related sectors, similar to mainly medium-sized and large housing corporations. Still, the evaluation of the implementation of the guidelines in 2003 highlighted that 75 per cent of the Supervisory Boards of corporations used the advice to determine the terms of employment (Peters Committee, 2003).

The Izeboud-Committee

In 2003, the Izeboud-Committee was formed to revise the recommendations of the Peters Committee and generalize its findings to include all types of housing corporations. The committee outlined several recommendations regarding executive pay conditions and terms of employment in the sector. Moreover, a revised compensation scheme was proposed that would create more transparency with regards to the determination of executive pay in the sector. This scheme determined a base scale of an executive based on size, market dynamics, and (value of) the tenant turnover rate (Izeboud et al., 2004). Although the guidelines were not obligatory to use, a comply or explain principle was introduced for all members of Aedes. Because of the high degree of organization, more than 90 per cent of all housing corporations now had to comply to the detailed guidelines, or explain.

(20)

Aedes and the Vereniging Toezichthouders Woningcorporaties (VTW), the association of supervisory board members of housing corporations, commissioned an evaluation of the guidelines proposed by Izeboud in 2009 (van Waning et al., 2009). The evaluation illustrated that more than 70 % of the corporations employed the guidelines to determine the managerial incentives and its associated pay level (van Waning et al., 2009). The executive pay data retrieved from the Ministry of the Interior shows that by 2010, 56 per cent of all executives were hired after 2004 (N=331). Hence, more than half of all housing corporations used the Izeboud-guidelines in 2009. The main arrangements used by those that do not employ the Izeboud guidelines are the CLA and possibly its evaluation handbook (11.8 %), the recommendations made by the Peters I Committee (8.7%), or make use of compensation consultants such as HayGroup (4.1 %).

Even though a benchmark was used to determine the base scale of executive pay, pay levels of other housing corporations seem to influence average executive pay too. In 2006, regulation was instated that obliged housing corporations to present its executive salaries in its annual reports. This Act on Disclosure of Publicly Funded Remuneration of Senior Officials (WOPT) increased transparency of executive earnings and may have contributed to the ratcheting up executive pay: average fixed income has risen from 124,000 euros in 2008

to 145,000 euros in 20128 _{(Data Ministry of Interior Affairs & Kingdom Relations, 2013).}

From 2010 onwards, Aedes obliges all of its members to employ its Sector Remuneration Code when new chief executives are appointed (VTW and Het Directeurencontact, 2010). The Sector Remuneration Code can be considered to be an extension of Izeboud. The per-sonal competencies and job characteristics that determine executive pay vary by arrangement over time.

Government intervention

The WOPT Act was instated in March 2006 to promote transparency of executive pay in the

(semi-) public sector9. It required organizations to dispose individual compensation packages

if they exceeded a Minister’s pay. In December 2010, the Minister of the Interior and Kingdom Relations published a report that described executive pay of all individuals that exceeded this benchmark (Ministry of the Interior and Kingdom Relations, 2010). A public and political outcry followed this report, arguing that executive pay in the semi-public sector should be moderated (Ministry of the Interior and Kingdom Relations, 2011).

8_{Inflation only accounts for 7.4% of the increase in executive pay in this period, whereas the increase in}

executive pay is 17 %.

(21)

Table 2.3: Maximum executive pay social housing sector 2014

0 - 8,001 - 14,001 - 24,001 - 40,001 - 60,001 - 100,001 - 150,001 - 375,000 8,000 14,000 24,000 40,000 60,000 100,000 150,000 375,000 & more 0 - 750 e082.1 e082.1 e082.1 e082.1 e082.1 e082.1 e093.0 e093.0 e093.0 751 - 1500 e093.0 e093.0 e093.0 e093.0 e093.0 e103.9 e103.9 e103.9 e103.9 1,501 - 2,500 e103.9 e103.9 e103.9 e103.9 e112.4 e112.4 e112.4 e112.4 e112.4 2,501 - 5,000 e112.4 e112.4 e130.6 e130.6 e130.6 e130.6 e130.6 e130.6 e130.6 5,001 - 10,000 e130.6 e148.8 e148.8 e148.8 e148.8 e148.8 e148.8 e148.8 e148.8 10,001 - 25,000 e167.0 e167.0 e167.0 e167.0 e167.0 e167.0 e167.0 e185.2 e185.2 25,001 - 50,000 e185.2 e185.2 e185.2 e185.2 e185.2 e185.2 e203.4 e203.4 e203.4 50,001 & more e203.4 e203.4 e203.4 e203.4 e203.4

