The journey towards the Economic and Monetary Union (EMU) in Europe was underpinned by the Maastricht criteria, which prioritized nominal convergence as a prerequisite for a stable common currency. These criteria focused on aligning nominal variables such as inflation, interest rates, exchange rates, and public deficits among prospective member states. However, parallel to this nominal alignment, an academic discourse was burgeoning, emphasizing the significance of real convergence. This perspective, rooted in the optimal currency area (OCA) theory, posited that for a common currency to function optimally, participating countries should exhibit fundamental similarities and economic integration. This would mitigate the impact of asymmetric shocks and ensure flexible markets capable of adapting without the buffer of nominal exchange rate adjustments. Therefore, the pre-EMU academic debate centered more on real economic convergence, focusing on the structural adjustments needed to harmonize economic institutions, reduce susceptibility to divergent economic shocks, and minimize adjustment costs within a monetary union. This understanding of real euro convergence was seen as crucial for the long-term stability and prosperity of the Eurozone.
In the initial decade of the Eurozone, nominal and real euro convergence appeared to progress in tandem. Financial integration facilitated the convergence of nominal interest rates, driven by a reduction in perceived financial risks. The anticipation of a stable currency, devoid of redenomination risks, led to a significant decrease in both the average and variability of ten-year government bond rates across the initial 12 Eurozone countries between 1994 and 1997, as depicted in Figure 1. This trend of convergence persisted for approximately a decade before the European sovereign debt crisis of 2010-2012 disrupted it. The crisis triggered a surge in the variance of bond rates, reaching levels reminiscent of the pre-1990s era. While real euro convergence in per capita incomes was observed across the broader Eurozone, it was notably absent among the EA11 group (the original Eurozone members excluding Luxembourg, but including Greece). Post-crisis, divergent economic patterns became increasingly prominent, challenging the initial hopes of unified real euro prosperity.1
Figure 1Nominal convergence in the euro area
Alt Text: Graph showing nominal convergence in the euro area with decreasing variance of 10-year government bond rates from 1994 to 1997, followed by a spike during the European sovereign debt crisis.
Source: AMECO database, European Commission.
Figure 2Real convergence in the euro area
Alt Text: Chart illustrating real convergence trends in the euro area, highlighting divergent patterns post-financial crisis, particularly within the EA11 group.
Source: Penn World Tables 9.1.
The dynamics of real euro convergence are intricately linked to capital flows and current account balances within the Eurozone. In the lead-up to the EMU, capital predominantly flowed from the Eurozone’s ‘core’ economies to its ‘periphery’. These current account divergences were generally viewed as a positive aspect of the convergence process, reflecting capital movement from higher-income to lower-income nations (Blanchard and Giavazzi, 2002). However, the global financial crisis prompted a reassessment of risk, leading to a sudden capital withdrawal from the periphery. These ‘sudden stops’ in current accounts precipitated a significant contraction in domestic demand and a reversal of the real euro convergence trend (Diaz del Hoyo, 2017).
This divergence in external positions was mirrored by contrasting unemployment trends across the Eurozone (Figures 3 and 4). Between 1999 and 2007, the periphery experienced a relatively robust cyclical phase characterized by declining unemployment rates. Conversely, the core economies faced weaker cyclical conditions, with unemployment increasing between 2002 and 2005. This pattern inverted in the post-crisis period, as current account reversals in the periphery were followed by a surge in unemployment.
These economic shifts had repercussions not only on cyclical dynamics but also on long-term growth potential. Prior to the crisis, investments directed towards the periphery were largely concentrated in the non-tradable sector. According to the Walters critique of the EMU, this meant that persistent real interest rate differentials manifested not just in cyclical disparities, but also in divergent economic structures, notably in the composition of output between tradable and non-tradable sectors (Buti and Turrini, 2015).2 The expansion of the non-tradable sector in the Eurozone periphery – in some instances fueled by substantial housing market bubbles – was generally accompanied by a decline in cost competitiveness, undermining the prospects for sustainable, export-driven growth and hindering genuine real euro convergence.
Figure 3External balances in the euro area
Alt Text: Chart depicting external balances within the euro area, contrasting core and periphery countries based on their external positions during 1999-2009.
Note: Core includes BE DE LU NL AT FI. Periphery includes EE IE EL ES FR IT CY LV LT MT PT SI SK. Core and periphery euro area countries grouped according to their external position over the 1999-2009 period (GDP weights).
Source: Eurostat.
