The Sources of Business Cycles in a Low Income Country*
  • 1 0000000404811396https://isni.org/isni/0000000404811396International Monetary Fund
  • | 2 0000000404811396https://isni.org/isni/0000000404811396International Monetary Fund

Contributor Notes

We examine the role of global and domestic shocks in driving macroeconomic fluctuations for Ghana. We are able to study the impact of exogenous shocks including productivity, credit supply, and commodity price shocks. We identify the shocks with a combination of sign and recursive restrictions within Bayesian VAR models. As a benchmark we provide results for South Africa to document the difference between two economies with similar structures but different levels of development. We find that global shocks play a more dominant role in South Africa than in Ghana. These shocks operate through three channels: trade, credit and commodity prices.

Abstract

We examine the role of global and domestic shocks in driving macroeconomic fluctuations for Ghana. We are able to study the impact of exogenous shocks including productivity, credit supply, and commodity price shocks. We identify the shocks with a combination of sign and recursive restrictions within Bayesian VAR models. As a benchmark we provide results for South Africa to document the difference between two economies with similar structures but different levels of development. We find that global shocks play a more dominant role in South Africa than in Ghana. These shocks operate through three channels: trade, credit and commodity prices.

1 Introduction

Developing policies for stabilizing macroeconomic fluctuations has been the subject of many papers in both advanced and emerging markets. A prerequisite for building the structural models to develop these policies is knowledge of the main sources of fluctuations in these economies. In this paper we take the first step towards developing stabilization policies in a low-income country by documenting the driving forces of fluctuations in this type of economy.

Empirical work for low-income countries has unique challenges. First, data is typically of poor quality and the time series length is short relative to those available for more advanced economies. Second, the economic structure of these economies is typically quite different from more advanced economies. In this paper we address the data issue by constructing our own time series for GDP. This allows us to create longer time series than is available historically. We take a first pass at issue two by studying a more advanced economy with a similar production structure.

In our case we study Ghana. Ghana is a good test case because it is the only Low-income Country (LIC) that is currently operating with an explicit inflation targeting framework in Sub-Saharan Africa. Moreover, the Bank of Ghana has been granted full independence since the bill was passed in December 2001 and the Monetary Policy Committee was established in September 2002.

Our objective is to develop a set of stylized empirical facts about Ghana in terms of the shocks that drive fluctuations. Our purpose is to investigate the type of shocks and model features that one would want to consider in building a structural dynamic stochastic general equilibrium (DSGE) model. The empirical facts here can then be used in such a model building exercise. This objective guides us in the type of shocks we consider. We are interested in the extent to which Ghana is exposed to international versus domestic shocks. This will motivate us to consider global counterparts to all of our shocks. We study a sequence of shocks that may be of importance to a country such as Ghana. First, we consider productivity shocks to measure the extent to which business cycles are driven by real factors. Second, we study credit market shocks to understand the importance of financial sector disruptions. Third, we consider commodity price shocks to measure the importance of such price changes of a dominant primary goods sector. We do not consider monetary policy shocks due to changes in monetary policy over the sample period. We leave this latter issue to a more structural investigation.

To identify shocks we use a standard macroeconomic tool–the Vector Autoregression (VAR), which we estimate with Bayesian methods. Identification of shocks is through a set of sign restrictions in the spirit of Uhlig (2005) for both the credit and productivity shocks. These use minimal but robust implications of structural models to impose restrictions on impulse response functions. For commodity price shocks we exploit the exogeneity of commodity prices for a small economy such as Ghana and use a recursive setup, with commodity prices ordered first. Global shocks are estimated using a similar set of restrictions on data constructed using the first principal component of G7 country level data.

In addition to studying the impact of shocks in Ghana we also study the impact of the same set of shocks in South Africa. The important question here is whether or not Ghana, which has a similar production structure and a monetary policy framework as South Africa, is fundamentally different. That is, in building a structural model for Ghana how much of our experience with models from more advanced and stable economies can we import?