Source: Minister of the Interior and Kingdom Relations (2013). Notes: the horizontal axis represents the largest number of inhabitants living in the municipality where the corporation is active and has more than 20% of its real estate located. The vertical axis represents the number of dwellings owned by the corporation. Executive pay is reported in thousands of euros.

The WNT Act replaced the WOPT Act in January 2013. The WNT Act was accepted by Parliament in December 2011 and introduced a cap of 130 % of the wage of a Minister on executive pay of publicly funded senior officials10_{. This Act is applicable to hospitals, the}

social housing sector, and other institutions that are established by virtue of the constitution. Although a transition period of 7 years was granted, the cabinet already proposed a new cap of 100 % of the Minister’s pay to be instated in 2015 (which includes a transition period for incumbent executives too). Ultimately, no public official is allowed to earn more than a minister’s pay in 2022 (Ministry of the Interior and Kingdom Relations, 2014a).

For housing corporations, specific scales were recently proposed that will lower the cap based on the size of the corporation and the biggest municipality in which it is active

(Minister of the Interior and Kingdom Relations, 2013). In general, this means that

ex-ecutives can buy a higher pay by acquiring more homes or merging with other corporations. In Table 2.3, the scales proposed in November 2013 are displayed. Here, 10 scales differ-entiate between the number of dwellings (horizontal axis). Using this method, executives may develop more homes or build in particular municipalities to receive or justify a high pay level. Not only may this be a waste of financial resources, Veenstra et al. (2014) reports that large housing corporations are usually less scale efficient than smaller corporations. Hence, if executives use their resources to increase their pay, average pay levels will increase and the internal efficiency of housing corporations decreases.

A short summary of all executive pay policies influencing executive pay since the grossing and balancing operation is presented Table 2.4.

(22)

T able 2.4: Timeline Rem uneration P olicies Managing Directors Housing Corp orations Date Do c u m en t Metho d Set increases? Comparison group Up un til 1999 CLA W o ondiensten The CLA do es not adv is e on pa y if th e job profile exceeds a certain thresholds that is defined b y th e typ e of resp onsibilities of the executiv e. In this case, the Sup ervisory Board then has to decide on executiv e pa y . Ann ually n/a Jan uary 2000 P e ters I Committee 4 job pr ofiles , based on p eer grou ps for small and large corp orations. No S m al l corp orations: real estate, es-tate agen ts and arc hitects. Large corp orations: construction and house bu ilding. Jan uary 2003 P e ters II Committee Comparison d iffe re n t p ee r groups to up date pa y lev els job profiles as sp ec-ified b y P eters I. Ann ually Health care d irec tor s, Go v ernmen t and Non-Profit, R eal Estate, Insti-tutional In v estors, and Health Care Insurers. July 2004 Izeb oud Committee Guidelines to determine e xecutiv e pa y lev el Biann ually Based on the median of the general mark et pa y lev els (executiv es) in the Netherlands July 2010 Sector Rem uneration Co de for Executiv es F ollo ws the ev aluation of Izeb oud’s metho d Biann ually See Izeb oud et al. (2004). Jan uary 2013 WNT Act caps pa y of emplo y ees in the semi-public sector to 130 % of a Min-ister’ pa y . Ann ually Minister Jan uary 2015 WNT (2) Cap reduced to 100 % of a Minister’ pa y . The Act reduces the maxim um pa y , based on the scale of the firm and the m unicipalit y in whic h it is activ e. Ann ually Minister (prop osed)

(23)

2.4 Concluding remarks

In conclusion, the social housing sector has been subject to several developments. The

history, market structure and executive pay policies describe the organizational structure in which housing corporations act. In order to be able to fully understand the way pay and performance work in this sector, it is necessary to acknowledge and take into account the role and function of housing corporations over time.

