Information about Asia and the Pacific Asia y el Pacífico
Chapter

6 Philippine Fiscal Policy: Sustainability, Growth, and Savings

Editor(s):
John Hicklin, David Robinson, and Anoop Singh
Published Date:
July 1997
Share
  • ShareShare
Information about Asia and the Pacific Asia y el Pacífico
Show Summary Details
Author(s)
Philip Gerson and David Nellor 

Over the last several years, there has been a dramatic improvement in the Philippines’ fiscal accounts. The consolidated public sector deficit has been reduced by about 5 percentage points of GNP since 1990 and reached approximate balance in 1996. However, further reform is needed to preserve the fiscal balance and achieve the government’s growth and macroeconomic objectives. To reach these goals, the government is seeking a comprehensive reform of the tax system and a streamlining of the civil service.

The paper analyzes Philippine fiscal policy in two steps. The first step is to determine whether fiscal policy can meet two minimum conditions without discretionary changes. First, the policy should be sustainable, which means that the public sector remains solvent or that the present fiscal stance permits the public sector to satisfy its debt obligations. Second, it should be possible to maintain public investment at a minimum level without sacrificing fiscal balance.

The second step is to define a fiscal policy that goes beyond the minimum conditions and enables the government to meet its broader medium-term economic goals. In particular, economic growth objectives call for a significant increase in public investment outlays. This increase in investment needs to be matched by higher saving so as to avoid pressure on the external current account and ensure macroeconomic stability.

Following these two analytic steps, we hope to have addressed several topical fiscal policy questions.

(1) Why is reform required when the present fiscal position is so strong? Fiscal reform is necessary because a number of prospective fiscal developments could lead to a violation of the two minimum conditions for fiscal policy. Moreover, economic growth objectives call for a significant increase in public investment outlays, which will threaten macroeconomic stability without fiscal reform.

(2) What size of fiscal reform is necessary to secure the government’s macroeconomic objectives? Analysis suggests that government infrastructure investment and savings need to increase significantly in the coming years. A growth accounting framework prepared for illustrative purposes suggests that government infrastructure investment would need to increase by at least 1½ percent of GNP to sustain rapid growth over the next ten years. At the same time, public savings should increase by about 6 percent. An illustrative scenario suggests that revenue or expenditure measures of some 2½ percent of GNP will be required on top of the increases in saving that will result from reduced interest expenditure (relative to GNP) and the natural elasticity of the tax system.

(3) How can higher public savings be achieved? Measures to increase savings will need to concentrate primarily on increased revenues rather than on reduced expenditure. Expenditure levels are relatively low by the standards of the Association of South East Asian Nations (ASEAN), and—other than through civil service reform to limit the growth of personnel expenditure—there is little room for adjustment. By contrast, a comparison with other ASEAN countries suggests that there is considerable scope for increasing tax collections in the Philippines. For these reasons, the government is planning to implement a comprehensive tax reform that aims to broaden the tax base, simplify the system, and improve administration.

Some Minimum Conditions for Fiscal Policy

This section analyzes whether fiscal policy meets the two minimum criteria established above: sustainability and the maintenance of minimum levels of investment.

Fiscal Sustainability

A fiscal policy is sustainable if it can be maintained indefinitely without leading the government into insolvency, whereas a fiscal policy is unsustainable if the stance is expected to result in an outcome that makes fiscal reform inevitable. Yet, fiscal sustainability is just a minimum condition for fiscal performance because a fiscal stance can be sustainable without necessarily being desirable. For example, a policy that entails wasteful government expenditure financed by poorly designed taxes might be sustainable but inferior, from an efficiency and growth perspective, to a policy characterized by lower spending and taxes. Moreover, the crowding out of private investment, current account deficits, and inflation concerns make public debt an important policy issue apart from the narrow issue of fiscal sustainability. Thus, tests of fiscal sustainability examine only whether a fiscal policy could be continued indefinitely, not whether it should be continued.

In this section, we review some ratios that indicate the improvement in the fiscal stance in recent years and provide a history of fiscal developments. Although the indicators suggest that fiscal sustainability may not be a problem in the Philippines, they apply to only a few recent years and are impressionistic. Accordingly, we undertake an econometric exercise to determine if the sustainability of fiscal policy can be demonstrated empirically over a longer period. This econometric analysis draws on data for 1970-93. We find that while both the stock of debt and the debt-to-GNP ratio are stationary, we cannot conclude that fiscal policy is sustainable. This result emphasizes the importance of maintaining and building on the recent improvement in the fiscal accounts.

Informal Indicators of Fiscal Sustainability

The prospects for sustainability have improved in recent years. The ratio of debt to GNP has fallen, and both the national government and the consolidated public sector have moved from deficit to near balance. While the ratio of interest to total expenditure in the Philippines is among the highest in the world, this situation has persisted for many years and, although a consideration for fiscal sustainability, is not of immediate concern. Developments in some key fiscal indicators can be summarized as follows.

Overall and primary budget deficits. The elimination of the consolidated public sector deficit and the growing primary surplus of the national government both suggest that current fiscal policy in the Philippines is sustainable. The overall deficit in the consolidated public sector has improved markedly in recent years. As recently as 1990, it stood at nearly 5 percent of GNP. Between 1991 and 1993, it dropped to an average of 2 percent of GNP before falling to 1 percent of GNP in 1994 and to near balance in 1995.

The national government budget deficit has also declined significantly. Between 1985 and 1991, the deficit averaged 2.6 percent of GNP and was never below 2 percent in any year. After a two-year period during which the deficit was about 1.5 percent of GNP, the national government budget has remained largely in balance; deficits were on the order of 0.3 percent of GNP in 1994 and 1995. The primary surplus has also improved. In 1985, it was less than 1 percent of GNP, and in 1986 the national government ran a primary deficit of more than 1½ percent of GNP. Subsequently, the primary surplus ranged between 3 percent and 4 percent of GNP before improving to 5 percent of GNP in 1995.

Level of public indebtedness. Public debt ratios have generally stabilized or come down in recent years. In relation to GNP, public debt peaked at almost 150 percent of GNP in the mid-1980s before declining to about 100 percent of GNP in 1995.

Figure 1 shows the time path of public sector debt during 1970-95: it roughly tripled in dollar terms every five years between 1970 and 1985 and then more than doubled between 1985 and 1995. Figure 2 graphs the discounted stock of public sector debt during 1970-93 and suggests a division into three subperiods.1 Between 1970 and 1984, the discounted value of the stock of debt increased steadily. Beginning in 1984, coincident with the Philippines’ debt crisis, the stock of debt grew rapidly, doubling in only two years. The large increase in total public sector debt was due primarily to large increases in the national government’s external and, especially, domestic debt (although the debt of the Central Bank of the Philippines also increased sharply), reflecting the assumption of significant amounts of guaranteed private sector obligations from debt rescheduling. Finally, since about 1986, the discounted value of the stock of public sector debt has remained relatively constant, with a sharp increase between 1990 and 1992 offset by declines in other years.

Figure 1.Public Sector Debt

(In billions of U.S. dollars)

Source: Philippine authorities.

Figure 2.Discounted Public Sector Debt

(In billions of 1970 U.S. dollars)

Sources: Philippine authorities; World Bank, World Debt Tables; and authors’ calculations.

A roughly similar pattern prevails for the ratio of public sector debt to GNP (Figure 3). Between 1970 and 1979, the debt ratio was essentially constant, with the ratio in 1979 actually marginally lower than the ratio at the beginning of the decade. Between 1979 and 1986, the debt ratio increased more than sevenfold, growing by more than 80 percent of GNP between 1984 and 1986. Finally, the debt ratio has generally declined since about 1986.

Figure 3.Public Sector Debt

(In percent of GNP)

Source: Philippine authorities.

As with the consolidated public sector, the debt of the national government has increased sharply in recent years, growing by a factor of nearly five between 1985 and 1995 (Figure 4).2 Although the debt-to-GNP ratio for the national government was lower in 1995 than at its peak in 1986, it was not significantly lower than its level at the beginning of the 1990s and as recently as 1993 was close to its peak level (Figure 5).

Figure 4.National Government Debt

(In billions of pesos)

Source: Philippine authorities.

Figure 5.National Government Debt

(In percent of GNP)

Source: Philippine authorities.

Interest expense. National government interest expenditure relative to GNP has remained fairly constant in recent years, remaining between 5 percent and 6 percent of GNP every year since 1987 despite the decrease in the debt-to-GNP ratio. Interest constitutes a large share of total national government spending, accounting for about 30 percent of total expenditure every year since 1987. In 1993 (the last year for which data are available for all Asian countries), interest accounted for a higher percentage of central government expenditure in the Philippines (27 percent) than in any other Asian country (IMF, Government Finance Statistics, various years). Among 98 countries, interest expense constituted a higher share of total central government expenditure in only four countries: Mexico, Brazil, Bulgaria, and Greece.

