International Monetary Fund

Please send us your feedback

Macro Research for Development: An IMF-DFID Collaboration

Topic 2: Public Investment, Growth, and Debt Sustainability

Last Updated: April 02, 2014

IMF economists, Rafael Portillo (left) and Felipe Zanna (photo: IMF)

IMF economists, Rafael Portillo (left) and Felipe Zanna (photo: IMF)

Many low-income countries (LICs) are working to deliver on an ambitious promise to scale up public investment to meet huge needs in infrastructure, energy, and other critical sectors.

With aid assistance stagnating, authorities are turning to non-concessional loans, which can bridge financing gaps but can also threaten macroeconomic stability and create heavy repayment burdens. To assess these risks, the World Bank and the IMF together created the Debt Sustainability Framework (DSF), an instrument for guiding LIC's borrowing decisions and reducing the chances of excessive debt accumulation. Although the framework is widely used, it has been criticized for a number of reasons. The Bank and the IMF have responded to some of these criticisms, for example, by proposing greater use of models such as that described below to better capture investment-growth linkages (see Revisiting the Debt Sustainability Framework for Low-Income Countries).

A Model-Based DSA (MBDSA)

Recently Buffie et al. (2012) have put forward a dynamic LIC-specific macroeconomic model that complements the IMF-World Bank DSF by addressing the following two criticisms:

(1) the DSF lacks a consistent analytic framework, which makes it difficult to systematically incorporate and assess the relationship between public investment and growth; and

(2) the existing DSF does not take into account the possibility of authorities making fiscal adjustments in reaction to rising debt levels.

To correct for these shortcomings, Buffie et al. develop an internally consistent model with productive sectors that use public capital as an input, different borrowing schemes (external concessional, external commercial, and domestic) and various fiscal rules that react to debt paths.

The authors conclude that an increase in infrastructure investment can produce striking benefits for the real economy in the long run because of output expansion and revenue gains. They note, however, that even highly productive investments may require long-run tax increases to finance recurrent costs of sustaining new public capital, because much of the benefit goes to the private sector and average tax rates are low. Even these positive results are contingent upon the country's structural conditions. Public investment inefficiencies and absorptive capacity constraints for example, can imply that the increases in private capital and GDP that result from increased public investment may be disappointing.

Public Investment Efficiency and Debt Sustainability

Simulations of the calibrated model show how lowering efficiency of public investment can translate into less effective capital, lower GDP, lower fiscal revenues and potentially unsustainable debt dynamics when countries contract external commercial debt.

Even if the long-run outcome is promising, transition problems can be formidable, especially when concessional borrowing does not fully cover the cost of the ambitious public investment plans, resulting in a substantial fiscal gap. Buffie et al. analyze the medium-term risks and trade-offs associated with different financing schemes:

(1) unconstrained tax and spending adjustments combined with concessional loans only;

(2) smooth tax adjustment supplemented with concessional and external commercial borrowing; and

(3) smooth tax adjustment supplemented with concessional and domestic borrowing.

Covering the fiscal gap with tax increases or government spending cuts requires sharp macroeconomic adjustments, crowding out private investment and consumption and delaying the growth benefits of public investment. Covering the gap with domestic borrowing is not helpful either: higher domestic rates increase the financing challenge and private investment and consumption are still crowded out. Supplementing with external commercial borrowing, on the other hand, can smooth these difficult adjustments, reconciling the scaling up with feasibility constraints on increases in tax rates or spending cuts. But borrowing in external commercial terms may be also risky. With poor execution (e.g., low public investment efficiency), sluggish fiscal policy reactions, or persistent negative exogenous shocks (e.g., terms-of-trade shocks or natural disasters), this strategy can easily lead to unsustainable public debt dynamics (see figure above).

The confluence of ambitious, front-loaded investment programs and weak structural conditions (such as low returns to public capital and poor execution of investments) make the fiscal adjustment more challenging and the risks greater.

Box 1. IMF-WB DSF Nuts and Bolts

Under the DSF, debt sustainability analyses (DSAs) are conducted regularly. They consist of:

  • a baseline set of 20-year projections for borrowing, GDP growth, exports, and other key macroeconomic variables that underpin an analysis of key debt ratios as well as the vulnerability assessment to external and policy shocks—baseline and stress tests
  • an assessment of the risk of debt distress over the 20 year period, based on indicative debt burden thresholds that depend on the quality of the country's policies and institutions; and
  • recommendations for a borrowing (and lending) strategy that limits the risk of debt distress.

Source: The Joint World Bank–IMF Debt Sustainability Framework for LICs

As a complement to the IMF-World Bank DSF, this fully articulated, dynamic macroeconomic model model developed by Buffie et al. provides a sophisticated debt sustainability analysis for LICs. This model enables country authorities, IMF country teams, and others to build a wide variety of logically consistent scenarios for public investment surges and other shocks. Its application in particular cases requires a number of country-specific assumptions. Many of these, such as the share of imports in consumption, can be based on available data. Others such as the rate of return to future public investments are more speculative. However, the use of the model should allow country teams and policymakers to understand better the implications of various assumptions and should drive demand for and eventually generate better information.

