Must-Must-Must Read in Its Entirety: Pierre-Olivier Gourinchas (2002): Comment on Blanchard and Giavazzi
Must-Must-Must Read in Its Entirety: The reason that I find the Princeton Economics Department's refusal to tenure Pierre-Olivier Gourinchas in the early 2000s incomprehensible, save as the sign of a massive collective cognitive disfunction:
Comment on Blanchard and Giavazzi: "This is a very nice paper...(2002):
...It is simple and intuitive and elegantly fits an interesting fact to the theory. Blanchard and Giavazzi argue that large current account deficits in Portugal and Greece, two small and relatively poor members of the European Union, are exactly what the neoclassical growth model predicts should happen when such economies integrate their financial and goods markets with the rest of the world. And that these large deficits are not cause for worry.
Why should current account deficits in poorer countries increase with integration? Theory emphasizes two channels:
Faster conditional convergence and catch-up: Financial market integration, coupled with monetary union, reduces the cost of capital and eliminates currency risk. Cheap capital stimulates investment, while low interest rates and increased future wealth lower saving. Meanwhile product market integration reduces the adverse terms-of-trade effect that accompanies the need to generate a given trade surplus in the future, effectively making borrowing even cheaper.
Productivity catch-up: Through increased competition or better discipline, integration improves domestic total factor productivity (TFP), which increases the country’s growth prospects.
The authors present evidence that largely supports the theory: the dispersion of current account deficits across European countries has increased in the last five years. Poorer European countries tend to run larger deficits, and more so now than in the past, so that, finally, the high correlation between national saving and private investment—the Feldstein-Horioka puzzle—has largely disappeared for this group of countries. The authors provide detailed evidence for Portugal and Greece that emphasizes the importance of financial integration working through a decline in real interest rates (Portugal) or through an easing of the credit constraints on firms (Greece).
The paper provides a very convincing account and delivers a welcome piece of good news. After all, many emerging market economies have experienced a rather bumpy ride as they liberalized their goods and financial markets. Globalization, it seems, has not been a smooth process. Evidence that the standard theory performs well, at least for some European countries, is therefore reassuring. It leads to the natural conclusion that the gains from integration are there and that they should be driven by the channels mentioned above: conditional convergence and productivity catch-up. It also underlies the authors’ normative conclusion that the recent current account developments are—to a first order—optimal.
My comments will address each point in turn.
First, I will argue that conditional convergence and productivity catch-up have quite different welfare implications. In particular, the estimated welfare benefits from conditional convergence are relatively small for Portugal and Greece compared with the potential benefits from productivity catch-up.
Second, I will show that more is at play than the simple conditional convergence story. Productivity growth in Greece and Portugal has been faster than in other European countries. More generally, poorer OECD countries have experienced faster TFP growth. This strengthens the argument for laissez-faire, since productivity growth provides first-order welfare gains.
Third, to the extent that income per capita converges among members of the European Union, a central implication of the theory is that greater cross-country dispersion in current accounts should be matched by lower income inequality. Here the data do not oblige: the evidence indicates that income inequality has increased, not decreased, over the recent past. This casts some doubts on the mechanism that the paper emphasizes.
All this indicates that the normative conclusions that the authors reach may not be warranted. Large current account deficits, even when a consequence of credible financial integration, may lead to situations of illiquidity. Some strictly positive amount of insurance, in the form of a government surplus, may be necessary.
Conditional convergence and the benefits of open capital markets: The paper emphasizes the benefits of both product and financial market integration. Yet it should be clear that financial integration is the key ingredient. Product market integration is only relevant in the paper insofar as it facilitates intertemporal lending and borrowing.
So we may ask a simple question: how much benefit can a small, open economy like Portugal or Greece reap from financial integration? In the standard neoclassical growth framework, open financial markets bring about faster convergence toward a country-specific steady state. How beneficial is this conditional convergence? As it turns out, this simple question has a simple answer: not very.
To understand why, consider two extreme scenarios. First, consider a small, open economy under financial autarky, with no intertemporal trade. Alternatively, think of the same economy as a financially integrated economy with perfect and frictionless capital mobility. The latter scenario, of course, involves potentially very large current account deficits. The welfare difference between the two scenarios should set an upper bound on the true welfare gains that can accrue to a country like Portugal or Greece; after all, neither country was in a state of financial autarky before adopting the euro, nor is either currently experiencing unfettered capital flows.
