1. Introduction
Following the deep recession caused by the onset of the international financial crisis and the collapse of a domestic property bubble, the Irish economy has experienced a robust recovery. Since 2013, when the country exited the EU-IMF financial support programme, domestic demand has grown at an average annual rate of 5.2 per cent, the unemployment rate has fallen from just over 16 per cent to 6 per cent and the budget deficit has declined from a record 32 per cent of GDP in 2010 to close to a balanced budget position in 2017. Given the depth and severity of the 2008–12 crisis, the pace of this recovery has surprised many forecasters, and current projections for the Irish economy envisage a continuation of strong growth over the short and medium term. However, while uncertainty is an ever-present dimension of all forecasting exercises, the outlook for the Irish economy is currently clouded by the presence of a large number of risks related to the external environment.
Because of the uncertainty about the future path of the economy, it is useful to explore the possible effects on future growth of changes in a selection of key external variables. As an exceptionally globalised economy, with exports amounting to more than 120 per cent of GDP, the Irish economy is particularly exposed to external shocks. This exploration can help us gain a better understanding of what drives the economy in the medium term and what are the important factors that will affect the actual outturn.
In considering the impact of external shocks on the Irish economy, it is important to ensure that the modelling approach chosen takes into account in a consistent manner all of the key channels through which such shocks could impact Ireland. Taking the example of an increase in interest rates, such a shock would directly increase the cost of borrowing for Irish firms and households and could reduce investment, and this effect would be captured in a standard model of the Irish economy. However, this channel captures only part of the likely impact of an interest rate increase on the Irish economy. This is because a rise in interest rates would affect growth in the Euro Area and to a lesser extent in Ireland's other key trading partners, the US and UK. Moreover a rise in rates would impact the euro exchange rate and therefore the relative competitive position of the Irish economy. An approach which only captures the direct effect of higher interest rates is therefore likely to underestimate the ultimate impact of the shock significantly.
To estimate the impact of potential shocks as fully as possible, in this paper we adopt a two-step approach. Using the NiGEM (National Institute Global Econometric Model) of the National Institute for Economic and Social Research (NIESR) in the UK, we first estimate the impact of external shocks on the international environment. This allows us to quantify the effect of external shocks on key exogenous variables for the Irish economy such as trading-partner demand, exchange rates and interest rates. In the second stage, we then take the simulation results from the NiGEM model and run these through a separate macroeconomic model of the Irish economy called COSMO (Core Structural Model of the Irish Economy).Footnote 1
Our work builds on previous research for Ireland by Reference Bergin, Conefrey, FitzGerald, Kearney and ŽnuderlBergin et al. (2013) and Reference Bergin, Morgenroth and McQuinnBergin et al. (2016), extending this work by considering some new shocks and building in the two-stage modelling approach described above. Our work is also complementary to Reference Bergin, Conroy, Garcia-Rodriguez, Holland, McInerney, Morgenroth and SmithO'Grady et al. (2017) who use a GVAR framework to investigate a similar set of shocks to the Irish economy, including shocks to interest rates and exchange rates.
The paper is structured as follows. Section 2 provides an overview of the methodology and assumptions. Section 3 discusses the main simulation results, describing firstly the results for key trading partners and then the estimated response of the Irish economy. Section 4 concludes.
2. Methodology and assumptions
COSMO is a detailed multisectoral structural model of the Irish economy. The behavioural equations are estimated econometrically. The model contains three sectors: traded, non-traded and government, with the sectors defined based on the input-output tables.Footnote 2 The long-run equilibrium in the model is driven by the supply side. The supply block consists of a 3-factor normalised nested constant elasticity of substitution (CES) production function with labour augmenting technical progress. The estimation approach for each sector follows that of Reference Barrell and PainBarrell and Pain (1997). The short-run dynamics in the model are determined by the error correction system. The long-run equilibrium in the model, determined by optimisation, ensures that the variables in the model eventually converge on their long-run path as specified by theory. This is achieved through the price-wage system, whereby the prices of the factors of production adjust to guide the economy back to its long-run potential level when there are short-run deviations.
