Abstract
This paper explores, from an investment-saving perspective, the factors underlying the persistent widening of the current account surplus in the Netherlands since the early 1980s. Standard intertemporal models, even appropriately extended to incorporate specific features of the Dutch economy, do not appear to fully account for this development. Accordingly, the paper focusses attention on the production side of the economy to gain further insight into the trends of the current account. Empirical evidence suggests that changes in relative factor prices and a relative demand shift toward non-tradable goods account for the bulk of the observed widening of the current account surplus. In turn, the impact of these factors on the current account appears to reflect both changes in factor proportions and deviations from perfect competition in the Dutch sheltered sector.
I. Introduction
In recent years, a number of European countries have experienced substantial swings in their current account position, whose magnitude would appear difficult to explain by cyclical factors alone. Since the late 1980s or early 1990s, France, Italy, Denmark, and Ireland, among others, saw their current account swing from deficits into significant surpluses. At the same time, countries such as Belgium, Ireland, and Switzerland experienced a sizeable widening of their current account surplus. Under these conditions, it would appear useful to analyze the underlying factors, and potential policy implications, of this development.
The Netherlands has also experienced a similar significant widening of its current account surplus. Relative to other European countries, however, this occurred much earlier, specifically in the early 1980s. This difference in timing raises the question of whether special factors have been at work in the Netherlands, while at the same time providing adequate degrees of freedom that greatly facilitate empirical investigation of this development. This paper explores the main factors underlying the long-run trends of the current account in the Netherlands. In particular, it examines a number of factors that may have affected saving and investment to better understand the persistent large current account surpluses registered since the early 1980s. This current account development poses the question of structural changes that may have depressed investment relative to saving in the Netherlands.
The emergence of persistent current account surpluses such as those experienced by the Netherlands since the early 1980s would appear puzzling from the perspective of the simplest intertemporal models of the current account, which emphasize consumption smoothing in response to transitory fluctuations in income. This paper explores the extent to which a number of considerations within the framework of more general formulations of the intertemporal model could account for current account surplus persistence in the Netherlands. However, it also departs from the intertemporal model in seeking explanations for such persistence relating to the production side of the economy.
From a policy perspective, it would be important to distinguish between factors that are related to underlying economic trends and others that primarily reflect the impact of distortions, whether policy-induced or resulting from market deviations from perfect competition. It is only in the case of factors emanating from the presence of distortions that a welfare-improving role for public policy may exist. For the case of the Netherlands, considerable attention has been paid to distortionary factors that may have acted to bias saving upwards. For the purposes of this paper, however, attention will only be paid to the role of distortionary factors, both on the saving and investment side, relevant to the break that apparently occurred in the early 1980s, rather than to the longer term level of the saving rate in the Netherlands.
The plan of the paper is as follows: Section 2 summarizes the long-term trends in saving, investment and the current account in the Netherlands. Section 3 explores potential factors within the framework of the intertemporal model that could account for the structural increase in the current account surplus since the early 1980s. Sections 4 and 5 explore factors relating to the production side of the economy, respectively relative factor price trends and relative demand shifts, that could account for this development. Section 6 concludes.
II. Background
Over the last three decades, the trends of the Dutch current account have exhibited considerable variations. During the 1960s, it was essentially in balance each year, as the deficit in the trade account was roughly offset by an invisibles surplus. During the 1970s and early 1980s, the trends of the current account became more mixed, with surpluses in the first half of the 1970s followed by deficits in the latter half of the decade, registering an average annual surplus of about 1 percent of GDP. Since the early 1980s, the current account has persistently recorded surpluses which averaged 3 1/4 percent of GDP per year during 1982-94. These trends are even more striking if one focuses on the trade balance, which swung from an annual deficit during 1970-81 to an average annual surplus of 3 3/4 percent of GDP during 1982-94.
In order to better understand these long-run developments in the current account of the Netherlands, it is useful to examine the trends of saving and investment (see Chart 1). In the late 1960s and early 1970s, the Dutch economy was characterized by high saving and investment, of about 27-28 percent of GDP. Thereafter, the situation changed significantly, in the context of what has come to be known as the “Dutch disease”. 2/ Until the early 1980s, the saving rate fell sharply, reaching a low of 20 percent by 1980. A sharp deterioration of the fiscal balance was the main reason, accounting for almost the entire decline in the saving rate between 1970 and 1982. Business and household saving also weakened during the same period, 3/ mainly under the impact a the squeeze on business profitability and the two successive international recessions. 4/ During the same period, the investment rate also fell until 1976, apparently affected by the economic downturn and the marked rise in the labor income share. After a brief swing during 1977-80, it resumed its sharp decline, reaching a low of 18 percent of GDP in 1982.
NETHERLANDS Saving and Investment
(Percent of GDP)
Citation: IMF Working Papers 1996, 042; 10.5089/9781451975772.001.A001
Source: CBS, National Accounts.NETHERLANDS Saving and Investment
(Percent of GDP)
Citation: IMF Working Papers 1996, 042; 10.5089/9781451975772.001.A001
Source: CBS, National Accounts.NETHERLANDS Saving and Investment
(Percent of GDP)
Citation: IMF Working Papers 1996, 042; 10.5089/9781451975772.001.A001
Source: CBS, National Accounts.Since the early 1980s, these trends have only partially been reversed. On the one hand, the saving rate increased significantly, peaking in 1989-90 at around its 1970 level. Efforts at fiscal consolidation constituted part of the story. 5/ Much more substantial, however, was the contribution of private saving, and in particular its business component, which rose markedly as business profitability improved. On the other hand, investment, while also rising, failed to keep up with saving, peaking in 1990 at only its mid-1970s level. 6/
This asymmetric recovery of saving and investment underlies the persistence of large current account surpluses in the Netherlands since the early 1980s. Thus, between 1970-81 and 1982-92 investment as a percent of saving fell on average by over ten percentage points, from 96 to 85 1/2 percent. With saving being in the range of 20-25 percent of GDP, this decline in the investment-saving ratio is equivalent to around 2 1/2 percent of GDP, or broadly speaking the observed widening of the current account surplus.
III. Current Account Surplus Persistence and the Intertemporal Model
The limitations of modeling the current account within a one-period framework are by now widely recognized. Instead, it can be argued that developments in the current account can be better understood as an outcome of the solution to an intertemporal optimization problem by households, which determines a time path for the economy-wide levels of consumption and saving, 7/ This family of intertemporal models typically views the current account as a buffer through which private agents can smooth out consumption over time.
While the intertemporal model emphasizes consumption smoothing, its simplest variants, which assume among other things stationary demographics, stable preferences, and perfect capital mobility, cannot readily account for persistent current account imbalances. This section examines whether relaxing some of these strong assumptions could explain the apparent structural widening of the Dutch current account surplus since the early 1980s. Specifically, it briefly explores the potential implications of: (a) distortionary factors affecting saving; (b) non-stationary demographics; (c) shifts in business saving behavior; and (d) imperfections in international capital flows.
1. Distortionary factors affecting saving
Potential distortions that may have biased upward private saving in the Netherlands, both on the business and on the household side, emanating from certain features of the Dutch fiscal system as well as certain aspects of the Dutch pension scheme, have received considerable attention. 8/ It could be argued that these distortions may have contributed to the persistence of current account surpluses, even within the general framework of the intertemporal model.