While national government interest expense is high, it has remained fairly constant relative to GNP and to total expenditure for several years. This pattern suggests that interest expense is not a threat to sustainability. Nevertheless, the large share of resources that must be devoted to paying interest hampers the flexibility of the national government to respond to economic shocks or to increase spending on priority items. Thus, the large interest expenditures of the national government contribute to the possibility of more rapid increases in debt in case of an economic shock.

To sum up, current and historical data indicate a sharp improvement in the fiscal accounts over the last several years, with the overall balance improving and the stock of debt declining relative to GNP. These findings suggest that fiscal policy may now be sustainable.

A Formal Test of Fiscal Sustainability

In this section, we undertake a formal econometric exercise to test whether the qualitative finding of the previous section—that fiscal policy may be sustainable—can be demonstrated through quantitative techniques. The test developed in this section depends on the current expectation about the size of the stock of public debt in the infinite future. To calculate this expected value, we look at historical data for 1970-93 to model the data-generating process for public debt. The approach implicitly assumes that no fundamental shift in fiscal policy has occurred, so that the historical process remains valid. However, it does allow for the effects of a onetime exogenous shift in the process owing to the public assumption of private sector debts during the debt crisis. At the end of the section, we consider the possibility that a fundamental shift in the fiscal regime has in fact occurred in the last few years.

Formal statistical tests of fiscal sustainability are based on a multi-period application of the single-period government budget constraint.3 The government budget constraint, equation (1), is an accounting identity stating that the increase in government debt in any period t is equal to the sum of the primary surplus (revenue less noninterest expenditure), interest expenditure, and any change in government cash balances. Thus, the constraint facing the government takes the form

where Bt is domestic-currency-denominated government debt; Bt1* is foreign-currency-denominated government debt in foreign currency; Xt is the current period exchange rate; rt-1 is the ex post interest rate on domestic-currency-denominated debt; rt1* is the ex post interest rate on foreign-currency-denominated government debt; St is the primary surplus of the government; and ΔCt, is the change in government cash balances.4

The budget constraint must be satisfied over time to avoid insolvency. Consequently, if debt is incurred in one period, the budget will have to generate a primary surplus in another period to satisfy that debt obligation. The next step is to reformulate the preceding simple budget constraint to derive a testable condition—that the expected value of future primary surpluses must equal the current stock of debt—or, in other words, to satisfy the basic solvency condition. This analysis, shown in Appendix I, develops the following testable equation:

where dt is the discounted (by the interest rate) value of public sector debt at time t, and Et is the expectations operator conditional on information available at t.

Equation (2) provides a test of fiscal solvency: if the current fiscal policy is sustainable, the unconditional expectation of the discounted public sector debt should be zero. In other words, the statistical requirement is that the public debt series is stationary and zero-mean, which establishes the conclusion that the present fiscal policy can be sustained indefinitely without the need for reform. By contrast, if the expectation of the public debt series is nonzero over the infinite horizon, this means that the public debt series is, ultimately, not sustainable. A change in fiscal policy is inevitable in this case because the public sector will eventually be unable to service its debt. But, solvency is not established even if the process is covariance stationary. The presence of a deterministic component—for example, a time trend or a nonzero intercept—means that the unconditional mean of the series is not equal to zero, and this implies eventual insolvency if policies are not changed.

From Wold’s Representation Theorem, we can assume that dt follows a multivariate ARIMA process of the form

where

μ(L) = [I - θ(L)]-1 [I-ρ(L)],

and where ρ(L) is a pth-order polynomial, θ(L) is a qth-order polynomial, and γ0 is the unconditional mean of the stationary series (I - L)mZt. The vector Zt includes dt as its first—but not necessarily only—element; and et is a vector white-noise process. The solvency conditions are therefore twofold. First, Zt must be stationary. Second, the first element of γ0 must be equal to zero. We use a special case of equation (3), namely,

In this case, discounted public debt will be nonstationary if α ≥ 1. In addition, even if α < 1, discounted public debt will be nonstationary if there is a nonzero deterministic trend (i.e., if (β ≠ 0). Finally, if μ ≠ 0, the solvency condition is still violated (because the expectation of dt+N is not zero) even though the time series is stationary.

Statistical tests suggest that both the discounted public sector debt and public sector debt ratio for 1970-93 are nonstationary, which implies that fiscal policy is not sustainable (Table 1).5 This result underscores the importance of maintaining the recent improvement in the fiscal accounts reported in the previous section. Unit root tests for the discounted stock of public sector debt and the ratio of debt to GNP were conducted using both Augmented Dickey-Fuller (ADF) and Phillips-Perron tests (See Dickey and Fuller, 1979 and 1981; and Phillips and Perron, 1988). While the Phillips-Perron test admits a broader class of errors than the ADF test, its small sample properties are poor. Because the results from the twotests are similar, only the ADF results are reported here. The null hypothesis for the ADF test is that α = 1 and β = 0. We are unable to reject the null hypothesis of a unit root in the time series, leading to the conclusions that both the discounted public sector debt and the public sector debt ratio are nonstationary and that fiscal policy is not sustainable.

Table 1.Unit Root Tests for Discounted Public Sector Debt and Debt Ratio Model with Time Trend: dt + μ + βt + αdt-1 + ut
ValueCritical Value1
TestDebt levelDebt ratio95 percent99 percent
ɸ32.492.931.080.74
tα-2.20-2.42-3.60-4.38
tβ2.202.122.853.74
n(α - 1)-8.10-6.76-17.90-22.50

Critical values are based on a sample si/.c or’25 and are derived from Fuller (1976) and Dickey and Fuller (1981). To evaluate these critical values, consider the test statistic ta. This statistic is atest of whether the true value of α is 1, based on the derived estimate of its value. The estimated equation contains random error terms, so the estimated value of a will never precisely equal its true value. The further the estimate from 1, the less likely that the true value of a is 1. Based onan assumed distribution of fa, if the true value of a were 1, we would observe the values of ta that are less than the 99 percent critical value less than 1 percent of the time. We would observe values less than the 95 percent critical value less than 5 percent of the time. These probabilities are small enough that they reject the hypothesis that the true value of α is 1. However, given that our ta is greater than either of these values, we cannot reject the hypothesis that a is equal to 1.

Critical values are based on a sample si/.c or’25 and are derived from Fuller (1976) and Dickey and Fuller (1981). To evaluate these critical values, consider the test statistic ta. This statistic is atest of whether the true value of α is 1, based on the derived estimate of its value. The estimated equation contains random error terms, so the estimated value of a will never precisely equal its true value. The further the estimate from 1, the less likely that the true value of a is 1. Based onan assumed distribution of fa, if the true value of a were 1, we would observe the values of ta that are less than the 99 percent critical value less than 1 percent of the time. We would observe values less than the 95 percent critical value less than 5 percent of the time. These probabilities are small enough that they reject the hypothesis that the true value of α is 1. However, given that our ta is greater than either of these values, we cannot reject the hypothesis that a is equal to 1.

The ϕ3 statistic is a test of the joint hypothesis that α = 1 and β = 0. Our inability to reject the null hypothesis implies that both assumptions hold, a result corroborated by the other three test statistics. The tα and tβ statistics are the t-statistics resulting from the regression (with tα calculated as a test for α = 1), while the n(α - 1) statistic is based on the ordinary least squares (OLS) estimate of α and on the sample size. Again, on the basis of these statistics, we cannot reject the null hypotheses that α = 1 and β = 0.

Additional support for the result that sustainability cannot be established is contained in Table 2. Having “proved” that β = 0, we repeat the ADF test for a unit root in discounted public sector debt and the public sector debt ratio using the model

dt = μ + αdt-1 + ut,

that is, omitting the time trend. We find again that we cannot reject the null hypothesis that α = 1. In addition, we find no evidence of a nonzero drift. The ϕ1 statistic is used to test the joint hypothesis that α = 1 and μ = 0. The failure to reject the null hypothesis is buttressed by the results from the tests on the t-statistics and on n(α - 1). These tests fail to reject the hypothesis that the series contains a unit root with no drift. The finding—an inability to reject the presence of a unit root in the discounted public sector debt series—requires two qualifications.

Table 2.Unit Root Tests for Discounted Public Sector Debt and Debt Ratio Model without Time Trend: dt = μ + αdt-1 + ut
ValueCritical Value1
TestDebt levelDebt ratio95 percent99 percent
ϕ12.280.930.490.29
tα-0.33-1.07-3.00-3.75
tμ1.501.362.613.41
n(α - 1)-0.42-1.50-12.50-17.20

The critical values are based on a sample size of 25 and are derived from Fuller (1976) and Dickey and Fuller (1981). The critical values differ from those in Table 1 because the model is estimated without a time trend.

The critical values are based on a sample size of 25 and are derived from Fuller (1976) and Dickey and Fuller (1981). The critical values differ from those in Table 1 because the model is estimated without a time trend.

