At this time each year, many Canadians have only. What s the cost of the foreign property limit in Canada s pension rules?

Transcription

1 What s the cost of the foreign property limit in Canada s pension rules? Kevin Milligan and Michael Smart FOREIGN INVESTMENT BY CANADIAN PENSION PLANS IS CURRENTLY RESTRICTED TO 20 PER CENT OF THE BOOK VALUE OF THE PLAN S PORTFOLIO. THIS POLICY INCREASES NON-DIVERSIFIABLE RISK IN PORTFOLIO RETURNS, WHICH ENTAILS COSTS FOR PENSION PLAN BENEFICIARIES. TO ESTIMATE THESE COSTS, WE EXAMINE THE EFFECTS OF AN EARLY-1990S INCREASE IN THE LIMIT ON THE PERFORMANCE OF MUTUAL FUNDS THAT HAD BEEN CONSTRAINED BY IT. THE EVIDENCE SUGGESTS THE LIMIT HAS HAD A SIGNIFICANT NEGATIVE EFFECT ON RISK-ADJUSTED RETURNS TO RETIREMENT SAVINGS IN CANADA. WE CONCLUDE THAT AN INCREASE IN THE LIMIT FROM 20 TO 30 PER CENT WOULD RAISE AVERAGE ANNUAL RETURNS BY 0.31 PERCENTAGE POINTS, OR $1.4 BILLION ANNUALLY. EN VERTU DE LA LOI ACTUELLE, LE MINIMUM QUE L ON PEUT INVESTIR EN PLACEMENTS ÉTRANGERS EST FIXÉ À 20 P. 100 DE L ACTIF TOTAL D UN RÉGIME DE RETRAITE. CETTE POLITIQUE ACCROÎT LA PART DE RISQUES NON-DIVERSIFIABLES DANS LE RENDEMENT DES PORTEFEUILLE, CE QUI ENTRAÎNE DES COÛTS POUR LES BÉNÉFICIAIRES DES RÉGIMES. POUR ÉVALUER CES COÛTS, LES AUTEURS ÉTUDIENT LES EFFETS QUE LA HAUSSE DE LA LIMITE DE RENDEMENT AUTORISÉE AU DÉBUT DES ANNÉES 1990 A PRODUIT DANS LE RENDEMENT DES FONDS MUTUELS. LES CHIFFRES MONTRENT QUE, EN CE QUI CONCERNE LES REVENUS CANADIENS DE RETRAITE, CETTE HAUSSE A EU UN EFFET POSITIF TRÈS NET SUR LE RENDEMENT AJUSTÉ AU RISQUE. ILS CONCLUENT QUE, EN PORTANT À 30 P. 100 LA LIMITE ACTUELLE DE 20 P. 100, ON AUGMENTERAIT DE 0,31 POINT DE POURCENTAGE (SOIT DE 1,4 MILLIARDS DE DOLLARS) LE RENDEMENT ANNUEL MOYEN DES FONDS DE PENSION. At this time each year, many Canadians have only recently made their annual retirement savings decisions, and many may prefer not to think about the matter again for another year. But this may be an opportune time to reconsider retirement saving policy for the nation as a whole. The federal minister of finance has been under considerable pressure in the past year to relax the 20 per cent limit on foreign property in both Registered Retirement Savings Plans (RRSPs) and employer-provided Registered Pension Plans (RPPs). Indeed, many commentators expected such a measure to be part of the latest federal budget (which had not been tabled when this issue of Policy Options went to press.) In this article, we review the arguments in favour of an increase in the limit, and we survey previous estimates of the cost of the policy to Canadians. Finally, we examine the impact of an earlier increase in the limit, from 10 to 20 per cent during the early 1990s. We argue that the previous reform had a substantial impact on returns to portfolios previously constrained by the limit, which suggests that the gains to further liberalization are also likely to be large. RRSPs and RPPs are the two main pillars of the private pension system in Canada. RPPs are employer-provided pensions, predominantly of the defined-benefit type. RRSPs are discretionary, individual savings vehicles administered by private financial institutions. The tax treatment of the two plans is largely identical: Contributions to plans are initially tax-deductible for plan beneficiaries (and employers in the case of RPPs), accrued income is tax-exempt, and withdrawals are taxable as ordinary income. Each taxpayer is currently permitted to make a combined annual contribution to RRSPs and RPPs up to the lesser of $13,500 or 18 per cent of earned income. In 1997, the book value of assets in the two types of plans exceeded $800 billion. RPPs and RRSPs are subject to foreign property restrictions. As of this writing, no more than 20 per cent of the book value of assets in these plans could be held in assets deemed to be foreign property. (The limit was raised from 10 per cent in two-per cent annual increments between 1990 and 1994.) As such, the current policy limits diversification possibilities for pensions and 36

2 retirement savings, and therefore entails real costs for plan beneficiaries. The ability to diversify portfolios among many assets and many countries is a form of insurance for investors. Since national stock market indexes typically do not rise and fall in unison, an investor holding an internationally diversified portfolio can obtain his or her desired level of expected return on assets while facing considerably less risk than would be possible with a portfolio of domestic stocks alone. Although the foreign property limit is therefore costly to investors, it is possible that the rule generates other gains for Canadians that are large enough to justify its existence. It is sometimes suggested, for example, that the limit makes additional funds available for domestic investment and reduces the cost of capital for Canadian firms. While this view may have had some validity when the current policy was enacted in 1971, with the increasing integration of global capital markets in recent years, these effects have largely been attenuated. Canada currently maintains no international capital controls other than the pension foreign property limit. Since Canada is small relative to the world market, the domestic cost of capital is largely determined by world market conditions, rather than by domestic supply and demand. It follows that the additional capital confined to the domestic market as a consequence of the foreign property rule merely displaces foreign investment funds, with no impact on either the cost or aggregate supply of capital. Moreover, the policy s impact on Canada s balance of payments is likely to be small, as many domestic investors can move non-pension wealth offshore