Pension Funds Allocation to Real Estate Equities

Pension Funds Allocation to Real Estate Equities:

The Past, Present and Future

Matt Godley

and

Thomas G. Thibodeau

University of Colorado-Boulder

RealEstateCenter

June 28, 2005

Financial support for this research was provided by Essex Financial Group. The views in this paper are those of the authors and do not necessarily represent Essex Financial Group.
Executive Summary

This research has two objectives. The first is to examine the growth in US pension fund total assets and pension funds’ allocation to real estate equities over the past twenty year period. The second objective is to forecast expected future pension fund demand for real estate equities through 2020. Total pension fund assets for the 1,000 largest pension funds increased from $1.07 trillion in 1985 to $5.35 trillion at the end of 2004. This represents an annual growth rate of 8.36% over this twenty-year period. This rate of growth has been volatile, however. The annual rate of change in total pension fund assets ranged from a -12.7% in 2001 to +24.8% in 1986. The standard deviation in the growth rate of pension fund assets was 9.72% for the 1985-2004 period--higher than the average rate of growth.

For the past twenty years, pension funds have been allocating about 3.67% of total assets to real estate equities. This percentage has ranged from a low of 3.0% in 1999 and 2000 to a high of 4.8% in 1990. From 1985 through mid 1994, pension funds held more cash than real estate equity. Since 1995, pension funds have cut their cash holdings in half. In 2004, the top 1,000 pension funds allocated 3.6%, or $192.6 B, to real estate equities.

The second part of this study forecasts pension funds’ future demand for real estate equities for the next 15 years. Changes in pension funds’ demand for real estate come from two sources: (1) changes in pension funds’ total assets and (2) changes in pension funds’ allocation to real estate. We examine both sources of changes in future pension funds demand for real estate.

Changes in total pension fund assets consist of contributions (both member and employer contributions) plus investment income (both dividend and capital gain income) less distributions and expenses. We forecast future pension fund total assets using historical benchmarks obtained from CalPERS. During the 1995-2004 period, contributions to CalPERS were 2.63% of assets; investment income averaged 9.28% of assets; distributions 4.06% of assets and expenses 0.11% of assets. Using these historical relationships, total pension fund assets for the 1,000 largest pension funds are forecast to increase to $17.6 trillion in 2020. If pension funds leave their allocation to real estate equities unchanged, pension funds will hold about $623.6B in real estate equities in 2020. This amounts to a 7.73% annual rate of increase in pension funds’ demand for real estate equities every year for the next 15 years. If pension funds gradually increase their allocation to real estate equities to 6% over the next 15 years, pension funds’ real estate holdings will total $1.0 trillion in 2020. This amounts to an annual rate of increase in pension funds real estate holdings of 11.32% per year.

Finally, we examined the sensitivity of our projections by altering the assumptions regarding contributions, investment income, distributions and expenses. Exhibit 1 summarizes the results. As mentioned earlier, the base case assumes (CalPERS) historical percent of total asset averages for contributions, investment income, distributions and expenses, pension fund and predicts total assets will grow to $17.6 trillion in 2020. Assuming the historical allocation to real estate, pension funds will hold $623.6 B in real estate equities. If contributions are one standard deviation higher than the historical rate, pension fund total assets will grow to $20.3 trillion in 2020 and pension funds will hold $716 billion in real estate equities. The sensitivity analysis indicates that the base case forecasts are most sensitive to changes in expected investment income.

Exhibit 1: Summary of 15-Year (2005-2020) Pension Growth Model Scenarios (in $ billions)

Introduction

The nature of the project was to first determine the growth of total pension fund assets over the next 15 years, and second, the growth of the real estate equity asset class within the pension funds over the same time period. A predictive model was constructed to attempt to forecast these numbers. The model used historical total asset value, asset allocation, and income and expense data from various sources. An analysis of the allocation data provided the insight that the real estate equity asset class weight has not deviated much over the past 20 years, which includes a downward trend the last couple of years. So, with the steadiness in allocation weight of this asset class does it still present a growth opportunity for real estate funding? Through the analytical model results, the pension fund values, and therefore, the real estate equity asset class, will continue to grow over the next 15 years. Negative growth rates were displayed in only one of thirteen model scenarios. The majority of the compound annual growth rates were above 7% for both total pension fund assets and real estate equity. These growth rates produce pension funds with total assets in excess of $15 trillion and real estate equity positions of over $600 billion (if we use the historical real estate asset allocation of 3.7%). With these model outcomes, the pension fund industry still represents a healthy source of real estate funding.

