The analysis so far has considered the impact of individual reforms, rather than packages of multiple reforms, on economic growth. In practice, policymakers will wish to act on the basis of a reform strategy that takes into account sequencing issues as well as possible complementarities among reforms on expected growth outcomes. While political constraints may often be paramount in determining what can be achieved and when, policymakers will opt for reform strategies that are likely to have the most favorable impact on economic welfare or growth subject to the other constraints. Although there is a large literature on the design and sequencing of reform packages, including, for example, whether “big bang” or piecemeal approaches to reform yield greater economic returns, the empirical evidence on the growth benefits of alternative sequencing strategies, and indeed, as a prior matter, the “stylized facts” of reform sequencing strategies—what countries have actually pursued in practice—are relatively scarce.14 This section examines the cross-country evidence on sequencing strategies and their growth effects, linking it where appropriate to the existing theoretical/normative work on these issues, which holds that
International trade should be liberalized before the external capital account. McKinnon (1973) argued that liberalizing capital inflows before trade was likely to amplify the distortions caused by tariffs and reduce the competitiveness of domestic firms through real appreciation. Liberalizing capital outflows before trade would be equally undesirable if trade restrictions misallocate resources and depress domestic returns to the point that domestic capital would leave the economy. For both reasons, McKinnon, and others following him, have advocated a “trade reform first” strategy, contending that the growth benefits of reform would be higher under such a strategy than under alternative sequencing strategies.
The domestic financial sector should be liberalized before the external capital account. In the presence of regulated interest rates and other financial system distortions, capital mobility is likely to be destabilizing: capital inflows could lead to overborrowing in foreign currency, which a dysfunctional domestic financial sector would misallocate, and capital outflows could erode the domestic deposit base (McKinnon, 1973). There is some evidence that capital account liberalization may increase volatility and crisis risk in the absence of a sufficiently reformed domestic financial sector (Dell'Ariccia and others, 2008). If such volatility leads to an inefficient allocation of resources, growth should be higher when the domestic financial sector is reformed before the external capital account than under the reverse sequencing strategy.
Trade should be liberalized before the domestic financial sector. Opening the economy to international trade first has been argued to make subsequent reform of the domestic financial sector more likely because greater competition in product markets (through trade) is likely to weaken the influence of monopolistic incumbents who may oppose financial development (Rajan and Zingales, 2003). While this argument may explain why trade reform is more likely to precede domestic financial sector reform than vice versa, it does not necessarily imply that growth should be higher under the first sequence than the second.
To what extent do countries actually follow the sequencing prescriptions advocated in the normative literature? Table 6.1 presents evidence on actual sequencing practices, by testing whether some reform indicators are leading indicators of—that is, generally precede— changes in other reform indicators. Specifically, five-year changes in the indicators of domestic financial sector liberalization (column 1), external capital account liberalization (column 2), and the tariff-based trade liberalization index (column 3) are regressed on five-year lags of all other reform indicators, controlling for a variety of other determinants of liberalization. The results suggest that trade liberalization does indeed help to predict future reform of both the domestic financial sector and the external capital account (first row), while it is not itself predicted by either of the other reforms (last column), consistent with the “trade-first” strategy advocated in the normative literature. The data, however, do not speak loudly on whether domestic financial sector liberalization leads or lags external capital account liberalization. The estimated coefficient on domestic financial sector liberalization in the external capital account reform regression (second column/second row) is borderline significant, providing only weak evidence that countries tend to reform the domestic financial sector before opening up to foreign capital.