to counteract the rule. These arguments notwithstanding, the foreign property rule might be held to serve other public policy objectives important to Canadians. For example, there might be concern that raising the limit would reduce pension funds demand for bonds, making it more difficult for Canadian governments to place their issues. Or, it might be felt that a Buy Canadian policy for pension funds and RRSPs somehow benefits the nation even if it does not increase domestic investment. If that is the case, it is important to assess the cost at which these objectives can be achieved. To summarize, by restricting international diversification the current policy induces investors to hold portfolios that are more risky and earn lower expected returns than they would otherwise choose. Several authors have attempted to develop quantitative estimates of this impact of the policy. The foreign property limit The simplest approach is to compare the historical returns to popular domestic and foreign securities. Thus a 1997 study by Ernst and Young considered investments in Canadian and world stock market indexes for the period. It found that an all-equity portfolio that invested 30 per cent in the world index would have realized annual returns 34 basis points (that is, 0.34 percentage points) higher than a portfolio that was only 20 per cent invested in the world index. While this result is suggestive, the methodology behind it has serious flaws. To begin with, the study makes no adjustment for difference in risk or transactions costs between domestic and foreign equities. But beyond that, and more importantly, the Ernst and Young estimate mainly reflects the poor performance of Canadian stock markets during the 1990s relative to US and most European exchanges. This did result in losses for investors constrained by the policy, but there is no reason to expect Canadian markets to continue to underperform in future. Thus a reform in the limit today may not yield gains similar to what could have been realized over the last two decades. Quite the contrary, the best predictor of future returns is likely the subjective beliefs of current investors, and Canadians have continued to make significant investments in Canadian equities, which suggests they expect returns to be quite high. An alternative approach ignores historical differences in expected returns to domestic and foreign stocks, and instead emphasizes differences in risk. In a 1995 paper, Keith P. Ambachtsheer, an investment advisor to several major pension funds, assumes that Canadian and foreign equities will earn exactly the same returns over the long haul. Even so, if Canadian and foreign stocks don t mimic each other exactly in the short run, an increase in the foreign property limit would allow investors to increase the international diversification, and reduce the risk exposure, of their portfolios. Ambachtsheer assumes that investors would respond by holding fewer low-risk, low-return bonds and more domestic and foreign equity. The estimated net effect of the changes is an increase in expected returns of 10 to 20 basis points per annum, with no change in portfolio risk. Last year, after a similar calculation, David Burgess and Joel Fried of the University of Western Ontario estimated that relaxing the limit would raise annual expected returns by 18 basis points. This approach to the problem is appealing in many ways but it requires strong assumptions about the behaviour of investors in response to an increase in the limit assumptions which do not appear to be borne out in reality. The idea is that the Since Canada is small relative to the world market, the domestic cost of capital is largely determined by world market conditions. The additional capital confined to the domestic market as a consequence of the foreign property rule merely displaces foreign investment funds, with no impact on either the cost or aggregate supply of capital. POLICY OPTIONS MARCH

3 Kevin Milligan et Michael Smart Our estimates are based on the performance of actual portfolios, and so reflect the actual practice of investors, such as home bias and active trading strategies gains to relaxing the foreign property rule will result from reductions in holdings of Canadian bonds, rather than the substitution of foreign for Canadian stocks. The large share of bonds in pension and retirement portfolios is certainly something of a puzzle, given the large premium historically earned by equity and the substantial evidence that stocks dominate bonds over longer holding periods. Because observed portfolios are difficult to justify using standard theories of investor behaviour, it seems especially hard to predict how asset allocations will change in response to an increase in the foreign property limit. Nor does this approach account for several other aspects of observed behaviour. For example, investor portfolios typically exhibit substantial home bias, even in the absence of regulatory restrictions on foreign content. In the presence of such bias, optimal portfolios will differ from those predicted by the standard risk-return analysis, so that responses to the policy reform are difficult to predict. As a result, it is probably better Table 1 Estimated impact of raising the foreign property limit dence of the policy s effect can be adduced by examining the impact of the doubling of the foreign property limit in the early 1990s on the relative returns to the two classes of portfolios. This approach offers several advantages over those taken by previous researchers. First, our estimates are based on the performance of actual portfolios, and so reflect the actual practice of investors, such as home bias and active trading strategies, rather than the hypothetical