With regard to business saving, certain aspects of the Dutch fiscal system have been considered as potential sources of upward bias. Among them, the most noteworthy among such fiscal aspects is probably the double taxation of dividends. While double taxation of dividends is by no means uncommon internationally, in most other countries dividends are taxed at a much lower rate relative to the income tax rate, or benefit from other favorable tax provisions. This is not, however, the case in the Netherlands, resulting in the highest rate of taxation on paid out corporate income among EU countries. 9/ It can be assumed that this feature of the Dutch tax system introduces an upward bias in the incentive of Dutch corporations to withhold earnings. An additional feature of the Dutch fiscal system that can be expected to work in the same direction is the non-existence of a tax on capital gains. The absence of a capital gains tax should further increase the incentive to withhold corporate earnings with an eye to driving up share prices.
The relevance of these factors for the overall level of private saving hinges crucially on the extent to which the resulting higher business saving is offset by lower household saving. In turn, this depends on the extent to which households take into account in the calculation of their permanent income the higher share prices resulting from higher retained earnings. However, empirical evidence from other countries in this area, while generally suggesting high offset rates, is mixed.
Potential distortions that may create an upward bias in the household saving rate in the Netherlands are typically associated with the prevalence of the fully-funded supplementary pension schemes. Supplementary pensions, which cover over 90 percent of Dutch employees and entail very high replacement ratios by international standards, are set in the context of sectoral collective bargaining agreements and are then applied to each sector in its entirety, in the context of the Dutch practice of declaring collective bargaining agreements generally binding. This “forced” nature of contractual household saving has sometimes been thought of as inducing households to oversave.
The extent to which the above factors act to increase household saving beyond its socially desirable level depends on the importance of liquidity constraints, which would prevent households from reducing their non-contractual saving in order to offset any excess contractual saving. The very low level (by international standards) of non-contractual, or “free” household saving is often regarded as indicating that such liquidity constraints are indeed important. It should be emphasized, however, that the level of free saving in the Netherlands is not readily comparable to that in other EU countries, where a much larger portion of pensions is unfunded. In fact, it would be reasonable to expect higher non-contractual saving under a pay-as-you-go pension system, to the extent that households are taking into account likely future increases in contribution rates, especially in view of the prospective aging of the population.
To the extent that the factors referred to above are in fact quantitatively important, they may have had an impact on the level of household saving in the Netherlands, which is indeed high by international standards. However, these factors cannot directly account for the apparent sharp shift in the trends of the Dutch current account since the early 1980s, which is the focus of this paper, as they had been more or less present in previous periods as well. 10/ If anything, one could reasonably expect the significance of some of these factors to have declined over time. For instance, financial innovation during the 1980s appears to have made available a range of credit instrument to finance consumption much wider than traditional consumer credit, thereby reducing the extent of liquidity constraints. 11/ In addition, access (and recourse) to early retirement, which can be viewed as effectively constituting a means of reducing higher than desired pension saving, has considerably increased.
2. Demographic factors
The assumption of stationary demographics on which the simple intertemporal model is partially predicated is clearly not applicable to the case of the Netherlands. As in most other European countries, the Dutch population is projected to age rapidly early in the next century. In this setting, intertemporal optimization would imply persistent current account surpluses while the population is still relatively young, as private agents attempt to smooth out consumption over time.
While the demographic profile of the Netherlands shares many of the features of its major trading partners, there are also a number of important differences. For one thing, the Dutch population is currently younger relative to other European countries. Secondly, while the old age dependency ratio is projected to increase during the first quarter of the next century in the Netherlands as in other European countries, the big jump in the dependency ratio is projected to occur somewhat later, and to be sharper (see tabulation below).
Old-Age Dependency Ratios
(In percent)
Old-Age Dependency Ratios
(In percent)
| 1990 | 2000 | 2010 | 2020 | 2030 | |
|---|---|---|---|---|---|
| Netherlands | 18.6 | 20.6 | 23.7 | 32.8 | 44.1 |
| Belgium | 21.9 | 24.7 | 25.0 | 30.2 | 38.9 |
| France | 20.8 | 23.6 | 23.6 | 30.2 | 37.5 |
| Germany | 21.5 | 24.1 | 30.6 | 33.4 | 41.5 |
| U. Kingdom | 23.7 | 23.6 | 24.4 | 29.2 | 36.8 |
| EU average | 21.2 | 23.6 | 26.0 | 30.3 | 37.5 |
Old-Age Dependency Ratios
(In percent)
| 1990 | 2000 | 2010 | 2020 | 2030 | |
|---|---|---|---|---|---|
| Netherlands | 18.6 | 20.6 | 23.7 | 32.8 | 44.1 |
| Belgium | 21.9 | 24.7 | 25.0 | 30.2 | 38.9 |
| France | 20.8 | 23.6 | 23.6 | 30.2 | 37.5 |
| Germany | 21.5 | 24.1 | 30.6 | 33.4 | 41.5 |
| U. Kingdom | 23.7 | 23.6 | 24.4 | 29.2 | 36.8 |
| EU average | 21.2 | 23.6 | 26.0 | 30.3 | 37.5 |
While demographic factors could conceivably have contributed to the widening of the Dutch current account surplus since the early 1980s, a number of considerations would suggest that their impact has probably been rather limited. In the first place, the timing and magnitude of the observed structural widening of the Dutch current account surplus would appear difficult to explain by demographics alone. In particular, it is not clear why, while the bulk of the deterioration of the old age dependency ratio in the Netherlands is projected to occur later than in other European countries, the widening of its current account surplus took place much earlier. At the same time, the projected eventual deterioration of the dependency ratio is projected to be only marginally more severe than in other countries (especially Germany). Secondly, and more importantly, any impact of demographic factors should have primarily made itself felt on the household component of private saving. Instead, in the case of the Netherlands, it was the business component that was mainly responsible for the recovery of the saving rate since the early 1980s, while the household component remained virtually flat (Chart 2). 12/
NETHERLANDS Saving
(Percent of GDP)
Citation: IMF Working Papers 1996, 042; 10.5089/9781451975772.001.A001
Source: CBS, National Accounts.NETHERLANDS Saving
(Percent of GDP)
Citation: IMF Working Papers 1996, 042; 10.5089/9781451975772.001.A001
Source: CBS, National Accounts.NETHERLANDS Saving
(Percent of GDP)
Citation: IMF Working Papers 1996, 042; 10.5089/9781451975772.001.A001
Source: CBS, National Accounts.3. Business saving behavior
The substantial increase in the business saving rate, particularly evident since the mid-1980s, has sometimes been interpreted as, at least partly, pointing to a sharp change in business behavior. Specifically, it has been suggested that Dutch enterprises, in response to their unfavorable experience related to large-scale reliance on bank credit during the recession of the early 1980s, may have opted for a stronger liquidity position, which led them to reduce the proportion of earnings devoted to investment. 13/ This is essentially an argument that points to the emergence of a precautionary motive of business saving since the early 1980s. Such a change in “tastes” on the part of the enterprise sector could conceivably have entailed some persistent effect on the current account balance, even within the framework of the intertemporal model, especially to the extent that adjustment costs in the area of enterprise investment are substantial.
This line of argument has the advantage of focusing attention on the business saving surplus, which has indeed been the major factor underlying the structural widening of the current account surplus in the Netherlands during the period under consideration. On the other hand, quite apart from the fact that it is extremely difficult to test empirically, it does not appear entirely persuasive. In particular, it is not clear why the response to a lower desired exposure by the enterprise sector to commercial banks should be a reduction in the overall level of investment (relative to profits), rather than simply a reduction in the share of investment financed by bank loans, as opposed to, say, equity.