Sequencing of Structural Reforms

Sequencing of Structural Reforms
| Domestic Financial Sector Liberalization | External Capital Account Liberalization | Trade Liberalization (Tariff) | |
|---|---|---|---|
| Dependent Variable: Reform Index (t)–Reform Index (t–5) |
(1) | (2) | (3) |
| Trade liberalization (tariff) (t–5) | 0.107*** | 0.235** | -0.578*** |
| (0.04) | (0.10) | (0.08) | |
| Domestic financial sector liberalization (t–5) | -0.694*** | 0.189 | -0.030 |
| (0.07) | (0.13) | (0.06) | |
| External capital account liberalization (t–5) | -0.031 | -0.839*** | 0.034 |
| (0.03) | (0.08) | (0.03) | |
| Observations | 353 | 353 | 352 |
| Number of countries | 74 | 74 | 74 |
| Adjusted R-squared | 0.44 | 0.43 | 0.32 |
Sequencing of Structural Reforms
| Domestic Financial Sector Liberalization | External Capital Account Liberalization | Trade Liberalization (Tariff) | |
|---|---|---|---|
| Dependent Variable: Reform Index (t)–Reform Index (t–5) |
(1) | (2) | (3) |
| Trade liberalization (tariff) (t–5) | 0.107*** | 0.235** | -0.578*** |
| (0.04) | (0.10) | (0.08) | |
| Domestic financial sector liberalization (t–5) | -0.694*** | 0.189 | -0.030 |
| (0.07) | (0.13) | (0.06) | |
| External capital account liberalization (t–5) | -0.031 | -0.839*** | 0.034 |
| (0.03) | (0.08) | (0.03) | |
| Observations | 353 | 353 | 352 |
| Number of countries | 74 | 74 | 74 |
| Adjusted R-squared | 0.44 | 0.43 | 0.32 |
What about the growth effects of alternative sequencing strategies? Figure 6.1 illustrates the evolution of the indices of domestic financial sector and external capital account liberalization, as well as the tariff-based index of trade liberalization, before and after growth breaks. Overall, the trade-first sequence seems to be associated with growth upbreaks, while a trade-last sequence seems to characterize growth downbreaks. Specifically:
In the run-up to growth upbreaks, economies have generally already introduced trade reforms, with the trade liberalization index above the country average (top panel of Figure 6.1). In contrast, in the run-up to downbreaks, economies have yet to open to trade (bottom panel). This suggests that a liberal trade regime is involved both in igniting growth and in sustaining it—the latter result is also strongly supported by the analysis of growth duration in Berg, Ostry, and Zettelmeyer (2008).
During acceleration episodes, the top panel of Figure 6.1 shows that the index of domestic financial sector liberalization and that of external capital account liberalization are on a rising trend three to four years before the upbreak. Thus, an open trade regime, together with an increasingly liberalized domestic financial sector and external capital account, appear to be an integral part of acceleration episodes, with the latter two reforms mostly progressing together. In contrast, a relatively open external capital account, combined with a relatively closed trade account and domestic financial sector reforms about equal to country averages, seems to be a common feature of growth decelerations.


Growth Breaks and Sequencing of Reforms
Source: IMF staff estimates based on Penn World Tables version 6.2.Note: The figures plot average liberalization indices for the period beginning five years before a growth break (year 0 on the horizontal axis) and ending five years after the growth break. The plots capture the within-country evolution of the liberalization indices obtained from a panel regression of each index on country fixed effects (to remove country averages) and year fixed effects (to remove global trends). As a result, the zero value on the vertical axis corresponds to the sample average of the liberalization indices for the countries considered. The number of countries used to compute each average is the same across the three indices.