To measure this upper bound, one needs only some estimate of the current and steady-state levels of physical and human capital per capita. Such estimates are provided and discussed in work I have done with Olivier Jeanne.1 Using these estimates, table 1 below reports compensating variation for twenty OECD countries as of 1995. Compensating variation measures the constant fraction of annual consumption that the typical household would have to give up to be indifferent between financial integration and financial autarky. For Greece this compensating variation is about 0.76 percent of annual consumption. The figure for Portugal is larger, at 2.67 percent. Those numbers are quite representative: compensating variation averages 0.91 percent of consumption for all countries in the sample, and it ranges from 0.10 percent for Norway to Portugal’s 2.67 percent.
How should we think of these numbers? I would argue that they are quite small. First, they are upper bounds on the true welfare benefits, and they are likely to be considerably smaller after adjustment costs, incentives, and intertemporal terms-of-trade effects are factored in. Second, even when taken at face value, they are small compared with the welfare benefits from, for instance, productivity improvements or the elimination of domestic distortions.2 These small numbers reflect the fact that, taken alone, financial integration is unlikely to remove domestic distortions or inefficiencies. This result weakens Blanchard and Giavazzi’s claim: if this is all that is going on, we should not worry about large current account deficits because they do not matter much for welfare, just as it does not matter much for welfare whether the capital account is open or closed.
Domestic efficiency gains in Portugal and Greece: Of course, conditional convergence is not the whole story. EU members exhibit convergence in income per capita. Equivalently, we observe a productivity catch-up. A simple look at labor productivity and TFP over the second half of the 1990s confirms that productivity growth is an important part of the story. Table 2 reports labor productivity for 1991–95 and 1996–2000 as well as TFP for 1990–95. There is clear evidence that both Portugal and Greece experienced strong labor productivity growth (at least 3 percent a year) between 1996 and 2000. When one compares 1991–95 with 1996–2000, it is also clear that Portugal and Greece (and Ireland) did break away from the pack. TFP also rose sharply in Ireland and Portugal, but less sharply in Greece.3
More systematic analysis finds strong evidence that productivity growth in OECD countries is linked—negatively—to the initial level of development. Regressing TFP growth from 1965 to 1995 on the initial level of output per capita, I obtain the following results:
These results indicate that low initial levels of output per capita are associated with faster TFP growth among the OECD countries. No such pattern is present for the non-OECD countries. This evidence in favor of productivity gains reinforces the message of the paper: after all, if poorer countries are also catching up in terms of TFP, so much the better, and the associated current account deficits should be even less of a concern.
One is left wondering, however, where these productivity gains are coming from. The paper mentions increased goods market competition and market discipline. Yet the discussion of Portugal and Greece does not revisit the issue as extensively. So we are left wanting more: is it purely a financial story whereby access to the international bond market (Portugal) or financial disintermediation (Greece) improves the efficiency of the domestic financial sector? Does it have to do with increased competition in goods markets? or with the discipline effect? These are important—and difficult—questions to answer.
Current account deficits, cross-country output, and catch-up: Take as given that the euro area countries are converging in terms of the level of GDP per capita, as suggested by the previous evidence. According to the neoclassical growth model, the cross-sectional variance for the logarithm of output per capita at time t, σ
2t , should follow:
2ut represents the variance of unexpected changes in preferences or production conditions across countries at time t, and β is the “speed of convergence” taken from the neoclassical growth model.4
This σ-convergence expresses the idea that we should expect to see less and less dispersion in output per capita as countries converge to their common steady state. For present purposes, observe that any factor that speeds up convergence (that is, increases β) should also lead to a faster decline in the cross-country dispersion of income. It is then a direct implication of the theory that financial and product market integration should lead simultaneously to an increasing dispersion, in the cross section, of the ratio of current account to GDP, and a decline in the dispersion of log output per capita.
Is this implication supported by the data? Figure 1 below reports the cross-country dispersion of log output per capita since 1975 for the OECD, the European Union, and the euro area, as defined in the paper. One can see a large decline in this measure of income inequality for all three groups, especially for the European Union and the euro area, where it has fallen from a peak of 0.32 in 1984 to a trough of 0.23 in 1997. At first glance, this massive reduction in income inequality appears consistent with the convergence hypothesis just described. Most of this decline can in fact be traced to the growth performance of only three countries: Ireland, Spain, and Portugal.
However, when we examine the joint evolution of output per capita and the ratio of the current account balance to GDP, the evidence appears less trenchant. Figure 2 is a scatterplot, for the euro area, of the cross-country dispersion in the ratio of the current account to GDP against income inequality. The figure shows three distinct phases. In the early 1980s the dispersion of current account balances was very large, reflecting the large budget deficits in Portugal and Ireland.
