A study of the relation between risk allocation and profits for corporate pension funds: Is the riskiness of the pension plan affected by the profitability of the sponsor?

(1)

1 A study of the relation between risk allocation and profits for corporate pension funds: Is

the riskiness of the pension plan affected by the profitability of the sponsor?

Abstract

This thesis researches how profits of the sponsor, proxied by profits in the corporate sector, influence the allocation of risky assets in pension funds. A dataset for the period 2000-2014 is constructed which consists of 230543 observations on 84907 corporate pension plans. The results from this study find that there is a relationship between the risky asset allocation of the pension funds and the profitability of the corporate sector and that pension funds allocate 0.0032 percentage points more towards risky assets if the corporate sponsor makes a profit. This outcome is estimated in a standard regression and with an estimated coefficient of 0.284 for a regression with the profitability of the corporate sector expressed in log. Both estimates were statistically significant at a 5 percent level.

Keywords:

Risky assets; corporate profits; funding ratio; asset allocation; pension funds

By: Max ten Velden Supervisor: Dr. W.E. Romp

Student ID: 5809398 Co-reader: Dr. C.A. Stoltenberg

October 2016 Faculty of Economics and Business

(2)

2 Verklaring eigen werk

Hierbij verklaar ik, Maximiliaan Jules ten Velden, dat ik deze scriptie zelf geschreven heb en dat ik de volledige verantwoordelijkheid op me neem voor de inhoud ervan.

Ik bevestig dat de tekst en het werk dat in deze scriptie gepresenteerd wordt origineel is en dat ik geen gebruik heb gemaakt van andere bronnen dan die welke in de tekst en in de referenties worden genoemd.

De Faculteit Economie en Bedrijfskunde is alleen verantwoordelijk voor de begeleiding tot het inleveren van de scriptie, niet voor de inhoud.

(3)

3

1. Introduction

The Funding Ratio of pension schemes in the U.S. fell from 123% to 82% in 2002, because both the equity market crashed and the bond yields dropped at the same time (Leibowitz, Stanley and Ilmanen, 2016). After this event, U.S. pension funds shifted to riskier investment strategies in order to maintain high discount rates and to present lower liabilities (Andonov and Cremers, 2012). Consequently this contributed to the financial uncertainty of pension funds and left most Defined Benefit funds underfunded.

To create a better understanding about the risk taking behaviour of pension funds, this thesis researches how profits of the sponsor, proxied by profits in the corporate sector, influence the allocation of risky assets in pension funds over the period from 2000 to 2014. Therefore the following hypothesis will be answered: Is the riskiness of the pension plan affected by the profitability of the sponsor?

As a consequence of the developments in 2002, doubts among plan sponsors and investors have been raised about the financial sustainability of Defined Benefit (DB) pension schemes and caused a shift from Defined Benefit to Defined Contribution (DC) schemes since the past few years. DC pension schemes are associated with more stable contributions and therefore with more securities for the corporate site (Broadbent, Palumbo and Woodman, 2006). This association has to do with the difference in working mechanism of the two types of schemes.

DB pension schemes promise a retirement benefit that depends on the employee’s earnings history, tenure of service and age. The benefit is determined by a formula that takes this information into account and calculates the optimal financial path (Bodie, 1990). Therefore the level of the contributions of DB plans are driven by the need to balance the plan’s assets and liabilities (Dvorak, 2012). When the value of the assets fall relatively to the value of the liabilities, the sponsor of a DB plan has the obligation to make this up with additional contributions. Consequently, the sponsors bear the risk of the funding and the investments.

Alternatively, contributions in DC plans are tied to employee’s salaries or their profits. Under a standard 401(k) plan, a fixed percentage of the employee’s salary is contributed and accordingly matched by the employers up to a certain percentage (Dvorak, 2012). These contributions are made on a regular basis and are used to make investment earnings, which make DC plans in effect tax-deferred retirement savings accounts held in trust for the employees (Bodie, 1990).

(5)

5

However, DC schemes are not 100% safe either. Because of the investment earnings DC schemes use to provide in the retirement payments, DC schemes are more dependent on market returns. Therefore retirees face more insecurity with regard to their pensions and the risk of the DC pension scheme is thus more with the individual employees than with the corporations.

The difference in risk between DB and DC schemes also has effect on the risk taking behaviour, which is described in economic theory. If a pension plan is safe and well-funded there would be room to invest in more risky securities. And if a plan is underfunded, it would invest in less risky assets (Mohan and Zhang, 2014). Increases in pension risk also increase the volatility in corporate finances, which are part of risk budgeting. Therefore risky assets are a part of pension plans and the investments.

When looking at how pension funds balance their portfolios and evaluate risks, it is important to look at how the market regards corporations funding decisions. According to empirical investigations of how pension funds adjust their portfolios over time, it seems that past investments returns are important drivers of pension funds’ investment policy. Rauh, 2009, finds that a one-year lagged investment return of corporate pension funds in the U.S. is positively correlated with the next time period investment in equity. And pension funds strongly rebalance their portfolios to counteract the impact of return on their portfolios (Bams, Schotman & Tyagi, 2016).