Growth Breaks and Sequencing of Reforms
Source: IMF staff estimates based on Penn World Tables version 6.2.Note: The figures plot average liberalization indices for the period beginning five years before a growth break (year 0 on the horizontal axis) and ending five years after the growth break. The plots capture the within-country evolution of the liberalization indices obtained from a panel regression of each index on country fixed effects (to remove country averages) and year fixed effects (to remove global trends). As a result, the zero value on the vertical axis corresponds to the sample average of the liberalization indices for the countries considered. The number of countries used to compute each average is the same across the three indices.Growth Breaks and Sequencing of Reforms
Source: IMF staff estimates based on Penn World Tables version 6.2.Note: The figures plot average liberalization indices for the period beginning five years before a growth break (year 0 on the horizontal axis) and ending five years after the growth break. The plots capture the within-country evolution of the liberalization indices obtained from a panel regression of each index on country fixed effects (to remove country averages) and year fixed effects (to remove global trends). As a result, the zero value on the vertical axis corresponds to the sample average of the liberalization indices for the countries considered. The number of countries used to compute each average is the same across the three indices.Econometric evidence corroborates the main results from Figure 6.1 regarding the growth effects of different reform sequencing strategies. Table 6.2 presents results on the effects of alternative pairwise sequencing strategies, controlling for standard growth determinants and the direct effects of reforms.15 The positive and statistically significant coefficient on the trade-before-external-capital-account liberalization sequencing term (first row/ second column) indicates that liberalizing trade before the capital account yields a more favorable growth outcome than the reverse sequence.16 By contrast, no clear ranking—in terms of growth outcomes—emerges between domestic financial sector liberalization and the opening of the external capital account (column 3), or between trade and domestic financial sector liberalization (the sequence with trade first is only borderline significant in column 1).
Growth Effects of Alternative Reform Sequencing Strategies

Growth Effects of Alternative Reform Sequencing Strategies
| Trade Liberalization (Tariff) Versus | Trade Liberalization (Tariff) Versus | Domestic Financial Sector Liberalization Versus | |
|---|---|---|---|
| Dependent Variable: | Domestic Financial Sector Liberalization | External Capital Account Liberalization | External Capital Account Liberalization |
| Annual Per Capita GDP Growth (t) | (1) | (2) | (3) |
| Reform sequence (first reform before second) (t–1) | 0.03 | 0.043** | 0.052 |
| (0.019) | (0.018) | (0.046) | |
| Direct effect (first reform) (t–1) | 0.019* | 0.026** | 0.071*** |
| (0.010) | (0.011) | (0.016) | |
| Direct effect (second reform) (t–1) | 0.062*** | 0.058*** | 0.034** |
| (0.023) | (0.017) | (0.014) | |
| Reform complementarity (t–1) | -0.018 | -0.051** | -0.054** |
| (0.024) | (0.020) | (0.021) | |
| Observations | 1,991 | 1,991 | 2,114 |
| Number of countries | 88 | 88 | 88 |
| Adjusted R-squared | 0.06 | 0.06 | 0.08 |
Growth Effects of Alternative Reform Sequencing Strategies
| Trade Liberalization (Tariff) Versus | Trade Liberalization (Tariff) Versus | Domestic Financial Sector Liberalization Versus | |
|---|---|---|---|
| Dependent Variable: | Domestic Financial Sector Liberalization | External Capital Account Liberalization | External Capital Account Liberalization |
| Annual Per Capita GDP Growth (t) | (1) | (2) | (3) |
| Reform sequence (first reform before second) (t–1) | 0.03 | 0.043** | 0.052 |
| (0.019) | (0.018) | (0.046) | |
| Direct effect (first reform) (t–1) | 0.019* | 0.026** | 0.071*** |
| (0.010) | (0.011) | (0.016) | |
| Direct effect (second reform) (t–1) | 0.062*** | 0.058*** | 0.034** |
| (0.023) | (0.017) | (0.014) | |
| Reform complementarity (t–1) | -0.018 | -0.051** | -0.054** |
| (0.024) | (0.020) | (0.021) | |
| Observations | 1,991 | 1,991 | 2,114 |
| Number of countries | 88 | 88 | 88 |
| Adjusted R-squared | 0.06 | 0.06 | 0.08 |
The benefits from appropriate reform sequencing can be economically significant. Based on the regression results in Table 6.2, column 2 (trade versus external capital account liberalization), Table 6.3 provides numerical examples of the growth effects of alternative reform sequencing strategies over a 15-year horizon. To keep alternative sequencing strategies comparable, growth paths were calculated for three scenarios. In each scenario, both trade and capital account indices are initially at the 25th percentile of their respective sample distribution. In the simultaneous reform (“big bang”) scenario, starting in the 5th year, both the trade and capital account indices increase linearly until they reach their respective 75th percentiles in the 11th year. In the “preferred sequencing” (trade before capital account liberalization) scenario, trade moves up to the 75th percentile during years two through eight, while capital account liberalization moves from the 25th to the 75th percentile during years 9 through 15. Finally, in the trade-after-capital-account-liberalization scenario, the reform sequence is the reverse of the second scenario.