Figure 4Unemployment in the euro area
Alt Text: Graph showing unemployment trends in the euro area, differentiating between core and periphery regions as previously defined.
Note: Core and periphery defined as before.
Source: Eurostat.
This analysis aims to investigate the hypothesis that pre-crisis macroeconomic imbalances significantly impacted real euro convergence within the Eurozone. While previous research has explored the connection between business cycle synchronization and macroeconomic imbalances,3 this study specifically focuses on the role of these imbalances in explaining the pace of real euro convergence.
In pursuing this objective, this paper extends beyond existing studies in several key aspects.4 Firstly, it benchmarks the Eurozone’s convergence patterns against the experiences of other country groups, both within and outside the EU and Eurozone. A broad panel of advanced and emerging economies is considered for this comparative analysis. Secondly, the study examines convergence across multiple dimensions, not only in terms of per capita GDP but also in total factor productivity (TFP), offering a more nuanced view of real euro convergence. Both ‘sigma’ and ‘beta’ convergence methodologies are employed. Thirdly, it correlates deviations from projected convergence paths with a range of variables indicative of macroeconomic imbalances. These include government and private debt levels, current account balances, net international investment positions (NIIP), credit flows, and the growth of the non-tradable sector, providing a comprehensive assessment of factors influencing real euro convergence.
Real convergence in the euro area
The dataset employed for this analysis comprises a comprehensive panel of advanced and emerging economies, largely sourced from the Summers–Heston Penn World Tables (PWT) version 9.1.5 This database provides harmonized data on variables expressed in purchasing power parity, facilitating robust cross-country comparisons over time.
Before examining the evidence of real euro convergence, it is important to revisit the concepts of beta convergence and sigma convergence, which are central to this analysis.
Unconditional beta convergence is observed when the growth rate of real per capita GDP exhibits an inverse relationship with its initial level. In simpler terms, poorer economies tend to grow faster than richer ones, suggesting a catch-up mechanism.
Conditional beta convergence expands on this by acknowledging that the growth rate of real per capita GDP is still inversely related to the starting level but is also influenced by other factors. These factors, or conditioning variables, account for differences in steady-state growth rates across countries. This relationship can be formally expressed as:
∆ logYit = α + β logYit-1 + ζ Zit+ uit (1)
Here, ∆ logYit represents the average growth rate of country i over a period t, approximated by the log difference in GDP per capita. This growth rate is related to the initial GDP per capita level (Yit-1), a set of conditioning variables (Zit), and a random error term (uit).
Beta convergence, whether conditional or unconditional, implies that economies with lower initial income levels will eventually converge towards those with higher incomes by growing at a faster rate. This expectation is reflected in a negative coefficient for β in equation (1). Conditional beta convergence further refines this by suggesting that countries converge towards their own unique steady-state growth paths, which are determined by their specific characteristics.
Sigma convergence, on the other hand, focuses on the dispersion of income across a group of countries. Sigma convergence occurs if the income dispersion within a specific group diminishes over time, indicating a narrowing of income disparities. Abstracting from the conditioning variables (Z), equation (1) can be reformulated as (Barro Sala-i-Martin, 2004):
∆ logYit = α – (1 – e -λ ) logYit-1 + uit (2)
If λ > 0, equation (2) implies beta convergence, where poorer countries grow faster than richer ones. Defining the variance of logYit as σt2, equation (2) also implies:
σt2 = e -2λ σt-12 + σt2, (3)
where σt2 represents the variance of uit. Equation (3) reveals that beta convergence is a necessary but not sufficient condition for sigma convergence. While beta convergence is a prerequisite for reducing income dispersion, other factors, such as shocks or policy changes, can influence sigma convergence independently.
Sigma convergence
To assess sigma convergence within the real euro context, Figures 5 and 6 illustrate the standard deviation of log GDP per capita for the Eurozone, the broader EU, the EA11, and a wider group of high-income countries. Figure 5 presents data from 1995 onwards to avoid data gaps for former transition economies, while Figure 6 extends the horizon back to 1960 by excluding transition countries from the country groupings. The trends in income dispersion suggest that sigma convergence has been more pronounced in the EU and the Eurozone compared to other high-income countries. This finding supports previous research identifying the EU as a ‘convergence club’.6 However, the EA11 group demonstrates convergence primarily up to the mid-1970s (when considering the longer time frame), followed by divergence after the global financial crisis. This divergence is particularly evident when including countries that implemented macroeconomic adjustment programs in response to the crisis, further complicating the path towards real euro convergence for this core group.