A subtler version of the precautionary business saving argument could relate to the exchange rate regime. While a tight link of the guilder to the D-mark as a target of Dutch monetary policy predates the inauguration of the European Monetary System (EMS) in 1979, Dutch-German long-term interest rate differentials would suggest that the peg was not entirely credible up to the mid-1980s. 14/ A perceived tightening of the exchange rate regime since the mid-1980s would imply that Dutch enterprises could no longer rely on nominal exchange rate adjustment to accommodate shocks impinging on their competitive position. This in turn could have led them to strengthen their liquidity position in order to be in a position to cushion such shocks, which may have partially involved lowering the share of profits devoted to investment.
This second line of argument also raises a number of questions. On a theoretical level, the related literature on the effect of uncertainty on business investment 15/ would suggest that the impact of a tightening of the exchange rate regime, and hence increased variability of the guilder price Dutch exporters or import-competing firms would be faced with, on investment is ambiguous at best. On an empirical level, we are not aware of strong cross-country evidence pointing to a link between the exchange rate regime and the business saving surplus.
4. International capital flows
It could be argued that the observed widening of the current account surplus in the Netherlands may have little to do with domestic developments, but may primarily reflect developments in international markets. In particular, the very large increase in the degree of capital market integration worldwide during the 1980s has been well-documented, with the growth of international capital flows far outpacing the growth in the volume of trade in goods and services. It may be argued that capital movements had been fairly liberalized in the Netherlands well before the 1980s, and that therefore any impact of this factor should not have been felt exclusively since the early 1980s. It should be pointed out, however, that some restrictions regarding capital movements (though mostly relating to outflows rather than inflows) persisted well into the 1970s and early 1980s in some of the Netherlands’ main partners. Perhaps more importantly, the concept of capital market integration is much wider than the liberalization of the legal framework governing capital movements. It has thus been argued that innovation, both technological and financial, may have contributed at least as much as capital flow liberalization to the observed large increase in international capital mobility.
A large increase in international capital mobility could in principle have a substantial and persistent impact on the relative saving-investment trends in a small, open economy like the Netherlands. In terms of the intertemporal model, barriers to capital movements (whether legal or otherwise) prior to the 1980s could have in effect forced Dutch households to adopt a sub-optimal consumption/saving path. By contrast, increased capital mobility could have enabled Dutch residents to diversify their portfolios by increasing the share of foreign assets in order to maximize expected returns and/or hedge more effectively in the face of uncertainty. This would imply a structurally higher saving surplus, and hence a widening of the current account surplus. 16/ The argument would suggest a causal relationship running from the capital account to the current account.
Given that the relationship between saving, investment, and the capital account is an accounting identity, direct empirical investigation of the impact of increased capital market integration internationally on the ratio of investment to saving is not straightforward. However, the underlying theory has certain testable features. Thus, among others, Feldstein and Horioka (1980) suggest that perfect capital mobility should have strong implications for the correlation between saving and investment. In their model, while perfect capital mobility could imply a structural change in the saving surplus, it would also imply zero correlation between domestic saving and domestic investment. While it has been demonstrated that this extreme result depends crucially on a number of rather strong assumptions of the Feldstein-Horioka model, 17/ it would be reasonable to postulate that, to the extent that the degree of capital market integration has an impact on saving and investment decisions, increased capital mobility should imply reduced correlation between domestic saving and domestic investment and, more generally, that domestic saving should be a poorer predictor of domestic investment.
This intuition is exploited below by breaking the available sample in 1981, when persistent current account surpluses started to emerge, and running separate regressions of the ratio of domestic investment to GDP (INV) on the ratio of domestic saving to GDP (SAV) over the two sub-samples. The estimation results were as follows (t-statistics in parenthesis):
| Sample: | 1970-1981 | |
| R2 = 0.54 SE = 1.857 | F(1,10) = 13.7 | DW = 1.6 |
| Sample: | 1970-1981 | |
| R2 = 0.81 SE = 0.486 | F(1,9) = 39.0 | DW = 1.4 |
| Sample: | 1970-1981 | |
| R2 = 0.54 SE = 1.857 | F(1,10) = 13.7 | DW = 1.6 |
| Sample: | 1970-1981 | |
| R2 = 0.81 SE = 0.486 | F(1,9) = 39.0 | DW = 1.4 |
The estimation results suggest that increased capital mobility over time is not likely to have played a major role in explaining the observed trends of investment relative to saving in the Netherlands since the early 1980s. For one thing, the estimated coefficient of the saving rate is virtually identical across the two sub-samples. Even more striking are the conclusions regarding the predictive power of the saving rate with respect to the investment rate. Based on the (adjusted) R2, standard error, and F-statistic, it is clear that the equation fit in fact substantially improves during the later period, implying that the domestic saving rate has been a better predictor of the domestic investment rate when capital was presumably more mobile. In particular, while the variation of the saving rate explains only about half of the variation of the investment rate during the 1970s and early 1980s, its explanatory power increases to over 80 percent in the period since the early 1980s.
The Importance of this result should not be exaggerated, however. What the estimation results suggest is simply that outward investment opportunities in the case of the Netherlands were not significantly affected by international developments towards closer capital market integration in the more recent period. Thus, the results should be interpreted as indicating that the trends of the ratio of domestic investment to domestic saving, and hence of the current account, during the period under consideration cannot be accounted for to any substantial extent by increased capital mobility, or at least that the impact of this factor has been dominated by more important developments. 18/
Quite apart from the issue of saving-investment correlation, the estimation results presented above help quantify the extent of the problem under consideration. Thus, while the estimated positive constant term implies that investment would increase less than proportionally with saving, or that the investment-saving ratio should decline as the saving rate increases, it turns out that this factor does not account for any significant portion of the widening of the current account surplus observed since the early 1980s. Instead, it is the decline in the constant term itself, implying that given the saving rate investment in 1982-92 was 2.6 percent of GDP lower on average relative to the previous decade, which accounts for the bulk of the widening of the current account surplus.
IV. Relative Factor Prices
The discussion in the previous section suggests that relaxing some of the assumptions of the simple intertemporal model, in light of considerations relevant to the Dutch experience, does not appear to account fully for the persistent widening of the current account in the Netherlands since the early 1980s. The remainder of the paper examines the implications of a more fundamental extension of the intertemporal model. In particular, we explore a number of factors that could explain this development relating to the production side of the economy, to which most intertemporal models of the current account pay only scant attention. This would appear warranted, in view of the fact that the main source of the widening of the current account surplus during the period under consideration has been the business, rather than the household, saving surplus.
Thus, the remainder of this paper focuses on potential factors on the production side that may have kept investment in the Netherlands low during the period since the early 1980s. An obvious question relates to the basis of comparison: low relative to what? It should be emphasized that the relevant comparison is not between the level of investment since the early 1980s relative to its own historical trends. After all, during this period investment actually rose significantly relative to the immediately preceding years. Rather, the relevant issue for the purposes of the study would be to explore the factors that may have kept investment low relative to saving, thus contributing to a significant increase in the saving surplus that finds its counterpart in the widening of the current account surplus.
The paper will investigate two sets of factors that would appear to characterize especially the period under consideration. In particular, attention shall be focused on the possible implications of relative factor price movements and of real sectoral shifts within the Dutch economy.
