Resilience and Growth in the Small States of the Pacific

Chapter 11. Estimating Fiscal Multipliers Using a Simplified General Equilibrium Model of Small States, with Application to Kiribati and Palau

Hoe Khor, Roger Kronenberg, and Patrizia Tumbarello
Published Date:
August 2016
  • ShareShare
Show Summary Details
Sergei Dodzin and Xuefei Bai 

The aim of this chapter is to help policymakers and economists better comprehend the link between fiscal policies and growth and to use this understanding operationally to design better macroeconomic policies. The links between fiscal policy and growth are highly significant in small states, where the government sector plays a major role in the economy (Figure 11.1). The importance of the government sector is even greater in the most remote and dispersed Pacific island countries. Moreover, for some Pacific islands (Kiribati, Marshall Islands, Micronesia, Palau, Tuvalu) fiscal policy is a single main economic policy instrument, as they do not have a central bank, and some could face a significant loss of fiscal resources (Marshall Islands, Micronesia, Palau) when grants from their Compacts of Free Association with the United States expire after 2023/24. This makes assessing the impact of fiscal policy on growth critical for helping governments design fiscal policies that take into account output dynamics and ensure long-term fiscal sustainability.

Figure 11.1Government Expenditure in Pacific Island Countries (PICs), 2014

(Percent of GDP)

Sources: Country authorities; and authors’ estimates.

Designing such policies requires a thorough understanding of the major contribution that government taxation, spending, and labor demand make to GDP growth. In particular, the design of fiscal adjustment policies needs to take into account the impact of adjustment on growth and to what extent and how fast private sector demand can offset the decline in government demand. To fully understand the impact of changes in government fiscal policies on growth and the overall economy, economists and policymakers need to understand the two-way linkages between government spending and taxation on one side and GDP growth on the other.

In general, however, these two-way linkages are not modeled explicitly in current practical frameworks. The common approach is to make the growth assumption and estimate the deficit, revenue, and expenditure aggregates without linking these aggregates or their changes explicitly to growth. The absence of these linkages makes it difficult to answer critical questions on the impact of changes in fiscal policies on the real sector, such as what the impact of a proposed fiscal adjustment would be on growth.

This chapter presents an operational and practical framework that incorporates these two-way linkages between the government sector and the overall economy that can be used by economists and policymakers to better answer such questions. The framework is based on a general equilibrium model that connects government spending, taxation, government employment, and wage setting with foreign demand, private sector production and employment, and demand for domestic and foreign goods. It is calibrated and linked to country economic sectoral accounts and other country data, rather than based on estimates from cross-country regressions. The framework—which this chapter applies to two Pacific island countries, Kiribati and Palau—allows specific policy questions to be addressed. These include the growth impact of a change in public wages versus a change in other government expenditures, and the size of private sector growth needed to offset the impact of fiscal consolidation. Such questions are answered through the differential treatment of different components of government spending and their changes, such as a change in the wage bill and changes in other components of public spending. The framework also allows estimation of separate fiscal multipliers for separate components, which are demonstrated in the application section of this chapter.

The model and framework rely on explicit integration and links between the fiscal sector and other sectoral accounts and projections. The most relevant literature is from those approaches that evaluate fiscal policy and fiscal multipliers in the general equilibrium calibrated models. Baxter and King (1993), Gali, Lopez-Salido, and Valles (2007), and Uhlig (2010) analyze fiscal multipliers in the general equilibrium nonstochastic framework. However, these works do not incorporate the impact of differential government spending on imports and domestic demand, and they do not explicitly model employment by government sector. Another strand of the literature is the Global Integrated Monetary and Fiscal model type of global equilibrium model (Kumhof and others 2010) in multicountry settings. This model, however, does not explicitly incorporate government sector employment, and a global setting is not needed given the size of the Pacific economies considered. The other extension is the calibrated dynamic stochastic general equilibrium type models, as observed by Baksa, Benk, and Jakab (2010). Similarly, the models of this type that we are aware of do not explicitly model government employment and are too complex for the types of economies examined in this chapter. Galstyan and Lane (2009) consider the effects of composition of government spending on the real exchange rate in the open economy model with tradable and nontradable goods; however, they do not investigate fiscal multipliers or separate the public and private sectors.