The results produced by this study, as a result from the econometric model, statistically confirm the hypothesis. This means that there seems to be a relationship between the risky asset allocation of the pension funds and the profitability of the corporate sector, where pension funds allocate assets with 0.0032 percentage points towards risky assets if the corporate sponsor makes a profit.

This study makes the following contributions to the literature. First, this study makes a link between the funding status of a pension scheme and the allocation of risky assets based on the profitability of the sponsors. Secondly, the study uses panel data analysis for the period 2000-2014. The added value to the field is that most researchers use cross-section data. Furthermore, I will use data from the Form 55000, presented by the U.S. Department of Labor (DOL), and I will use data on the profits of the corporate business sectors, provided by the Bureau of Economic Analysis, to serve as a proxy for the profitability of the sponsors. The Form 5500 provides a large sample of over 7000 corporate

(6)

6

pension plans and the data of the Bureau of Economic Analysis provides the profits of the corporate branches in the U.S. of the past 20 years.

The remainder of this thesis will be structured as follows: In chapter 2 the literature on the research subject will be reviewed. In chapter 3 the methodology will be explained. Chapter 4 will show the dataset and the empirical results will be shown in chapter 5. Chapter 6 and 7 will be reserved for the discussion and the conclusion.

(7)

7

2. Status of the Literature

This section reflects the literature that describes characteristics of the corporate pension market of the United States. It begins with an overview of general characteristics of pension plans. Next the literature about different perspectives between sponsors and pension schemes is described. And the last part reviews the literature about the determinants of risk behaviour of pension funds. The purpose of this section is to get an overview of all the useful information about pension funds and plan sponsors, and to get an overview of which risk determinants are relevant for this research.

2.1 General Characteristics of Corporate Pension Plans

As stated in the introduction, pension funds have had problems with their funding levels since 2002. A pension fund is defined as fully funded when the current value of the financial assets is equal to the present value of the liabilities aggregated across all scheme members. In a typical pension scheme, pension funds generally choose to run a mismatch risk, which means that future surpluses and deficits will occur. Therefore there is an asymmetric allocation of residual risk within this system between sponsors and retirees (Exley et al., 1997). The asymmetric allocation causes a difference between the borne risks and it is not always clear who bears the investment risk and who does not. Therefore the mismatch risk leads to a situation wherein pension fund trustees have to accept the possibility of volatility in the financial conditions. This also affects the sponsor, which is supported by considerable evidence (Bartram, 2016).

Findings in the literature show that the funding level of the DB pension plan is reflected in the market value of the sponsor (Exley et al., 1997). From there it follows that the market risk of the equity of the sponsor reflects the risk level of the pension plan. According to Pfau (2009) this could be a result from the cases where there is a funding deficit and the sponsor may have had the legal or moral obligation to increase contributions.

Another characteristic of pension schemes is the lifecycle asset allocation. The strategies involved with this asset allocation are counterproductive to the retirement saving goals of individual investors. The typical portfolio grows the most in the years just before the retirement, when lifecycle funds have already switched to a more conservative asset allocation (Pfau). There has been developed a solution to this problem, called Target-Date-Fund (TDF’s). These funds start with

(8)

8

allocating a high proportion of assets to risky investments during the early period of the pension scheme and then shift to more conservative assets along the way. However, because of the financial crisis, there has been increasing critique on TDF’s by supervisors. The risk exists that when a pension schemes switches too late to a TDF, which has more conservative assets, investors miss their opportunity for capital gains because the value of the schemes assets decreases in comparison to the return of investment.

2.1.1 Defined Contribution Plans

In a DC pension plan, workers accrue funds in individual accounts administered by the plan sponsor. Commonly, pension pay-outs in a DC pension plan are in a structure where the employers make a stated contribution on a regular basis to each of their covered employees. This could be, for example, an annual amount of 3% of their salary (CAIA, 2013). So, the pension benefit accumulated during the employee’s working career depends on the contributions made while working and the investment returns earned on the plan balances (Broadbent, Palumbo and Woodman). This means that the retirement income that will be provided with DC pension schemes is unknown in advance, since the investment returns are unknown. Under pension legislation, DC plan assets belong to the worker. This means previous contributions are portable across employers or through employment spells: a worker may leave the plan assets under the administration of a previous employer, transfer the assets to a new employer’s plan or transfer the assets to an individual retirement savings account (Broadbent, Palumbo and Woodman). This portability clearly stimulates employees who work multiple jobs in a career, and it is more beneficial for employees who work for firms that may not have the financial strength to pay long-term pension benefits in the amount promised. When leaving an employer, the employee is able to roll over the balance in the DC plan into the plan offered by the next employer or into an individual retirement account.