Figure 5Sigma convergence: Standard deviation of log GDP per capita, 1995-2015
Alt Text: Graph showing sigma convergence in GDP per capita from 1995-2015, comparing standard deviation of log GDP for Euro area, EU, EA11, and high-income countries.
Sources: Penn World Tables 9.1.
Figure 6Sigma convergence: Standard deviation of log GDP per capita, 1960-2020
Alt Text: Chart illustrating sigma convergence in GDP per capita from 1960-2020, comparing standard deviation of log GDP for Euro area, EU, EA11, and high-income countries over a longer period.
Sources: Penn World Tables 9.1.
The neoclassical growth model (Solow, 1956; Swan, 1956) posits that output convergence is primarily driven by the convergence of capital stock. In this framework, countries with lower capital stock per capita and consequently higher marginal productivity of capital offer more attractive investment opportunities. Figure 7 examines convergence patterns in capital stock per capita to assess whether empirical data aligns with the neoclassical model’s predictions regarding real euro convergence. The chart compares the EA11 group and a broader set of advanced non-transition economies since 1960, and Eurozone countries since 1995 (due to data limitations). The data reveals that convergence is considerably more pronounced when focusing on capital per capita rather than GDP per capita, even within the EA11 group. This observation lends empirical support to the standard neoclassical growth theory mechanism of convergence, suggesting that capital accumulation plays a significant role in driving real euro convergence.
Figure 7Sigma convergence: Standard deviation of log capital per capita
Alt Text: Graph depicting sigma convergence in capital per capita, showing standard deviation of log capital per capita for Euro area, EA11, and advanced non-transition economies.
Sources: Penn World Tables 9.1.
Discrepancies between GDP per capita and capital per capita dynamics can often be attributed to the influence of Total Factor Productivity (TFP).7 Modern growth theories emphasize TFP growth, driven by innovation and the diffusion of advanced technologies, as a key driver of income convergence (Aghion and Howitt, 2006). Countries lagging in technological advancement have greater potential for growth by adopting existing, superior technologies. Figure 8 presents the standard deviation of TFP within the EA11 group, offering insights into real euro convergence from a technology perspective. Limited TFP convergence appears to have occurred up to the 1990s. However, TFP dispersion fluctuated in subsequent decades. In contrast, more consistent convergence is observed for the broader set of advanced economies and the Eurozone as a whole, despite the relatively short data series available for the latter. This suggests that while capital convergence may be occurring, TFP convergence, a crucial component of sustained real euro convergence, has been more uneven, particularly within the EA11 group.
Figure 8Sigma convergence: Standard deviation of log TFP
Alt Text: Chart illustrating sigma convergence in TFP, showing standard deviation of log TFP for Euro area, EA11, and advanced economies.
Sources: Penn World Tables 9.1.
Figure 9 compares sigma convergence for GDP per capita with GDP per employee for the EA11 group. The graph clearly demonstrates a close co-movement between the dispersion of these two variables over the period starting in the 1960s. However, a notable upward spike in GDP per capita dispersion occurred post-crisis, while divergence in GDP per employee remained considerably more limited. This finding offers a refined understanding of the post-crisis divergence process within the real euro area. It indicates that the divergence was not primarily driven by significant disparities in capital per employee or TFP, but rather by substantial divergence in employment rates. This suggests that the post-crisis income divergence was largely a labor market phenomenon, potentially transient and concentrated in countries most severely impacted by post-crisis recessions and the unwinding of macroeconomic imbalances and debt crises.
Figure 9Sigma convergence: GDP per capita vs GDP per employee – standard deviations, EA11
Alt Text: Graph comparing sigma convergence of GDP per capita and GDP per employee for EA11, showing standard deviations and highlighting post-crisis divergence in GDP per capita.
Sources: Penn World Tables 9.1.
Overall, the evidence suggests that sigma convergence in the Eurozone has occurred at rates comparable to those observed across other country groups. For the EA11, sigma convergence appears to have been slow and largely confined to the period up to the mid-1970s. The relatively sluggish pace of GDP per capita convergence can be partly attributed to the EA11 group’s initial high degree of income homogeneity. Furthermore, the lack of TFP convergence in recent decades has contributed to the stagnation of income convergence. The post-crisis divergence in income per capita is predominantly linked to divergent employment rates, a phenomenon likely to be temporary and concentrated in countries most affected by post-crisis economic adjustments.