This paper is related to the literature on the sources of macro-economic fluctuations in developing countries. One strand of this literature employs univariate methods to estimate cyclical variations in macroeconomic series (e.g. Agenor et al. (2000), Cashin (2004), du Plessis (2006) and Male (2011)). For instance, Agenor et al. (2000) employ de-trended methods to a set of macro-aggregates for 12 developing countries and find procyclical real wages, suggesting that productivity shocks play a dominant role in macroeconomic fluctuations of these economies. They also find countercyclical government expenditure, which implies that the government plays a significantly role in these economies. Finally, their analysis shows that the business cycle of advanced countries has a significantly positive impact on economic activity of developing countries. However, Male (2011) recently challenged this last finding. She applies the classical cycle analysis, of Harding and Pagan (2002), to 32 developing countries and finds no clear patterns in the co-movement between business cycles of developing and advanced countries.

Another strand of the literature employs structural VAR (SVAR) or theoretical models (e.g. see for instance, Broda (2004) Deaton and Miller (1996) Hoffmaister and Roldós (1997); Hoffmaister et al. (1998), Houssa (2008, 2009), Houssa et al. (2010), Houssa et al. (2013), Kose (2002), Kose and Riezman (2001), Mendoza (1995), Chia and Alba (2006)) and highlights the importance of two types of shocks: productivity shocks and terms of trade shocks.

This paper contributes to this literature by documenting the roles of credit supply and productivity shocks in developing countries, comparing domestic versus foreign nature of these shocks and their role in economies with a similar structure but at different stage of development. Especially, the study of macroeconomic impacts of credit supply shocks fills an important gap in the literature of Emerging Markets and LICs. To the best of our knowledge only Tamasi and Vilagi (2011) explicitly identify credit supply shocks for emerging markets. They employ a Bayesian VAR (BVAR) model with sign restrictions and report that credit supply shocks account for a larger share of output fluctuations in Hungary. Other related studies have mainly used regression techniques to document a positive co-movement between credit and real activity (e.g. Akinboade and Makina (2010)). However, in such a framework is it difficult to distinguish correlation from causality. Moreover, such techniques do not systematically analyze the impacts of non-expected shocks. For research studying international transmission of credit shocks to emerging markets; see Schnabl (2012) and Cetorelli and Goldberg (2010)).

The remainder of this paper is organized as follows. Section 2 presents a historical background on macroeconomic conditions in Ghana. Section 3 introduces the VAR model and discusses the structural identification strategy. Section 4 presents empirical results. Section 5 concludes.

2 Background on Macroeconomic Conditions in Ghana

Figure 1 and 2 present selected macroeconomic indicators for Ghana in 1980 : 1 – 2012 : 4. Quarterly data on real activity is only available from 2006 for Ghana. As such, we estimate a measure of real activity growth for Ghana that covers a much longer time period. For this purpose we pool data from different frequencies (yearly, quarterly and monthly) and experienced with two approaches. We use the common component and the now-cast measures of real output (Giannone et al. (2008)). A plot of the series based on the former is on the left top panel of Figure 1.

Figure 1:
Figure 1:

Selected Macroeconomic indicators for Ghana (YoY percent)

Citation: IMF Working Papers 2015, 040; 10.5089/9781498310147.001.A001

Figure 2:
Figure 2:

Exchange Rates and International Reserves (YoY percent)

Citation: IMF Working Papers 2015, 040; 10.5089/9781498310147.001.A001

Prior to 1983 Ghana suffered from macroeconomic instability which was most characterized by large internal and external deficits, limited foreign exchange reserves, high inflation, overvalued exchange rate, and negative economic growth (see Figures 1 and 2). In 1983 a Structural Adjustment Program was initiated under the denomination of Economic Recovery Program (ERP). ERP covers the period from April 1983 till March 1992. Its main components include a reduction of government deficit, tight (credit and) money growth, and a number of liberalization programs (in trade, prices, exchange rates). This break in policy leads to the start date of our sample.1

Monetary policy from 1983 switched from a direct control of credit and interest rates to a monetary targeting regime. The liberalization of interest rates charged by banks takes a gradual route. Minimum and maximum deposit (except saving) rates were removed in 1987; minimum lending rates for commercial banks were abolished in 1988 and commercial banks were given the right to fix their own rates in 1989 (Sowa and Acquaye (1999)).