Initially, housing corporations focused the maintenance and refurbishment of homes and received large subsidies to do so. In principle, housing corporations were expected to coun-teract the shortage in housing and to refurbish old homes. In the 1990s, core responsibilities included both the provision of adequate housing, the quality of the neighborhood and the ten-ants’ well-being. Today, its core responsibilities are the provision of adequate and affordable social housing, whilst minimizing the external effects of its tenants.

The grossing and balancing operation acquitted all outstanding loans of housing corpo-rations, who became financially independent organizations. The operation resulted in fewer, yet wealthier, housing corporations. Since the dissolution of direct subsidies in 1995, the social housing market became less competitive; entry and exit are regulated, and the strong segmentation in the rental sector leaves less-endowed households with few choices. Long waiting lists illustrate the stable demand for social housing dwellings.

The increase in scale of corporations and the professionalization of the sector after the grossing and balancing operation also resulted in higher pay levels of executives (See also Chapter 4). This trend was visible throughout the (semi-)public sector. In order to pro-mote transparency concerning the pay levels of senior public officials, the WOPT Act was introduced. In January 2013, the WNT Act replaced the WOPT Act and stipulated that all (semi-)public organizations should disclose its executive pay, and it moderated executive pay for newly hired executives.

In general, three periods can be identified that characterize executive pay guidelines in the social housing sector: (1) from the grossing and balancing agreement to the distinction in the common labor agreement between executives and all other employees in the housing sector (1994-1999), (2) The establishment of guidelines regarding executive pay in the social housing sector (2000-2004), (3) General sector guidelines regarding executive pay, according to a comply or explain principle, and (2004-2011) (3) Government intervention that promotes transparency of pay in the semi-public sector and caps executive pay (2011-present).

(24)

Chapter 3 Pay and performance

In order to establish the link between pay and performance, a framework to measure perfor-mance should be developed. This chapter will firstly describe the empirical relation between pay and performance, according to the literature. Second, the performance of housing corpo-rations is defined. Third, the methodology of the study is discussed, and the base empirical model is established. The results are discussed in the final section of this chapter.

3.1 The empirical relation between pay and

perfor-mance

Most studies that focus on the empirical relationship between top executive pay and per-formance of enterprises regard listed companies. In contrast, few studies consider non-profit organizations or other not-listed companies, mainly because of the lack good data and the difficulty to compare the diverse institutions in the non-profit sector (Oster, 1998).

In the Netherlands, WOPT requires public and semi-public institutions to disclose the pay of its executive(s). The Act only required public disclosure if executive pay exceeded the pay of a minister. Since 2013, the WNT stipulates firms to disclose its executive pay regardless of the level1_.

Profit Sector

Both Murphy (1999) & Core et al. (2003) provide a comprehensive overview on the existing empirical literature on the pay-performance relationship for listed companies. In general, the literature documents a positive relationship between top executive pay and shareholder

(25)

value, and variation between executive pay levels in different sectors. More recent studies in this field, such as Conyon and He (2011), focus on particular sectors or countries, and usually confirm pay-performance sensitivity.

The few studies that do report a weak pay-performance sensitivity argue that managerial power and rent extraction determine pay (Fried and Bebchuk, 2003). Duffhues and Kabir (2008), a case study that concerns the Netherlands, report that managers seem to influence their own compensation. Here, higher pay is awarded even when firm performance does not increase.

Non-profit sector

Oster (1998) and Hallock (2002) analyze the nonprofit sector in the United States, and find a strong link between performance and pay. Oster (1998) concludes that executive pay also depends on the industry in which the nonprofit is active, and whether the organization

has religious affiliations. Hallock (2002) reports also that nonprofits who receive higher

government grants pay their executive more, when firm fixed effects are not accounted for.

3.2 The performance of housing corporations

In order to establish the link between pay and performance, a framework to quantify per-formance should be developed. Because the business environment and duties of housing corporations are considered to be highly diversified, difficulties arise creating a uniform mea-sure of good performance. Due to the multitude of non-contractible goals, several scholars have expressed their doubts concerning the possibility and the strength of a uniform per-formance measurement in the social housing sector (Conijn, 2005; Algemene Rekenkamer,

2014). After the grossing and balancing operation, the deliberate choice was made not

to introduce an aggregate performance measure that could be regulated. The Ministry of Housing, Spatial Planning and the Environment (VROM) decided to strengthen the corpo-rations’ independence, rather than to monitor and evaluate them on a regular basis (VROM in Algemene Rekenkamer, 2014).