The absence of sigma convergence does not necessarily preclude beta convergence. In other words, it does not rule out the possibility that, in general, countries with lower per capita incomes have experienced faster growth rates. The occurrence of specific shocks or crises can induce dispersion, even in the presence of underlying beta convergence trends driving real euro convergence.
Beta convergence
Beta convergence, a crucial aspect of real euro convergence, occurs when countries with lower initial per capita income levels exhibit faster growth rates over the medium to long term. Figure 2 provides initial visual evidence of beta convergence within the Eurozone. However, a more rigorous analysis necessitates accounting for factors beyond initial income levels that influence long-term growth performance. Growth regressions offer a method to incorporate these additional variables and test for ‘conditional’ convergence. In addition to initial income per capita, growth rates are analyzed in relation to other explanatory variables using panel regression techniques, leveraging both cross-sectional and time-series variations. The specific regression model employed is:
∆5 logYit = α + β logYit-5 + ζ Zit + γi + δt + εit , (4)
where Yit represents either output per capita (in PPP) or TFP, and Zit is a vector of control variables. The subscripts i and t denote countries and time periods, respectively. Following standard practice in panel data growth regressions, annual observations are averaged over five-year, non-overlapping sub-periods to mitigate short-term fluctuations (Barro Sala-i-Martin, 2004).
The control variables (Zit) are chosen to capture factors impacting steady-state growth, as informed by the neoclassical growth model. Population growth is included to account for the dilution of capital stock per capita, with an expected negative coefficient. The investment-to-GDP ratio serves as a proxy for the savings rate, expected to have a positive association with capital accumulation. Human capital accumulation, measured by the PWT 9.1 index based on schooling years and returns to education, is included to reflect improvements in labor input and is expected to have a positive coefficient.8 Two additional variables capture factors influencing TFP growth: trade openness (imports plus exports as a share of GDP) reflects technology and knowledge transfer (Edwards 1998; Frenkel and Romer, 1999); and institutional quality, measured by the Fraser index of economic freedom, accounts for the role of institutions in fostering innovation and risk-taking (Glaeser et al., 2004). Region-specific (γ) and time-specific (δ) fixed effects are included. Region fixed effects control for unobserved geographical factors without losing cross-section variation (Temple, 1999).
Table 1 presents the estimation results for conditional beta convergence in output per capita. Results are shown for the full sample of 66 advanced and emerging economies, as well as for country sub-samples and sub-periods.9 The control variables generally exhibit the expected signs, although significance varies across regions and periods. A common issue in growth regressions is the potential endogeneity of investment. However, instrumental variable estimation, using the price of investment goods as an instrument, yields qualitatively similar results, suggesting endogeneity is not a major concern here.10
For the entire sample, there is evidence of beta convergence, as indicated by the negative and statistically significant coefficient on initial GDP per capita. This supports the catching-up hypothesis. This finding holds for the EU (column 3) and the Eurozone (column 4), but not for the EA11 (column 5). Examining the post-euro adoption period (columns 6-8), the results remain consistent for the Eurozone, EU, and EA11. However, when focusing on the post-2007 period, encompassing the global financial crisis and European sovereign debt crisis, evidence of convergence weakens for the Eurozone and EU (columns 9 and 10). For the EA11, the post-2007 period suggests divergence rather than convergence, with the coefficient on initial per capita income becoming positive, though statistically insignificant in column 11.11
Table 1Conditional beta convergence: Output per capita