One of the salient features of the Dutch economy since the early 1980s has been a marked shift in wage trends relative to the preceding period. In particular, since the early 1980s nominal (and real) wages rose much more moderately, both relative to other industrial countries and relative to the trends of the preceding decade. Indicative of the magnitude of this shift is the evolution of labor’s share in total income, which declined from a peak of over 95 percent in the beginning of the 1980s to 80 percent by the end of the decade, before edging up again at the beginning of the 1990s to reach 84 percent by 1994.
Given the Netherlands’ status as a small, open economy, it would be reasonable to postulate that the price of capital is entirely determined exogenously from abroad. As the credibility of the Dutch monetary policy attained during the period under consideration enabled the Netherlands to avoid the substantial fluctuations in interest rates that characterized other EMS countries, the real price of capital during this period can be expected to have experienced much greater stability relative to the trends of the real price of labor. Accordingly, the period since the early 1980s likely witnessed a major change in factor prices, with the price of labor rising much more moderately relative to the price of capital. 19/ In turn, these trends in relative factor prices can be expected to have had an important impact on factor proportions, with individual firms switching to more labor-intensive techniques or, at the level of the economy as a whole, with production shifting to more labor-intensive sectors since the early 1980s. This pattern of factor substitution would suggest a potentially important link between wages and the degree of capital accumulation in the Netherlands. With the economy adjusting to a lower capital-output ratio, one could expect a depressing impact on business investment which, given adjustment costs, could be persistent.
Once again, it should be stressed that what is at issue is not whether a decrease in the relative price of labor would lead to a reduction in investment in absolute terms (or as a share of GDP). In fact, it has been well established that the recovery of investment since the early 1980s mainly reflects improved business profitability. Rather, what is relevant for the purposes of this chapter is whether in fact lower wages can be expected to result in a reduction in investment relative to saving. The argument will be that, while lower wages would tend to encourage both higher (business) saving 20/ and higher investment, the impact on the investment side is partly offset by a pattern of substitution away from capital toward labor; the reverse would be expected to hold in periods of higher relative price of labor. The quantitative importance of this factor would thus appear to hinge crucially on the strength of these patterns of factor substitution.
A problem in investigating the issue of factor proportions directly relates to the non-existence of capital stock data for the Netherlands. However, there are indirect indications that the effects involved could be substantial. This can be seen from the relative trends in employment and production between the early 1970s and the early 1990s. Thus, the OECD (1994) estimates that between 1970 and 1985 employment in the sheltered sector stagnated, while production increased by 45 percent in real terms. During approximately the same period, developments in the exposed sector were even more extreme, with a similar production growth of around 45 percent accompanied by a decline in employment of the order of over 20 percent between 1970 and 1984. These trends are estimated to have been sharply reversed in the ensuing period. Thus, between 1985 and 1992, a production growth of around 20 percent in the sheltered sector was accompanied by a 20 percent increase in employment; similarly, the (more capital-intensive) exposed sector experienced a 25 percent production growth and a 7 percent employment growth. 21/ Such sharp swings in labor productivity would provide strong prima facie evidence that changes in factor proportions have in fact been substantial between the early 1970s and the early 1990s.
As a starting point in the empirical investigation of the impact of relative factor prices, a simple equation linking a real wage variable to the investment-saving ratio is postulated. Given the assumption of near-constancy of the price of capital (at least compared to the price of labor) the real wage will serve as a proxy of relative factor prices. The wage variable chosen was a real compensation per employee index (RW), as the most representative of the entire economy, 22/ normalized around 100 so as to be of the same order of magnitude as the investment-saving ratio.
A specification imposing instantaneous adjustment would be unrealistic, given that the underlying adjustment process involves switches in production techniques which can be expected to be costly and take time. Accordingly, a standard partial adjustment process was postulated, according to which an equation of the form
is regarded as a long-run relationship, whereas the actual investment-saving ratio only partially adjusts to deviations from its long-run desired level, according to the equation: 23/
This specification introduces some rudimentary dynamics into the model, and has the advantage of distinguishing between the short-run and long-run coefficient of RW. At the same time, it allows us to investigate the question of whether the impact of changes in relative factor prices has a persistent impact on the saving-investment balance. Straightforward manipulation of equations (1) and (2) yields the final equation to be estimated:
where
c0 = ka 0
c1 = ka 1
c2 = 1-k
As can be easily verified, all of the structural parameters of the model are identified.
Using the actual level of the real wage as explanatory variable would not adequately capture the impact of relative factor prices, however. The main problem stems from the fact that RW is non-stationary. Thus, the time-path of RW (almost steadily increasing, but at a much slower pace since the mid-1980s) and the time path of the INV/SAV variable (exhibiting a structural jump since the early 1980s but otherwise displaying moderate fluctuations), suggest that the two variables are not co-integrated. In this context, the high R2 could be regarded as mainly capturing the sub-period up to the early 1970s, when the variability of RW was much higher. In fact, a Chow test rejects the stability of the specification employing RW as explanatory variable over different sub-periods.
In order to test for robustness of the impact of relative factor prices on the investment-saving ratio, two variants of the RW variable that are not subject to the co-integration problem referred to above were used as explanatory variables: a de-trended real wage, and the rate of change of the real wage (rather than its level). 24/ With regard to the first variant, the explanatory variable (RWDEV) is defined as the deviation of the real wage from its Hodrick-Prescott trend. The estimation results of the instantaneous adjustment and the partial-adjustment specifications using RWDEV as explanatory variable are presented below:
For the case of the rate of change of the real wage (DRW) as explanatory variable, the estimation results were as follows (for brevity, only the partial adjustment specification is presented):
The estimation results using the de-trended real wage and the rate of change of the real wage as explanatory variables lend considerable support to the conclusion that relative factor prices explain a substantial part of the variation in the saving-investment ratio. For both variables, the R2 was high, around 75 percent in the partial adjustment specification. At the same time, the estimated adjustment coefficients, measuring the extent of instantaneous adjustment of the investment-saving ratio, at 44 percent in the de-trended real wage equation and at 39 percent in the real wage growth equation, do not appear particularly large in view of the likely cost of switching to an alternative production technique (or expectation inertia). It would thus appear that the impact of the shift in relative factor prices that occurred in the early 1980s entails sufficient persistence to account for the observed trends in the saving-investment balance and the current account.
The results of the estimations performed above suggest that the impact of relative factor prices on the investment-saving ratio is strong. 25/ For instance, between 1970-81 and 1982-92 investment as a percent of saving fell on average by 10.3 percentage points, from 95.9 percent to 85.7 percent. The fall in the de-trended real wage and in the real wage growth variables during this period account for the bulk of this decline, estimated at 9.7 and 9.4 percentage points respectively. It should be cautioned that these figures probably overestimate the true contribution of the wage variables as other relevant explanatory variables have not been considered. They suggest, all the same, that factor prices are an important determinant of the saving-investment ratio.
As a final point, it may be worthwhile to inquire whether our particular interpretation of the impact of the real wage variables on the investment-saving ratio is supported empirically, which would require estimating saving and investment equations separately. 26/ Recall that the postulated mechanism was one according to which a fall in the relative price of labor, through factor substitution away from capital, tends to raise investment relatively less than saving, hence lowering the investment-saving ratio and increasing the current account surplus. Alternatively, it could be argued that the impact of relative factor prices on the investment-saving ratio works primarily through business saving. Our formulation would suggest that the impact of a relative factor price change on the current account should be less than its impact on saving, as the latter would be somewhat dampened by an increase in investment.