Given that DC plans are personal accounts, the employee contribution, investment gains and vested portion of employer contributions can be given to the employee’s heirs should the employee die before retirement (Pension Fund Portfolio Management, CAIA Association, 2013). Another feature of a DC pension plan is that, by definition, it is always fully funded and the employer typically has no financial obligation other than to make periodic payments into the plan. Also, relative to a DB plan, the accounting treatment is quite simple: payments to a DC plan are treated as any other corporate expense (Yermo, 2003).

(9)

9 2.1.2 Defined Benefit Plans

A traditional Defined Benefit (DB) plan is based on a formula linked to the salary of an employee, from which regular payments are promised to employees from the date of their retirement until they die. This in effect links the pension benefits to the employees’ salary and ensures a continuing pay after they retire to ensure a certain standard of living. Both employers and employees make contributions to the DB pension plan, but it is the employer’s duty to make additional payments to cover the risk if the scheme is underfunded. Therefore the employer faces the risk that the pension benefits costs may exceed the initial predictions, if the retiree lives longer than expected. Also, the DB plan has different types of risks towards employers and employees. As said, employees face longevity risk of their employees, but the employees face inflation risk and accrual risk. Any changes towards the final years of labour can result in accrual benefits falling short of the employee’s expectation (Wiegmann, 2012).

2.1.3 Hybrid Plans

Next to the standard DB and DC pension plans, there also exist in between forms called Hybrid Pension Plans. These plans are a combination of both DB and DC plans. The mechanism of hybrid pension plans is that they use the DB plans tax arrangements and the regulations for these plans, mostly for the benefits these provide. An example is the cash-balance plan, where participants hold national accounts that build annual pay credits plus interest. The pension benefit is then expressed as a lump sum amount and can be resolved whenever is optimal. Either with retirement, or if the plan just has to be terminated or when a worker changes employers. This is much like DC pension plans (Broadbent, Palumbo and Woodman).

2.2 Pension Schemes and Sponsors

According to the Corporate View, pension assets are part of the firm’s assets. This means that any surplus or deficit of a plan belongs to the firms shareholders (Bodie et al., 1985). This is shared by the modern corporate finance theory that states that there are a couple of core activities for corporations, for which one is that corporations are pass-through entities and mostly are managed to maximize shareholder value. The value of the corporation is what matters most to the shareholders and their vision on the corporation determines everything.

(10)

10

An important consequence of the corporate view, with regard to funding status of pension funds, is that it views that pension deficits are also a form of corporate debt. To explore the effect of a firm’s pension deficit on its share price a study was performed by Feldstein and Seligman’s (1981) and with the use of a sample of 200 U.S. firms they showed that the deficit is rapidly incorporated into the share price. This suggests that the unfunded pension liability is being recognised by shareholders as equivalent to corporate debt. Black (1980) emphasises that a firm’s pension fund is legally separate from the firm but, because pension benefits are normally independent of fund performance, pension assets affect the firm as though they were firm assets. From multiple studies it also becomes clear that the relative greater volatility of pension debts have a greater impact on the credit rating of corporations than other long-term corporate debts ( Mckillop & Pogue, 2009).

The whole ‘pension deficit is a corporate debt’ analysis adds an interesting dimension to the theory of which factors are involved with the risk taking behaviour of corporations and pension funds. Another research on the subject of risk-management in corporations, is the research done by Anantharaman & Lee, 2014. They found that managerial risk aversion could be a possible explanation for the absence of risk-shifting in financially distressed firms. They concluded that stockholder-controlled firms have greater underfunded pension plans and invest a greater proportion of pension assets in risky classes.

2.3 Risk Determinants of Pension Schemes.

In the analysis up to this point, some factors of risk for pension schemes have been put forward. In this part the analysis will continue with a focus on what the empirical evidence determines what the risk or risk taking behaviour in pension schemes is. Jin (2006) has examined if the systemic equity risk of 4,500 U.S. firms from the years 1993-1998 reflected the risk of their pension plans. In the article it was found that the equity beta’s of firms indeed do appear to accurately reflect the beta’s of their pension assets and they also appear to accurately reflect the beta’s of the liabilities. It was also found that the cost-of-capital calculations, used internationally in corporate finance, didn’t distinguish between the operating-asset risk and the pension plan risk.

Other determinants of risk are perhaps more plain, but also important to distinguish. The following definitions are given by the CAIA association in their book, CAIA level II. First there is the mortality risk, which is defined as the risk that someone dies at a certain age. This risk is highly uncertain for

(11)

11

individual investors but can be more predictable when it is averaged over a large number of people, e.g. the members of a pension plan. Longer lifetimes are associated with larger retirement assets. Then there is longevity risk, the risk that an individual will live longer than anticipated. This risk is heterogeneous, since it affects different investors in different ways. Pension plans bear the risk that lifetimes sometimes will be longer than anticipated, which requires them to have a larger number of monthly benefit payments or months of retirement spending (Pension Fund Portfolio Management, CAIA Association, 2013).