The Framework: Linking the Government Sector and Growth

The framework is based on a model that links government spending, taxation, government employment, and wage setting with foreign demand, private sector production and employment, and demand for domestic and foreign goods. This allows the relationships between economic growth and government spending, taxation, as well as other factors such as external demand, to be estimated. These estimates can then be used to assess the impact of fiscal policy on the economy, including evaluating fiscal multipliers. The framework can also be used to directly evaluate the implications for fiscal sustainability.

Our main approach is to estimate the demand for domestic output from government, domestic residents, and foreigners, and then link it to domestic output and GDP on the production side. This is conceptually divided into government and private GDP. Government GDP mainly consists of wages of public employees, which the government determines. Private GDP is the remaining GDP corresponding to private output, which is modeled as a Cobb-Douglas production function (Figure 11.2).

Figure 11.2GDP on the Production Side

Source: Authors’ representation.

This underlying model is the general equilibrium model in which markets for goods, labor, and financial assets clear (detailed specifications are provided in Annex 11.1). At the same time some elements, such as household saving decisions, are modeled on the reduced form specified by economists, rather than within the fully optimizing framework used, for example, by the dynamic stochastic general equilibrium model. The specifications can vary depending on the circumstances of a particular country, but we believe this approach increases the framework’s applicability and flexibility.

Demand for private output critically depends on consumption; saving and investment decisions by government, domestic residents, and foreigners; and on the share of spending on domestic versus imported goods. Private sector wages and employment are determined by demand for the domestic product. The private wage wpt is set with some discount compared to the government wage, consistent with the observed practices in the countries considered in the example (Kiribati and Palau). Private employment is determined by the optimization of production given demand for private sector output.1

Government consumption, saving, and investment are determined by fiscal policy and the government budget. Government is also responsible for decisions on public sector hiring. Public and private employees pay taxes and allocate income between consumption and saving. Saving decisions are significantly affected by pension rules that are prevalent in a particular country. Taking into account the economy of the countries considered and to enhance tractability, nonpension-related saving is determined in the reduced form specified by economists and teams applying the framework rather than in the full optimization setting.

Businesses invest in the local economy after paying taxes to the government, wages to domestic employees, and interest obligations and dividends to domestic and foreign investors.

Foreign consumption and investment form a portion of spending on private output (Figure 11.3). Foreign investment in a small country normally occurs through foreign direct investment and bank loans. Foreign investors decide on the share of dividends that is reinvested as well as on new investments. The share of imports in consumption and investment spending is determined by individual preferences and an economy’s production capacity. For example, in an economy with abundant agricultural resources and significant agricultural production, the share of spending on domestic agriculture will be higher.

Figure 11.3Private Output and GDP

Source: Authors’ representation.

The parameters of the model are calibrated and depend on a country’s particular circumstances as well as data availability. For example, in some countries, where input-output statistics are available, the share of spending on domestic versus imported products can be found directly from these data. When such data are not available, shares can be estimated, to the extent possible, using income and expenditure surveys, weights of imported and domestic goods in the consumer price index, and other sources. If data are unavailable, the shares will be based on those of other economies that are similar in structure. For example, to the extent data are available, future extensions of the model beyond current limitations could include modeling of the real sector in multi-output settings and the extension of modeling for monetary-fiscal interactions.

After the model is calibrated, it can be applied to estimate the government’s impact on economic dynamics, such as the calculation of fiscal multipliers. This estimation can be made by comparing changes in simulated economic outcomes to changes in government policy, such as spending or taxation.

Applications of the Framework: Kiribati and Palau

We now apply the framework to Kiribati and Palau, and calibrate the model and conduct simulations in each case (Annex Table 11.1 shows the parameters used in our simulation). These countries are representative of small states: they are constrained by small size, limited economic scale, and geographic remoteness. According to 2013 data, Palau’s population was 20,918 and Kiribati’s 102,531. Because of these constraints, they suffer narrow production bases, limited opportunities for diversification, and high transportation costs.