Methodology

In order to produce the analytical results, the research process followed the steps:

Table 1 and Figure 1 provide historical pension fund total assets for the past 20 years (1985-2004). This information was used to calculate the historical growth rates of pension funds (Table 2).

Table 3 and Figure 3 provide historical asset allocation information. This data was gathered through a variety of sources that included public pension funds (i.e. CalPERS and Texas Retirement System (Appendix 1)). The historical data indicates that over one-half of pension fund assets are allocated to stocks, about one-third to bonds and the remaining to cash, real estate equity, mortgages, and private equity. The allocation to stocks has increased significantly from 41.6% in 1985 to 61.5% in 2004 while the allocation to bonds has been decreasing. The allocation to real estate equities has remained fairly constant.

The model to analyze the growth of the pension funds and the real estate equity asset class was then constructed. The model included the major sources of income and expense that the pension funds incur on an annual basis. These consist of the members’ and employers’ contributions, the investment return (income), the distributions, and the administration expenses. CalPERS’ past ten years (1995-2004) of cash flows were utilized as a proxy for the entire pension fund industry’s income and expenses (Tables 5 & 6). Once the historical data was reconciled, the income and expense categories were calculated as a percent of total assets (Tables 7 & 8). These ratios were then used as the basis for the pension growth model.

After the model was completed, the growth analysis commenced through a series of sensitivity analyses. The baseline growth model was constructed using the historical average ratios. The other sensitivity analyses employed variable changes of +/- 1 standard deviation incontributions, income, distributions, and real estate equity allocation. The next couple of models examined the scenario if contributions decreased and distributions increased--a likely scenario with baby boomers retiring and increasing withdrawals from pension funds.

Finally, we examined what would occur if the pension fund managers increased their real estate equity asset class allocation from the historical 3.67% to 6%, 8% and 10%.

Results

The historical pension fund total asset data for the past 20 years (1985-2004) displayed the steady growth of the largest 1,000 domestic (both public and private) pension funds (Table and Figure 1). At the end of 2004, the pension funds had an estimated $5.3 trillion under management compared with approximately $1.1 trillion under management at the end of 1984. This represents an annual rate of 8.37% over the 20 year period. During the period, the funds only experienced two years of negative growth, which happened to occur concurrently with the stock market down years of 2001 and 2002 (Table and Figure 2). Overall, the pension fund growth rate correlates highly (.70 correlation coefficient) with the S&P 500 annual returns.

This relationship is built on the fact that the largest asset class held by the pension funds has historically been equities. As the stock market moves, the pension funds follow. At the end of 2004, as Table 3 displays, equity allocation was near a 20-year high at 61.5% of total assets. This high allocation of stocks usually does not bode well for the other asset classes. As the managers move into equities, they have to move out of the other major asset classes (bonds, cash, and real estate). These negative correlations are shown in Table 4. Since real estate equity does not have an extremely strong negative correlation with stocks, the real estate equity allocation has not fluctuated much over the past 20 years. The weighting has even declined during the past 5 years as the stock market tried to find a bottom. The data shows that the managers preferred the bond market over the real estate market as they moved their money out of the stock market. Based on the historical data, real estate equity allocation within the pension funds moves more with the real estate cycles than with the stock market. This lower correlation with the equity market might be as a result of the fact that corporate profits and jobs drive real estate demand. So, when the stock market declines as a result of a poor economy, the pension fund managers do not move into an asset class that will suffer in a year or two. As a result, the pension funds probably hold real estate because of its consistent income stream versus its diversification properties.

If the data describes a situation where the pension fund managers do not want more exposure to the real estate asset class, then why continue to explore the subject? The reason is the predicted growth in pension fund total assets over the next few decades. According to the pension growth model, the only scenario that provides negative growth rates for both pension fund total assets and real estate equity allocation is if the funds’ investment income consistently returns one standard deviation (9.57%) below the historical mean (9.28%). The following descriptive table outlines the results of the different scenarios used in the pension growth model.