Cumulative Growth Effects of Alternative Reform Sequences: A Numerical Example
(In percent)

Cumulative Growth Effects of Alternative Reform Sequences: A Numerical Example
(In percent)
| Reform Sequence of Trade Versus External Capital Account | |||
|---|---|---|---|
| Simultaneous (“big bang”) | Trade openness (tariff) before external capital account openness | External capital account openness before trade openness (tariff) | |
| Sequencing | 0.0 | 7.5 | -11.7 |
| Direct | 43.6 | 27.3 | 57.9 |
| Interaction | -31.5 | -20.0 | -34.8 |
| Total | 12.1 | 14.8 | 1 1.4 |
Cumulative Growth Effects of Alternative Reform Sequences: A Numerical Example
(In percent)
| Reform Sequence of Trade Versus External Capital Account | |||
|---|---|---|---|
| Simultaneous (“big bang”) | Trade openness (tariff) before external capital account openness | External capital account openness before trade openness (tariff) | |
| Sequencing | 0.0 | 7.5 | -11.7 |
| Direct | 43.6 | 27.3 | 57.9 |
| Interaction | -31.5 | -20.0 | -34.8 |
| Total | 12.1 | 14.8 | 1 1.4 |
The trade-before-capital-account reform sequence has by far the most favorable impact on per capita income levels, raising per capita GDP growth by an annual 1.4 percent during the 15-year horizon compared to a no-reform scenario (in which both reform indicators remain at their 25th percentiles), and, notably, by almost a ¼ percentage point relative to the “big bang” approach where reforms are pursued simultaneously. By contrast, the reverse sequence of capital account liberalization, while still better than no reform, performs worse than the simultaneous reform case, and yields annual growth about ⅓ percentage point lower than the trade-first sequence. Thus, when possible, implementing reforms in the correct order is likely to yield substantial growth benefits.
Bhattacharya (1997) provides a review of the theoretical literature. Previous empirical work has focused on the sequencing of product and labor market reforms for OECD countries (Fiori and others, 2007), but has generally ignored the broader sequencing issues among the different sectors covered in this paper. A related issue, well outside the remit of this paper, concerns the appropriate sequencing between macroeconomic stabilization and structural reforms; see Zalduendo (2005) for an analysis.
Specifically, the regressions control for several standard variables, such as per capita GDP, terms of trade shocks, tertiary education, political institutions, and country and time dummies. In addition, the regressions estimate the direct growth effects of reforms X and Y, possible complementarities between reforms through an interaction term X*Y, and a sequencing effect represented by a second interaction term X*Y*(%X–%Y), where %X denotes the percentile of a country's level of reform X in a given year in the regression sample.
The sequencing term reflects the extent to which more progress in one reform (i.e., a higher percentile) than in the other reform affects the complementarity of two reforms. Thus, a positive coefficient implies that more reform progress on X than Y is preferable to the reverse. The interaction between the product of the indices and their difference helps distinguish between the case in which the same product of the indices is achieved with a high value of the first liberalization index and a low value of the second (sequencing from the first to the second reform), vice versa (sequencing from the second to the first reform), or similar values of both indices (no sequencing). In fact, as Table 6.1 shows, the coefficient on the complementarity term is typically negative, that is, reforms typically are partial substitutes in their effects on growth. A positive coefficient on the sequencing term then implies that proper sequencing reduces the substitution effect and raises the total growth benefits of reform.
To make the level of liberalization comparable across sectors, the indices were transformed into percentiles of the distribution of each index. Estimating the regressions with the raw indices yields similar results.