First, the extremely small sample size produces large standard errors. This feature of the test is unlikely to change the conclusion that fiscal policy is unsustainable because it appears that the series contains either a unit root or a time trend. Consider the evidence of a unit root in the series. In the first equation, the OLS estimate for a is 0.65 for discounted public sector debt and 0.69 for the public sector debt ratio, neither of which is very close to 1. However, the relatively large standard error on the estimate limits the size of the r-ratio. For example, for public debt, the OLS estimate for a would have to be less than 0.43 to reject the presence of a unit root at the 95 percent level. Thus, estimates closer to zero than to one could still fail to reject a unit root. The implication is that, given the small sample size, rejecting the null hypothesis of a unit root even for many series that are actually stationary would be very difficult. Yet, even if we reject the hypothesis of a unit root, we still cannot establish sustainability. The reason is that, in this case, standard t-ratios would apply and the hypothesis of no time trend could be rejected at the 5 percent level (using a critical value of 1.96).

Second, the possibility of a structural break in the series during the early or mid-1980s is a qualification to the result that fiscal sustainability cannot be established using the unit root test. As noted earlier, public sector debt increased sharply between 1984 and 1986, owing in large part to the assumption of guaranteed private sector debts. Perron (1989) notes that the presence of a structural break in a series can lead to a dramatic loss of power in standard unit root tests. Perron develops an alternative procedure for testing for a unit root in the presence of a single break in the level or slope of the trend function. The procedure is to introduce two new variables: a dummy variable that has a value of zero for all periods up to the date of the structural break and of one for all periods after that, and a trend function that begins in the period following the break. The time series is “demeaned” and “detrended” by regressing it on a constant, a time trend, and the two new variables, and an ADF test is performed on the residual to test for the presence of a unit root.6 We allow the data to identify the date of the structural break by choosing the date that maximizes the F-statistic for the dummy level and trend variables. This procedure suggests that the break in the data occurred in 1984—for both the public debt and the public debt ratio—which is in fact the year immediately before the large jump in the level of discounted public sector debt.

Conducting the ADF test on the residuals from the regression involving the discounted stock of public sector debt, we obtain a t-statistic on the detrended and demeaned series of -4.05, which is between the 90 percent and 95 percent critical values for the test (of-3.86 and -4.18, respectively). Given that the coefficient on the lagged value of the detrended and demeaned public sector debt is only 0.16, rejecting a unit root in this series seems reasonable. Moreover, a Wald test on the estimated coefficients for the two time trends cannot reject the hypothesis that the post-break series contains no time trend (i.e., that the coefficients on the two time trends sum to zero). However, the hypothesis that the post-break series has no intercept term is rejected at the 1 percent level, which again leads to the conclusion that fiscal policy is, ultimately, unsustainable.

After detrending and demeaning the data on the public sector debt ratio, we used an ADF test on the residual, producing a t-value of -4.27, which is sufficient to reject the hypothesis of a unit root at the 5 percent critical value. Using a Wald test, the hypothesis that the post-break series has no time trend cannot be rejected at the 5 percent level. In fact, the estimated coefficient on the post-break trend (i.e., the sum of the coefficients on the two trend variables) is extremely close to zero. However, the hypothesis that the post-break series has no intercept can be rejected at the 1 percent level. Thus, the series again seems to imply that policy would not lead ultimately to the repayment of public debt.

This analysis, based on the period 1970-93, cannot prove that fiscal policy has been sustainable. Once allowance is made for the presence of a structural break, sustainability is violated not because the stock of debt is nonstationary—which would imply that it would eventually spiral out of control—but rather because both series appear to have converged to long-run, positive values. Technically, this violates the sustainability condition because it means that at some point policy will need to be changed to repay the debt. However, the finding that the series are stationary means that no changes in fiscal policy would be required to prevent the stock of debt from increasing without bound.

What conclusion can be drawn from these results? The econometric evidence suggests that during 1970-93 fiscal policy was not on a sustainable path. The informal review of fiscal indicators, however, suggests that sustainability may no longer be a concern now that the overall deficit has disappeared and the ratio of debt to GNP has fallen. One possibility is that a fundamental change has recently occurred in the fiscal regime. Another is that the recent developments are merely temporary deviations from the long-run path. The fact that no major revenue or expenditure measures have been introduced in recent years gives some support to this latter interpretation.7 Instead, much of the improvement in the fiscal position of the public sector has been due to privatization receipts (which in 1995 totaled 1.3 percent of GNP for the public sector), the rundown in national government cash balances (which between 1993 and 1995 reduced the nominal stock of debt by almost P80 billion, more than 4 percent of 1995 GNP), and rapid economic growth (which reduced the interest expenditure of the national government by more than 2 percent of GNP between 1992 and 1995 even as it remained almost constant in peso terms).8 If rapid output growth persists, the ratio of debt to GNP will probably continue to fall. However, in part because privatization receipts are nonrecurrent and because government cash balances are close to minimum working levels, it is by no means clear that any fundamental break with past fiscal policy has occurred. The government itself has recognized that the privatization program is temporary, stating that in addition to enhancing economic efficiency, the program would provide a window of opportunity during which a sustained fiscal reform (such as the proposed comprehensive tax reform) could be initiated.

However, even if the fiscal regime had changed fundamentally in the last few years, so that the first minimum condition for fiscal policy—sustainability—had been achieved, it would not follow that there was no need for further fiscal reform. The reason is that prospective fiscal developments may lead to the violation of the second minimum condition: maintaining government infrastructure investment at 3 percent of GNP while preserving fiscal balance. The following section examines whether the current fiscal stance is compatible with prospective developments.

Prospective Fiscal Developments and Present Fiscal Stance

This section examines whether, given the identifiable trends, the current stance of fiscal policy can be maintained. We characterize this fiscal policy as requiring that government infrastructure investment be equivalent to at least 3 percent of GNP and that fiscal balance be maintained.9 We find that prospective fiscal developments will require major fiscal reform, such as the government’s comprehensive tax reform plan, to maintain this minimum fiscal policy requirement.

Prospective Fiscal Developments

Action may be needed to preserve the current fiscal stance for several reasons. Much of the recent improvement in the fiscal accounts has been based on nonrecurrent revenue sources, such as privatization, and these receipts might decline; revenues from international trade taxes might drop because tariffs are being reduced; civil service salaries are rising; and increases in transfers to local government units are required under the local government code.

Nonrecurrent revenues. About half of the 3 percentage points of GNP increase in revenues of recent years stems from nontax revenues that are nonrecurrent. In the coming years, however, these receipts will decline, reducing revenue by about 1 percent of GNP beginning in 1997 and by about 2 percent of GNP over the medium term. This decline will stem from three sources. First, in 1994, 1995, and 1996, privatization contributed significantly to government revenues. However, the privatization program was largely completed by 1996, and in future years these receipts are likely to be much smaller.10

Second, the government has benefited from large interest earnings on its deposits at the central bank. Until recently, the national government maintained a policy of building up its cash balances at the central bank to help the bank absorb liquidity created through purchases of foreign exchange and through the bank’s own operating losses. However, following the central bank’s financial restructuring in late 1993, the national government has run down its balances to low operating levels, thereby reducing future interest earnings. In addition, even if the national government’s cash balances increase somewhat over time, interest earnings are still likely to decline as a share of GNP as long as recent rapid economic growth persists.

Third, nontax revenues, in the form of fees and charges, are increased only occasionally and may not be sufficient to reverse a steady erosion in real rates. Unless future discretionary adjustments are made to keep fees and charges in line with prices, their contribution to revenues as a share of GNP will fall sharply. Even if fees are adjusted to keep pace with price inflation, fees and charges will fall as a percentage of GNP because of their low elasticity.

International trade taxes. The program of tariff reduction is another important fiscal development. By the year 2003, the maximum standard tariff rate will decline from the current 30 percent to 10 percent. By 2004 it will fall further, to 5 percent. The potential impact on customs revenues can be illustrated using 1995 import data. In 1995, total nonoil imports (c.i.f.) amounted to P690.1 billion. A uniform 5 percent tariff levied comprehensively on these imports would have raised P34.5 billion in customs collections, compared with the P61.1 billion that was actually assessed.11 The difference between the two amounts, of P26.6 billion (1.4 percent of GNP), represents the minimum that would have been lost if the new tariff regime had applied in 1995. It is a minimum revenue loss because the 5 percent tariff is unlikely to be comprehensive. In 1995, about 44 percent of non-oil imports were exempt from duties. At this exemption level, the revenue loss from a 5 percent tariff, compared with prevailing 1995 tariffs, would have equaled 2.1 percent of GNP.12

Transfers to local government units. The growing mandated transfers from the national government to local government units are another important fiscal development. In 1991, the Philippine congress passed a local government code specifying that, by 1994, transfers to local government units would be 40 percent of the internal revenue (tax revenues excluding international trade taxes) collected by the national government three years previously. As this formula has been phased in, transfers to local government units have increased from 0.5 percent of GNP in 1991 to 2.7 percent in 1995, making them the fastest growing item of national government expenditure. The share of local government units in the budget will continue to grow in the next few years, because of a recent change in the tax treatment of oil, which reduced import duties but increased excise taxes (which are part of internal revenue).