The model initially focusses on the output/production relationships after which the expenditure/income relationships are determined. Given the structure of the Irish economy, where the majority of goods and services output is exported and with overall exports amounting to over 100 per cent of GDP, the equation for traded sector output is a key mechanism in the model. Traded sector output is modelled as a function of world demand and competitiveness. Output in the non-traded sector is driven by domestic demand. On the labour market, wages are modelled in a bargaining framework that includes variables that affect both the supply and demand for labour. There is a long-run relationship for the producer wage and another for the consumer wage. Each of these represent either side of the bargaining process. From the producers' perspective, producer prices matter along with productivity. From the consumer perspective, consumer prices and the unemployment rate – representing the relative bargaining strength of workers – is included. Labour supply is modelled as a function of participation and migration. Migration has been a key channel influencing labour supply in Ireland over a long period. In COSMO, the emigration flow is modelled as a function of the relative attractiveness of alternative labour markets.
A new feature of the COSMO model is the inclusion of a banking sector based on Reference McInerneyMcInerney (2016). The model consists of a series of supply and demand equations for different components of bank credit (mortgages, consumer credit, commercial property and non-property NFC loans) as well as housing and commercial property. In addition, impairments (arrears and insolvencies) are also modelled. The inclusion of a banking sector means that the model can capture the interaction between credit markets, macroprudential policy and the property market. The inclusion of real-financial linkages is an important innovation in the COSMO model following the criticisms of macro models during the financial crisis.
To explore the impact of external shocks, we first solve the model to produce a baseline scenario for the Irish economy. The baseline scenario is generated using out-of-sample projections from COSMO out to 2030. The international assumptions used to generate the baseline in COSMO come from a ‘no-Brexit’ scenario from NIESR. Experience with the model indicates that results are relatively invariant to the type of baseline used.
Each simulation is performed in the following manner. First, we shock a variable (or variables) in NiGEM and this produces a set of outputs for the key exogenous variables used in COSMO (foreign demand, exchange rates, interest rates etc.) These outputs are then used as inputs into COSMO. After the shock variables from NiGEM have been incorporated in COSMO, we then calculate the deviations from the baseline, and these values represent the impact of the shock.
In all cases, no fiscal feedback rule is imposed in order to identify the impact of the external shocks on the public finances. In discussing the impact of the shocks, we generally focus on the long-run effects on the economy after ten years.
We model the separate effect of three key external shocks: a reduction in external demand due to Brexit, a 1 percentage point increase in interest rates and a 10 per cent depreciation of the sterling exchange rate. Taken together, these three shocks encompass a large amount of the uncertainty in the current external environment for Ireland.
3. Results
3.1 A reduction in external demand due to a hard Brexit
Given the reliance of the Irish economy on exports, assumptions on external demand are a key input into any forecast of the economy. These external demand assumptions in turn are critically dependent on the forecasts for growth in Ireland's three key trading partners: the US, UK and Euro Area. In the case of the UK, there is a risk that a hard Brexit could reduce growth and hence UK demand for Irish exports, as well as having an impact on Euro Area and US output.
With the UK being Ireland's single largest trading partner, Brexit has the potential to have a significant effect on the Irish economy. Just under 13 per cent of Irish goods exports and 16 per cent of services are sold into the UK market. Around 16 per cent of Irish imports come from the UK. For specific sectors, such as the employment intensive agriculture and food sector, the dependence on the UK market is even larger than suggested by these aggregate figures (Reference Donnellan and HanrahanDonnellan and Hanrahan, 2016). Around 40 per cent of Irish agri-food exports are destined for the UK market (Department of Agriculture, 2017). Reference Lawless and MorgenrothLawless and Morgenroth (2016) show that, because some of the highest WTO tariffs apply to agricultural products, a hard Brexit could have severe negative consequences for this sector of the Irish economy.