In the framework of the new monetary policy three targets were set: i) the reserve money as the operating target; ii) the money supply as the intermediate target; iii) and the general price level as the ultimate target (Abradu-Otoo et al. (2003)). The implementation of monetary policies with these targets was undertaken within the Financial Programming Framework. Explicitly the implementation of monetary policy involved four phases. In the first step the money supply target is determined, with the quantitative money equation, given the target for the price level and the level of activity. In the second step, the money base target is determined given the money multiplier. Third, the level of net domestic asset (NDA) is determined using the balance sheet identity of BOG given targets of reserve money and the net foreign asset (NFA). Fourth, an open market operation is used to achieved the level of NDA. These procedures were in place till 2001 where the Bank of Ghana (BOG) started the transition to inflation targeting regime, which was formally adopted in 2007. These changes in policy throughout the sample motivate us to not consider VAR based monetary policy shocks. We conjecture that a structural approach will be more appealing for these shocks.

Prior to ERP Ghana operated a fix exchange rate regime but adjustments of the cedi were frequent (1967, 1971, 1972, 1976). In 1983 a liberalization of the exchange rate begun with a gradual approach in six phases where full liberalization was achieved in 1990 when for the first time the exchange rate was set in the inter-bank market (Sowa and Acquaye (1999)). Since then the currency is in principle allowed to fluctuate. As most small open economies, however, authorities intervened in the foreign exchange market. With flexible exchange rate global shocks are likely to impact Ghana, in part motivating our consideration of augmenting our VAR with global factors.

Ghana exports three main commodities: cocoa and gold are traditional commodities and from 2011 Ghana started exporting crude oil. Our commodity price shocks will focus on the former two as they are produced for the entire sample.

3 Methodology

3.1 Bayesian Vector Auto-Regressive (BVAR) Model

Consider the following Vector Auto-Regressive (VAR) model,

Yt=A0+A1Yt1++APYtP+μt,(1)

where Yt is a n × 1 vector of home and foreign real, nominal and financial indicators; the Ai are n × n auto-regressive coefficients, A0 contains the constant terms, and μt is a n × 1 vector of Gaussian white noise with covariance matrix Ψ=E(μtμt). The G7-countries represent our foreign economy whereas for the home economy we focus on either Ghana or South Africa.

The series included in Yt depend on the type of structural shocks being estimated. We identify two categories of shocks: credit and productivity shocks, on the one hand; commodity price shocks on the other hand. For credit and productivity we consider global as well as domestic versions of the shocks.

3.2 Credit and Productivity Shocks

The productivity shocks we consider are interpreted as exogenous shocks to total factor productivity. In the context of a DSGE model this would be the standard Solow residual. The credit shocks we study are exogenous shocks to the supply of credit. To study credit and productivity shocks we estimate a VAR model with 16 variables. In particular, Yt includes six indicators of the foreign economy: G7-real GDP; G7-inflation; G7-real credit; G7-short-term interest rates; US-credit spread; and US-default rates. The four G7-factors are estimated by extracting the first principal component from the series of G7 countries. For the remaining two indicators we use US series because of data limitation.

The remaining 10 series of Yt relate to the home economy. We include six indicators of the domestic economy analogous to those of G7-countries. Moreover, the home- block contains four indicators allowing to capture the transmission of foreign shocks to the domestic economy: exports, imports, primary commodity prices, and the real exchange rate. The VAR is setup so that global shocks impact Ghana (or South Africa), but domestic shocks have no impact on global aggregates.

Table 1:

Data

article image
article image

We use quarterly data from Ghana, South Africa and G7-countries in 1985 : 1 − 2010 : 3.2 Where appropriate we transform the series in year to year growth rates. Table 1 reports detailed information on the dataset and the transformation applied to each series. For the South African economy we measure the credit spread by the difference between the yield on Eskom and the US baa bond. As a proxy for the default rate we make use of data on the number of insolvencies on loans.

We use the US corporate credit spreads (baa-aaa) as the measure of the price of risky credit for G7-countries. To measure default on credit for G7-countries we also use a proxy for the US economy. In particular, we use the distance to default measure proposed by Gilchrist et al. (2009). We take the inverse of this indicator and transform it to a year to year growth rate.

We estimate Eq. (1) using Bayesian methods with 3 lags. In all BVAR models we employ a combination of two types of priors: i) a Normal-inverted Wishart prior; and ii) a Minnesota type prior that assigns low weights on off-diagonal AR coefficients and specifically zero weights on coefficients related to the home economy’s indicators in the block defined by (primary) commodity prices and G7-factors.3 The later restrictions are consistent with the Small Open Economy (SOE) assumption for Ghana and South Africa.