portfolios assumed in past research. Moreover, our methodology allows us to adjust returns for differences in risk between eligible and ineligible portfolios, and to control explicitly for the performance of the domestic stock market over the sample period. We believe that, by not confusing short-run differences in market returns with the long-run effects of the policy, this technique yields superior estimates. Still, our approach also has limitations, and some caution is therefore warranted in interpreting our results. By restricting our comparisons of Change in expected return Standard error Constrained vs. unconstrained funds 26.7* (9.9) With controls for limit and year effects 24.0* (9.2) With controls for limit and market risk 31.7* (6.7) Excluding unconstrained funds 14.1 (47.9) Notes: Estimaed impact on expected return to a constrained portfolio of a 10 percentage point increase in the limit, in basis points per annum. Estimated robust standard errors in parentheses. * Significantly different from zero at the 95 per cent confidence level. to estimate the effect of the policy by observing the actual behaviour of investors. That s what our research does. We examine returns to publicly traded equity mutual funds in Canada. Mutual funds provide a good practical test for the effects of the policy, since some funds (those deemed RRSP-eligible) are constrained to hold no more than 20 per cent of their assets in foreign property, while other (RRSP-ineligible) funds are unconstrained by the limit and so hold internationally diversified portfolios. Since the foreign property restriction is the only apparent difference in the two classes of funds, its impact can be estimated by comparing risk-adjusted returns of eligible and ineligible funds. Moreover, additional evi- returns to equity mutual funds alone, we are ignoring the possibility that investors would respond to reduced risk by increasing their exposure to equity markets. For this reason, our estimates are probably a lower bound on the likely impact of reform on expected returns. Moreover, we rely crucially on the assumption that, in the absence of the foreign property limit, eligible and ineligible mutual funds would experience similar returns. We do perform a number of checks on the sensitivity of our results to this assumption, which are discussed in what follows. The results of our comparison-based estimation techniques are presented in Table 1. While our estimates are based on a comparison of 38

4 actual returns over the period, for the sake of concreteness the figures reported in the Table indicate the predicted impact on annual expected returns of portfolios of increasing the foreign property limit to 30 per cent from 20 per cent. The simplest application of our technique merely involves comparing the observed average return to eligible and ineligible funds over the sample period. Recall that eligible funds may hold 20 per cent and ineligible funds 100 per cent of their assets in foreign securities. One very simple view is that an increase in the limit to 30 per cent would close one-eighth (i.e., 10 per cent divided by 80 per cent) of the gap in average returns between the two classes of funds. On average, annual returns to ineligible funds were 2.14 percentage points higher than ineligible funds. By this calculation, then, an increase in the limit to 30 per cent would increase eligible returns by 26.7 basis points annually (since 26.7 is one-eighth of 214). There are several possible objections to this comparison and the implied estimate of the policy s effect. First, the foreign property limit did rise during the sample period. If our theory is correct, the difference in returns of the two classes of funds should be smaller after the reform than before it. To allow for this possibility, we conducted a regression analysis of fund performance. In the second row of Table 1, we report results for a specification that examines how returns to eligible funds changed in proportion to the increase in the foreign property limit, while controlling for unobserved factors affecting returns to all funds that may vary from year to year. By including these year effects, we are recognizing that fund returns may have risen after 1990 due to factors unrelated to the policy. In this specification, however, we assume that any change in the performance of eligible funds relative to ineligible funds can be attributed to the reform. Our results in this case imply that a 10-percentage point increase in the limit raises expected returns by 24 basis points per annum. As noted, our approach assumes that the only relevant difference between the two classes of funds is the effect of the limit itself. In practice, this is not the case. First, because RRSP-eligible funds are less diversified, they are riskier and so less valuable to investors than a simple comparison of average returns would suggest. Ignoring this difference in systematic risk therefore causes a downward bias in our estimate of the policy s impact. Second, our results so far reflect the actual return to Canadian and foreign stocks during the sample period. As we argued earlier, there is no reason to expect that the past performance of Canadian mar- The foreign property limit kets is likely to be repeated. Since Canadian stock returns fell relative to foreign markets during the 1990s, and since eligible funds are more strongly correlated with the domestic market, eligible funds would naturally have underperformed ineligible ones at the same time as the limit was being raised. For this reason, we have likely further underestimated the impact of the reform on expected future returns. To deal with these sources of bias, we expanded our analysis to control for actual returns to the TSE 300 