| Dep var: GDP p.h. growth,5-year averages | (1)All sample | (2)All sample, IV | (3)EU | (4)EA | (5)EA11 | (6)EU>1999 | (7)EA>1999 | (8)EA11>1999 | (9)EU>2007 | (10)EA>2007 | (11)EA11>2007 |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Ln GDP p.h. PPP, 5 lags | -2.266**[-6.80] | -2.566** [-7.04] | -3.072** [-6.91] | -2.802** [-6.43] | -0.749 [-0.86] | -3.905** [-7.97] | -4.149** [-7.11] | -2.291+ [-1.86] | -0.547 [-0.56] | -0.423 [-0.53] | 7.148 [1.40] |
| Human capital, 5 lags | 0.436 [1.16] | 0.566 [1.61] | 0.366 [1.11] | 0.479 [1.47] | -0.317 [-0.47] | 1.355** [2.99] | 1.541** [3.34] | 1.311+ [2.11] | 2.293** [3.84] | 2.655** [3.94] | -1.434 [-0.47] |
| I/GDP, avg | 10.422** [6.76] | 15.589** [7.48] | 8.211** [3.66] | 5.003+ [1.90] | 2.846 [0.86] | 2.559 [0.46] | 3.747 [0.54] | 9.090 [0.69] | 5.976 [0.55] | 14.611 [1.46] | 5.449 [0.27] |
| Pop growth, avg | -0.449* [-2.27] | -0.404* [-2.17] | -0.437+ [-1.98] | -0.583* [-2.72] | -0.051 [-0.17] | -0.387 [-1.48] | -0.415 [-1.44] | 0.173 [0.31] | -1.934** [-2.92] | -2.516** [-6.34] | -0.750 [-0.20] |
| Economic freedom, avg | 0.514** [3.71] | 0.470** [3.45] | 0.945** [3.53] | 0.498 [1.25] | 0.344 [0.64] | 1.303* [2.62] | 2.398** [3.52] | 1.740+ [1.97] | 3.356** [2.87] | 3.177* [2.29] | 3.089 [1.18] |
| Openness, avg | 0.794* [2.29] | 0.565+ [1.81] | 0.870+ [1.78] | 1.350** [2.98] | 2.386 [1.64] | 1.445** [3.07] | 1.316* [2.59] | 1.670 [1.12] | 2.161** [3.06] | 2.216* [2.41] | -1.376 [-0.54] |
| Observations | 516 | 516 | 203 | 143 | 99 | 84 | 57 | 33 | 28 | 19 | 11 |
| Countries | 66 | 66 | 28 | 19 | 11 | 28 | 19 | 11 | 28 | 19 | 11 |
| R-squared | 0.41 | 0.40 | 0.57 | 0.64 | 0.62 | 0.70 | 0.74 | 0.54 | 0.71 | 0.83 | 0.80 |
Alt Text: Table 1 displaying conditional beta convergence results for output per capita across different samples and time periods, highlighting coefficients and statistical significance.
Note: Constant, time effects and regional effects included. Robust (clustered) t-statistics in brackets. **p
Source: Authors’ estimation.
Growth regressions were also performed to assess beta convergence in TFP growth, another critical dimension of real euro convergence. Table 2 presents these results. Initial TFP, human capital, investment, institutional quality (Fraser index), and trade openness were considered as control variables. Initial TFP is expected to be negatively correlated with TFP growth, reflecting the catch-up potential of laggard countries. Human capital accounts for the innovation capacity of a skilled workforce. Institutional quality captures the incentive environment for innovation and entrepreneurship. Trade openness reflects technology diffusion. While initial TFP levels and institutional quality appear significant in explaining TFP growth, other control variables are generally statistically insignificant (column 1). Restricting the specification to only initial TFP and institutional quality (column 2) does not substantially alter the coefficient for initial TFP. Columns 3-9 utilize this restricted specification.
Table 2Conditional beta convergence: TFP
| Dep var. TFP growth | (1)All sample | (2)All sample | (3)EU | (4)EA | (5)EA11 | (6)EA1999-2007 | (7)EA>2007 | (8)EA11, 1960-2007 | (9)EA11>2007 |
|---|---|---|---|---|---|---|---|---|---|
| Log TFP level PPP, 5lag | – 1.537** [-4.88] | -1.499** [-4.73] | -2.064** [-4.41] | -2.058* [-2.76] | 0.500 [1.04] | -3.900** [-5.10] | -1.983+ [-1.98] | 0.035 [0.07] | 3.539+ [2.06] |
| Avg. schooling, 5 lags | 0.337+ [1.75] | ||||||||
| I/GDP, avg | -0.496 [-0.45] | ||||||||
| Economic freedom, avg. | 0.201* [2.09] | 0.241** [2.93] | 0.386** [3.01] | 0.340 [1.31] | 0.880** [4.06] | 0.389 [0.98] | 2.580+ [2.05] | 0.845** [4.19] | 1.827+ [1.94] |
| Openness, avg. | 0.153 [0.59] | ||||||||
| Observations | 502 | 502 | 203 | 143 | 99 | 38 | 19 | 88 | 11 |
| Countries | 64 | 64 | 28 | 19 | 11 | 19 | 19 | 11 | 11 |
| R-squared | 0.25 | 0.25 | 0.30 | 0.36 | 0.55 | 0.64 | 0.35 | 0.55 | 0.81 |
Alt Text: Table 2 showing conditional beta convergence results for TFP across different samples and periods, highlighting coefficients and statistical significance.
Note: Constant, time effects and region effects included. Robust (clustered) t-statistics in brackets. ** p
Source: Authors’ estimation.