Several economists have evaluated the performance of public and semi-public institutions by defining its effectivity and its efficiency, because profit maximization is not their primary goal (Hakfoort et al., 2002). Such a measure determines the extent to which a corporation maximizes output given input, or minimizes cost given output. This method has been used by de Graaf et al. (2001) and Veenstra et al. (2013) to approximate the efficiency of housing

(26)

In principle, effectivity refers to the extent to which targets of an institution are achieved. For example, the percentage of lower income households living in its dwellings could be an indicator of the enterprise its effectivity. Efficiency relates to the output, relative to the inputs the organization uses. It increases when a firm uses less inputs to produce the same (or more outputs with the same inputs). Alternatively, efficiency increases if costs can be reduced while keeping output constant.

3.3 Methodology

3.3.1 Base empirical model

The empirical specification used in this study is derived from the cost minimization problem. Housing corporations are assumed to be rational producers who aim at cost-minimization, given output and input prices. The main hypothesis of this study is that there is a significant relation between pay and performance. The base empirical model follows the cost minimiza-tion specificaminimiza-tion of de Graaf et al. (2001) and Veenstra et al. (2013). Using this model, a housing corporation performs well if it keeps its costs as low as possible, for a given output.

The base empirical model will be estimated by use of the ordinary least squares method (OLS) following a log-log specification. This specification is desired, because a percentage increase in one of the independent variables is expected to lead to a constant percentage change in the total costs (TC). Both time and corporation fixed effects will be taken into account in the specification. Equation 3.1 outlines the base empirical model.

𝑙𝑛(TC𝑖𝑡) = 𝛼+𝛽1𝑙𝑛(pay𝑖𝑡)+𝛽2𝑙𝑛(q𝑖𝑡)+𝛽3𝑙𝑛(ttr𝑖𝑡)+𝛽4𝑙𝑛(urban𝑖𝑡)+𝛽5𝑙𝑛(ℎ𝑣𝑖𝑡)+𝛾1𝑇𝑡+𝛾2𝐺𝑖+𝜀𝑖𝑡

(3.1)

Here, 𝑞𝑖𝑡 is a measure of output of firm 𝑖 in terms of the number of homes owned over time

𝑡. 𝑡𝑡𝑟𝑖𝑡 represents the annual tenant turnover rate of the organization. Both 𝑞𝑖𝑡 and 𝑡𝑡𝑟𝑖𝑡

are aggregate measures of output. 𝑢𝑟𝑏𝑎𝑛𝑖𝑡 marks the weighted average of the number of

inhabitants of the municipalities in which the corporation is active. It serves as a control

for the local environment in which the corporation operates. ℎ𝑣𝑖𝑡 reports the quality of the

home. 𝑝𝑎𝑦𝑖𝑡, the base pay of executives, is the main variable of interest.

In this specification, 𝑇𝑡 and 𝐺𝑗 are included to control for time- and corporation fixed

effects. 𝛾1approximates the effect of input prices, productivity growth and common shocks on

annual basis. Finally, the coefficient of the corporation fixed effects, 𝛾2, estimates

corporation-specific unobserved characteristics that are constant over time. An example of this type of fixed effects is the particular performance agreement some corporations agree upon with

(27)

the municipality. This contract influences its cost structure. A second example is the cost structure a new executive inherits when (s)he is appointed. For instance, the new executive of housing corporation De Woonbron was appointed in a period after the investment in the former cruise ship SS Rotterdam cost 230 million more than expected (Aedes, 2014b). More disaggregated models than the one described in Equation 3.1 can be found in Appendix A, Table A.3.