At the same time, the two differ in income, economic structure, and level of development. Palau is a high-income country (GDP per capita was US$14,086 in 2013), with the government having a relatively smaller share in the economy than in Kiribati. In 2012, government spending in Palau was about 30.9 percent of GDP, with the central and local government accounting for about 20 percent of the total. In low-income Kiribati (GDP per capita was US$1,583 in 2013), the government plays a more significant role. In 2012 government spending accounted for 56.4 percent of GDP and the government sector for 28.1 percent of total GDP (Figure 11.4). In Palau, tourism is the most important source of external earnings and foreign exchange, while in Kiribati fishing license fees and remittances are the major sources of external income. The two governments also differ in how they finance fiscal deficits: Palau mainly relies on external debt for fiscal financing, while Kiribati draws down the Revenue Equalization Reserve Fund, its sovereign wealth fund.2

Figure 11.4Kiribati and Palau: Government Spending and Share of Government Spending in GDP

(Percent of GDP)

Sources: Kiribati and Palau authorities; and authors’ estimates.

The conclusions and suggestions from our case studies could shed light on a variety of small state cases, including other Pacific island countries as well as nonregional, low-income states and small states with large tourism sectors in the Caribbean and Africa.

The calibration of the framework covers the main sectors of economies and key channels of interaction between them (Figure 11.5), taking into account all relevant components of investment and consumption demand, public and private outputs, GDP, and production inputs. Using parameters and actual public wage and expenditure data since 2006, we can conduct a simulation of the model for both Kiribati and Palau.

Figure 11.5Interactions in the Simulation System

Source: Authors’ representation.

Because of data limitations, we had to estimate and calibrate some of the key parameters based on the data available. For example, Kiribati and Palau do not compile input-output tables, so the shares of domestic goods and services in government spending, public and workers’ consumption, and total investment were estimated based on historical fiscal statistics, household spending statistics, national account statistics, and balance of payments data. Since Kiribati’s agricultural sector is significantly larger than Palau’s, its shares of domestic goods and services in consumption and investment are generally higher. For example, the share of domestic goods and services in government workers’ expenditure in Kiribati is 50–60 percent, while in Palau it ranges from 20–30 percent.3 For public workers, the saving rate is mainly determined by the provident contribution; for private workers, it is estimated based on financial statistics due to the data limitation.

Another key aspect of the calibration is to estimate the share of wages, profits, and intermediate goods in private output.4 Detailed data on intermediate goods, which allow the direct calculation of shares, are only available for Palau. For Kiribati we assumed that the share of intermediate goods in each detailed industry is similar to that of Palau. We confirmed that the estimates are reasonable by fitting them to historical data.

Once historical parameters have been calibrated, parameters for future years can be projected based on the historical parameters and available forward-looking information. We assume the government sets the public wage and other expenditure levels, while external demand is also exogenous. The model then endogenously determines private output and workers compensation. Future public expenditure and the external demand path under the baseline scenario are in line with IMF fiscal and balance of payments projections, and GDP growth is simulated based on the calibrated model. In our baseline scenario, real GDP growth in the medium term is about 2 percent for both Palau and Kiribati.

Calculating the Fiscal Multipliers

To examine the interaction between fiscal adjustment and GDP growth, we simulated different scenarios, including public wage spending shock (both nominal wage increases and real or additional public hiring);5 public expenditure shock scenarios (using expenditure other than wages, including current expenditure and development expenditure); and compound public spending shock scenarios. We then calculated the corresponding fiscal multipliers under each scenario. Fiscal multipliers can be measured in several ways. They are defined here as the ratio of a change in GDP output (ΔY) to a discretionary change in government spending (ΔG) (see IMF 2013a for more details). Here GDP is in real terms, so the multiplier means the effect of a US$1 increase in spending on the real GDP level. Depending on the time frame, two types of fiscal multipliers are considered:

It is worth noting that different kinds of public wage expenditure shocks have different impacts on real GDP. If the government increases public wages without changing public employment levels, the impact on real output is indirect, mainly transmitting through consumption and investment channels. In contrast, under the real public wage shock, the wage bill increases because the government sector employed more workers. In addition, real public GDP will increase. The magnitude of the increase will depend on the degree of crowding out of private sector employment. If the crowding-out effect is low (in cases in which unemployment is high), the corresponding fiscal multiplier would be close to or larger than 1. In the case of high crowding out, the multiplier will be significantly less than 1. We examined multipliers in both cases.