Transfers to local government units are also likely to grow rapidly over the medium term because of the decline in revenues from international trade taxes that will accompany the anticipated tariff reform. If total tax revenues are kept constant through increases in domestic tax revenues that are exactly equivalent to the declines in international trade taxes, only 60 percent of the revenue generated by the new domestic taxes will be available for discretionary spending because the balance is transferred to local government units. Accordingly, domestic tax revenues would need to increase by 1.7 percentage points of GNP for every 1 percentage point of GNP decrease in international trade tax revenues if the current, already low levels of discretionary expenditure are to be maintained.

Public debt and interest rates. Over the past few years, the government’s debt-service burden has declined significantly. Interest payments dropped from 6.2 percent of GNP in 1994 to 4.7 percent in 1996, owing in part to the strong fiscal position and in part to a stabilization of nominal interest rates.13 Together, these developments have allowed interest expenditures to remain steady in nominal terms. Meanwhile, rapid growth has enabled the debt-to-GNP ratio to fall.

A reversal of recent favorable interest rate trends could, however, have significant implications for the budget. External debt is largely bilateral fixed-rate debt, and so is not subject to interest rate changes (although exchange rate changes affect debt service). Domestic debt, however, accounts for more than half of the stock of debt and nearly four-fifths of interest payments. This domestic debt is nearly all short term, making the budget position extremely sensitive to shifts in monetary policy or in investor confidence. With national government domestic debt totaling about 37 percent of GNP at the end of 1995, a 1 percentage point increase in domestic interest rates could increase government interest expenditure by nearly 0.4 percent of GNP.

Figure 6 suggests that there is a link between real domestic interest rates and the consolidated public sector deficit. If the factors discussed above—the decline in revenues from nonrecurrent resources or from tariff reform and the continued increase in nondiscretionary expenditure—were to lead to an increase in the deficit, domestic interest rates would most likely increase, placing further pressure on the budget.

Figure 6.Consolidated Public Sector Deficit and Real Interest Rates

Sources: Philippine authorities; and IMF, International Financial Statistics.

Civil service salaries. Under the Salary Standardization Law, civil service salaries are to roughly double between 1994 and 1997. The implementation of the law has already led the government wage bill to increase from about 5.2 percent of GNP in 1993 to a projected 6.0 percent of GNP in 1996, despite rapid real GNP growth. In 1997, expenditure on personnel services is forecast to increase further, to some 6.6 percent of GNP. Although 1997 is the final year of implementation of the law, because much of the 1997 wage increase will be granted only in November, personnel expenditure will increase sharply again in 1998. Thus, between 1993 and 1998, expenditure on personnel services may increase by 1½ percent of GNP. Moreover, even after full implementation, civil service wages are still likely to lag behind those in the private sector, especially for managers. Therefore, unless steps are taken to contain employment, the wage bill could well continue to grow. For this reason, the government plans to streamline the civil service.

Implications of Prospective Developments

An illustrative medium-term scenario has been prepared to examine the impact of these prospective developments. This scenario shows that, without tax reform (but with civil service streamlining), cuts to capital expenditure will be necessary to maintain budget balance. National government infrastructure expenditure falls from about 3.0 percent of GNP in 1995 and 1996 to just 2.2 percent of GNP in 1998 and does not recover to its previous levels until 2000. National government savings remain stagnant, with only minimal improvement for most of the projection period. This means that while the debt-to-GNP ratio continues to decline, because fiscal balance will be maintained, essential infrastructure investment will need to be cut, and improvement in national savings may be retarded.

Table 3 presents a medium-term scenario that illustrates the effects of the developments discussed above. The scenario is predicated on the following assumptions:

  • a nominal elasticity of 1.12 for domestic tax revenues;

  • a revenue-neutral shift of oil taxation from international to domestic taxes beginning in 1997;

  • a gradual decline in revenues from international trade taxes (relative to GNP), reflecting the impact of tariff reforms;

  • a sharp decline in nontax revenues (again, relative to GNP) beginning in 1997, reflecting the cessation of the privatization program, with a gradual decline thereafter, as other forms of nontax revenue fail to keep pace with GNP growth;

  • continued implementation of the local government act, with transfers to local government units representing 40 percent of domestic tax revenues three years earlier;

  • no increase in the real wage bill after 1998. Given that civil service wages are likely to be lower than private sector wages, even after the Salary Standardization Law is implemented, this assumes a reduction in civil service staffing that precisely offsets the impact of further real wage increases;

  • refinancing of the existing stock of debt at the average interest rate prevailing in 1996 (about 6 percent);14

  • cuts in capital expenditure as needed to keep the budget in balance; and

  • constant (relative to GNP) expenditure on all other items.

Table 3.Hypothetical Medium-Term Scenario(In percent of GNP)
19951996199719981999200020012002200320042005
Revenue and grants19.019.318.618.418.418.418.318.318.218.218.3
Tax revenue15.716.316.616.716.716.816.816.916.917.017.0
Domestic-based taxes10.711.412.212.412.512.712.813.013.113.313.4
International trade taxes4.94.84.34.24.14.03.93.83.73.63.5
Other offices0.10.10.10.10.10.10.10.10.10.10.1
Nontax revenue3.33.01.91.71.61.61.51.41.31.31.2
Privatization1.20.90.20.00.00.00.00.00.00.00.0
Central bank interest, rebates, and dividends0.70.70.50.50.40.40.40.30.30.30.2
Other1.41.41.21.21.21.21.11.11.01.01.0
Total expenditure and net lending19.119.118.618.318.318.318.318.318.218.318.2
Current operating expenditure15.315.215.315.114.714.213.713.112.612.111.7
Personnel services5.66.06.66.76.36.05.75.45.14.84.5
Maintenance and operations2.22.22.22.22.22.22.22.22.22.22.2
Interest payments5.14.73.93.53.22.92.62.32.11.91.7
National government3.73.52.92.62.32.11.91.71.51.41.2
Central Bank Board of Liquidators1.41.21.01.00.90.80.70.611.60.50.5
Subsidies0.30.30.40.30.30.30.30.30.30.30.3
Allotments to local governments2.12.02.22.42.62.82.92.92.93.03.0
Capital expenditure and net lending3.83.93.33.23.74.14.65.25.66.16.6
Capital expenditure3.43.53.02.83.33.74.24.85.25.76.2
Infrastructure and other2.82.92.42.22.63.03.54.14.55.05.4
Allotments to local governments0.60.60.60.60.70.70.70.70.70.70.8
Equity and net lending0.40.40.30.40.40.40.40.40.40.40.4
Balance-0.10.20.00.00.00.00.00.00.00.00.0
Financing0.1-0.20.00.00.00.00.00.00.00.00.0
Memorandum item:
GNP1,971.02,266.72,547.72,838.23,150.43,496.93,881.54,316.34,804.05,351.75,961.8
Inflation6.55.55.05.05.05.05.05.05.0
Real GNP growth5.95.96.06.06.06.26.36.46.4
Debt stock1,313.01,316.21,310.81,310.81,309.71,308.41,307.51,307.91,310.31,311.31,312.3
Marginal interest rate5.512.012.012.012.012.012.012.012.012.012.0
Debt-GNP66.658.151.546.241.637.433.730.327.324.522.0
Savings-GNP2.53.23.13.23.74.14.65.25.66.16.6
Sources: Philippine authorities; and authors’ estimates.
Sources: Philippine authorities; and authors’ estimates.

Under this scenario, cuts in capital expenditure will be necessary to preserve fiscal balance. In 1997, revenue declines by 0.7 percent of GNP, owing to the drop in privatization receipts, along with modest declines in other nontax revenues. Meanwhile, current expenditure remains largely constant, with the combined effects of higher expenditure for personnel services and an increased allotment for local government units that offsets the benefits of continued declines in interest expenditure. As a result, capital expenditure would have to be cut by ½ of 1 percent of GNP to maintain the fiscal balance. Additional cuts would be required in 1998. By 1999, declines in interest expenditure and personnel services would begin to outstrip the fall in trade taxes and nontax revenues, and capital expenditure would begin to increase. Nevertheless, only in 2000 would capital expenditure recover to its projected 1996 level.

The scenario illustrates that present fiscal policy, defined as fiscal balance with infrastructure investment of at least 3 percent of GNP, cannot be maintained. Fiscal reforms will be necessary if one or both of these conditions are not to be violated. This reduced infrastructure investment would probably have implications for output growth, although it is difficult to quantify them precisely.15

Furthermore, there are three reasons why—without reform—the reduction in capital expenditure and growth might be larger than envisaged. First, the scenario is predicated on continued rapid output growth throughout the projection period. Growth could be substantially lower than assumed if private investment, labor supply, or total factor productivity does not increase by the strong margins assumed in the scenario. Slower growth would imply lower revenues and lower capital spending.