Testing the above proposition necessitates regressing the investment rate on the real wage variable and the saving rate. Our formulation would imply that, given the saving rate, the investment rate should respond positively to the real wage. The estimation results of such an equation (using the detrended real wage as the wage variable), which are presented below, lend support to this interpretation:
The increase in the current account surplus related to these factor price developments should not be cause for policy concern, whatever the origin of the factor price changes. 27/ To the extent that the dominant underlying reason relates to relative factor supply developments, notably an expansion of labor supply associated with an increase in female labor force participation during the period under consideration, a fall in the price of labor relative to the price of capital should be viewed as part of the transmission mechanism towards a new, welfare-improving equilibrium. Even if, however, one views the moderation of wage trends since the early 1980s as mainly reflecting a policy of wage restraint, it would be difficult to view this policy as distortionary. In fact, to the extent that this view of wage developments is accurate, the wage moderation policy probably contributed to easing excess supply conditions in the labor market, thus reducing already existing distortions.
V. Real Sectoral Shifts
Shifts between the various sectors of the Dutch economy may also have had an impact on the level of investment relative to saving during the period under consideration, which could account for the persistent widening of the current account surplus since the early 1980s, contrary to the implications of the simple intertemporal model. The importance of such sectoral shifts is reflected in the long-term trends in relative prices. In particular, the Netherlands has experienced a trend rise in the relative price of nontradable relative to tradable goods during the last two decades, an experience very much shared by most other industrial countries. Thus between 1981 and 1993 the overall increase in the price of nontradable goods was 37 percent; during the same period, the increase in the price of tradables was only 13 percent. The effect is so big that, in recent years, many countries have found that nontradables inflation accounts for almost the total inflation.
For the purposes of this paper, determining the origin of these relative price changes is crucial. In particular, the impact on the investment-saving ratio would likely be quite different depending on whether supply side or demand side shocks have been at the source of the observed relative price trends. A supply-side modelling of relative price changes has a long tradition in economics, going back to Harrod (1938) and formalized by Samuelson (1964) and Balassa (1964). The theory rests on the assumption, which has received wide empirical support, of faster productivity growth in the tradable goods sector relative to the nontradable goods sector, and predicts a rise in the relative share of tradables along with a fall in their relative price over time. If such supply side factors have been the principal cause of the observed relative price changes, then sectoral considerations would be of little usefulness in accounting for the trends in the investment-saving ratio in the Netherlands during the period under consideration. Indeed, the dominance of supply side factors would likely imply an increase in the investment-saving ratio, given that the tradable goods sector (principally manufacturing) is much more capital intensive relative to the non-tradable goods sector (typically services). 28/ This would of course be the reverse of the trend observed in the Netherlands since the early 1980s.
However, recent evidence indicates that in a number of countries, including the Netherlands, an increase in the relative price of nontradables has been accompanied by a rise in their relative share in the economy’s total value added. 29/ Thus, the share of nontradable goods in total value added rose steadily from 65 1/2 percent in 1970, to 71 percent in 1980, and to 75 1/2 percent in 1993. This would suggest that a demand shift towards nontradables could be at least as important a part of the relative price trends described above. Two broad types of explanations for such a relative demand shift towards nontradable goods have been suggested in the literature. The first relates to consumer tastes, and in particular is based on the hypothesis that the income elasticity of demand is higher for services than for manufactured goods. This hypothesis has received a fair amount of empirical support from both aggregate time series and household consumption cross section analysis, and suggests that the relative demand for nontradables should be expected to increase in line with the rise in per capita income. The second, and probably more important, explanation relates to the trend rise in the share of the public sector in the total economy over time, in the sense that government expenditure directly produces a range of nontradable commodities, ranging from public safety to health care. 30/
To the extent that such a relative demand shift away from tradables and towards nontradable goods has been an important feature of the Dutch experience during the period under consideration, it may be expected to have had a significant impact on the investment-saving ratio, and hence the current account. The impact of the relative demand shift may be viewed as working via two distinct channels. The first, more direct, channel once again concerns factor proportions. As services production is significantly less capital intensive than goods production, 31/ a demand shift towards nontradable goods can be expected to have a negative impact on capital accumulation.
The second link between relative sector demand and the investment-saving ratio is somewhat more subtle (and much more difficult to test empirically). While the tradable goods sector, being exposed to international competition, is usually treated as conforming to the perfect competition (or at least, to the extent that economies of scale are important, to the monopolistic competition) paradigm, it could be argued that the nontradable goods sector is less than perfectly competitive. In fact, the non-competitive aspects of the sheltered sector have received wide attention in the Netherlands, in view of the predominance of cartel arrangements, barriers to entry, complex licensing requirements, and a host of other distorting factors. The presence of these distortions could act as a strong impediment on investment in this sector in the event of a relative demand shift towards nontradables, thus biasing the overall investment-saving ratio downwards, and the current account surplus correspondingly upwards. This impact on the current account could conceivably be reinforced by a distortion on the saving side. To the extent that monopoly rents are important in the sheltered sector, a relative demand shift toward nontradables could bias overall business profitability, and hence business saving, upward. In the presence of these distortionary factors, structural policies aiming at increasing product market competition could have an important impact on the current account.
The quantitative importance of real sectoral shifts for the issues in question is tested empirically on the basis of a simple model that brings together the supply side Balassa-Samuelson features and the relative demand considerations outlined above. 32/ On the supply side, the model postulates constant returns to scale, Cobb-Douglas production functions for tradable and nontradable goods, while allowing for differences with regard to factor shares and total factor productivity growth between the two sectors. In log-linear form, the production functions for each type of good are as follows:
where Y denotes output, L denotes labor input, K denotes capital input, G denotes the level of technology, and a denotes the (constant) share of labor. Subscripts T and N denote, respectively, the tradable and nontradable good. It is assumed that aT<aN, and that the rate of total factor productivity growth is higher in the tradable sector than in the nontradable sector: D(GT)>D(GN).
In addition, the law of one price is assumed to hold in the traded goods sector and, given perfect capital mobility, the price of capital is assumed to be exogenously given to domestic producers by the international capital market. Under perfect competition, the change in the relative price of nontradables relative to tradables (P) can be readily shown to be completely supply-determined, and to depend only on the difference in total factor productivity growth, weighted by the relative factor shares in the two sectors.
Equation (6) summarizes the Samuelson-Balassa hypothesis: higher rate of total factor productivity growth in the nontradable sector (together with higher labor intensity in the production of the nontradable good) would tend to raise the relative price of the nontradable good over time. In addition, to the extent that the markets for both the tradable and nontradable good are perfectly competitive, this conclusion would be unaffected by relative demand conditions. 33/
Relaxing the assumption of perfect competition for at least one of the two sectors would allow demand factors to have a role in the determination of relative prices. The model derives a higher income elasticity of private consumption of the nontradable good by introducing a (positive) subsistence level of consumption of the tradable good; with regard to government consumption (G), it is assumed that it falls entirely on the nontradable good. 34/ Following De Gregorio, Giovannini and Wolf (1994), a representative consumer is assumed to maximize the present discounted value of:
where CN and CT stand for the consumption levels of the nontraded and traded good, respectively, and C0 denotes the subsistence level of consumption of the tradable good. The analysis is considerably simplified by assuming that the consumer maximizes equation (7) on a period by period basis. On the assumption that government expenditure is financed through taxation, the economy-wide budget constraint (with the tradable good as numeraire) can be written as:
where Y stands for real GDP, YT+PYN.