Next Step: Application

So far, six applications of this model have been completed: Togo, Côte d'Ivoire, Burkina Faso, Cape Verde, Rwanda and Liberia.

IMF teams working in all six countries have presented the model to the authorities in the context of Article IV consultations and discussions related to the HIPC Initiative and assessed the macroeconomic implications of investment scaling up. Fund staff has also provided one-week training about the model and how to use it to the Rwandan and Liberian authorities.

Box 2. Togo

As part of the rehabilitation process following a protracted domestic, social, and political crisis, Togolese authorities have initiated a plan to increase public investment in infrastructure rapidly over the next 10 years. Andrle et al.(2012) employ the Buffie et al. model to analyze a set of public investment scaling-up strategies that would allow Togo to reap all the benefits associated with public investment, particularly GDP growth, while ensuring favorable debt-dynamics.

The model suggests that despite the significant positive macroeconomic effects generated by a very large public investment increase in the long-run, the accompanying increase in tax rates necessary to ensure debt sustainability would be unrealistically high. Thus, a more moderate increase in public investment would be advisable, which would allow for a reasonable increase in tax rates in the medium-run, conditional upon an improvement in the efficiency of tax collection. However, authorities would have to monitor the speed and size of the fiscal adjustment carefully.

Another financing route that could help alleviate the immediate pressures of fiscal adjustment would be the use of external commercial borrowing to complement concessional loans, although this strategy heightens the risks from high interest rates and liquidity shocks that could bring forth unfavorable debt dynamics. Determining the timing and pace of scaling-up remains at the crux of the discussion, particularly whether Togo should frontload increases in investment or proceed more gradually to reap greater benefits from improving the efficiency of investment through "learning by doing" and reforming fiscal governance.

Lastly, when using the model to assess the growth projections underlying the IMF-World Bank debt sustainability analysis for Togo, Andrle et al. find that despite public investment shocks, a large component of these growth projections is actually explained by TFP shocks.

Togo. Fund staff examined the relationship between public investment, growth, and debt sustainability in Togo using the model developed in Buffie et al. The model was used to inform the policy dialogue with the authorities on sustainable investment scaling up plans. The application suggested that a larger investment scaling up could promote growth without jeopardizing debt sustainability, under certain conditions that would nevertheless require continued strong reform efforts on improving the efficiency and quality of public investment, obtaining additional external resources in the form of grants, and increasing revenue collection. IMF staff presented the results to the authorities, who found that the application provided a useful analytic tool for comparing medium-term policy alternatives. The authorities have also expressed their intention to give priority to mobilizing more revenue and accelerating public financial management reforms. (See Box 2 for additional details of this application.)

Burkina Faso. The application of the model to Burkina Faso revealed that, under certain conditions, scaling-up investment to levels envisioned in the new PRSP could raise economic growth substantially in the medium to long term, boost private investment, and increase consumption (see Appendix III of the Art. IV staff report). The model's results also suggested that, taking into account the country's membership in a customs and monetary union that limits VAT adjustments, better macroeconomic outcomes would be achieved if investments were scaled up only moderately and financed with higher grants, concessional loans, and increased government revenues from enhanced collection of user fees. The IMF country team used the results as the basis for policy recommendations on scaling up public investment, which the country authorities generally agreed with. The authorities found that the proposed financing option was in line with the amount of investment they had expected. Because the country is committed to VAT harmonization, the authorities reaffirmed their commitment to mobilizing more revenue.

Côte d’Ivoire. Fund staff applied the model developed by Buffie et al. to study the relationship between public investment, growth, and debt sustainability in Côte d’Ivoire (see HIPC completion point and MDRI document). Simulation results advised against a massive and sustained surge in public investment growth due to the likelihood that debt sustainability problems will arise even if structural conditions improve. The results suggested, instead, a moderate scaling up of public investment alongside structural reforms in such areas as public investment efficiency, the ability to collect more revenue, and the return to investment and productivity on both the traded and non-traded sectors.

Cape Verde. To evaluate public investment scaling-up strategies in Cape Verde, various scenario and sensitivity analyses were developed using the model. The results suggested that a moderate scaling-up of public investment—combined with ensuring the quality of public investment to raise productivity—was more likely to contribute to stable and sustained growth over the medium to long term (see Box 2 of the Art. IV staff report). These results served to discuss with the authorities that the focus of some of this public investment plans needs to move from capital accumulation to improvements in the quality and efficiency of public infrastructure. More ambitious plans might be feasible if some of the structural bottlenecks were solved combined with a concerted effort to revamp tax policy and administration in order to generate resources to finance these plans.