Secondly, as mentioned, Defined benefit (DB) plans provide a guaranteed income to retirees. This can be risky for employers. In a defined benefit plan, the employer takes all of the investment risk while offering a guaranteed, formulaic benefit to retirees (Pension Fund Portfolio Management, CAIA Association, 2013). Thirdly, there is the surplus risk. The surplus of a pension plan is the amount of assets in excess of the projected benefits, which are the present value of the benefits assumed to be paid to all future retirees of the firm (Pension Fund Portfolio Management, CAIA Association, 2013).

Another determinant of risk is the agency problem. Pension plan sponsors have conflicting goals when designing the asset allocation of the plan. The first goal is to earn a high return on pension assets, which will be used to reduce the employer’s long-term contributions required to fund employee benefits. The second goal is to minimize the degree of underfunding or the amount of surplus risk incurred in the plan (Pension Fund Portfolio Management, CAIA Association, 2013). But also the employees themselves pose a risk. Employees do not generally allocate DC plan assets in the same careful way that professional managers allocate DB plan assets. Employees often invest in just a single fund, resulting in a 100% equity or 100% cash allocation, or they diversify contributions equally across all investment choices. Employees also do not rebalance frequently or change allocations when their investment needs become more conservative as retirement approaches. This lack of rebalancing results in a drifting asset allocation, where the highest-returning asset classes grow as a share of the portfolio. Due to a lack of investor sophistication as well as regulatory restrictions, most DC plan participants do not have the ability to directly invest in alternative investments. When alternative investment choices are offered in DC plans, they typically focus on commodities or real estate. It is quite rare for individual employees to be able to invest in private equity or hedge funds through DC plans (Pension Fund Portfolio Management,

(12)

12

CAIA Association, 2013). This means that the focus of the DC plans’ investments should be on equity or cash allocations, since these will be the dominant drivers of the portfolio’s.

2.4 Possible Other Determinants of the Risk Portfolio

There are also determinants of the risk portfolio that are not that straightforward. From the literature it becomes clear that one way to maximize value is by using profitable tax arrangements. Tax effects are mentioned as a potential major influence on how a firm delivers and on how it copes with the pension plan. In line with the law, firms can earn pre-tax rates of return on any assets that they hold in their pension funds and they can pass these returns on to the shareholders. If the fund is invested in only the most heavily taxed assets then it maximizes tax savings, which means that pension funds’ investments should exist entirely out of taxable bonds rather than common stock or other assets (Bodie et al., 1985). However, there is a timing difference, since the employees’ and employers’ contributions are not taxed in the year they are made and the portfolio’s gains are not taxed in the year they are earned. The taxes are paid when the employees retire and withdraws the assets (Pension Fund Portfolio Management, CAIA Association, 2013). Firms maximize tax-benefits by overfunding and investing in taxable bond, but this also shifts their relative weight in investment to risky assets (Bodie et al., 1985). This could be a problem if a firm is not profitable.

(13)

13

3.

Data and Sample

3.1 Form 5500 and the Bureau of Economic Analysis

To answer the hypothesis of this study, two sources will be used for the implementation of the data. The first source is a sample of the corporate pension plans of the Form 5500 Private Pension Plan Research Filings, provided by the Department of Labor (DOL). These filings contain datasets ranging from 2000 to 2015. All pension sponsors must file the IRS 5500 form on an annual basis and are required by law to hand over information about their financial and actuarial status, about active participants, contributions and assets of retirement benefit plans. The filings also consist of many schedules and attachments, which are made available electronically. Because of the extensive amount of data, the starting point of this thesis is a dataset used by Borst (2015), which is up to date for the years 2000-2007. In this research the dataset of Borst is combined with the remaining necessary data from the website of the Form 5500 to complete the dataset up till the year 2014. The other dataset that will be used for answering the hypothesis, is the analyses of corporate profits from the Bureau of Economic Analysis. This source contains information about all the profits per industry of the private sector in the U.S. by year, measured in billions of Dollars and indexed for inflation1_,

and ranges from 1998 to 2014. The profits per industry will be used as a proxy for the profitability of the corporate sector.

The data from the Form 5500 consists of the total active participants, total contributions, the business codes, the type of plan, the liabilities, the assets and current status of the plans. The forms used in this study to filter the data are the Form 5500 general schedules and some of the attached schedules; SB, MB and H (http://www.dol/gov/ebsa/foia/foia-5500.html). The general actuarial information, the plan type, tax information, the number of participants and the amount of contributions are all taken from the general form. Information about debt and asset allocation is taken from Schedules MB and SB. Schedule H provides the financial information concerning the earnings on investments, expenses of the plan and the total assets and liabilities of the plan.

1_{On the website of the BEA it states that the GDP-by-industry accounts are measured in both current and}

(14)

14 3.2 Data Description

In order to produce a workable dataset to resolve the hypothesis it was necessary to assemble different sources of information to the dataset from Borst (2015), which was used as a starting point. Added to this database were all the general files, merged with the SB, MB and H schedules of the same year, from 2008 up to 2014. This created a dataset which contained all the corporate pension plans known to the DOL, from 2000 to 2014, with their unique identity numbers, their financial information, information about the participants, the business sectors, total assets and total liabilities. After these merges, useless observations were removed in order to have a small and comprehendible dataset. The variables that were kept are the before mentioned in this paragraph.