Constrained maximization of (7) yields the total (private and public) demand for the traded and the nontraded good. After straightforward algebraic manipulation, these can be written as:
where g denotes the share of government expenditure in GDP.
Combining the demand and supply features of a model such as the one described above, De Gregorio, Giovannini and Wolf (1994) estimate the following reduced-form equation:
where prod stands for relative factor productivity in the two sectors (weighted by the ratio of the labor shares), gov is government expenditure over GDP (both in real terms), y is per capita income, and D(infl) is the first difference of the rate of inflation. This latter term is included to capture the impact of sluggish price adjustment in the nontradable goods sector. The equation was estimated via a panel method, combining time series and cross section observations for a number of OECD countries. The authors found that demand-side factors, in addition to supply-side factors, are an important determinant of relative prices. Over the longer-term, however, the importance of demand-side factors in explaining relative prices is considerably reduced, and supply-side factors become dominant, as should be expected. Time series analysis confirmed the authors’ results for the specific case of the Netherlands.
For the purposes of this paper, the main direction of influence that is of interest is not the impact of relative demand and supply factors on the tradable goods/nontradable goods relative price. Rather, the direction of influence relevant for our purposes concerns the impact of the above factors on the composition of output, which is already implicit in the findings of De Gregorio, Giovannini and Wolf (1994), as this is postulated to affect the investment-saving ratio. 35/ Accordingly, a variant of equation (11) was estimated, employing the investment-saving ratio as the dependent variable. We begin by examining the extent to which supply factors and relative demand shifts alone, to the exclusion of other relevant explanatory variables, can account for variations in the (INV/SAV) variable. In the second stage of the investigation, the joint impact of these factors with that of the real wage variables discussed above will be examined, in order to obtain a fuller picture of the main determinants of the variable in question, as well as of their relative importance.
A practical problem concerns the actual definition of the two sectors for the purposes of econometric analysis, as in reality goods and services cannot be grouped into tradables and nontradables as neatly as theory would have it. For the purposes of this paper, two alternative breakdowns were considered as the operational definition of the two sectors: the traditional one between goods and services, and one which groups transportation with the tradable goods category, in line with the De Gregorio, Giovannini and Wolf (1994) methodology. 36/ As the estimation results under the two definitions were virtually identical, the results presented below relate to the goods/services breakdown.
The equation to be estimated includes (INV/SAV) as the dependent variable, and relative productivity in the two sectors, weighted by the average respective labor shares (PROD), general government current expenditure as a percent of GDP (GOV), and real GDP per capita (GDPPC) as explanatory variables. 37/ A number of shortcomings should be mentioned. A first problem relates to the sample size: real value-added data on a sectoral basis go back only to 1977, thus reducing the sample size by almost one third relative to the equations estimated above. A more fundamental problem concerns the definition of the relative productivity variable. As capital stock figures are not available for the Netherlands, total factor productivity could not be computed. Instead, labor productivity in each of the two sectors was used as a proxy for total factor productivity.
The estimation results are as follows (t-statistics in parenthesis):
The results suggest that relative demand shifts have been an important determinant of the investment-saving ratio during the period under consideration. With regard to the demand-side variables, both government expenditure and per capita GDP are statistically significant and correctly signed. On the other hand, the coefficient of the PROD variable, while correctly signed, turned out to be statistically insignificant. While this could be a reflection of the fact that labor productivity is not a perfect proxy for total factor productivity, it may also suggest that technical change in the Netherlands in recent years may have been of a predominantly capital saving nature. 38/ However, to the extent that the insignificance of the PROD coefficient is related to the use of labor productivity as an (imperfect) proxy for total factor productivity, this could raise some doubt over the interpretation of per capita GDP as a predominantly demand-side variable.
The above results thus suggest that a demand shift toward nontradables during the period under consideration (as captured by growth in per capita income and, more importantly, a higher share of government current expenditure relative to the previous decade) partially explain the persistent widening of the Dutch current account surplus since the early 1980s, which would appear difficult to explain in the context of the simple intertemporal model. The sources of the persistence of the relevant impact would appear to be rather wider relative to those that were discussed for the case of relative factor price changes. In addition to adjustment costs and expectations inertia, the process underlying a relative demand shift would entail adjustment to a new long-run equilibrium, essentially involving entry of new firms to the sheltered sector(or the expansion in capacity of existing firms). This adjustment process can be reasonably expected to be far from instantaneous.
While the results presented above point to the importance of relative demand shifts in explaining the investment-saving ratio, it is very difficult to test directly to what extent this effect operates via the impact on factor proportions, rather than reflecting the presence of distortions in the nontradable goods sector. However, the implications of the underlying model suggest a simple test that could lead to a useful international comparison. It should be recalled that the benchmark case of perfect competition suggests that in the long run relative prices should be independent of demand-side factors, on the assumptions of a constant returns to scale technology in the nontradable goods sector, an exogenously determined price of capital, and price-taking behavior in the tradable goods sector. In that sense, the extent of the actual increase in the relative price of nontradables could proxy the extent of the deviation from perfect competition in the sheltered sector. De Gregorio, Giovannini and Wolf (1994) plot a best-equation fit of the relation between relative total factor productivity growth and the average annual rate of increase in relative prices (see Chart 3). On the assumption of comparable relative demand shifts between industrial countries, it can be calculated that the average annual growth of the nontradables/tradables relative price in the Netherlands was three quarters of a percent higher than would be predicted on the basis of the industrial countries sample. This would suggest that the extent of distortions in the Netherlands’ sheltered sector may be significantly higher relative to its main trading partners.
Netherlands Differential Factor Productivity Growth and Relative Price of Nontradables
(percent)
Citation: IMF Working Papers 1996, 042; 10.5089/9781451975772.001.A001
Source: De Gregorio, Giovannini and Wolfe (1994)Netherlands Differential Factor Productivity Growth and Relative Price of Nontradables
(percent)
Citation: IMF Working Papers 1996, 042; 10.5089/9781451975772.001.A001
Source: De Gregorio, Giovannini and Wolfe (1994)Netherlands Differential Factor Productivity Growth and Relative Price of Nontradables
(percent)
Citation: IMF Working Papers 1996, 042; 10.5089/9781451975772.001.A001
Source: De Gregorio, Giovannini and Wolfe (1994)Separate estimation of an investment and a saving equation against the variables discussed above could shed additional light on the question of the relative importance of factor proportions versus market deviations from perfect competition underlying the impact of relative sectoral shifts on the current account. Specifically, while estimating an investment equation cannot disentangle the impact of these two factors, there is no obvious reason why the factor proportions explanation could account for an impact of a demand shift toward nontradables on the saving rate. Accordingly, to the extent that such a demand shift tends to raise overall business profitability (and business saving), this should be interpreted as evidence that the nontradables sector is less than perfectly competitive. A regression of the saving rate on the variables mentioned above, holding the investment rate constant, yielded the following results (t-statistics in parentheses):
The coefficients of the demand shift variables (GOV and GDPPC) turned out to be correctly signed and statistically significant. Accordingly, the estimation results presented above point to a positive link between a demand shift toward nontradables and overall business profitability. As such, it provides support for an interpretation of the impact of such a demand shift on the current account as partially reflecting a less than competitive market structure in the nontradables sector.