For the structural identification, we employ a set of zero and sign restrictions. These restrictions are implemented with the penalty function approach proposed by Mountford and Uhlig (2009) and Uhlig (2005). We explicitly modify the objective function in order to also allow for zero restrictions. The results reported in the paper are based on the following sequential ordering: G7-credit shocks, G7-productivity shocks, domestic (SA or GHA)-credit shocks and domestic (SA or GHA)-productivity shocks. However, using a different ordering does not change the main results of the paper. Table 2 reports the identification restrictions for the four shocks. In all cases, the restrictions assume negative shocks and are imposed over the first four quarters.

Table 2:

Indentification Restrictions

article image
Notes: The sign restrictions were imposed on the first quarters.

Zero restrictions allow to disentangle South African shocks from global shocks. In particular, we assume that G7 countries do not respond to shocks originating from South Africa. On the other hand, sign restrictions help to distinguish exogenous credit supply shocks from the endogenous response of credit to macroeconomic conditions. The identification of credit supply shocks is based on the recent literature (e.g. Helbling et al. (2011) and Meeks (2012)). We assume that an adverse credit supply shock is characterized by an increase in the credit spread and a decrease in real credit. In addition, we require that default rates on corporate bonds do not increase. This additional restriction helps to isolate the endogenous response of credit to fundamental macroeconomic shocks (see Meeks (2012)). Note that we leave unrestricted the IRFs of other series in Yt including Real GDP, inflation, the monetary policy rate etc …

For adverse productivity, we impose that real credit does not increase. Moreover, we require for these shocks that default rates do not decrease. Again, this later restriction is employed to discriminate credit supply shocks from endogenous responses of credit to fundamental (productivity and demand) macroeconomic shocks. We identify negative productivity shocks as decreases in output that increase inflation, as implied by a New Keynesian DSGE model. For advanced economies one can also include productivity itself in the VAR. Helbling et al. (2011) do so and find that the additional data does not affect the estimated responses. As we do not have productivity data for Ghana we do not include any measure of productivity in the VAR.

3.3 Commodity Price shocks

To identify commodity price shocks we consider a VAR model with six variables: commodity price; CPI, GDP, Credit; short-term interest rate and the real effective exchange rate. We follow the standard approach of the literature and impose the recursive identification scheme on these shocks, with commodity prices ordered first (e.g. Broda (2004) and Mendoza (1995)).

Analysis focuses on gold price allowing to understand how Ghana and South Africa respond to similar commodity price shocks. We deflate the nominal commodity price series by the US producer price index (PPI). Moreover, we estimate the economic impact of an increase in commodity prices over three sample periods: i) the full sample (1985 : 1 – 2010 : 3); i) before the introduction of the inflation targeting framework (IT); and iii) after the introduction of the inflation targeting framework. For South Africa, the before and after IT periods are: 1985 : 1 – 1999 : 4 and 2001 – 2010 : 3. For Ghana these are 1985 : 1 – 2001 : 4 and 2002 – 2010 : 3.4 In each case we fix the lag length to 3.

4 Empirical Results

4.1 Global Shocks and Fluctuations in Ghana and South Africa

In this section we study the impact of global productivity and credit shocks on Ghana and South Africa. Figures 3 and 4 report the medians together with the 16th and 84 percentile responses to G7 shocks. A full analysis of G7 responses to these shocks can be found in Helbling et al. (2011). The results can be briefly summarized as credit supply shocks have negative but relatively modest impacts on global output, while productivity shocks have sharp and persistent impacts on global output.

Figure 3:
Figure 3:
Figure 3:

G7 Credit Shocks

Citation: IMF Working Papers 2015, 040; 10.5089/9781498310147.001.A001

Figure 4:
Figure 4:
Figure 4:

G7 Productivity Shocks

Citation: IMF Working Papers 2015, 040; 10.5089/9781498310147.001.A001

The IRFs to the global credit supply shocks in Figure 3 show that there is a significant impact on output in South Africa, but not in Ghana. Looking at the responses to other variables we see that the credit shock has a modest reduction in credit in both countries. This is suggestive that while South Africa is financially integrated such that global credit conditions affect the domestic economy, Ghana is not. In other words, global credit supply is not important for financing the production of Ghanian GDP. The difference in impact of the shock can also be seen on commodity prices. The credit supply shock lowers commodity prices in South Africa, while raising them in Ghana. Understanding the differences in the financing of the production of the two types of commodities will be important for understanding the role of credit markets in these primary goods producing countries.