Second, the scenario assumes no increase in domestic interest rates. Although the cuts in investment assumed in the scenario are substantial, they would be much larger were it not for the persistent decline in interest expenditure over the projection period. Higher interest rates would require either additional cuts in capital expenditure or an increase in the deficit. If the latter occurred, this fiscal deterioration could provoke another increase in interest rates, swelling the deficit even further and creating a vicious circle. In addition, debt ratios could rise if interest rates exceed nominal output growth, even though the national government would be running a primary surplus.

Third, the scenario assumes no real increase in expenditure for personnel services. As a result, all of the decline in the ratio of personnel expenditure to GNP after 1998 is available for increased capital expenditure. If real wages are to be increased further at any point after 1998, the scenario assumes enough of a reduction in the level of civil service employment to keep real expenditure constant. Otherwise, if capital expenditure is not reduced, the deficit will increase.

This scenario sheds some light on the key question raised earlier—whether the recent improvement in fiscal indicators represented a new regime or a temporary deviation from the long-term path. Although not conclusive, the scenario suggests that it may well represent the latter. Certainly, it illustrates that without fiscal reform it will be impossible to maintain the current fiscal stance. Instead, investment will need to be cut, possibly by large amounts, if growth falters, interest rates rise, or personnel spending increases.

Growth and Savings Objectives

The previous section focused on minimum conditions for fiscal policy. It found that even if fiscal policy is sustainable—in the sense that public debt obligations can be met—the present policy of maintaining fiscal balance while keeping government investment expenditure at 3 percent of GNP cannot continue. Accordingly, a shift in the fiscal regime will be necessary. This section looks at the fiscal policy that would be required to help achieve the government’s broad macroeconomic goals, especially for growth and savings.

Investment and Growth

This section looks at public investment and the capital stock in the Philippines and its neighbors and concludes with an estimate of the level of government capital expenditure that would be required to support continued rapid output growth.

The required level of public investment depends on the overall economic strategy. In countries where the public sector provides nearly all the required infrastructure, public investment will naturally tend to be higher than in those—such as the Philippines—where the private sector plays an important role. Still, the gap between public infrastructure spending in the Philippines and its ASEAN neighbors is too large to be explained solely by different economic roles of the state.

On average, capital expenditure in other ASEAN countries is more than 2 percentage points of GNP higher than in the Philippines. These countries also tend to devote a larger share of government expenditure to capital spending than in the Philippines, largely because of the significant interest expenditure incurred by the Philippine government. Table 4 shows data on average levels of central government capital expenditure as a share of GDP in five ASEAN countries during 1990-93 (the last year for which International Monetary Fund Government Finance Statistics data are available for all countries).

Table 4.Government Capital Expenditure in ASEAN Countries(Averages, 1990–93)
As Share of
Government spendingGDP
Indonesia47.58.5
Malaysia20.85.8
Singapore21.94.5
Thailand23.23.6
Average28.45.6
Philippines17.03.3
Source: IMF, Government Finance Statistics Yearbook, 1995
Source: IMF, Government Finance Statistics Yearbook, 1995

Low current-period expenditure on capital investment would be less of a concern if the stock of public capital were adequate. However, the relatively low level of capital expenditure has persisted in the Philippines for some time: government investment averaged just 2.8 percent of GNP between 1977 and 1995. As a point of reference, national government investment averaged 3.9 percent of GNP in Thailand over the same period. Based on reasonable assumptions, this means that the stock of capital is about $500 a person higher in Thailand than in the Philippines. Allowing for a moderate rate of depreciation, the stock of national government capital in Thailand equaled about $757 a person at the end of 1995, compared with about $268 a person in the Philippines.16 The accumulated shortfall of about $500 a person represents a difference of some $33 billion in national government investment.17

To close the gap between the per capita public capital stocks in the Philippines and Thailand in the next ten years, national government investment would need to equal about $5.3 billion a year at current exchange rates, or double its current level. To close the gap over the next twenty years would require national government investment of about $4.7 billion a year, an increase of nearly 80 percent over current investment levels (to 6.3 percent of GNP from 3.4 percent of GNP in 1995). Moreover, these figures assume no growth in population. Rapid population growth could dramatically increase the investment requirement. Even without population growth, it would take ten years of doubled national government investment to reach the per capita stock of national government capital that now prevails in Thailand. As current investment is already much higher in per capita terms in Thailand than in the Philippines, under existing policies the gap will only widen.

Anecdotal evidence also suggests that the stock of public capital in the Philippines lags behind that of other lower-middle-income countries. For example, the paved-road density in the Philippines (the number of kilometers of paved road for every 1,000 inhabitants) was only 242 in 1992 according to the World Bank (1995b), compared with 841 in Thailand, 509 in the Republic of Congo, 476 in Ecuador, and 542 in Senegal.18 That same year, according to the World Bank, the number of telephone mainlines for every 1,000 inhabitants was only 10.3 in the Philippines, compared with 112.0 in Malaysia and 31.0 in Thailand.

The energy sector will also place considerable demands on public investment. Although independent power producers have alleviated the electricity shortages of the early 1990s, capital requirements for power generation are expected to equal about $10 billion between 1994 and 2000. Although much of this expenditure will be undertaken by independent producers or newly privatized elements of the National Power Corporation, investment in transmission (which will remain a public sector monopoly) is expected to exceed $3 billion over the same period. In sum, it appears unlikely that prevailing rapid rates of real GNP growth can persist without an increase in government capital expenditure.

The implications of low levels of current investment for the stock of national government capital are probably compounded by low maintenance and operations expenditure. Low maintenance and operations outlays accelerate rates of depreciation of the capital stock. For example, about 40 percent of national concrete roads are reported to be in only fair-to-poor condition, along with 75 percent of national asphalt roads and 100 percent of gravel roads. Maintenance expenditure on roads was $33,500 a kilometer in 1994, less than half the amount required. Similarly, operations and maintenance expenditure by the Metropolitan Waterworks and Sewerage System is estimated at $0.04 a cubic foot of water, about half of what is spent in Bangkok and the lowest amount in the region. As a result, the water system is expected to require substantial amounts of new investment in the near future.19

By how much must national government investment increase to achieve the government’s growth target of about 7 percent? Using the growth accounting framework set out in Appendix II, it would appear that an increase in government investment to about 9 percent of GNP by the end of the forecast period would be called for. Such a large increase in investment is probably not feasible. Of course, such estimates of the desirable government investment levels are very sensitive to the assumptions of productivity growth, implementation capacity, and the ability of the private sector to take over some investment projects from the public sector. For example, if total factor productivity growth increases from the assumed 1 percent a year over the projection period to 1.5 percent, national government investment needs to rise only to about 5 percent of GNP by the end of the projection period—an increase of about 1½ percent over its 1995 level—in order to maintain real growth of 7 percent annually. At 5 percent of GNP, national government investment in the Philippines would still be below the ASEAN average for 1990-93, which should be seen as a minimum target. Indeed, given the critical shortage of infrastructure capital in the Philippines, the rate of return on national government investment is likely to be quite high, particularly if it is accompanied by fiscal policies that help preserve macroeconomic stability and confidence.

Requirement of Higher Savings

Fiscal policy will play a key role in ensuring that macroeconomic stability is preserved through its implications for national savings and the external accounts. As the domestic saving-investment gap widens, foreign savings must necessarily increase. However, if substantially increased access to foreign savings is unlikely or undesirable, fiscal policy will be inconsistent with macroeconomic stability. This section discusses the interaction between public and private savings to develop a sense of the increase in public savings required to ensure that fiscal policy is compatible with macroeconomic objectives, including external policy. It concludes with some indications of how the required increase in savings can be achieved.

Savings and Growth

Higher domestic saving rates may have direct, favorable implications for economic growth for two reasons. The first reason is that the current favorable debt dynamics in the public sector would be preserved. Although an exact rule for determining a country’s maximum sustainable current account deficit is unavailable, it seems that large increases in the current account could not be financed without a significant increase in domestic interest rates. This would undo the favorable debt dynamics the Philippines has enjoyed in recent years, potentially creating a vicious circle in which higher interest rates increase the public sector deficit, lowering confidence and further increasing interest rates. In addition, the foreign savings attracted by higher interest rates would probably be short term, increasing the economy’s vulnerability to shocks. The second benefit of higher domestic savings is suggested by the considerable econometric evidence pointing to the relationship between saving and growth rates in other ASEAN countries. This result may seem surprising for an open, modern economy. Of course, in a closed economy, savings and investment are inextricably linked, and if higher rates of investment imply higher long-run growth rates (as in some recent models of endogenous growth), a correlation between saving rates and output growth follows.20 However, in an open economy there is no reason for domestic saving and investment rates to be equal or, it would appear, for high domestic saving rates to correlate with rapid output growth in the long run. Nevertheless, research by a number of economists shows that even in highly open economies with ready access to international capital markets, domestic savings and domestic investment are closely linked.21

National Government Savings Versus Private Savings

An increase in government investment matched by a similar increase in government savings would most likely have implications for the external current account balance, because declines in private savings may offset part of the increase in national government savings. For example, if full Ricardian equivalence holds, private savings will fall enough to offset the entire increase in government savings. In that case, any increase in national government investment would directly worsen the current account. The larger the extent of the offset, the greater the degree to which higher government savings reduce private savings, and the more significant the impact of higher national government investment on the current account.