From 1985 to roughly 1995 there was a massive reduction in income inequality, without any significant change in the dispersion of current account positions. Lastly, from 1996 to 2001 there was a large increase in current account dispersion, accompanied by a modest yet significant increase in income inequality.5 This last segment is the focus of the paper. Yet the associated increase in income inequality contradicts the view that convergence is driving the process. A look at the time plot of log GDP per capita for the EU countries (figure 3) confirms that convergence seems to have stopped, except for Ireland, over the period when current account deficits were driven apart.
Of course, it is possible that country-specific shocks were sufficiently large to counteract the convergence process and drive incomes per capita apart over that period. However, one is left wondering what exactly these shocks were.6
Should we worry? This slowdown or even reversal in convergence suggests that we should look at the components of the current account for further insight. Blanchard and Giavazzi state that “The channel [for the increased external deficit] appears to be primarily a decrease in saving— typically private saving... rather than an increase in investment.”
According to the paper’s table 1, private saving in Portugal decreased by 5.6 percent of GDP from 1985–91 to 2000–01. In Greece private sav- ing decreased by 7.7 percent of GDP between 1981–91 and 2000–01 (table 2). By contrast, investment increased in Portugal by a modest 2.8 percent of GDP and remained more or less constant in Greece. In both countries public saving is not an essential part of the story.
The experience of these two countries—up to this point—is very reminiscent of that of many Latin American countries that have adopted exchange rate–based stabilization programs.7 Stabilization of the exchange rate, renewed access to international capital markets, and some euphoria at the prospect of steady future growth combined to generate a strong consumption boom—that is, a decline in saving—which may or may not have been accompanied by an investment boom. Growth was initially solid and everything looked benign. Over time, however, clouds gathered on the horizon: the currency appreciated in real terms, competitiveness plummeted, and foreign investors became worried as growth performance failed to meet expectations. The endgame is well known: with a fixed exchange rate, restoring competitiveness required an adjustment in relative prices. Often this was too little and too late. Eventually capital pulled out, forcing a devaluation.
European countries, too, have experienced similar dynamics in the past. For instance, France’s experiment with “competitive disinflation” presents a number of similarities: a strong peg, a currency that appreciated over time relative to the deutsche mark, sustained external imbal- ances, a failure of wage and price moderation to restore competitiveness, and eventually an abrupt adjustment at the time of the 1992 crisis in the exchange rate mechanism.8
How do these episodes differ from the current situation in Portugal and Greece, and what lessons do they offer? I see two important differences. First, the Latin American countries in the early 1990s and the European countries in the late 1980s had a checkered inflation record. As a conse- quence, the real appreciation was relatively rapid. Inflation inertia was key to that process. Portugal and Greece today are in a different situation: there is virtually no risk that inflation will get out of control. Yet their strong economic performance, as well as the impact of convergence on the price of nontraded goods (the Balassa-Samuelson effect), implies that one can expect higher inflation in both countries than in the rest of the euro area. Annual consumer price inflation in September 2002 was 3.7 percent in Portugal and 3.4 percent in Greece, against only 2.1 percent for the euro area as a whole.9
Taken together, these considerations imply that real overvaluation may happen relatively slowly in Portugal and Greece. But there are signs that it is coming. In time, this will require an adjustment in relative prices, which may prove painful.
Second, to the extent that both countries belong to European monetary union, there is no escape clause: the risk of devaluation is also nonexistent. This means fewer reasons for financial markets to worry. Indeed, the fact that neither Portugal’s nor Greece’s debt carries substantial spreads over that of other European countries can be taken as a sign of market confidence in these countries’ ability to honor their international obligations. But this does not mean that capital cannot or will not pull out. Even with a relatively evenly distributed maturity structure, markets could refuse to finance additional increases in debt. At current levels this would mean a sudden stop on the order of 5–7 percent of GDP, which would surely raise the specter of default. In other words, although a common currency may eliminate concerns that capital flight will force a devaluation, it does not ensure against situations of illiquidity.
What should governments do? Certainly, provision of full insurance is unwarranted. As the authors argue, this would completely eliminate the benefits of greater integration. But it seems that no insurance at all is not the answer either. This discussion teaches us that there is some, probably strictly positive level of insurance that governments should purchase: a buffer, in the form of a larger government surplus, would prove useful should markets become less confident.