In Figure 4 we plot the responses to G7 productivity shocks. For both Ghana and South Africa we see similar shapes and magnitudes for the response. Given that Crucini et al. (2011) find that the global business cycle is primarily driven by productivity shocks this result is not surprising. From a modelling perspective it suggests that standard formulations for global productivity shocks are appropriate for Ghana. The response of inflation is also in line with the prediction of a New Keynesian model, where drops in productivity are inflationary. Note that while global inflation is constrained to rise after the contractionary shock, there are no restrictions on Ghanian inflation. In terms of commodity prices we see that prices fall in both South Africa and Ghana, though the impact is not significant in Ghana. Again, highlighting that Ghana is not as integrated with the world as is South Africa.

Finally, we see that the world gold price increases following adverse productivity and credit shocks, though the impact is modest.

4.2 Domestic Shocks and Fluctuations in Ghana and South Africa

Domestic credit supply shocks are shown in Figure 5. They generate similar inflationary pressures in South Africa and Ghana as did the global shocks. They have a quantitatively larger impact in Ghana. The inflation pressures of credit supply shocks are in line with the models presented in Atta-Mensah and Dib (2008) and Gerali et al. (2010). The effects of the shock leads to an increase in interest rates in Ghana and South Africa, though the impact is only significant for a short period of time. This is suggestive that inflation is more important for both countries than the small output losses caused by financial market disruptions. The similarities in responses suggest that monetary policy rules in South Africa can be exploited in building a model for Ghana.

Figure 5:
Figure 5:
Figure 5:

Domestic Credit Shocks

Citation: IMF Working Papers 2015, 040; 10.5089/9781498310147.001.A001

Figure 6:
Figure 6:
Figure 6:

Domestic Productivity Shocks

Citation: IMF Working Papers 2015, 040; 10.5089/9781498310147.001.A001

Domestic productivity shocks, shown in Figure 6, have a significant impact on Ghana, and a more muted impact on South Africa. By construction these shocks lower output and raise inflation in both countries. The interesting result is the difference on the other variables in the VAR. In response to the shock short term interest rates rise in Ghana and fall in South Africa. These responses cause a real exchange rate appreciation in Ghana, lowering exports and raising imports. In South Africa real exchange rates do not move, while exports fall and imports rise. This points to monetary policy focusing on inflation pressures more than output drops in Ghana then in South Africa. A interesting question is if a policy rule in the context of a DSGE model can justify such a strong response. If not, a lack of credibility for the central bank may be leading to aggressive responses of monetary policy to inflation in order to build credibility.

4.3 Commodity Price Shocks

Figure 7 displays the dynamic responses of various indicators to gold price shocks in both countries. Considering first the full sample results shows that output decreases on impact in South Africa but latter increases significantly and turns to positive values during several quarters. On the contrary, output increases significantly on impact following an increase in gold prices. A similar result can be observed before IT but during the IT period the impact response is negative in Ghana as well. One explanation of the different impact period responses could capture the earlier result that G7 adverse shocks cause an increase in gold price but also a much stronger recession in South Africa.

Figure 7:
Figure 7:
Figure 7:

Commodity Price Shocks

Citation: IMF Working Papers 2015, 040; 10.5089/9781498310147.001.A001

Comparing the results across sub-periods shows that for South Africa output response to gold shocks have become less persistent during the IT period. This result may have to due to do with the different responses of inflation and the policy rate to gold price shocks across the two periods. In particular, before the IT inflation increases on impact but decreases significantly following the shocks. During IT, on the other hand, inflation show positive responses. As such, the central bank raises the policy rate to respond to inflationary pressure during the IT period and this policy action also helps in containing the output boom. The response of the bank of Ghana during IT follows a similar line in responding to inflation that decreases on impact and increases later. As such, the policy rate also decreases first and later increases. The BOG before IT is more difficult to explain as inflation and output increase on impact but the policy rate is reduced.