The simulation results show that different spending shocks have different multipliers (Figure 11.6). In line with the above discussion, values of impact multipliers under real wage shocks are close to 1 (assuming a low crowding-out effect). In contrast, as the shares of imported goods and services in household consumption and government expenditure are high in both countries, the stimulus effects of fiscal policy on domestic GDP are limited, and fiscal multipliers under these shocks are relatively small compared with bigger states and economies with more developed and larger domestic industries (IMF 2013a).

Figure 11.6Simulation Results: Impact Multipliers and Cumulative Multipliers

Sources: Kiribati and Palau authorities; and authors’ estimates.

Under a positive real public wage shock, the impact multiplier is 1.25 for Kiribati and 0.98 for Palau, while the multiplier under a positive nominal public wage shock is 0.44 and 0.14, respectively. Under a public expenditure (other than wage) shock, impact multipliers are also relatively small (0.47 for Kiribati and 0.16 for Palau), which means that the effect of expansionary fiscal policy stimulus on GDP growth is limited in these two countries.

Moreover, the stimulus effect of expansionary public spending (both wage and other expenditure) on private GDP in Kiribati is higher than in Palau. As already noted, the rationale behind this phenomenon is that in Kiribati, more domestic goods and services are used by households and the government for consumption and investment.

In our results, the wage shocks have higher fiscal multipliers than the nominal wage shocks and the public expenditure shocks, under the assumption of low crowding out. However, hiring more public workers may not be an effective way to increase growth. First, there may not be enough skilled workers in these countries and that can lead to a crowding-out effect that would negatively impact private GDP growth. Simply hiring more public workers may lead to low efficiency and a waste of resources if the scope of hiring exceeds capacity. Furthermore, the spending increases may not sustainable, a point that applies to various types of expansionary public expenditure as the fiscal multipliers under public expenditure shocks are smaller.6

The fiscal sustainability of government policy is an essential component of analysis. As noted earlier, to finance its public deficit, Palau mainly relies on external debt, while Kiribati draws down its sovereign wealth fund. The nominal fund per capita dynamic in Kiribati and the public debt dynamic in Palau under expansionary fiscal policy shocks are shown in Figures 11.7 and 11.8, respectively. These indicate that, if other conditions are kept unchanged, expansionary fiscal policy can worsen fiscal sustainability. So fiscal sustainability should be taken into account when designing fiscal policies.7

Figure 11.7Kiribati: Sovereign Wealth Fund Dynamic under Positive Spending Shocks

Sources: Kiribati authorities; and authors’ estimates and projections.

Figure 11.8Palau: Public-Debt-to-GDP Ratio under Positive Public Spending Shocks

Sources: Palau authorities; and authors’ estimates and projections.

How to Offset the Negative Growth Impact of Fiscal Consolidation

The framework allows us to analyze various related aspects of economic dynamics in addition to building up the two-way link between the fiscal and real sectors and investigating the fiscal impact. For example, we look at the question of the extent to which an economy should improve productivity and external demand in order to offset the possible negative effects of fiscal consolidation. Because productivity is difficult to measure for small states, the analysis in the following section mainly focuses on external demand. It calculates the external demand gap that is needed to offset the negative impact of fiscal contraction on private output based on our model.

The results suggest that the necessary increase in external demand may be significant (Figure 11.9). In Kiribati, to offset the negative impact of a US$1 reduction in the deficit, external demand for domestic goods and services needs to increase by about US$0.60 if the fiscal contraction is from declining nominal public wages and by US$0.55 if the consolidation is achieved by cutting public expenditure. In Palau, external demand needs to rise by US$0.56 and US$0.32, respectively, to maintain the same scale of consolidation.