Hence, executive pay (which may be endogenous), and exogenous controls are the ex-planatory variables of the base empirical model. The dependent variable, total costs (𝑇 𝐶), can be determined by use of several definitions. Below, the different definitions of total costs are discussed, as well as a more elaborate definition of the other variables and executive pay. Dependent variable

The dependent variable, total costs, can be defined in several ways. First, the log of total variable costs (TVC ) includes the labour costs, maintenance costs and ‘miscellaneous oper-ational expenses’ (miscellaneous) defined by BZK. The latter category includes for instance service costs. Second, the total operational costs variable (TOC ) includes all the variable costs described above, but adds depreciation, ground lease costs, revaluation of current assets and the industry levy. Finally, TOIC adds the interest charges to the total operational costs. Please note that all measures of total costs exclude the variable executive pay, in order to isolate its effect on all other costs instead of its own. Below, you can find an overview of the debit entries that make up the total costs for each of the variables described above.

Table 3.1: Summary total cost definitions

TVC TOC TOIC

Labor costs Depreciation assets Interest charges Maintenance costs

+ Revaluation assets +

Miscellaneous Ground lease

- Executive pay + Industry levy +

TVC is used in the base specification, because it includes all costs that an executive is able to influence in the short-run. Moreover, a similar definition is used by Veenstra et al. (2013). Yet, the activities of the corporation are not solely defined by its current property management and maintenance. TOC and TOIC are used as robustness checks

and to alleviate some of the concerns of the CFV2. These additional specifications include

2_{CFV has expressed its concerns with the data quality and variable choice in Veenstra et al.’s model.}

(28)

costs resulting from endeavors such as commercial activities or real estate development in the dependent variable. TOC includes all operational expenses and covers its other, possibly more commercial, activities too. Because novel ways of financing your debt appear to be an important aspect of the job activities of some executives too, TOIC may also be an appropriate measure (Some cases are highlighted by the Parliamentary Committee, see Aedes (2014b)).

Independent variables

Size is one of the firm characteristics that is likely to explain higher total costs. It is measured by the number of dwellings, because the provision of an adequate number of homes is part of

its core activities3_{. It is expected that the ownership and management of more homes results}

in higher maintenance costs. Moreover, it is likely that more labor is necessary to provide its services and accommodation to its tenants.

The allotment of its dwellings is a second responsibility housing corporations are account-able for. Homes can only be adequately allotted if they are vacant, or more dwellings are surrendered and at the corporation’s disposal. The tenant turnover rate measures the dy-namics of your current housing stock and are likely to affect total costs. For instance, some corporations carry out checks before the tenant leaves the room, and possibly renovate the dwelling.

The degree of urbanization is used as an exogenous control for local market conditions. This is a measure based on the weighted average number of inhabitants of the municipalities in which it is active. Different types of dwellings are found in cities than in rural areas. Moreover, different types of tenants in different urban environments may lead to diverse cost structures for housing corporations.

The quality of rental homes is measured by use of a national home valuation method. This method, called the Woningwaarderingsstelsel, accredits a number of points to each home, based on the size, location, facilities and quality of the home (het Rijk, 2014). The Dutch government established this system to determine the value of residential real estate. Executive pay

Executive pay is defined as the sum of all financial incentives an executive receives (total cash). Total cash includes the variable wage (bonus) of year t+1 and all other financial incentives in year t. Bonuses are usually awarded based on performance in the year before.

(29)

As a robustness checks, the total ex ante financial incentives offered to the executive are used (base pay). In the instructions sent to the housing corporations, corporations are told to submit the executive base pay that includes the payable long-term incentives, social security premium, and vacation allowance (Ministry of the Interior and Kingdom Relations, 2014b). Excluded from the base pay definition are variable and severance pay. The information is retrieved form the Ministry of the Interior, who retrieves the data from all corporations annually.

Total cash is chosen, because it is expected that our instruments are more appropriate to estimate total cash, rather than base pay. The different regimes had different view points regarding the total cash the executive could receive. For instance, the Izeboud guidelines un-derlined and promoted variable pay as an important part of the total compensation package. Data

BZK has kindly shared its data that regards the characteristics of all property of housing corporations on a municipality level, ranging from 2007-2012. Financial firm characteristics are determined on the firm level.