4.4 Variance Decomposition Analysis

Tables 3 and 4 report the median percentage variance shares of selected series that are due to each of the shocks. Considering the 3-year horizon for credit and productivity shocks shows that the global shocks are more important in both countries although these shocks have a larger impact in South Africa than in Ghana (Table 3). Moreover, these two global shocks explain about the same share of variation of the macroeconomic aggregates in both countries. For instance, each of these global shock explains about 8 and 11% of output variation in Ghana and South Africa, respectively. Helbling et al. (2011) report a similar finding indicating that the global credit and productivity shocks explain each about 12% of output fluctuation in G7-countries.

Table 3:

Variance Decomposition for Credit and Productivity Shocks

article image
Table 4:

Variance Decomposition for Commodity Price Shocks

article image

In the short-term domestic credit and productivity shocks dominate their global counterparts for inflation and credit. Especially, domestic productivity shocks are the main drivers of output and inflation fluctuations in the short run. For real credit domestic credit shocks account for a large portion of the volatility, with productivity shocks contributing as well. For the short-term interest rates and the real effective exchange rates the global credit and productivity shocks are still important, explaining 10 percent of the volatility. This result may reflect the use of those two instruments for stabilizing external shocks.

Finally, commodity price shocks are also more important for South Africa than for Ghana (Table 4). Moreover, the share of variance due to this shock is in general a bit larger than the one of productivity or credit shocks for both countries. In addition, there has been an increase of the role of commodity price shocks in macroeconomic fluctuation in Ghana and South Africa in recent years. In the inflation targeting regime commodity price shocks account for 20-25 percent of the volatility of all variables in the VAR. This reflects the dependence on the commodity sector in these economies.

Taken together our five shocks (using the IT period of commodity price shocks) account for half the fluctuations in our main macroeconomic variables. A full accounting of the macroeconomic volatility in Ghana will likely require both fiscal and monetary policy shocks in addition to our shocks. In this study we have chosen to focus on exogenous shocks rather than policy shocks. Limits on data on fiscal policy make studying the impact of Fiscal shocks in a VAR framework difficult. As noted earlier, the frequent changes in monetary policy structure do not lend themselves well to a VAR analysis. The former would be best studied using a narrative approach, while the latter with a regime switching structural model.

5 Conclusion

In this paper, we analyze the role of domestic and global shocks in explaining business cycles in Ghana and South Africa. For this purpose, we use a medium-scale Bayesian Vector Auto-Regressive (BVAR) model that captures the dynamics macroeconomic indicators in G7-countries and in each of these two countries.

We find that global productivity and credit shocks have a greater impact on South Africa than in Ghana. The fact that global credit market shocks do not affect Ghana while productivity shocks do is suggestive that Ghana’s integration with the world is more through trade channels, and less through financial channels. Domestic versions of both shocks highlight a key difference between South Africa and Ghana in that monetary policy in Ghana has responded more strongly to inflation than drops in output. Finally, we find that commodity shocks are an important driver in business cycles in both countries. This result highlights the importance of a primary goods sector in developing a model for business cycle stabilization in Ghana. In contrast, shocks traditionally emphasized in advanced economies (i.e. productivity and credit) play a more muted role.

Appendix: Nowcasting

The detailed exposition on nowcasting can be found in Giannone et al. (2008) and related studies.

A sequence of nowcasts of ytQ at different vintages v, v + 1; v + 2 is:

E[ytQ|Ωv];E[ytQ|Ωv+1];E[ytQ|Ωv+2];

where E is the expectation operator and the Ωv represent the information set used to make the forecast.

A Dynamic Factor Model (DFM) is employed to compute E [• | : ]. DFM can be summarizing the dynamic of N series can be written as follows:

Xt=μ+λft+ξt,(2)

where Xt = (x1,t, …, xN,t)′, ξt = (ξ1,t, …, ξN,t)′ is the vector of idiosyncratic component, ft = (f1,t f2,t, · · ·, fr,t) are the r common factors, and λ, is a N × r matrix of series-specific factor loadings. The model is estimated with the two-step approach proposed by Doz et al. (2011). The method relies on Kalman filter and Kalman smother which helps to estimate missing values. The latter occur because of the mixed frequency nature of the data set. For instance, a monthly definition of a quarterly data is one observation with two missing observations. Table 5 contains the data used.