Empirical studies that have attempted to estimate the size of the offset between public and private savings have obtained different results, depending on the countries and periods covered, the variables included, and the econometric techniques employed. Dayal-Gulati and Thimann (see Chapter 7 in this volume) examine a cross section of ASEAN countries and find an offset of only about 0.2 (meaning, for example, that a P1 increase in national government savings would lead to an increase of 80 centavos in national saving). The same study finds an offset of about 0.7 for Latin American countries, which—with their relatively high debt stocks—may share some of the Philippines’ characteristics. In preliminary empirical work using only Philippine data for 1978-95, we obtained an offset of 0.76. A recent cross-sectional study of saving in countries with IMF programs obtained an estimated offset of 0.70 (see Savastano, 1995).

The high domestic saving rates of other ASEAN countries will not be achieved in the Philippines in the near future. However, the larger the extent of the offset between public and private savings, the more public savings will need to be increased to mobilize a given level of national savings. If the offset is, indeed, as large as about 0.7, a very significant increase in government savings would be called for.

How Much Should Public Savings Increase?

A strong case can be made for increasing government savings substantially in the Philippines: between 1991 and 1995, public savings averaged only about 3 percent of GNP, compared with 7 percent of GDP in Indonesia and 12-13 percent in Malaysia, Singapore, and Thailand. It is unlikely that the Philippines can achieve the high saving rates of its neighbors in the next several years, but it should nevertheless be possible to close much of the gap relative to the ASEAN norm. Higher savings would help preserve the favorable debt dynamic and would also reduce the risk to the economy of excessive reliance on foreign savings. An increase in government savings of 6 percent of GNP over the next ten years (relative to the 1996 level of 3.2 percent of GNP) would bring the Philippines much closer to the average level of public savings in ASEAN countries. In addition, even if the offset between public and private savings is as large as about 0.7, higher national government savings of about 6 percent of GNP would help ensure that an increase of 1½ percent of GNP in national government investment does not lead to a worsening of the current account deficit. To the extent that the offset is less than 0.7, an increase in government savings of about 6 percent of GNP would strengthen the current account balance.

An illustrative scenario, shown in Table 5, achieves 7 percent real output growth supported by higher government savings and higher investment. Under the scenario, national government investment increases by 1½ percent of GNP between 1996 and 2005, while national government savings increase by 6 percent over the same period. The government runs a substantial fiscal surplus throughout the projection period, which reaches nearly 4 percent of GNP by the end of the projection period. In the scenario, these surpluses are used to reduce the national government debt, so that by the end of the projection period the debt-to-GNP ratio of the national government falls to 11 percent, half the level in the scenario described earlier (Table 3). The scenario does not assume that domestic interest rates fall because of the fiscal surplus. If this were to happen, government interest expenditure would decline, further increasing the surplus and possibly creating a virtuous circle.

Table 5.Alternative Hypothetical Medium-Term Scenario(In percent of GNP)
19951996199719981999200020012002200320042005
Revenue and grants19.019.319.119.219.419.719.820.120.320.420.6
Tax revenue15.716.317.217.517.818.118.418.719.019.219.4
Domestic-based taxes10.711.412.713.113.513.914.314.715.115.415.7
International trade taxes4.94.84.44.34.24.14.03.93.83.73.6
Other offices0.10.10.10.10.10.10.10.10.10.10.1
Nontax revenue3.33.01.91.71.61.61.41.41.31.21.2
Privatization1.20.90.20.00.00.00.00.00.00.00.0
Central bank interest, rebates, and dividends0.70.70.50.50.40.40.30.30.30.20.2
Other1.41.41.21.21.21.21.11.11.01.01.0
Total expenditure and net lending19.119.119.219.018.718.518.217.917.517.116.9
Current operating expenditure15.315.215.315.014.514.113.513.012.411.911.4
Personnel services5.66.06.66.76.35.95.65.24.94.64.3
Maintenance and operations2.22.22.22.22.22.22.22.22.22.22.2
Interest payments5.14.73.93.53.12.82.52.11.81.51.2
National government3.73.52.92.62.32.01.81.51.31.00.8
Central Bank Board of Liquidators1.41.21.00.90.80.80.70.60.50.50.4
Subsidies0.30.30.40.30.30.30.30.30.30.30.3
Allotments to local governments2.12.02.22.32.62.93.03.13.21.33.4
Capital expenditure and net lending3.83.93.94.04.14.44.75.05.15.25.4
Capital expenditure3.43.53.63.63.74.04.34.64.74.85.0
Infrastructure and other2.82.93.03.03.13.33.53.83.94.04.2
Allotments to local governments0.60.60.60.60.60.70.80.80.80.80.8
Equity and net lending0.40.40.30.40.40.40.40.40.40.40.4
Balance-0.10.20.00.20.81.11.72.22.73.33.8
Financing0.1-0.20.0-0.2-0.8-1.1-1.7-2.2-2.7-3.3-3.8
Memorandum items:
GNP1,971.02,266.72,563.62,876.33,212.93,588.84,008.74,481.75,010.55,606.86,268.4
Inflation6.55.55.05.05.05.05.05.05.0
Real GNP growth6.66.76.76.76.76.86.86.96.8
Debt stock1,313.01,316.21,310.81,311.51,306.51,278.41,232.61,158.31,049.9895.7687.7
Marginal interest rate5.512.012.012.012.012.012.012.012.012.012.0
Debt-GNP66.658.151.145.640.735.630.725.821.016.011.0
Savings-GNP2.53.23.74.14.95.66.37.17.88.59.2
Sources: Philippine authorities; and authors’ estimates.
Sources: Philippine authorities; and authors’ estimates.

Increasing National Government Savings

The previous section suggested that an increase in national government savings of about 6 percent of GNP over the next ten years would help the Philippines achieve its macroeconomic and growth objectives. It can secure some of this increase through a decline in interest expenditure relative to GNP as well as through the natural elasticity of the tax system. In addition, however, revenue or expenditure measures will be required. Given the relatively low amount of noninterest current expenditure in the budget (about 10 percent of GNP), it seems likely that most of the increase in savings will need to come from higher government revenues. (In addition, both scenarios discussed above already assume that some measures will be taken on the expenditure side, since expenditure on personal services is assumed to be constant in real terms.) To the extent that domestic tax measures are introduced to raise savings, the required amount of measures will increase, because of the associated rise in transfers to local government units. In the scenario presented in Table 5, revenue measures of about 2½ percent of GNP are required over the next ten years to enable the Philippines to achieve the required increase in national government savings. Together with the tax elasticity assumptions, this would increase domestic tax revenues from about 11½ percent in 1996 to close to 16 percent in 2005. The government will make significant headway in boosting national savings if it is successful in gaining legislative passage of its comprehensive tax reform plan.

In the illustrative scenario presented in Table 5, most of the revenue measures come in the first five years: by 2001, revenue measures of 1½ percent have been introduced. Even these are relatively front loaded, with measures of ½ of 1 percent of GNP in 1997 alone. By 2001, savings have increased by 3 percent of GNP, owing in equal parts to revenue measures and decreasing current expenditure. After 2001, however, most of the additional 3 percent of GNP increase in national government savings comes on the expenditure side. While revenue measures of about 1 percent of GNP are still required, current expenditure declines by 2 percentage points of GNP simply because of a continued reduction in interest and personnel expenditure. The scenario therefore depends on the early introduction of tax and civil service reform, along the lines currently proposed by the government. To the extent that the introduction of tax reform will create a system that is more elastic and easier to administer, the need for explicit measures in later years could be reduced or even eliminated. As the following paragraphs will illustrate, there is clearly scope for increasing tax revenues through improved administration in the Philippines.

National government revenues are lower in the Philippines than in other ASEAN countries, and there is considerable potential for increasing them (especially with respect to tax revenues). Total tax revenues were about 15.6 percent of GNP in the Philippines in 1994, compared with 21.4 percent of GDP in Malaysia, 17.3 percent in Thailand, 16.7 percent in Singapore, and 15.6 percent in Indonesia.22 Not only were total tax collections lower in the Philippines than in most other ASEAN countries, but the share of collections coming from international trade taxes was much higher than in any other country. The government is currently exploring a number of tax reform options.23

First, corporate income tax collections are lower in the Philippines than in most other ASEAN countries. In 1994, corporate income taxes totaled about 2.4 percent of GNP, compared with 6.9 percent of GDP in Malaysia, 6.5 percent in Indonesia, 3.8 percent in Thailand, and 2.5 percent in Singapore. Widespread exemptions, tax holidays, and evasion accounted for the low collections in the Philippines, as the corporate income tax rate prevailing at the time, 35 percent, was the same as in Indonesia and was higher than that in Malaysia (32 percent), Thailand (30 percent), and Singapore (27 percent).