The results provide evidence of TFP convergence across the entire sample of countries (columns 1 and 2), as well as for the EU and Eurozone (columns 3 and 4). However, no evidence of TFP convergence is found for the EA11, and TFP appears to diverge in the post-financial crisis period (columns 5, 8, and 9), hindering real euro convergence. Conversely, the EU as a whole shows evidence of TFP convergence even in the post-crisis sub-sample (columns 6 and 7), suggesting a more resilient convergence process outside the core Eurozone group.
Deviations from convergence paths: A role for macroeconomic imbalances?
What factors might explain the lack of real euro convergence within the Eurozone, particularly within the EA11? Could macroeconomic imbalances be a contributing factor? To address these questions, the first step is to quantify ‘convergence gaps’. These gaps represent the difference between actual growth rates and the growth rates predicted based on countries’ initial conditions and other growth-related characteristics. These predicted paths are derived from regressions estimated on the largest country panel and time period (column 1, Table 1 for GDP and column 2, Table 2 for TFP), ensuring robust and less biased estimates. Figure 10 illustrates actual GDP growth rates compared to predicted growth paths for Eurozone countries over five-year periods.
Figure 10Actual growth rates in GDP per capita and predictions from growth regressions
Alt Text: Graph comparing actual GDP per capita growth rates with predicted growth rates from regressions for Euro area countries over five-year periods.
Sources: PWT 9.1 and authors’ estimations.
The second step involves correlating these convergence gaps with variables reflecting macroeconomic imbalances. Figures 11 and 12 suggest a potential correlation between convergence deviations and private debt stocks and current account balances, respectively. To simultaneously assess the influence of various macroeconomic imbalances, multivariate regression analysis is employed. Regression (5) is estimated using OLS to formally test the role of imbalances in the real euro convergence process:
Figure 11Convergence gaps and private debt stocks
Alt Text: Scatter plot showing the correlation between convergence gaps and private debt stocks, indicating a potential negative relationship.
Source: Eurostat and authors’ estimations.
Figure 12Convergence gaps and current accounts
Alt Text: Scatter plot showing the relationship between convergence gaps and current account balances, suggesting a possible correlation.
Source: Eurostat and authors’ estimations.
εit = α + λ IMB it-5 + γi + δt + uit , (5)
Here, εit represents the convergence gap for GDP per capita or TFP growth (residuals from regressions in Table 1 and Table 2), and IMB represents a set of variables related to macroeconomic imbalances. Six variables are considered: (i) initial private debt-to-GDP ratio; (ii) initial government debt-to-GDP ratio; (iii) initial net international investment position (NIIP) as a percentage of GDP; (iv) credit to the private sector as a share of GDP; (v) current account gap; and (vi) the share of construction in total value added, as a proxy for the non-tradable sector.12 The credit and construction share variables are demeaned by country long-term averages to account for structural differences. The current account gap is calculated as the difference between the actual current account balance and its ‘norm’ based on economic fundamentals, following Coutinho et al. (2018). Data sources include Eurostat and AMECO, supplemented by World Economic Outlook and Bank of International Settlements data for non-EU countries.
Table 3 presents the regression results, separately for the Eurozone and a comparison group of all other countries in the sample. Time splits are also included: post-1999 (euro introduction) and post-2007 (financial crisis). Results are shown for both GDP per capita and TFP convergence gaps.