The baseline model used historical averages of the real estate equity allocation and the income and expenses to predict the funds’ and real estate equity growth. This model shows growth rates for pension fund balances and real estate equity of 7.73% and 7.23%, respectively. These rates of growth would produce a total pension fund balance of $17.6 trillion and a real estate equity position of approximately $625 billion. All of the other models were compared to this model in the analysis. As a result, the main variables that influence the overall growth of the pension funds and the real estate equity positions are the volatile investment income returns, the pension fund managers’ investment policies, and if the amount of distributions increase while contributions decrease. These variables were examined and the reasons and results are described below:

Contributions +/- 1 Standard Deviation – This model (Tables 11 & 12) was constructed to determine how much influence the amount of contributions had on the overall growth of the pension funds and the real estate equity. The standard deviation of the past ten years’ percent of total assets datafrom CalPERS (0.96%) was added and subtracted from the mean (2.63%). The results of the model (Exhibit 1) produced ranges of balances that did not deviate much from the historical average model. Therefore, this variable alone (unless contributions were substantially reduced) did not have much influence on the growth of pension funds and the real estate asset class.

Income +/- 1 Standard Deviation – This model (Tables 13 & 14) was constructed to determine how much influence the amount of investment income had on the overall growth of the pension funds and the real estate equity.The standard deviation of the past ten years’ percent of total assets data from CalPERS (9.57%) was added and subtracted from the mean (9.28%). The results of the model (Exhibit 1) generated substantial differences from the baseline model. The primary reasons for this deviation were that the income variable held the largest ratio of total pension fund assets versus the other variables and the inherent volatility of investment returns. The income variable further displays the strong relationship between pension fund growth and the various investment markets, primarily the equities market. As a result of these conclusions, this variable holds the strongest correlation with the growth rates of the pension funds and the real estate asset class.

Distributions +/-1 Standard Deviation – This model (Tables 15 & 16) was constructed to determine how much influence the amount of distributions had on the overall growth of the pension funds and the real estate equity.The standard deviation of the past ten years’ percent of total assets data from CalPERS (0.66%) was added and subtracted from the mean (4.06%). The results of the model (Exhibit 1) produced balance ranges and growth rates that did not differ much from the historical average model. Consequently, this variable alone (unless distributions were substantially increased) did not have much controlover the growth rates of pension funds and the real estate asset class.

Real Estate Equity +/- 1 Standard Deviation - This model (Tables 17 & 18) was constructed to determine how much influence the amount of real estate equity had on the overall growth of the pension funds and the real estate equity.The standard deviation of the past twenty years’ asset allocation data from Pensions & Investments (0.47%) was added and subtracted from the mean (3.67%). The results of the model (Exhibit 1)generated balance ranges with minimal deviations form the baseline model of historical averages. The real estate equity variable has substantial control over the amount of real estate equity held in pension fund portfolios, but not over the growth of total pension fund assets. Thus, the only way for this variable to influence the real estate equity would be for the pension fund managers to increase or decrease their allocation policies.

Distributions increasing at 0.25% per year and Contributions decreasing at 0.15% per year–This model (Tables 19, 20, & 21) was constructed to ascertain the effect of distributions increasing while at the same time contributions decreasing. The theory behind this scenario is the current demographic trends in the United States. As the “Baby Boomer” generation begins to retire, a greater amount of pension fund distributions will be required each year. Concurrently, the larger amount of retirees will not be contributing to the funds. The results of the model (Exhibit 1) produced balances considerablyless than the baseline model. If this scenario transpires, the primary method that the funds can offset these slower growth rates is through higher investment income returns. These higher returns cause an increase in risk which would probably disturb both the pension fund managers and their beneficiaries.

Real Estate Equity increases to a Target of 6%, 8%, & 10% - These models (Table 22, 23, & 24) were constructed to determinethe amount of real estate equity held by pension funds at the end of the 15 year growth period. Fund managers may begin to change their allocation policies as more of their beneficiaries retire. The policy changeswould most likelyaddress the fund managers attemptto match the income stream with the liabilities (increased distributions). Since real estate equity provides managers a stable source of income, the allocation may increase to meet their needs. The results of the models (Exhibit 1) display a growth rate that is significantlyhigher than the historical average model. This proves the real estate equity variable has considerable influence over the real estate equity asset class.

Table 1: Total Pension Fund Size (in $ billion)