Second, excise tax collections in the Philippines equal about 2 percent of GNP and are similar to those in Indonesia, Malaysia, and Singapore. However, relative to GDP, collections in Thailand are nearly twice as large as those in the Philippines, suggesting considerable potential for increase.

Third, collections from the individual income tax totaled about 1.7 percent of GNP in the Philippines in 1994, about the same as in Indonesia and Thailand. However, individual income tax collections equaled 5.2 percent of GDP in Singapore and 2.5 percent of GDP in Malaysia.

Fourth, the yield of the value-added tax (VAT) is much lower in the Philippines than in Indonesia and Thailand, the only other ASEAN countries with longstanding VATs. Collections from the VAT totaled 2.7 percent of GNP in the Philippines in 1994, compared with 4.8 percent of GDP in Indonesia and 3.3 percent of GDP in Thailand. Thailand was in fact able to collect slightly more from the VAT than the Philippines, although the basic rate on its VAT is only 7 percent compared with 10 percent in the Philippines. Put another way, the efficiency ratio for the VAT—collection as a share of GDP divided by the basic rate—was only 0.27 in the Philippines compared with 0.47 in Thailand and 0.48 in Indonesia. Administrative weaknesses and exemptions (which have since been narrowed by the introduction of the expanded value-added tax) account for the low efficiency of the tax in the Philippines.

Finally, nontax revenues in the Philippines are unlikely to reach the levels obtained in other ASEAN countries. Indonesia and Malaysia benefit from large petroleum-related receipts, while Singapore receives considerable investment income from its sizable foreign exchange reserves and overseas investments. (Excluding privatization, nontax revenues totaled about 11.4 percent of total revenues and grants in the Philippines in 1994, compared with 15.9 percent in Indonesia (in 1993), 25.5 percent in Malaysia, and 27.6 percent in Singapore.)

Conclusion

Fiscal trends in recent years in the Philippines are encouraging: deficits and the stock of debt have declined as a share of GNP and output growth has been strong. Based on these indicators, fiscal solvency would not seem to be a concern. Yet quantitative analysis, based on long-term trends, suggests that the fiscal stance may not be sustainable. Moreover, a number of prospective developments suggest that the current fiscal stance cannot be maintained into the future. Important sources of revenue are drying up, the wage bill is increasing, and the national government is transferring a significant portion of its resources to local governments. Consequently, fiscal reform, such as the government’s planned comprehensive tax reform and civil service streamlining, is needed to avoid cuts in infrastructure investment that would constrain the rate of sustainable economic growth.

The government wishes not only to consolidate the present fiscal position, but also to increase the rate of sustainable economic growth. This will require a significant increase in national government infrastructure investment. To secure this increase in public investment, alongside an appropriate level of government savings and a sustainable current account balance, requires meaningful increases in revenues, cuts in current expenditure, or both in the medium term. An illustrative scenario suggests that these objectives will require an increase in government infrastructure investment of at least 1½ percent of GNP over the next ten years and that national government savings should increase by about 6 percent of GNP. We estimate that measures of about 2½ percent of GNP would be required over the next ten years to achieve the targeted increase in savings. The low level of noninterest expenditure means that most of the adjustment will need to come on the revenue side. Fortunately, a comparison with other ASEAN countries suggests that there is substantial scope for increasing tax revenues in the Philippines.

Appendix I. Deriving a Testable Equation for Assessing Fiscal Sustainability

Equation (1), in the main text, is the government budget constraint identity. The constraint facing the government takes the form

where Bt is domestic-currency-denominated government debt; Bt* is foreign-currency-denominated government debt in foreign currency; Xt is the current-period exchange rate; rt-1 is the ex post interest rate on domestic-currency-denominated debt; rt1* is the ex post interest rate on foreign-currency-denominated government debt; St is the primary surplus of the government; and ΔCt is the change in government cash balances.

Defining Dt=Bt+XtBt* as the current stock of public sector (foreign and domestic) debt, valued in domestic currency units, we can manipulate equation (1) to obtain

where εt is the proportional change in the exchange rate in period t; that is, εt - ΔXt/Xt. Alternatively, we can express equation (2) as

where

and where S¯t is defined as the “augmented primary surplus.” The primary surplus is augmented to account for changes in debt that occur on account of deviations from uncovered interest parity and changes in the consolidated public sector’s cash balances.

Define qt as the discount factor from period t back to period 0. That is,

Multiplying equation (3) by qt to discount all variables to period 0 gives

or

where dt is the discounted value of the stock of debt at period t, and s¯t is the discounted period t augmented primary surplus. In other words, in each period the change in discounted debt is equal to the discounted primary surplus.

Rearranging equation (4) to take the form dt+s¯t=dt1 and then substituting forward iteratively yields

In an economy with a finite horizon, the government solvency condition requires that the stock of debt be nonpositive in the final period, that is, that dT ≤ 0. This is the “No Ponzi Game” condition that prevents the government from running up a large stock of debt and then defaulting in the final period. In a dynamically efficient economy with an infinite horizon, we instead impose the transversality condition

where Et equals the expectation operator conditional on information as of time t. After imposing this condition, we are left with the familiar present-value government budget identity

which states that the expected value of future augmented primary surpluses must be equal to the current stock of debt. If the stock of debt is positive, this means that the government will need to run primary surpluses at some point in the future to avoid insolvency.

In the empirical analysis, we assume that dt follows a multivariate ARIMA process (like Wilcox, 1989; and Buiter and Patel, 1992):

where ρ(L) is a pth-order polynomial, θ(L) is a qth-order polynomial, and γ0 is the unconditional mean of the stationary series (I - L)mZt. The vector Zt includes dt as its first—but not necessarily only—element; et is a vector white-noise process. Both ρ(L) and θ(L) are assumed to satisfy the requirements for stationarity and invertibility, so that equation (6) can be represented as

μ(L)[(I - L)mZt - γ0] = et,

where

μ(L) = [I - θ(L)]-1[I - ρ(L)].

The solvency conditions are therefore twofold: first, Zt must be stationary. Second, the first element of γ0 must be equal to zero. We use the special case of equation (2):

dt = μ + βt + αdt-1 + ut.

In this case, discounted public debt will be nonstationary if α ≥ 1. In addition, even if α<1, discounted public debt will be nonstationary if there is either a nonzero deterministic trend (i.e., if β ≠ 0) or nonzero drift (i.e., if µ ≠ 0).

Appendix II. A Growth Accounting Framework

A simple economic growth accounting framework is based on measures of the historical impact of increases in the real stock of capital and the supply of labor on real output. Based on investment, output, and labor supply data for 1961-95, we decompose real output growth into three sources: increases in the stock of capital, increases in the supply of labor, and changes in total factor productivity (TFP). Assuming that output is produced through a Cobb-Douglas production function of the form

Y = AKαL1-α,

where Y equals real GDP, A equals real TFP, K equals the real stock of capital, and L equals the supply of labor, we can show that

Thus, regressing the change in real output per worker on a constant and the change in real capital per worker gives an estimate of the average change in TFP over the period, and an estimate of α—which is the contribution of a 1 percent increase in the real stock of capital to the real growth rate of output. We find that the average change in TFP was negative during 1961-95 and that α = 0.60, which implies that 1 - α—the contribution to growth of a 1 percentage point increase in the supply of labor—is 0.40.24

Although the average change in TFP was negative during 1961-95, it is possible to identify three subperiods during which the change displayed varying behavior. Between 1961 and 1983, changes in TFP had relatively little impact on real output growth, averaging only -0.3 percent. Between 1984 and 1993, changes in TFP typically had a sharply negative effect on output growth, averaging -1.9 percent.25 Finally, in the last two years, the change in TFP has become positive, contributing an average of 1.2 percent to the real growth rate. The results of the growth accounting exercise also make it clear that investment was responsible for nearly all of the real output growth in the Philippines during 1961-95, accounting for 3.4 percent of the average real growth rate of 3.9 percent over that period.

References

    AdlerMichael and BernardDumas1983“International Portfolio Choice and Corporation Finance: A Synthesis,”Journal of FinanceVol. 38 (June) pp. 92584.

    BayoumiTamim1990“Saving-Investment Correlations: Immobile Capital, Government Policy, or Endogenous Behavior?”Staff PapersInternational Monetary FundVol. 37 (June) pp. 36087.

    BuiterWillem H. and Urjit R.Patel1992“Debt, Deficits, and Inflation: An Application to the Public Finances of India,”Journal of Public EconomicsVol. 47 (March) pp. 171205.

    CumbyRobert E. and MauriceObstfeld1984“International Interest Rate and Price Level Linkages Under Flexible Exchange Rates: A Review of Recent Evidence,” in Exchange Rate Theory and Practiceed. byJ.Bilson and R.Marston (Chicago: University of Chicago Press).

    DickeyDavid A. and Wayne A.Fuller1979“Distribution of the Estimators for Autoregressive Time Series with a Unit Root,”Journal of the American Statistical AssociationVol. 74 (July) pp. 42731.