The second data set used in this study is retrieved from the Dutch government. It gathers data on the liveability of neighborhoods an a biannual basis, published at www.leefbarometer.nl. A uniform score for every Dutch municipality is determined by use of a ranking based on population characteristics, housing stock, social demographics, and the safety of the neigh-borhood. The average of the biannual data is used to create an annual measure of liveability in the more disaggregated models. The corporation’s liveability score is determined by a weighted average of the data, according to the share of their housing stock in each munici-pality.

3.3.2 Instrumental variables method

One of the challenges of using the OLS method stems from endogeneity. Endogeneity affects the estimates of coefficients and could hence lead to biased estimates. On the one hand, the level of pay may influence the level of costs, because a better paid, more talented, executive may be able to run the organization more effectively. On the other hand, a higher level of total costs results in more responsibility for the executive, and hence a higher pay may be awarded to the executive. Using the Instrumental Variable (IV) method, appropriate instruments could alleviate this problem and allow me to estimate unbiased parameters. Two exogenous shocks that affect pay, but not total costs, have been chosen to serve as instruments.

(30)

As described in Table 2.4, different exogenous events determined the guidelines on ex-ecutive pay in social housing. First, Peters Committee (2000) introduced guidelines when the CLA was no longer in use for executives. The majority of supervisory boards started to use this method when the CLA was no longer applicable to executives. Only newly hired executives were valued by means of these guidelines. Second, the guidelines proposed by Izeboud et al. (2004) were based on a comply or explain principle and completely revised the Peters Committee (2000) guidelines. These guidelines were commissioned by Aedes and VTW, organizations that represent the housing corporations, and the supervisory board members of housing corporations respectively of the sector. Both events were exogenous effects to the sector, and individual firms had no say in the determination of the pay levels of the different groups defined in the guidelines. Two dummy variables are used as instruments to indicate whether an executive was hired in the period before 2000, between 2000 and July 2004, or afterwards.

Generally, valid instruments should satisfy two conditions: instrument exogeneity, and instrument relevance. Changes in remuneration policies can serve as effective instruments, because they immediately affect the pay of newly hired executives, yet do not directly influ-ence total costs (instrument exogeneity). The exclusion restriction rules out any direct effect of the instrument on the dependent variable: in this case, I do not expect that a change in remuneration policy directly affect totals costs.

No reverse effects of the dependent variable on the instrument are expected: costs in 2008-2012 cannot have an effect on the pay regime that holds for the director when (s)he is appointed. The only way reverse effects may occur is if the director is fired due to high costs during 2007-and the date of getting laid off. The changes in remuneration policies are expected to be highly correlated with the endogenous regressor executive pay, controlling for the exogenous regressors and control variables (instrument relevance). Indeed, the instru-ments its correlation coefficients indeed suggest no relation between the dummy variables and the independent variables (0.0052 < 𝜌 < 0.0937).

3.4 Findings

3.4.1 Descriptive Statistics

In Table 3.2, the number of observations, mean, standard deviation and minimum and max-imum of the data is reported. Excluded in the regressions are the approximately 200 ob-servations of corporations that have zero total costs. Almost all of these obob-servations are from 2011, so it seems like the observations that are missing are a random selection. BZK

(31)

Table 3.2: Summary statistics

obs mean sd min max

log(TVC) 1047 9.300 1.396 3.135 12.515 log(TOC) 1046 9.684 1.464 3.332 13.209 log(TOIC) 1047 9.981 1.417 3.332 13.475 log(q) 1047 8.163 1.311 1.386 11.260 log(ttr) 1047 2.000 0.374 0 4.544 log(hv) 1047 4.887 0.491 0 5.255 log(urban) 1047 11.375 1.795 0 17.738

starting date before 2000 937 0.314 0.464 0 1

starting date 2000-2004 937 0.197 0.398 0 1

starting date 2004-2012 937 0.489 0.500 0 1

Source: BZK Data, www.leefbarometer.nl.

confirmed that for these firms, their consolidated financial statements seem incorrect, and they are still trying to find out why that is. In general, corporations with on average 3000 dwellings are in my view very unlikely to have no maintenance costs, labor costs or other costs to maintain, refurbish and possibly develop their real estate. Hence these observations are excluded from the empirical base model and the instrumental variable models.