Table 5:

Data used in the Nowcasting

article image

References

  • Abradu-Otoo, P., B. Amoah, and M. Bawumia (2003): “An investigation of the transmission mechanisms of monetary policy in ghana: A structural vector error correction analysis,” Working Papers 2003/02, Bank of Ghana.

    • Search Google Scholar
    • Export Citation
  • Agenor, P.-R., J. McDermott, and E. Prasad (2000): “Macroeconomic fluctuations in developing countries: Some stylized facts,” World Bank Economic Review, 14, 251285.

    • Search Google Scholar
    • Export Citation
  • Akinboade, O. A. and D. Makina (2010): “Econometric analysis of bank lending and business cycles in south africa,” Applied Economics, 42, 38033811.

    • Search Google Scholar
    • Export Citation
  • Atta-Mensah, J. and A. Dib (2008): “Bank lending, credit shocks, and the transmission of canadian monetary policy,” International Review of Economics & Finance 17, 159176.

    • Search Google Scholar
    • Export Citation
  • Broda, C. (2004): “Terms of trade and exchange rate regimes in developing countries,” Journal of International Economics, 63, 3158.

    • Search Google Scholar
    • Export Citation
  • Cashin, P. (2004): “Caribbean Business Cycles,” IMF working papers 04/136, International Monetary Fund.

  • Cetorelli, N. and L. S. Goldberg (2010): “Global banks and international shock transmission: Evidence from the crisis,” Working Paper 15974, National Bureau of Economic Research.

    • Search Google Scholar
    • Export Citation
  • Chia, W.-M. and J. D. Alba (2006): “Terms-of-trade shocks and exchange rate regimes in a small open economy,” The Economic Record, 82, S41S53.

    • Search Google Scholar
    • Export Citation
  • Crucini, M., A. Kose, and C. Otrok (2011): “What are the driving forces of international business cycles?Review of Economic Dynamics, 14, 156175.

    • Search Google Scholar
    • Export Citation
  • Deaton, A. and R. Miller (1996): “International commodity prices, macroeconomic performance and politics in sub-saharan africa,” Journal of African Economies, 5, 99191.

    • Search Google Scholar
    • Export Citation
  • Doz, C., D. Giannone, and L. Reichlin (2011): “A two-step estimator for large approximate dynamic factor models based on Kalman filtering,” Journal of Econometrics, 164, 188205.

    • Search Google Scholar
    • Export Citation
  • du Plessis, S. (2006): “Business Cycles in Emerging market Economies: A New View of the Stylised Facts,” Working Papers 02/200, Stellenbosch University, Department of Economics.

    • Search Google Scholar
    • Export Citation
  • Gerali, A., S. Neri, L. Sessa, and F. M. Signoretti (2010): “Credit and banking in a dsge model of the euro area,” Journal of Money, Credit and Banking, 42, 107141.

    • Search Google Scholar
    • Export Citation
  • Giannone, D., L. Reichlin, and D. Small (2008): “Nowcasting: The real-time informational content of macroeconomic data,” Journal of Monetary Economics, 55, 665676.

    • Search Google Scholar
    • Export Citation
  • Gilchrist, S., V. Yankov, and E. Zakrajsek (2009): “Credit market shocks and economic fluctuations: Evidence from corporate bond and stock markets,” Journal of Monetary Economics, 56, 471493.

    • Search Google Scholar
    • Export Citation
  • Harding, D. and A. Pagan (2002): “Dissecting the cycle: a methodological investigation,” Journal of Monetary Economics, 49, 365381.

    • Search Google Scholar
    • Export Citation
  • Helbling, T., R. Huidrom, M. A. Kose, and C. Otrok (2011): “Do credit shocks matter? a global perspective,” European Economic Review, 55, 340353.

    • Search Google Scholar
    • Export Citation
  • Hoffmaister, A. W., J. E. Roldos, and P. Wickham (1998): “Macroeconomic fluctuations in sub-saharan africa,” IMF Staff Papers, 45, 132160.

    • Search Google Scholar
    • Export Citation
  • Hoffmaister, A. W. and J. Roldós (1997): “Are business cycles different in asia and latin america,” IMF working papers 97/9, International Monetary Fund.