Rwanda. Using the model by Buffie et al., the Rwanda team studied the challenge of scaling up public investment and the associated resource requirements. Model simulations showed that scaled-up investment could provide a boost to growth, but mobilizing the required resources for the scaling up would be challenging because of unfeasible fiscal and private sector adjustments. Simulations that incorporated a combination of increases in tax revenue, expenditure rationalization, and judicious resort to external borrowing were the most realistic in terms of implementation. The simulations also demonstrated the importance of relying, to the extent possible, on concessional loans, and of the efficiency of the public investment process. The authorities appreciated the discussion and said they would draw on these results in the development of their EDPRS II (see Box 4 of the Art. IV staff report).

Liberia. The model was used to examine the macroeconomic impact of scaling up public investment in Liberia. The simulations captured some of the costs and benefits of three potential investment strategies. The more ambitious investment program yielded the larger growth dividend over the medium term, but at the risk of unsustainable debt dynamics. Other results showed that a combination of increased public investment and efficiency of capital spending could lead to a 1 percentage point increase in the average annual growth rate of real per capita income over ten years. These results helped to discuss with the authorities the importance of step up efforts to tackle bottlenecks to project implementation and even more importantly to properly prioritize and select high return projects that effectively translate into productive capital (see Box 4 and supplement 1 in the Art. IV staff report)

Applications to Afghanistan, Senegal, Ghana, Tanzania and Ethiopia (see Box 3) are well underway. The table below lists the countries and the status of the application.

Box 3. Ethiopia

The model is currently being applied to Ethiopia. To ensure that the model fully captures the country's idiosyncratic features, the team made the following modifications and extensions:

  • include a state-owned electricity sector where domestic prices are set below the shadow price;
  • introduce a state-owned banking sector that sets deposit and lending interest rates exogenously;
  • allow for seigniorage.

The IMF Ethiopia country team and Enrico Berkes (Northwestern University and former Fund staff) presented a preliminary version of the model to the Central Bank and the Ministry of Finance authorities, pointing out potential short-term financing problems. The discussions with country's officials acknowledgedalso centered on the potential of this model to improve policy making by providing a "big picture" view of the economy's macroeconomic dynamics. Moreover, the same visit allowed the team to collect additional information and data on the Ethiopian economy to improve the model's calibration.

Looking Forward

The team is currently working with the Czech consulting firm OGR on extending the model to more systematically incorporate uncertainty and external shocks. The idea is to allow the user to provide ranges, rather than specific estimates, for key parameters such as maximum tax rates and the return to public capital. The user may also provide estimates of the risk of various shocks such as productivity growth and export prices. The model can then help assess how these uncertainties interact and ultimately determine the risk of unsustainable debt. The team is also working on assessing risks of debt sustainability in countries that discover natural resources and are considering borrowing against future export revenues. Finally, the team is working on applying the model to actual cases of investment scaling up.

This, along with other empirical work—e.g., to better quantify absorptive capacity constraints—should enrich the applications of the model.

Applications to LICs
Country Status
Afghanistan Completed
Burkina Faso Completed
Cape Verde Completed
Côte d'Ivoire Completed
Ghana Completed
Liberia Completed
Rwanda Completed
Senegal Completed
Togo Completed
Yemen Completed
Benin Ongoing
Ethiopia Ongoing
Guinea Ongoing
São Tomé and Príncipe Ongoing


  • Felipe Zanna and other Fund Staff, in collaboration with Professor Edward Buffie (Indiana University) are extending the model developed in Buffie et al. to make it applicable to Ethiopia (see Box 3)
  • Professors Christopher Scott Adam and David Bevan seek to explore dynamic medium-term macroeconomic model's implications for growth and debt sustainability in circumstances where public investment generates much higher recurrent costs in the form of operations and maintenance expenditures in addition to direct depreciation costs (where failure to maintain recurrent expenditures may erode the productivity of public capital), and where the government is reliant on a distortionary output tax.
  • Salifou Issoufou, Andrew Jewel (IMF), Mouhamadou Bamba Diop and Kalidou Thiaw (Direction des Previsions et des Etudes Economiques, DPEE) and Professor Edward Buffie (Indiana University) are introducing an electricity sector and more fiscal instruments in the model and applying it to Senegal
  • The paper by Buffie et al. (2012), which develops the model was presented in:
    • The African Economic Conference 2012, held in Kigali (Rwanda) from October 30th to November 2nd
    • The International Growth Centre workshop on fiscal and monetary policy organized by the London School of Economics and held in London (UK) on November 2-3, 2012
  • Andrew Berg and Felipe Zanna, in collaboration with the Fund’s Institute for Capacity and Development (ICD), taught a course of the model developed by Buffie et al. to Fund staff, including country desks.


Working Papers

Article IV Consultations