Table 3Deviations from convergence paths and macroeconomic imbalances
| | GDP growth residuals | | TFP growth residuals |
|—|—|—|—|—|—|—|—|
| | (1)EA>1999 | (2)Non-EA>1999 | (3)EA>2007 | (4)Non-EA>2007 | | (5)EA>1999 | (6)Non-EA>1999 | (7)EA>2007 | (8)Non-EA>2007 |
| Private debt/GDP, 5 lags | -0.009** [-3.78] | -0.014** [-4.13] | -0.014** [-3.33] | -0.001 [-0.18] | | -0.005* [-2.53] | -0.009** [-2.76] | -0.007* [-2.34] | 0.001 [0.32] |
| Gov. debt/GDP, 5 lags | -0.029** [-5.67] | -0.005 [-1.35] | -0.035* [-2.31] | -0.000 [-0.05] | | -0.025** [-5.39] | -0.005 [-1.57] | -0.027+ [-2.09] | 0.001 [0.21] |
| NIIP/GDP, 5 lags | 0.009* [2.72] | -0.002 [-0.79] | 0.015+ [1.97] | -0.001 [-0.15] | | 0.006+ [1.99] | -0.000 [-0.12] | 0.008 [1.27] | 0.002 [0.76] |
| Credit flow/GDP, 5 lags (relative to country long-term average) | 0.018+ [2.00] | 0.025+ [1.78] | -0.012 [-0.63] | -0.017 [-0.54] | | 0.016 [1.53] | 0.002 [0.23] | 0.006 [0.44] | 0.001 [0.07] |
| Current account gap, 5 lags | 0.018 [0.56] | 0.092+ [1.94] | 0.033 [0.34] | -0.030 [-0.34] | | -0.000 [-0.01] | 0.048 [1.64] | 0.005 [0.06] | -0.030 [-0.80] |
| Construction VA share, 5 lags (relative to country long-term average) | -0.579** [-3.21] | -0.228+ [-1.75] | -0.929** [-3.85] | -0.317 [-1.39] | | -0.500** [-3.58] | -0.133 [-0.96] | -0.623** [-2.88] | 0.047 [0.39] |
| Observations | 53 | 93 | 19 | 32 | | 53 | 93 | 19 | 32 |
| Countries | 19 | 32 | 19 | 32 | | 19 | 32 | 19 | 32 |
| R-squared | 0.53 | 0.36 | 0.72 | 0.36 | | 0.50 | 0.26 | 0.62 | 0.41 |
Alt Text: Table 3 presenting regression results on deviations from convergence paths and macroeconomic imbalances for GDP and TFP growth residuals, across Euro area and non-Euro area groups.
Note: Robust t-statistics in brackets. Constant, time effects and regional effects included. ** p
Source: Eurostat and authors’ estimation.
For the sample period starting in 1999, private debt, government debt, NIIP, and the construction share significantly explain Eurozone GDP per capita convergence gaps, with coefficients exhibiting expected signs. For non-Eurozone countries, most variables lose significance except private debt, while current accounts gain some explanatory power. Eurozone results remain similar when restricting the analysis to the post-crisis period. Wald tests do not reject the hypothesis that coefficients are equal across sub-periods at the 95% confidence level. However, for non-Eurozone countries, all variables lose significance in the post-crisis period.
Overall, the evidence suggests a significant negative association between high private debt and subsequent growth relative to predicted convergence paths. This finding is robust across country groups and time periods. For the Eurozone, high public debt and a large non-tradable sector also appear to significantly depress growth below expected convergence levels. External imbalances are less robustly linked to convergence gaps. For TFP convergence gaps in the Eurozone, private debt and the non-tradable sector remain significant, while only private debt is significant for non-Eurozone countries.
These findings indicate a clear link between convergence gaps and macroeconomic imbalances, with distinct features within the Eurozone. The stronger impact of government debt on Eurozone convergence gaps is consistent with the increased probability of bond market tensions associated with higher debt levels. Given that Eurozone countries, on average, have higher government debt, and bond market tensions significantly impede growth via credit availability and sovereign-bank feedback loops, a greater impact of government debt on real euro convergence paths is expected.13 Current accounts appear less critical for explaining Eurozone convergence deviations, possibly due to the European System of Central Banks’ liquidity provision mitigating the real effects of current account sudden stops (Merler and Pisani-Ferry, 2012). Finally, convergence paths in the Eurozone are comparatively more sensitive to non-tradable sector growth. The narrowing of interest rate differentials in the Eurozone periphery, facilitated by monetary union, led to capital inflows largely directed towards construction and other non-tradable activities. This resource allocation limited the periphery’s export-driven growth potential when domestic demand contracted due to deleveraging. Furthermore, as TFP growth is typically faster in tradable sectors, the expansion of non-tradable activities contributed to subsequent disappointing TFP growth rates, hindering real euro convergence.
Conclusions
This paper has examined convergence patterns within the Eurozone and investigated the role of macroeconomic imbalances in explaining the lack of real euro convergence, particularly in the post-crisis era. The analysis reveals that Eurozone convergence patterns are broadly similar to those of other country groups. However, convergence within the EA11 (original Eurozone members excluding Luxembourg) has lagged since the 1980s, primarily due to persistent TFP divergence, partly attributable to the group’s already high income homogeneity. Post-crisis employment rate divergence has further exacerbated this trend, impeding real euro convergence.
The study also demonstrates that deviations from predicted convergence paths are associated with pre-existing macroeconomic imbalances, notably private debt. Crucially, convergence gaps within the Eurozone appear to be uniquely influenced by factors such as government debt and the share of the non-tradable sector, which are less significant in comparison groups. Eurozone convergence gaps are also comparatively less sensitive to external imbalances, potentially due to the mitigating effects of intra-Eurozone financial mechanisms.