Based on instructions from the site of the DOL, an unique ID-code was constructed, called einpn. This ID is based on the ‘ein’ and ‘pn’ variables in the forms and combined together they form a 12-digit unique number, by which the plans can be identified. The variable ‘Total INVESTMENTS funds’, based on the article of Kisser, Kiff and Soto (2015), was used to put all the financial information from the forms together in one place. The variable consists of all the financial and actuarial information of the plan (e.g. common stock, real estate and corporate debt) and therefore summarizes what the total accrued value of the plans’ financials are. This variable was then used to determine the risk taking behaviour of the pension plans, by creating the variable ‘risk’ (see 3.3). Other variables that were created to make the dataset useful for the research, are variables describing the different private sectors e.g. Financial or Durable Goods and the profitabilities per private sector.

Using the insights of Bikker, Knaap and Romp (2014), again in line with ten Dam (2015) and Borst (2015), small pension plans which have less than 125 participants were dropped since these could be used as tax vehicles. Also excluded are observations with a return on investments under the -100% or over the 50%, in line with Borst (2015). This is because those observations can be caused by unrealistic low or high unrealized returns. In the same way observations with allocated risky assets under 0% or over 100% are dropped. Plans without a business code, with no information about the share of active participants and no information about the plan size were also dropped. The variable year is defined as time, ranging from 2000 to 2014. Business code is defined as the unique code that signifies the business sector a plan belongs to. For example, all business codes with 52 in the first two digits of the business code number belong to the financial sector. The number of total

(15)

15

participants are defined as the number of plans’ members who are still working. Total assets and total liabilities are linearly and in Dollars. The number of total assets being 0 or (-) 100 do exist. Risk is defined below, in chapter 3.3.

Furthermore the data of the profitability, provided by the bureau of Economic Analysis, is added to the dataset as well. It consists of the profits of the corporate sectors from the years 2000 to 2014 and is expressed in billions of dollars. The data is accrued per business sector to create one variable, called ‘profitability corpsector’, which contains all the corporate profits over the years in one variable. The final sample of plan-level data consists of 230543 observations on 84907 pension plans, based on the unique sponsors’ EIN and PN (einpn). Further there is a mean of 3254 participants for the total of pension funds. The average amount of assets is $ 5.52 * 10^7, whereas the average amount of liabilities is $ 1.15 *10^6. The mean for risk is 91.32 and the average of the Total Investments is $ 1.55 * 10^7.

Table 1: Summary of division of years across the dataset

Year Freq. Percent. Cum.

2000 13,264 5.75 5.75 2001 11,032 4.79 10.54 2002 10,896 4.73 15.26 2003 12,979 5.63 20.89 2004 13,314 5.73 26.62 2005 12,887 5.59 31.21 2006 12,653 5.49 37.70 2007 12,212 5.30 43.00 2008 26,511 11.50 54.50 2009 25,681 11.14 65.64 2010 14,416 6.25 71.89 2011 4,712 2.04 73.93 2012 1,994 0.86 74.80 2013 4,124 1.79 76.59 2014 53,979 23.41 100.00 Total 230,543 100.00 100.00

(16)

16 3.3 Variable Determination: Focus on Business Sector and Risky Assets

To conduct the research and to combine the datasets in a proper fashion, the data was divided in line with the largest business sectors. In accordance with the Form 5500, which describes the nature of the plan sponsor’s business, their classification codes were used to organize the data. Using the instruction file on the website of the DOL it is possible to identify the sponsors by the largest business sector and from there to divide them into 14 different industries.

Table 2: Variable construction of business sectors and their frequency in the dataset

Business Sector Code Frequency of funds over the years (expressed in thousands)

Utilities 22 2.549

Manufacturing 31; 33 68.845

Petroleum and coal products

324 0.861

Machinery 333 6.186

Computer and Electronic products

334 3.903

Durable goods 423 17.810

Nondurable goods 424 9.779

Motor vehicles, bodies and trailers, and parts

441 10.077

Food and beverage and tobacco products 445 3.171 Retail trade 44; 45 33.643 Transportation and warehousing 48; 49 10.651 Information 51 10.242 Financial 52 38.861

(17)

17

Since the hypothesis of this research focusses on the effect of the profitability on risk of pension plans, risk needs to be determined as a variable. To determine the riskiness of the pensions’ assets, the most convenient way would be to look at the volatility of the pension funds portfolio’s. But, as Borst (2015) points out, the Form 5500 does not provide this data. Therefore, the same line of reasoning in this research is used to measure the riskiness of the asset allocation as is put forward by Kisser, Kiff and Soto (2014):

Risk

=

(1−cash− Accounts receivable− US treasuries Corporate Debt)

(18)

18

4.