Instruments

Several remuneration policies were instated after the grossing and balancing operation. Before 2000, executive pay was based on the housing collective labor agreement. Between 2000 and 2004, the counsel of the Peters Committee was the leading policy used to determine the pay of newly hired executives. In 2004, the Izeboud Committee provided guidelines to determine pay, which were immediately implemented by 70% of the Supervisory Boards when they appointed a new executive. The policies in place at the moment a new executive is appointed determine the terms of employment for the new executive. These terms indicate a base scale of executive pay for its (future) employee. Table 3.3 illustrates the differences in pay when being installed in certain periods. Over time the average base pay increases, yet differences are found between the different remuneration policies in place when the executive is hired. The three groups are roughly comparable, Their home values, tenant turnover rate and urbanization scores are very similar. The biggest differences are found in size: the group ‘Before 2000’ is a bit smaller in terms of size (and consequently in terms of revenues and expenses).

(32)

Table 3.3: Average total cash by executive starting date

Starting date Before 2000 January 2000 - July 2004

-July 2004 December 2012 All executives

𝑁 415 263 608 1286 2009 156.9 157.8 153.1 155.4 2010 162.6 3.60% 156.8 -0.62% 153.8 0.44% 157.5 1.32% 2011 164.9 1.45% 158.0 0.78% 150.6 -2.07% 156.5 -0.63% 2012 169.5 2.76% 163.7 3.60% 154.3 2.41% 160.5 2.53% Average 163.1 159.0 152.9 157.4 Dwellings 5290 6328 6705 6119 st. dev. 7708 9190 10410 9278

Source: BZK Data. The percentages mentioned next to the average executive base pay represent the change in pay between year t and year t-1.

3.4.2 Results

Table 3.4 illustrates the outcome of the OLS and IV methods and its log-log specification. The first stage of the fixed effect IV regressions can be found in Appendix A, Table A.1. The results from the ordinary least squares method suggest that a ten percent increase in executive pay will add .2 percent to the total variable costs of a housing corporation (Specification (1), significant at a 10 percent level). In other words, pay your executive more leads in this

model to higher costs. The average total cash pay is e144,027. A ten percent increase in

executive pay would result in a e631,000 increase in TVC (added to an average TVC of

e10.938 million).

As expected, owning more dwellings results in higher costs. This is significant at the 1 percent level for the base specification, as well as all other robustness checks in Table 3.4 (Model 1-6). Surprisingly, a higher average tenant turnover rate reduces total costs in the base empirical model. The effect of the turnover rate is approximately the same for all OLS models: a ten percent increase reduces total costs with 0.01-0.08 percent. Surprisingly, the higher the quality of the real estate is, the higher the costs are. Possibly, the costs for developing and maintaining higher quality real estate continue to be higher. Finally, operating in more urban environments has a varying effect on total costs.

Thus, the OLS model reports that pay positively affects total costs. If one would draw immediate conclusions from this model, moderating pay would have no deteriorating effect on the performance of the corporations. Robustness checks, presented in Table 3.4, specification (3)-(6), show that the initial results hold for the different definitions of total costs, despite the varying significance of executive pay. In addition, the conclusions drawn above retain its

(33)

Table 3.4: Regression results cost model TVC TOC TOIC (1) (2) (3) (4) (5) (6) log(total cash) 0.0510* 0.0493 0.0513 0.553 0.0476 0.390 (0.0283) (0.246) (0.0713) (0.671) (0.0465) (0.431) log(q) 0.196*** 0.163*** 0.326*** 0.286*** 0.270*** 0.228*** (0.0344) (0.0356) (0.0866) (0.0971) (0.0564) (0.0624) log(ttr) -0.0101 0.00308 -0.0882 -0.114 -0.0506 -0.0583 (0.0321) (0.0370) (0.0809) (0.101) (0.0527) (0.0648) log(hv) 0.195 0.217 0.278 0.213 0.142 0.0614 (0.175) (0.189) (0.440) (0.519) (0.286) (0.331) log(urban) -0.00557 -0.00684 0.00850 0.0392 0.00414 0.0260 (0.00760) (0.0147) (0.0191) (0.0402) (0.0125) (0.0258)