We now proceed to incorporate the wage variables discussed above in order to estimate the joint importance of wage moderation and relative demand shifts in explaining the trends of the investment-saving ratio, and hence the current account. The estimation results, using the detrended real wage (RWDEV) as the wage variable 39/ are as follows (t-statistics in parenthesis):
The estimation results suggest that wage developments and real sectoral shifts both constitute important determinants of the investment-saving ratio during the period under consideration, with the explanatory variables accounting for over 90 percent of the variation of (INV/SAV). In addition, the equation fit, in terms of R2, standard error and F-statistic, significantly improves under joint estimation. While the estimated magnitude of the impact of both sets of variables is now lower than under separate estimation, as should be expected, the coefficients of all variables remain correctly signed and statistically significant (again, with the exception of the PROD variable). In terms of explanatory power, of the 10.3 percentage point decline in the investment-saving ratio during the period under consideration, the wage variable accounts for 3.1 percentage points, government expenditure for 7.3 percentage points, and per capita GDP for 2.6 percentage points. 40/
In contrast to the impact of wage developments, it could be argued that the impact of relative demand shifts on the investment-saving ratio may well entail distortionary features that call for welfare-improving policy actions. It would be difficult to assess the direct welfare implications of the increasing government share in the economy, as this trend may well reflect societal preferences. 41/ On the other hand, to the extent that the sheltered sector contains important non-competitive features, an increase in the relative demand for non-tradables would entail distortionary effects. In that sense, the structural reform policies to enhance competition in the sheltered sector that are being implemented, while beneficial in their own right, could also have a positive impact on the saving-investment balance and hence the current account.
Quite apart from the issue of potential distortions, however, the empirical results could have important implications for macroeconomic policy. It may be argued that, in view of the large current account surplus, the policies of fiscal consolidation being implemented by the Dutch authorities may be misplaced. Our empirical findings suggest that the link between deficit reduction and the current account would very much depend on the form that fiscal consolidation takes. To the extent that it is concentrated on expenditure reductions rather than tax increases, as is the case with the medium-term program that has been announced by the Dutch authorities, the resulting shift of relative demand towards tradable goods could offset the tendency of deficit reduction to increase the current account surplus.
VI. Summary and Concluding Remarks
This paper has investigated the potential determinants of the saving-investment balance in the Netherlands in order to gain insight into the nature of the persistence of large current account surpluses since the early 1980s. This development would appear somewhat puzzling in the context of the simplest intertemporal models of the current account, which emphasize consumption smoothing in the face of transitory shocks to income. At the same time, a number of extensions of the simple intertemporal model that would conceivably be capable of yielding persistent current account imbalances also did not appear to account fully for the Dutch experience.
The bulk of the paper focussed on identifying possible factors on the production side of the economy that may have affected the saving-investment balance during the period under consideration. Specifically, the potential impact of two such factors was explored. In the first place, it was postulated that relative factor price changes, mainly related to moderate wage developments since the early 1980s, could have led to lower growth in investment relative to saving, as they can be expected to have encouraged substitution away from capital. The empirical results suggested that relative factor prices have indeed been an important determinant of the investment-saving ratio, a result which proved to be robust under a variety of specifications. On the other hand, the resulting increase in the current account surplus should not be cause for policy concern, as the factor price changes during this period do not reflect the impact of distortions (and in fact they have in all probability contributed to a substantial alleviation of existing distortions).
Secondly, it was postulated that a relative demand shift towards nontradable goods, reflecting a higher income elasticity of demand for nontradables and, more importantly, the increased share of the public sector in the economy, has also been a relevant factor explaining the trends in the investment-saving ratio since the early 1980s. This postulated effect also received strong empirical support. The impact of relative demand shifts may partly work in a direct fashion via factor proportions, given that the capital intensity of nontradables production is much lower. In addition, it may reflect non-competitive features in the sheltered sector, for which some indirect empirical support was provided. To the extent that such distortions are important, structural policies to strengthen competition can be expected to raise investment relative to saving, and thus reduce the current account surplus. Quite apart from the issue of distortions, however, the importance of relative sectoral demand has strong implications for macroeconomic policy. In particular, it would suggest that the potential impact of budget deficit reduction to raise the current account surplus could be offset by a reorientation of relative demand towards tradable goods. This would be the case to the extent that fiscal consolidation is concentrated on the expenditure side of the budget.
Finally, a number of cautionary notes would appear in order. In the first place, persistence of the effects discussed in the paper does not imply that they are in any way permanent. All the same their magnitude and duration would appear to justify the empirical investigation undertaken in this paper. Secondly, no claim for universal validity of the channels investigated in this paper can be made. Evidently, a general move towards wage moderation or a universal demand shift toward nontradables cannot be expected to result in an improvement of every country’s current account. Still, empirical investigation of other countries’ response to a set of shocks such as those discussed in this paper would appear useful, at least in order to gain a better understanding of the underlying economic structure. Thirdly, the empirical results suggested that a departure from the one-good model, and an explicit focus on real sectoral shifts within the economy, can offer a much richer explanation of the trends of the saving-investment balance. Accordingly, the analysis would gain from an improvement in the quality of data available at the sectoral level, especially as regards sectoral factor proportions and sectoral productivity.
References
Agénor, P.-R., C. Bismut, P. Cashin and C. J. McDermott, “Consumption Smoothing and the Current Account: Evidence for France, 1970-94”, I.M.F. Working Paper, WP/15/119, November 1995.
Bakker, B. B., Saving in the Netherlands, Groningen, The Netherlands, 1993.
Balassa, B., “The Purchasing-Power-Parity Doctrine: A Reappraisal,” Journal of Political Economy, 72, December 1994.
Barro, R. J., N. G. Mankiw and X. Sala-i-Martin, “Capital Mobility in Neoclassical Models of Growth,” American Economic Review, Vol 85, No. 1, March 1995.
Bayoumi, T., “Saving-Investment Correlations,” I.M.F. Staff Papers, 137, 1990.
Broeder, C. den and C. Winder, “Financing Government Spending in the Netherlands: An Analysis from Ricardian Perspective”, De Economist, 140 (1), 1992.
De Gregorio, J., A. Giovannini and T. H. Krueger, “The Behavior of Nontradable Goods Prices in Europe: Evidence and Interpretation,” I.M.F. Working Paper, WP/93/45, May 1993.
De Gregorio, J., A. Giovannini and H. C. Wolf, “International Evidence on Tradables and Nontradables Inflation,” I.M.F. Working Paper, WP/94/33, March 1994.
Feldstein, M. and C. Horioka, “Domestic Saving and International Capital Flows,” Economic Journal, 90, 1980.
Gardner, E., “Taxes on Capital Income”, in “Tax Harmonization in the European Community”, I.M.F. Occasional Paper 94, July 1992.
Harrod, R., International Economics, Cambridge, U.K.: Cambridge University Press, 1939.
Kremers, J. J. M., “The Dutch Disease in the Netherlands,” in J. P. Neary and S. van Wijnbergen (eds.), Natural Resources and the Macroeconomy, Oxford: Blackwell, 1986.
McDonald, D., “Implications of Unification for Saving and Investment in West Germany,” in L. Lipschitz and D. McDonald (eds.), German Unification: Economic Issues, I.M.F. Occasional Paper No. 75, December 1990.
Micossi, M. and G. M. Milesi-Ferretti, “Real Exchange Rates and the Prices of Nontradable Goods,” I.M.F. Working Paper, WP/94/19, February 1994.