Methodology

This chapter specifies the methodological framework that was used to answer the main question of this study and describes the features of this framework, the used variables and the guiding direction of the literature.

4.1 The Econometric Model: Regression Baseline Definition

The main question of this study is whether the profitability of the business sector has an effect on the riskiness of pension plans. Therefore the following regression is estimated, with riskiness as the explained variable and profitability of the sector as the explanatory variable:

Risk

it

=

β

0

+

β

1

Profitability

i

+

β

2

Active Participants

it

+

β

4

Assets

it

+

β

5

Ratio Retired

Participants/Active Participants

it

+

β

6

Funding Ratio

it

+

ε

it

In the specification the dependent variable is the riskiness of the allocated assets of the pension funds and the independent variable is the profitability of the pension plans, expressed as the profitability of the corporations which serve as a proxy for the pension plans. Also included are, in line with Borst (2015), controls for the plan size measured as the assets at the beginning of the year and the share of active participants. And

ε

itdenotes the error term.

Additionally, it has become clear from the literature that both the age of the participants as the years that a pension scheme exist have an effect on the risk of the portfolio. The belief that younger individuals can assume greater risk than older individuals and therefore should invest more in risky assets like stocks have been generally well accepted by DC plan providers so much, that it now serves as an important investment principle anchoring most participant education programs in United States (Basu, 2007). Therefore, the ratio of retired members versus active members has been taken into account for the estimated regression to serve as a proxy for their influence on the risk of the pension fund. And an additional control variable for the funding ratio is submitted in the regression to check if there is influence on the risk allocation, where the funding ratio is defined as the ratio of assets over liabilities.

(19)

19

5. Results

In this section the empirical findings from the regressions are presented and analysed. The main results from the regressions are summarized at the end of next page.

5.1 Regression Results

Table 3: Regression results for the 2000-2014 period Regression results for the 2000-2014 period

Coefficient Std. Er. t P>|t| Profitability corporate sector 0.0032 0.000969 3.25 0.001 Total Active Participants -0.0000074 0.000034 -0.22 0.827 Total Assets 0.0000000022 0.0000000016 1.35 0.178 Ratio Active/Retired Participants 0.000049 0.000036 1.36 0.174

Funding Ratio 1.45e-07 1.42e-07 1.02 0.308

Regression results for the 2000-2014 period, with robustness

Coefficient Std. Err. t P>|t| Profitability corporate sector 0.0032 0.00095 3.30 0.001 Total Active Participants -0.00000074 0.000001 -0.73 0.463 Total Assets 0.0000000022 0.00000000085 2.58 0.10 Ratio Active/Retired Participants 0.000049 0.0000172 2.82 0.005 Funding Ratio 0.00000015 0.000000042 3.42 0.001

Regression results for period 2000-2014, with corporate profits in log

Coefficient Std. Err. t P>|t| Profitability corporate sector 0.284 0.1484 1.92 0.0055 Total Active Participants 0.000013 0.00002 0.68 0.469 Total Assets 0.00000000019 0.0000000084 2.23 0.026 Ratio Active/Retired Participants 0.000033 0.000027 1.22 0.000 Funding Ratio 0.000000013 0.00000035 3.55 0.000

(20)

20 5.2 Analysis of the Results for the Allocation of Risky Assets

In table 3 column 1 the estimated coefficient β1 for the influence on the allocation of risky assets of

pension plans has the expected positive sign and is statistically significant at a 5 percent level. This indicates that pension plans belonging to firms that have higher profits are associated with a higher allocation towards risky assets.

The comparison between the different regression results shows that for a standard regression the outcome indicates that pension plans have a 0.0032 percentage point growth towards risky allocations when the sponsor makes a profit (with variables for the profitability of the corporate sector, the number of participants ,the amount of assets of the pension funds, ratio active over retired and the funding ratio). The other variables in the first regression did have the expected signs, but were not statistically relevant. This outcome does change when there is robustness included to check for heteroscedasticity, as presented with the regression results with robustness. Furthermore, the effect of total active participants in the first regression is negative, but also very small and not statistically significant. The effect of total assets seems to be positive, but also very small. However, the coefficients for the ratio active over retired participants and for the funding ratio in the first regression are not statistically significant, which is a clear indication that the first simple regression is biased. For the regression with robustness the outcome differs and the signs of the coefficients for profitability of the corporate sector, the ratio active/retired participants and the funding ratio, are as expected and now all statistically significant.

Additionally a regression was estimated where the profits of the corporate sector are expressed in log , to check if it matters for the result of the regression that the profits of the corporate sector are expressed in billions of dollars or as a log of billions of dollars. The coefficient for the profitability of the corporate sector is still positive and statistically relevant. And with 0.284 percentage points towards the allocation of risky assets if the sponsor makes a profit.

(21)

21

6. Discussion

The results of the research done in this thesis, as discussed in the previous chapter, seem to be clear and have the expected outcome. There are still a lot of other factors however that are not taken into consideration, that could have influenced the acquired result.