Year fixed effects Yes Yes Yes Yes Yes Yes

Firm fixed effects Yes Yes Yes Yes Yes Yes

Model OLS Yes No Yes No Yes No

ModelIV No Yes No Yes No Yes

𝑅2 (overall) 0.846 0.891 0.821 0.738 0.905 0.819

𝑁 1047 937 1046 936 1047 937

Notes: Standard errors in parentheses. Significance at the 10%, 5%, and 1% levels is indicated by

*_, **_{, and}***_{, respectively. Source: BZK Data.}

significance when more control variables are added. Examples are when the liveability of the neighborhood, the percentage of homes allocated to special attention groups or the number of newly vacant homes allotted are added to the equation. These more disaggregated models can be found in Appendix A, Table A.3.

A somewhat different picture emerges from the robustness checks using using the instru-mental variable method (column 2, 4, and 6 in Table 3.4). Even though executive pay does not significantly influence total costs, this method is in this case superior to OLS. The va-lidity of the instruments is confirmed by the Davidson-MacKinnon test of exogeneity: the OLS regression is indeed endogenous in Models 1, 3, and 5. The starting date of executives is significant on a 5% and 1% level.

Notably is that next to size, no other variables significantly influence total costs. Most importantly, pay does not have a significant effect on costs when the IV method is employed (see model (2) as the base empirical specification). The possible interpretation of this regres-sion is that the total variable costs are insensitive to the type of compensation scheme of the executive.

(34)

One of the shortcomings of the used instruments is the potential reverse effect of perfor-mance on pay. If the former executive (financial) blunders and is fired, the future executive may get paid more to reorganize the corporation and repair the mistakes made. In this case, there would be reverse effect of cost efficiency on pay. However, it appears that getting fired in the social housing sector is a relatively rare phenomenon. To the best of my knowledge, only four executives were fired in the last few years (Vestia, Woonbron, Rochdale, WSG). In other words, it appears that the reverse effect of perforamnce of pay is limited, even if it may be prevalent. The results of Table 3.4 remain similar if the corporations in questions are removed from the regression analysis.

The insensitivity of total costs to executive pay in the instrumental variable method would imply that the moderation of executive pay in the social housing sector would have no dete-riorating effect on the cost efficiency of housing corporations. In other words, executive pay (and other managerial abilities associated with executive pay) does not translate into lower costs. The question that is left unanswered by this Chapter is what the actual determinants of executive pay are. If pay-performance sensitivity does not play a role, are there other firm characteristics or managerial abilities that do explain executive pay? Chapter 4 aims to provide an answer to that question.

(35)

Chapter 4 Executive pay

Economic theory argues that financial incentives can help to reduce the principal-agent prob-lem: monetary compensation can help to align the interests of the executive (the agent) with the interests of the (local) government (principal) (Laffont and Martimort, 2002). A better understanding of the dynamics of social housing labor market would yield important insights regarding the determinants of the remuneration policies of directors.

In contrast to the for-profit sector, estimating the determinants of executive pay in the nonprofit sector creates various complications: the conventional performance measures, such as profits or firm value, are not applicable to nonprofit organizations. Nonprofits do not maximize profits: rather, they typically aim to maximize output, conditional on its revenues outweighing costs. The output of nonprofits are mostly hard-to-evaluate goods, which cannot be measured in terms of stock prices, profits or enterprise value; conventional performance indicators in the for-profit sector. Instead, they aim to fulfill their missions in response to demands from stakeholders such as clients or the (tax-paying) public at large.

The Dutch social housing sector is an interesting case to study: the sector has resem-blances with both a for-profit enterprise and a nonprofit organization. In practice, the ex-ecutive pay in the Dutch social housing sector is low compared to highest pay levels in the health care sector. In comparison to the educational institutions, and public officials, the top directors of housing corporations receive high pay levels. Table 4.1 illustrates the highest base pays in different non-profit sectors in 2011.

This chapter will firstly describe the determinants of executive pay found in the literature. Second, the wage differential of executives of nonprofit and profit sector will be discussed. Third, the empirical strategy and methodology of the identification of the determinants are discussed. Finally, the findings are presented and concluding remarks are offered.

High executive pay may not pay off : case study of the Dutch social housing sector

Thesis submitted in partial fulfilment of the requirements

for the MSc Economics degree