Obstfeld, M., “Capital Mobility in the World Economy: Theory and Measurement,” Carnegie Rochester Conference Series on Public Policy, 24, 1986.
Obstfeld, M. and K. Rogoff, “The Intertemporal Approach to the Capital Account”, National Bureau of Economic Research Working Paper No. 4893, October 1994.
O.E.C.D., Economic Surveys: Netherlands, 1994.
Pindyck, R. S., “A Note on Competitive Investment Under Uncertainty,” American Economic Review, Vol. 83, No. 1, March 1993.
Razin, A., “The Dynamic Optimizing Approach to the Current Account: Theory and Evidence”, in P. Kenen (ed.), Understanding Interdependence: The Macroeconomics of the Open Economy, Princeton University Press, Princeton, NJ: 1995.
Sakellaris, P., “A Note on Competitive Investment Under Uncertainty: Comment”, American Economic Review, Vol. 84, No. 4, September 1994.
Samuelson, P., “Theoretical Notes on Trade Problems,” Review of Economics and Statistics, Vol. 46, March 1964.
I would like to thank, without implication, Uli Baumgartner, Willem Buiter, Paul Masson, Donogh McDonald, Ramana Ramaswamy, Max Watson, as well as participants in the European I Department Seminar Series, for helpful comments, and Susan Becker for research assistance.
For a discussion of the principal stylized features of the “Dutch disease” see Kremers (1986).
This is not inconsistent with the fact that household and business saving remained roughly constant between 1970 and 1982. In fact, one should have expected a substantial improvement in these two categories of saving, precisely in view of the fiscal deterioration. Empirical evidence for the Netherlands suggests that, while Ricardian equivalence in its extreme form is rejected, private saving typically offsets over 50 percent of a change in public saving; see, for example, den Broeder and Winder (1992).
In net terms, the private saving rate also fell during this period.
Note that the increase in government saving fell short of the magnitude of deficit reduction attained, as public investment bore part of the burden of fiscal consolidation.
The decline in both saving and investment in the early 1990s is attributable to cyclical factors.
For general formulations of such intertemporal models, see Obstfeld and Rogoff (1994) and Razin (1995).
For a survey of the relevant arguments and a description of the Dutch institutional structure see Bakker (1993).
Based on the 1991 tax laws in various countries, Gardner (1992) has estimated that the after-tax proportion of distributed profits reaching the shareholder is a mere 26 percent in the Netherlands, compared to 58 percent in Germany and 60 percent in the U.K.
For instance, the supplementary pension schemes were introduced in the 1950s and 1960s, while the double taxation of dividends has been a longstanding feature of the Dutch fiscal system.
For example, a substantial part of mortgage credit is widely regarded as having effectively served as a close substitute for consumer spending in recent years.
This, of course, need not imply that the impact of demographics on desired household saving has been negligible. To the extent that liquidity constraints had been important, it could suggest a reduction in the “forced” component of household saving.
For a detailed survey of the relevant arguments, see Bakker (1993).
In fact, the guilder was devalued vis-à-vis the D-mark in 1983.
See, for example, Pindyck (1993) and Sakellaris (1994).
A positive response of total saving to increased capital mobility, presumably reflecting a higher rate of return would strengthen this effect. However, empirical evidence for a number of countries has tended to suggest that the interest elasticity of saving is rather low.
See, for example, Obstfeld (1986), Bayoumi (1990), and Barro, Mankiw and Sala-i-Martin (1995).
The conclusion that the impact of closer capital market integration on saving and investment may have been smaller than is generally thought is corroborated by evidence from other countries. In a recent paper on France, Agénor et. al. (1995) directly test an intertemporal model, and conclude that strict capital controls up to the late 1980s did not prevent French households from attaining an optimal consumption path.
At this stage of the argument, it is not very important to determine the sources of the change in relative factor prices. In particular, the relevant linkages would be virtually identical whether the factor price changes can be best characterized as exogenous, or whether they themselves have been caused by underlying factor supply developments. This point will be addressed briefly below.
The impact on total private saving would be higher to the extent that, in the short-run, low wage growth lowers the household consumption to GDP ratio, thus increasing the household saving rate as well.
The employment figures are expressed in terms of full-time equivalents; the increase in the number of persons employed was substantially higher.
Estimations using the real hourly wage in manufacturing as the wage variable produced essentially identical results.
In addition to an argument involving adjustment costs, partial adjustment of this type can also be rationalized in terms of adaptive expectations about the independent variables, in this case relative factor prices. This consideration also appears pertinent to the problem at hand. It would be reasonable to assume that it took some time for Dutch producers to become convinced that wage moderation would become a permanent feature of the Dutch economy in the period since the early 1980s.
For both variables, a Chow test failed to reject stability across sub-samples.
In order to further test the robustness of the postulated effect, the labor income share in total value added was also considered as a potential explanatory variable. While this variable is in some ways a less reliable indicator of relative factor prices since it compounds changes in factor prices and changes in factor proportions, it should capture any bias caused by the assumption that the real price of capital has fluctuated much more moderately than the price of labor over the estimation period. The estimation results, while entailing lower explanatory power and lower significance levels were entirely consistent with the results presented above, suggesting that any mis-specification with regard to the relative factor price variable is not likely to be important.
This is the approach taken, among others, by McDonald (1990) in analyzing saving-investment trends in Germany.
More accurately, the current account surplus should not be viewed as a symptom of underlying distortions warranting policy intervention.
Strictly speaking, this is not entirely correct. The solution would depend on what form technical progress takes, i.e., whether it is “labor-saving” or “capital-saving.”
See, for example, De Gregorio, Giovannini and Krueger (1993), Micossi and Milesi-Ferretti (1994), and De Gregorio, Giovannini and Wolf (1994).
This effect is only partially offset by the reduction in the relative demand for nontradables (given their higher income elasticity) deriving from the reduction in household disposable income resulting from the government’s increased financing needs.
In fact, the CPB’s medium-term macroeconometric model FKSEC imposes the assumption that in the sheltered sector labor constitutes the only factor of production.
The structure of the model follows closely De Gregorio, Giovannini and Wolf (1994).
Obviously, the relative output of the two sectors will depend on both supply and demand conditions.
Obviously, this assumption does not involve loss of generality: all that we need to assume is that the share of the nontradable good in government consumption is higher than the corresponding share in private consumption.
It should be pointed out that the authors’ finding that the impact of a (permanent) demand shift towards nontradable goods on relative prices is lower in the long-run would imply that their impact on relative quantities should be higher in the long-run. The underlying mechanism is one in which entry of new firms into the sheltered sector over the long-run tends to return relative prices to their initial level, while resulting in further increases in the relative output of the sheltered sector.
The authors based the distinction on the actual degree of tradability (defined as the share of value added that is exported) of different categories of goods and services. They found that, while all categories of goods qualified as tradables, all service categories, with the exception of transportation, qualified as nontradables.
The inflation variable turned out to be insignificant under all specifications, and was consequently dropped.
To the extent that capital-saving technical progress is embodied in the more recent vintages, this could be a partial explanation of the higher rate of capital depreciation since the late 1970s.
The use of real wage growth as the relevant wage variable resulted in somewhat lower significance levels (especially with regard to the coefficient of DRW itself), but otherwise yielded virtually identical results.
The relative productivity variable would predict a 1.8 percentage point increase in the investment-saving ratio.
Note, however, that to the extent that the increase in government spending is financed by non-lump sum taxes, it would entail distortions even if it socially desirable.