Figure 11.9Increases in External Demand to Offset the Impact of Fiscal Contraction

Source: Authors’ calculations.

These results are in line with the estimated fiscal multipliers. First, for Kiribati, in which the government sector plays a larger role in the economy and fiscal multipliers are higher than those in Palau, a larger increase in external demand is needed to maintain private GDP growth when facing fiscal consolidation. Second, as the shares of domestic goods are higher in the consumption basket of households, more external demand is required when public spending consolidation is undertaken by cutting public wages.

Annex 11.1. Model of Fiscal Impact on GDP and the Overall Output Economy: Analytical Details

Private Sector Output

Private sector output is produced with three factors: capital, labor, and intermediate goods (intermediate goods are assumed to be partly domestically produced and partially imported).

The production function in the private sector is a Cobb-Douglas production function:

It can be shown from the standard cost minimization that the minimum cost of the given amount of output is then:

where r denotes return to capital, ωp private wages, and pI price of intermediate goods. Under the assumption that the price is equal to minimum costs, the price deflator for the output can be approximated as:


The government levies tax τL on labor (private and public), τK on capital, and hires labor LG at the wage ωG and also gets and spends on nonwage expenditure E. It also can finance the deficit with the change of net debt, as ΔFt = ΔDt - ΔAt, where D is gross debt and A is assets. The budget constraint of the government therefore is:

Public Workers

Wage ωgt and the number Lpt of public workers are set by the government. Public workers receive wages from the government and private sector, respectively. The wage of government worker ωgt is set by the government. In addition to wage income, government workers receive other noninterest income oiLgt and also receive income on investments in the form of interest and dividends.8

Denote CLgt,SLgt as total consumption and saving of government workers, and srLgt, srLpt corresponding saving rates. Then, total consumption and total saving by domestic workers are:

where rftNALg_ft, rdtNALg_dt is interest received on (net) assets invested abroad and domestically.

Private Sector Workers

Private sector wages and employment are determined by the demand for the domestic product. In the model it is assumed that the private wage ωpt is set with some discount compared to the government wage. The private employment is determined by the production optimization given demand for private sector output. As with government workers, private workers receive other noninterest income and interest and dividends on their investments. Total consumption and spending of private workers can be written as:

Domestic Private Firms and Entrepreneurs

Domestic firms and entrepreneurs receive their share of output (after paying for intermediate goods) and then pay for:


Foreigners’ demand for domestic output consists of consumption CFt of domestic goods and services and foreign savings IFt invested in the country. Foreigners receive dividends and interest rt on their investment NAft and decide on the rate of reinvestment, rinvt. In addition, they decide on new investments, nIFt. Therefore, total foreign investment demand IFt is determined as IFt = rinvt*rtNAFt + nIFt.

Intermediate Demand for Domestic Products

Intermediate demand for domestic products depends on the distribution between imports. If the share of domestic goods in intermediate goods is δd, then from a Cobb-Douglas production function the share of spending on intermediate goods is δ, and equating final and intermediate demand to the total demand to output, PY = DFin + δPY. Therefore:

Total Demand for Domestic Output

The demand for domestic output consists of (1) the share of government nonwage expenditure on domestic output, (2) the share of total consumption spent by domestic public and private workers on domestic output, (3) the share of investment goods in the total investment demanded by domestic residents, (4) foreign final demand for domestic output, and (5) intermediate demand for domestic output.

Where CLgt, CLpt denote total consumption by the government and private workers, respectively, SLgt, SLpt represent corresponding saving.

Let ϕE be the share of other government expenditure falling on consumption of domestic product, ϕLp the share of government workers’ expenditure falling on consumption of domestic product, ϕLP the share of private workers’ expenditure falling on consumption of domestic product, and ψ/ the share of investment expenditure falling on domestic product.