    DickeyDavid A. and Wayne A.Fuller1981“Likelihood Ratio Statistics for Autoregressive Time Series with a Unit Root,”EconometricaVol. 49 (July) pp. 105772.

    FeldsteinMartin S.1994“Tax Policy and International Capital Flows,”NBER Working Paper No. 4851 (Cambridge, Massachusetts: National Bureau of Economic Research).

    FeldsteinMartin S. and Charles S.Horioka1980“Domestic Savings and International Capital Flows,”Economic JournalVol. 90 (June) pp. 31429.

    FrenchKenneth R. and James M.Poterba1991“Investor Diversification and International Equity Markets,”NBER Working Paper No. 3609 (Cambridge, Massachusetts: National Bureau of Economic Research).

    FullerWayne A.1976Introduction to Statistical Time Series (New York: Wiley).

    HamiltonJames D. and Marjorie A.Flavin1986“On the Limitations of Government Borrowing: A Framework for Empirical Testing,”American Economic ReviewVol. 76 (September) pp. 80819.

    International Monetary FundGovernment Finance Statistics Yearbook (Washingtonvarious years).

    International Monetary FundInternational Financial Statistics (Washingtonvarious issues).

    International Monetary FundPhilippines: Recent Economic Developments (Washingtonvarious years).

    International Monetary FundThailand: Recent Economic Developments (Washingtonvarious years)

    MishkinFrederic S.1984“Are Real Interest Rates Equal Across Countries? An Empirical Investigation of International Parity Conditions,”Journal of FinanceVol. 39 (December) pp. 134557.

    PerronPierre1989“The Great Crash, the Oil Price Shock, and the Unit Root Hypothesis,”EconometricaVol. 57 (November) pp. 1361401.

    PhillipsPeter C.B. and PierrePerron1988“Testing for a Unit Root in Time Series Regression,”BiometrikaVol. 75 pp. 33546.

    SavastanoMiguel A.1995“Private Saving in IMF Arrangements” in IMF Conditionality: Experience Under Stand-By and Extended Arrangementsed. byS.SchadlerOccasional Paper No. 129 (Washington: International Monetary Fund).

    SmithRoger S.1989“Factors Affecting Saving, Policy Tools, and Tax Reform: A Review,”IMF Working Paper 89/47 (Washington: International Monetary Fund).

    WilcoxDavid W.1989“The Sustainability of Government Deficits: Implications of the Present-Value Borrowing Constraint,”Journal of Money Credit and BankingVol. 21 (August) pp. 291306.

    World BankWorld Debt Tables (Washingtonvarious years).

    World Bank1995a“Philippines: Public Expenditure Management for Sustained and Equitable Growth” (Washington).

    World Bank1995bWorld Development Indicators 1995 (Washington).

Note: Philip Gerson is an Economist in the IMF’s Fiscal Affairs Deparment, and David Nellor is the IMF’s Resident Representative in Manila. The authors thank, without implication, Gil Beltran, Neil Ericsson, Joshua Felman, John Hicklin, Sandy Mackenzie, David Robinson, Mark Stone, Amando Tetangco, and participants in a departamental seminar for their useful comments on earlier drafts. They would also like to thank the Philippines’ Department of Finance, especially Gil Beltran, for providing the data.

The discounted stock of debt refers to the stock of debt converted to constant 1970 dollars using the dollar interest rate applying each year to new public sector debt.

Including direct, assumed, and contingent liabilities, but excluding debts related to the restructuring of the central bank.

See, for example, Hamilton and Flavin (1986) and Wilcox (1989) on the United States and Buiter and Patel (1992) on India.

Bascand and Razin examine fiscal sustainability in Indonesia (see Chapter 5 in this volume). Their model begins from the same budget constraint as in this paper, but adopts a more static approach, calculating for each annual augmented primary surplus (as defined in Appendix I) the associated steady-state stock of debt, given interest rates and nominal output growth rates. This approach works well when the augmented primary surplus is relatively stable and the nominal rate of interest consistently exceeds the rate of nominal output growth. These conditions have not held in the Philippines.

Rather than attempting to characterize steady states, this paper examines time-series properties of the stock of debt in order to characterize the data-generating process and draw inferences about the long-run value of debt.

The tests examined both the discounted stock of public sector debt and the ratio of public sector debt to GNP using time series of 24 observations covering 1970-93 (see Figure 2 and 3). Data on public sector debt were provided by the Philippine authorities. The data were converted into dollar terms using end-of-period exchange rates. The debt figures are discounted using the “all creditors” rate from the World Bank’s World Debt Tables, which represents the weighted average interest rate on all new long-term debt contracted by the national government in a given year.

Perron (1989) provides the critical values for this test. Note that for this test to be appropriate the structural break should be exogenous. Any endogenous breaks should be captured by the data-generating process. As noted above, the break in the series is due to the assumption of public-sector-guaranteed debt. Thus, the analysis implicitly treats the debt crisis as an exogenous event, rather than as an element of fiscal policy.

Between 1992 and 1995, national government tax revenues increased only from 15.1 percent of GNP to 15.8 percent, while over the same period noninterest expenditure increased from 13.0 percent to 14.1 percent of GNP.

Excluding interest expenditure related to the restructuring of the Central Bank of the Philippines.

In recent years, national government infrastructure investment has averaged about 3 percent of GNP.

Ideally, privatization receipts should be treated as financing rather than as revenue.

Excluding the impact of tax credits on collections.

The impact of tariff reform on medium-term revenues could be mitigated by a number of factors. For example, a rapid increase in the tax base would offset some of the decline in tariff rates if imports were to grow more rapidly than GNP in peso terms. The tariffication of existing quantitative restrictions will contribute additional revenues.

This figure includes interest expense of the Central Bank Board of Liquidators.

This assumes, perhaps optimistically, that existing concessional debt can be rolled over at concessional terms.

A growth accounting framework—set out in Appendix II—allows us to determine the rate of real output growth consistent with these investment figures. However, these results should be viewed as illustrative only because they are subject to considerable uncertainty. The framework uses the following assumptions:

  • growth in total factor productivity (TFP) of 1 percent a year over the projection period (higher than the historical average but in line with the experience of the last two years);

  • labor supply growth of 3.3 percent a year (which exceeds the historical population growth rate and therefore implies a steady decline in the rate of unemployment);

  • an increase in nonnational government fixed investment—which is assumed to include build-operate-transfer investments—from about 19.3 percent of GNP in 1996 to 21.5 percent of GNP by 2005; and

  • a slow decline in the growth of net factor income from abroad.

In these circumstances, the cuts in capital expenditure could cause GNP growth to slow from its current rate of about 7 percent to an average rate of 6 percent during 1997-2005.

This comparison is intended not to suggest that the national government capital stock in Thailand is an appropriate target, but rather to assist in the interpretation of the Philippine data.

Real current investment data for both Thailand and the Philippines were obtained from International Monetary Fund, Recent Economic Developments reports for various years. These data were converted to 1995 national currency values and then converted into U.S. dollars at the end-1995 exchange rate. The capital stock was estimated by assuming 5 percent straight-line depreciation.

The shortfall in public investment is even greater: the per capita stock of public capital was about $1,432 in Thailand at the end of 1995, as opposed to about $458 in the Philippines.

In part, the lower paved-road density in the Philippines might be explained by the greater importance of sea transport because of the country’s many islands.

Figures in this paragraph come from the recent World Bank public expenditure review (World Bank, 1995a).

In neoclassical models with very slow convergence to the steady state, higher rates of investment would also correspond to higher rates of growth over a long period.

Feldstein and Horioka (1980), using data on member countries of the Organization for Economic Cooperation and Development (OECD) during the 1960s and 1970s, find that there is a strong, positive correlation between levels of savings and investment in countries, with changes in domestic savings leading almost dollar for dollar to changes in investment. Subsequent work by Smith (1989), Bayoumi (1990), and others has confirmed this result. Feldstein (1994) reports regressions of foreign direct investment and direct investment abroad on domestic saving rates among OECD countries and finds that neither is significantly affected by domestic savings. In addition, Mishkin (1984) and Cumby and Obsrfcld (1984), among others, have demonstrated the existence of persistent differences in real rates of return on investments internationally, which implies that capital is not perfectly mobile. Finally, Adler and Dumas (1983) and French and Poterba (1991) have found that individual investors are much more likely to hold domestic than foreign securities, despite the benefits of international diversification.

However, a large share of revenues in Indonesia was derived from taxes on oil.

Except for Singapore, or where otherwise indicated, all data are for 1994. Data for Singapore are for 1993. Data for the Philippines come from the Philippine authorities. All other data come from the International Monetary Fund, Government Finance Statistics.

As would be expected with a constant returns-to-scale production function, the estimated contributions to growth of 1 percentage point increases in the stock of capital (0.60) and labor (0.40) are very close to the shares of national income accruing to capital (0.61) and labor (0.39) as derived from the National Income Accounts.

However, between 1986 and 1989, changes in TFP were actually positive.

    Other Resources Citing This Publication