Overall, this analysis underscores the critical importance of macroeconomic stability and resilience for fostering economic convergence and realizing the promise of a truly unified real euro area. Preventing excessive private debt accumulation is paramount both within and outside the Eurozone. Furthermore, maintaining prudent public debt levels and fiscal policies is particularly vital within the Eurozone context. A key policy implication is that achieving sustainable real euro convergence necessitates addressing legacy macroeconomic imbalances and fostering balanced economic structures to ensure long-term prosperity and stability for all member states.
* The authors thank the participants of the European Central Bank Surveillance Workshop in Frankfurt on 19 October 2017, and in particular David Sondermann, for useful comments and suggestions. We also thank Marco Buti and José Leandro for useful comments on an earlier version of the paper. All remaining mistakes are the authors‘ solely. The views expressed in this paper are those of the authors and should not be attributed to the European Commission.
References
Aghion, P. and P. Howitt (2006), Joseph Schumpeter lecture appropriate growth policy: A unifying framework, Journal of the European Economic Association, 4(2-3), 269-314.
Barro, R. J. and X. Sala-i-Martin (2004), Economic Growth, MIT Press.
Blanchard, O. and F. Giavazzi (2002), Current account deficits in the euro area: the end of the Feldstein-Horioka puzzle?, Brookings Papers on Economic Activity, 2, 148-186.
Böwer, U. and A. Turrini (2010), EU Accession: A Road to Fast-track Convergence?, Comparative Economic Studies, 52, 181-205.
Buti, M. and A. Turrini (2015, 17 April), Three waves of convergence. Can Eurozone countries start growing together again?, VoxEU, https://voxeu.org/article/types-ez-convergence-nominal-real-and-structural (21 August 2020).
Coutinho, L., A. Turrini and S. Zeugner (2018), Methodologies for the Assessment of Current Account Benchmarks, European Economy Discussion Paper, 086.
De Grauwe, P. and Y. Ji (2013), Self-fulfilling crises in the Eurozone: An empirical test, Journal of International Money and Finance, 34, 15-36.
Diaz del Hoyo, J.-L., E. Dorrucci, F. F. Heinz and S. Muzikarova (2017), Real convergence in the euro área: a long-term perspective, European Central Bank Occasional Paper, 203.
ECB (2015), Real convergence in the euro area: evidence, theory and policy implications, European Central Bank Economic Bulletin, 5.
Edwards, S. (1998), Openness, productivity and growth: what do we really know?, The Economic Journal, 108(447), 383-398.
Estrada, Á. and D. López-Salido (2013), Patterns of convergence and divergence in the euro area, IMF Economic Review, 61(4), 601-630.
Frankel, J. A. and D. H. Romer (1999), Does trade cause growth?, American Economic Review, 89(3), 379-399.
Glaeser, E. L., R. La Porta, F. Lopez-de-Silanes and A. Shleifer (2004), Do institutions cause growth?, Journal of Economic Growth, 9(3), 271-303.
Inklaar, R. and P. Woltjer (2019), What is new in PWT 9.1?, University of Groningen.
Inklaar, R., R. Jong-A-Pin and J. de Haan (2008), Will business cycles in the euro area converge? A critical survey of empirical research, Journal of Economic Surveys, 22(2), 234-273.
Lukmanova, E. and G. (2017), Macroeconomic imbalances and business cycle synchronisation. Why common governance is imperative in the Eurozone, Economic Modelling, 62, 130-144.
Mankiw, N. G., D. Romer and D. N. Weil (1992), A contribution to the empirics of economic growth, Quarterly Journal of Economics, 107, 407-437.
Merler, S. and J. Pisani-Ferry (2012), Sudden stops in the euro area, Bruegel Policy Contribution, 2012/06.
Mundell, R. A. (1961), A theory of optimum currency areas, The American Economic Review, 51(4), 657-665.
Solow, R. (1956), A contribution to the theory of economic growth, Quarterly Journal of Economics, 70(1), 65-94.
Sondermann, D. (2014), Productivity in the euro area: any evidence of convergence?, Empirical Economics, 47(3), 999-1027.
Swan, T. (1956), Economic growth and capital accumulation, Economic Record, 32(63), 334-361.
Temple, J. (1999), The new growth evidence, Journal of Economic Literature, 37(1), 112-156.