Annex Table 11.1Selected Parameters Used in Simulation
Share of domestic goods in government expenditure (%)40–5020–30
Share of domestic goods in government workers’ consumption (%)50–6020–30
Share of domestic goods in private sector workers’ consumption (%)55–6530–40
Share of domestic goods in total domestic investment (%)35–1520–30
Government workers’ saving rate (%)15–2015–20
Private sector workers’ saving rate (%)3–1010–15
Government workers’ average wage/private workers’ average wage2.11.7
Sources: Kiribati and Palau authorities; and authors’ estimates.
Sources: Kiribati and Palau authorities; and authors’ estimates.

Let ψLg, ψLp, and ψLk denote shares of savings invested domestically by government workers, private workers, and private firms and entrepreneurs, respectively.

The total demand for domestic output therefore must be equal to output produced:

Using the discussion of multipliers above, this equation can be rewritten as:

Taking into account that the total amount of savings invested in the domestic sector is equal to total investment.

Substituting into the first equation above, we obtain:

Substituting consumption and saving calculations for public workers, private workers, and foreigners, and noting that ωLptLLpt = βptYpt, rkKt = αptYt and simplifying, we get the expression for ptYt:

where coefficients θLg θLp, ϑk are derived from the parameters of the model and correspond to the shares of domestic spending on consumption and investment by government workers, private workers, and private firms and entrepreneurs, respectively.


    BaksaD.S.Benk and M. Z.Jakab. 2010. “Does ‘The’ Fiscal Multiplier Exist? Fiscal and Monetary Reactions, Credibility and Fiscal Multipliers in Hungary.” Office of the Fiscal CouncilBangko Sentral ng PilipinasManila.

    • Search Google Scholar
    • Export Citation

    BaxterMarianne and Robert G.King. 1993. “Fiscal Policy in General Equilibrium.” American Economic Review8 (3): 31534.

    GaliJordiJ. DavidLopez-Salido and JavierValles. 2007. “Understanding the Effects of Government Spending on Consumption.” Journal of the European Economic Association5 (1): 22770.

    • Search Google Scholar
    • Export Citation

    GalstyanVahagn and PhilipLane. 2009. “The Composition of Government Spending and the Real Exchange Rate.” Journal of Money Credit and Banking41 (6): 123349.

    • Search Google Scholar
    • Export Citation

    International Monetary Fund (IMF). 2013a. “Guidance Note on Fiscal Multipliers.” Unpublished Washington.

    International Monetary Fund (IMF). 2013b. “Macroeconomic Issues in the Small States and Implications for Fund Engagement.” Washington.

    • Search Google Scholar
    • Export Citation

    KumhofM.D.LaxtonD.Muir and S.Mursula. 2010. “The Global Integrated Monetary Fiscal Model (GIMF)– Theoretical Structure.” IMF Working Paper 10/34International Monetary FundWashington.

    • Search Google Scholar
    • Export Citation

    UhligHarald. 2010. “Some Fiscal Calculus.” American Economic Review100 (2): 3034.


The wage setting function considered in this chapter is consistent with countries’ circumstances where the private unemployment rate is high enough to incorporate an increase in demand for domestic products. The framework flexibility allows the modeler to incorporate different employment functions.


There is no formal target for the Revenue Equalization Reserve Fund balance. An informal target for the fund authorities used in the past was to maintain the real per capita value of the fund at or above the level in 1996; that is, $A4,700.


The different economic structure may explain the difference between these two ranges. For example, the share of the agricultural sector in Palau’s total GDP was 2.1 percent (in 2012) and 17.5 percent in Kiribati’s (2009).


These shares are parameters α, β, and γ in our augmented Cobb-Douglas production function.


For nominal wage shocks, public wages increase only because the government decides to raise the average wage per employee. For real wage shocks, the payroll bills grow purely as the government decides to hire more people. Under both types of shock, we assume the public wage shocks are permanent.


However, the impact of government expenditure on productivity is not fully taken into account here. If public expenditure increases productivity, namely, “A” in our production function, the actual fiscal multiplier would be higher.


The model assumes yields on the Revenue Equalization Reserve Fund and public debt consistent with the observed actual rates.


The other income includes remittances, which are an important source of income.

    Other Resources Citing This Publication