First an evaluation of the regression results. The variable ‘active participants’ was taken up in the first estimated regression, but it’s coefficient was very small and not significant. This is also the case for the coefficients for the ratio active over retired participants and the funding ratio. This outcome indicates that there is omitted variable bias in the regression. Furthermore, there has not been made a link in the used methodology between the allocation of risky assets and the expected rates of return. It is indicated in the literature that this link is interesting to investigate, because it show how much the burden of the risky assets is on the pension funds and how it can potentially influence the pension promises to future beneficiaries and taxpayers (Andonov, Bauer and Cremers, 2012). Also not included is the effect between different values of profit. It is interesting to see that there is indeed a link between the profitability of the sponsor with the risk allocation of the pension fund, but it is for now unknown if this effect becomes stronger or weaker if a higher profit is obtained by the sponsor.

Secondly, also not included into this research is the conclusion of Kisser, Kiff and Soto (2014) that there is the possibility that pension funds manipulate their liabilities in a downward manner. They conclude that the effects they measure are strongest for underfunded pension funds. If this is taken into account, this would mean that the official indicated values for the liabilities are not entirely ‘up-to-date.’ If this is a solvable problem is unclear, but it is good to keep in mind.

(22)

22

7. Conclusion

This research presents an interesting answer to the question if there is an effect on the allocation of risky assets of pension plans by the profitability of corporate sponsors. There seems to be a relationship between the risky asset allocation of the pension funds and the profitability of the corporate sector. This research finds that pension funds allocate assets with 0.0032 percentage points towards risky assets if the corporate sponsor makes a profit. This outcome is estimated in a standard regression and with an estimated coefficient of 0.284 for a regression with the

profitability expressed in log. Both estimates were statistically significant at a 5 percent level. Hence, pension funds belonging to sponsors that have higher profits are associated with a higher allocation towards risky assets. However, the first regression probably suffered from omitted variable bias which indicates that the found relation can change with additional tests.

Towards future research and considering the limitations of this one, the following aspects are considered. First of all, the dataset with the profits of the whole corporate sector missed some of the business sectors, namely: Agriculture, Construction, Real Estate, Services, Health care and Entertainment. Future research could make more use of the data by trying to be more complete.. Towards further research that validates the positive effect of corporate profitabilities on risky asset allocation, policy recommendations that assert the impact of this risk allocation should have a focus on the amount of distribution. Although it is not strange to discover that pension plans use some of the additional profits to invest in risky assets, the distribution between non-risky assets and risky assets is unknown. Further research on this subject would make for an interesting follow-up.

(23)

23

References

Anantharaman, D., Lee, Y.G., Managerial Risk Taking Incentives and Corporate Pension Policy,

Journal of FIancial Economics, vol.111, issue2. Pp. 328-251, 2014

Ambachtsheer, K.P., Would the Real Investment Policy Please Stand Up?, Financial Analysts Journal, vol. 47, no. 3, pp. 7-15, CFA Institute, 1991

Andonov, A., Bauer, R., Cremers, M., Pension Fund Asset Allocation and Liability Discount Rates: Camouflage and Reckless Risk Taking by U.S. Public Plans?, Netspar, 2012

Andonov, A., Bauer, R., Cremers, M., Pension Fund Asset Allocation and Liability Discount Rates, 2016

Antolin, P., Pension Fund Performance, 2008

Bams, D., Schotman, P., Tyagi, M., Asset Allocation Dynamics of Pension Funds, 2016

Bartram, S.M., In Good Times and in Bad: Defined Benefit Pensions and Corporate Financial Policy,

Warwick Business School, 2016

Basu, A.K., Portfolio Size and Lifecycle Asset Allocation in Pension Funds, Queensland University of Technology, 2007

Bauer, R.M.M.J., Cremers, K.J.M., Frehen, R.G.P., Pension Fund Performance and Costs: Small is Beautiful, Center Tilburg University, 2010

Black, F., The tax consequences of long-run pension policy, Financial Analysts Journal, issue36, pp.25-31, 1980

Bikker, J.A., Dreu de, J. Operating Costs of Pension Funds: The Impact of Scale, Governance and Plan Design, De Nederlandse Bank, vol. 8, pp. 63-89, Cambridge Press, 2007

Bikker, J.A., Broeder, D.W.G.A., Hollanders, D.A., Ponds, E.H.M., Pension Funds’ Asset Allocation and Participant Age: A Test of the Life-Cycle Model, The Journal of Risk and Insurance, vol. 79, no. 3, pp. 595-618, 2012

Blake, D., How is Asset Allocation Influenced by the Valuation of Liabilities?, The Pensions Institute, 2001, ISSN 1367-580X

Blome, S., Fachinger, K., Franzen, D., Scheuenstuhl, G., Yermo, J., Pension Fund Regulation and Risk Management: Results from an Alm Optimisation Exercise, 2007

Bodie, Z., Pension Funds and Financial Innovation, Financial Management, vol. 19, no. 3, pp. 11-22, 1990