market index, allowing for differences in the covariances (or betas ) with the market of individual mutual funds. Results for this specification are reported in the third row of the Table. As expected, the estimated impact of the policy is larger: A ten percentage point increase in the limit is predicted to increase returns by 31.7 basis points annually. Even with its controls for differences in market risk, our approach relies on comparing returns to eligible and ineligible funds, before and after the policy reform. Our results will therefore be misleading if returns to the two classes of funds are simply incomparable, due to unobservable differences in the characteristics of the funds and the way they responded to changes in financial markets in the 1990s. If this is the case, a better approach is simply to exclude ineligible funds from the analysis and to examine how returns to eligible funds alone changed following the reform. This approach is probably extreme, since it discards potentially relevant information on fund returns. Still, it is a useful test of the robustness of our results. Accordingly, the fourth row of Table 1 reports results for this specification. The estimated effect of the policy is smaller in this case a ten percentage point increase in the limit is predicted to increase annual returns by only 14.4 basis points and the estimate has less statistical precision. This is to be expected, since the impact of the policy reform is small relative to other, extraneous factors that influenced funds returns over the sample period. Even so, the estimate has the predicted sign, and the magnitude is roughly comparable to previous specifications. In all, a fairly consistent picture emerges from our analysis. A ten percentage point increase in the limit would increase returns to constrained portfolios by about 15 to 30 basis points annually. While our approach is not fully comparable to previous methodologies, our estimate appears broadly similar to, if somewhat larger than, the findings of other researchers. In our view, the estimated cost of the foreign property limit is quite substantial. To Increasing the foreign property limit to 30 per cent would increase annual returns to these assets by 31.7 basis points, or approximately $1.4 billion annually. POLICY OPTIONS MARCH

5 It is important to recognize that the foreign property rule entails costs for Canadians. Kevin Milligan et Michael Smart place our estimates in perspective, Statistics Canada reports that the book value of assets in RRSPs and RPPs stood at $822 billion in 1997, with about 55 per cent, or $452 billion, held in equities. According to our preferred estimate, increasing the foreign property limit to 30 per cent would increase annual returns to these assets by 31.7 basis points, or approximately $1.4 billion annually. This is in fact a conservative estimate, since the market value of pension assets substantially exceeds their book value, and the calculation also ignores the increased return to future retirement saving. Canada s foreign property rule was conceived over a quarter-century ago, apparently as a means of increasing the supply of capital to Canadian firms and, with it, domestic investment. Because Canadian and world capital markets have become more integrated since then, there is little reason to believe that a policy of domestic preference has any effect on the cost of capital or level of investment of Canadian firms. Whatever other objectives the policy might be held to serve, however, it is important to recognize that the foreign property rule entails costs for Canadians. According to our estimates, these costs are substantial. We predict that increasing the limit to 30 per cent would raise returns to pension and retirement savings assets by as much as $1.4 billion annually. Kevin Milligan is a doctoral candidate in the Department of Economics at the University of Toronto. Michael Smart is Assistant Professor of Economics and Research Associate of the Institute for Policy Analysis at the University of Toronto. Their research received financial support from the Donner Canadian Foundation. A working paper containing a fuller description of the research is available at VAGUENESS SWEET AS HONEY Politicians of principle don t always win, but they know who they are. The norm is for politicians to focus on process more than principle. Ideas alienate as well as attract. Presidential candidates who win their party s nomination by appealing to the core principles of the right or left will, in the general election, follow their political consultant s advice and head for the murky, moderate middle. Political principle is tart in many mouths, whereas vagueness tastes like honey. It has always been dangerous... for an elected politician to take a position of principle on the issues of race or class. The commonsense result has been to avoid the clarity of conviction powerfully expressed. Even the mention of ideology. political philosophy, or principle has been discouraged; just get the politicians together, divvy up the economy pie, and keep the process moving forward toward an undefined end. In such a world, party has little content and becomes more an entry fee than a patriotic faith. Bill Bradley, Time Present, Time Past (1996) Vol I, No. II An Internet recruiting service founded as a volunteer project of staff, students and alumni at the Universities of Alberta, BC and Western Ontario. A technologically advanced employment information communications system that enables job seekers and employers to conact each other and then communicate directly and privately without having to give out contact information until they wish to take that step. Posting cost: $25 for would-be employers, free for job-seekers. 40