    • Search Google Scholar
    • Export Citation
  • Houssa, R. (2008): “Monetary union in west africa and asymmetric shocks: A dynamic structural factor model approach,” Journal of Development Economics, 85, 319347.

    • Search Google Scholar
    • Export Citation
  • Houssa, R. (2009): “Asymmetric shocks in the west african monetary union,” Research paper 187, African Economic Research Consortium.

    • Search Google Scholar
    • Export Citation
  • Houssa, R., J. Mohimont, and C. Otrok (2013): “Credit Shocks and Macroeconomic Fluctuations in Emerging Markets,” CESifo Working Paper Series 4281, CESifo Group Munich.

    • Search Google Scholar
    • Export Citation
  • Houssa, R., C. Otrok, and R. Puslenghea (2010): “A model for monetary policy analysis for sub-saharan africa,” Open Economies Review, 21, 127145.

    • Search Google Scholar
    • Export Citation
  • Kose, M. A. (2002): “Explaining business cycles in small open economies: ‘how much do world prices matter?’,” Journal of International Economics, 56, 299327.

    • Search Google Scholar
    • Export Citation
  • Kose, M. A. and R. Riezman (2001): “Trade shocks and macroeconomic fluctuations in africa,” Journal of Development Economics, 65, 5580.

    • Search Google Scholar
    • Export Citation
  • Male, R. (2011): “Developing Country Business Cycles: Characterizing the Cycle,” Emerging Markets Finance and Trade, 47, 2039.

  • Meeks, R. (2012): “Do credit market shocks drive output fluctuations? evidence from corporate spreads and defaults,” Journal of Economic Dynamics and Control, 36, 568584.

    • Search Google Scholar
    • Export Citation
  • Mendoza, E. G. (1995): “The terms of trade, the real exchange rate, and economic fluctuations,” International Economic Review, 36, 10137.

    • Search Google Scholar
    • Export Citation
  • Mountford, A. and H. Uhlig (2009): “What are the effects of iscal policy shocks?Journal of Applied Econometrics, 24, 960992.

  • Schnabl, P. (2012): “The international transmission of bank liquidity shocks: Evidence from an emerging market,” Journal of Finance, 67, 897932.

    • Search Google Scholar
    • Export Citation
  • Sowa, N. K. and I. K. Acquaye (1999): “Financial and foreign exchange markets liberalization in ghana,” Journal of International Development, 11, 385409.

    • Search Google Scholar
    • Export Citation
  • Tamasi, B. and B. Vilagi (2011): “Identification of credit supply shocks in a bayesian svar model of the hungarian economy,” MNB Working Papers 2011/7, Magyar Nemzeti Bank (The Central Bank of Hungary).

    • Search Google Scholar
    • Export Citation
  • Uhlig, H. (2005): “What are the effects of monetary policy on output? results from an agnostic identification procedure,” Journal of Monetary Economics, 52, 381419.

    • Search Google Scholar
    • Export Citation
*

Affiliations: CRED & CeReFiM, University of Namur; CES, University of Leuven and CESifo (Houssa); CRED & CeReFiM, University of Namur (Mohimont); University of Missouri and Federal Reserve Bank of St Louis (Otrok). This paper is forthcoming in the Pacific Economic Review special issue on the macroeconomics of low-income countries. We are grateful for helpful comments from Giovanni Melina and participants at the pre- and final conferences at IMF. This paper is part of an IMF research project on macroeconomic policy in low-income countries supported by the U.K.'s Department for International Development (DFID). The views expressed in this paper are those of the author(s) and do not necessarily reflect those of DFID, the IMF, or IMF policies or the Federal Reserve Bank of St Louis.

1

Actually we start in 1985 to account for this break in monetary policy in Ghana but also to account for the start of the globalization period in 1985.

2

We start in 1985 to account for the break in monetary policy in Ghana that occurred in 1983 but also to account for the start of the globalization period in 1985. We end the SVAR analysis sample in 2010:3 for data limitation on the coverage of the US default rate.

3

The results reported in the paper are based on 250 draws. Using a larger number of draws leaves qualitatively the results unchanged.

4

This change in monetary policy regime could have also changed the monetary policy responses to other shocks.

The Sources of Business Cycles in a Low Income Country
Author: Romain Houssa, Jolan Mohimont, and Mr. Christopher Otrok