(24)

24

Bodie, Z., Light, J.O., Morck, R., Taggart, R.A., Corporate Pension Policy: An Empirical Investigation, Financial Analysts Journal, volume 41, no.5, pp. 10-16, CFA Institute, 1985

Bodie, Z., Shoven J.B., Wise, D.A., Light, J.O., Morck, R., Funding and Asset Allocation in Corporate Pension Plans: An Empirical Investigation, University of Chicago Press, pp. 15-48, National Bureau of Economic Research, 1987

Borst, A., Effect of Cyclicality of Pension Funds on their Asset Allocation. Are Non_cyclical Pension Funds Performing Better than Cyclical Funds in Difficult Economic Times?,

http://www.scriptiesonline.uba.uva.nl, 2015

Broadbent, J., Palumbo, M., Woodman, E., The Shift from Defined Benefit to Defined Contribution Pension Plans: Implications for Asset Allocation and Risk Management, Global Savings and Asset

Allocation, 2006

Broeders, D., Valuation of Contingent Pension Liabilities and Guarantees Under Sponsor Default Risk, The Journal of Risk and Insurance, vol. 77, no. 4, pp. 911-934, 2010

CAIA Book,www.caia.org, 2013

Chen, J., Hong, H., Huang, M., Kubik, J.D, Does Fund Size Erode Mutual Fund Performance?: The Role of Liquidity and Organization, 2004

Dvorak, T., Timing of Retirement Plan Contributions and Investment Returns: The Case of Defined Benefit versus Defined Contribution, B.E. Journal of Economic Analysis & Policy Contributions, volume 12, issue 1, article 23, 2012

Dyck, A., Pomorski, L., Is Bigger Better? Size and Performance in Pension Plan Management, 2011

Exley, C.J., Mehta, S.J.B., Smith, A.D., The Financial Theory of Defined Benefit Pension Scheme, 1997

Feldstein, M.S, Seligman, S., Pension Funding, Share Prices, and National Saving, NBER Working Paper, no. 0509, 1980

Franzoni, F., Marin, J.M., Portable Alphas from Pension Mispricing, Universitat Pompeu Fabra,

2006

Haan de, L., Recovery Measures of Underfunded Pension Funds: Higher Contributions, No Indexation, or Pension Cuts?, De Nederlandsche Bank, no. 485, 2015

(25)

25

Huang, L., Lalani, M., Corporate Pension Risk Management and Corporate Finance: Bridging the Gap between Theory and Practice in Pension Risk Management, Society of Actuaries, 2015

Kisser, M., Kiff, J., Soto, M., Do Pension Plans Manipulate Pension Liabilities?, 2014

Leibowitz, M.L., Stanley, M., Ilmanen, A., U.S. Corporate DB Pension Plans: Today’s Challenges,

AQR Capital Management, 2016

Lucas, D.J., Stephen, P.Z., How Should Public Plans Invest?, American Economic Review, vol. 99:2, pp. 527-532, 2009

Getmansky, M., The Life Cycly of Hedge Funds: Fund Flows, Size and Performance, 2004 McKillop, D., Pogue, M., Pensin Plan Risk: The Impact upon the Financial Markets, The Association of Chartered Certified Accountants, 2009

McKillop, D., Pogue, M., The Influence of Pension Plan Risk on Equity Risk and Credit Ratings: a Study of FTSE100 Companies, Journal of Pension Economics and finance, volume 8, issue 4, pp. 405-428, 2009

Mohan, N., Zhang, T., 2014, An Analysis of Risk-Taking Behavior for Public Defined Benefit Pension Plans. Journal of Banking and Finance, 40(3), pp. 403-419.

Pfau, W.D., The Portfolio Size Effect and Lifecycle Asset Allocation Funds: A Different Perspective, National Graduate Institute for Policy Studies, Journal of Economic Literature, 2009

Poterba, J., Rauh, J., Venti, S., Wise, D., Defined Contribution Plans, Defined Benefit Plans and the Accumulation of Retirement Wealth, Journal of Public Economics, vol. 1, pp. 2062-2086, 2007 Racicot, F.E., Théoret, R., Procyclicality and Diversification in the Hedge Fund Industry in the Aftermath of the Subprime Crisis, 2014

Rauh, J., Risk Shifting versus Risk Management: Investment Policy in Corporate Pension Plans,

Review of Financial Studies, 2009

Sharpe, W.F., Budgeting and Monitoring Pension Fund, 2002

Turner, J.A., Hybrid Pensions: Risk Sharing Arrangements for Pension Plan Sponsors and Participants, Pension Policy Center, 2014

Yermo, J., Severinson, C., The Impact of the Financial Crisis on Defined Benefit Plans and the Need for Counter-Cyclical Funding Regulations, OECD Working Papers on Finance, Insurance and Private Pensions, no. 3, 2010

A study of the relation between risk allocation and profits for corporate pension funds: Is the riskiness of the pension plan affected by the profitability of the sponsor?

1