To model the impact of a hard Brexit scenario, we begin by replicating the results in Reference Ebell and WarrenEbell and Warren (2016) and Reference Ebell, Hurst and WarrenEbell et al. (2016) using NiGEM. They examine the impact of a hard Brexit scenario on the UK economy by considering four shocks: reductions in trade with EU member countries, an increase in tariffs, a reduction in inward FDI flows and the repatriation of the UK's projected net contributions to the EU budget. The size of each shock is based on a synthesis of the academic evidence.
While we replicate the hard Brexit scenario as described in Reference Ebell, Hurst and WarrenEbell et al. (2016) for the most part, we make a number of small modifications. Firstly, in Reference Ebell, Hurst and WarrenEbell et al. (2016), investment is assumed to be reduced in the UK due to uncertainty. This uncertainty effect is assumed to be short lived, with the impact dying out after around six quarters (by early 2018). As considerable uncertainty still exists in relation to the form of the UK's future relationship with the EU, we have modified the uncertainty effect in Reference Ebell, Hurst and WarrenEbell et al. (2016) so that it persists until the end of 2020. In addition, we assume that this uncertainty effect also impacts Irish firms so that business investment in Ireland is reduced by an equivalent amount as in the UK.Footnote 3
In a WTO scenario where the UK no longer had a free trade agreement with the EU, tariff barriers would rise on EU-UK trade. Reference Ebell, Hurst and WarrenEbell et al. (2016) incorporate the impact of tariffs of EU-UK trade by shocking UK export prices to the EU and EU export prices to the UK from the third quarter of 2018 (corresponding to the end of the Article 50 negotiating period). An additional minor modification we include is to increase the size of the tariff shock in Reference Ebell, Hurst and WarrenEbell et al. (2016) from 5 per cent to 9 per cent to match the size of the average WTO most-favoured-nation (MFN) import tariff.
Figure 1 shows the impact of this hard Brexit scenario on the UK economy. Taking all of the shocks related to leaving the EU together, UK GDP is projected to be around 3.5 per cent lower in the long run (after ten years) than in the baseline forecast in which the UK remains in the EU. As shown in figure 1, the decline in the UK export market share due to the reduction in UK trade with the EU is the main driver of the overall fall in UK GDP. Reference Ebell, Hurst and WarrenEbell et al. (2016) assume a decline in total UK trade in goods and services with the EU of 50 per cent. This resulting effect contributes to a decline in UK GDP of around 1.5 per cent in the long run. In addition to the impact on the UK economy, there would be a further depreciation of sterling versus the euro. GDP in Ireland's other main trading partners – the Euro Area and US – would also be reduced but by a smaller amount than the reduction in UK output.
Deviation of UK GDP from baseline: Brexit shock
Source: NiGEM, authors' calculations.
We feed these results from NiGEM into the COSMO model to assess the impact of this scenario on Irish output. In particular, the main impact of the hard Brexit shock on the Irish economy comes through its effect on the level of external demand for Irish exports. Table 1 shows the estimated impact of the shock on the Irish economy. In overall terms, output would be reduced by around 3.2 per cent after ten years. The traded sector would endure most of the negative impact of this external shock with the level of output 3.7 per cent lower in the long run. The reduction in traded sector output would arise due to the fall in demand for Irish exports (mainly from the UK), along with the deterioration in Ireland's relative competitiveness caused by the depreciation of sterling.
The reduction in output in the traded sector and resulting decline in firm profitability would lead to a fall in investment and hence output in the non-traded sector. The fall in traded and non-traded sector output would reduce employment by close to 2.5 per cent after ten years and the unemployment rate would be almost 1 percentage point higher. The increase in unemployment and fall in output would lead to a reduction in wages and households' disposable incomes. As a result, consumption would fall by just under 3 per cent after ten years.
Lower tax revenue due to a fall in company profits, combined with higher government expenditure due to the increase in unemployment, would also result in an increase in the government deficit and debt by the end of the period.
Overall, the scenario illustrates the potential negative impact of a hard Brexit on the economy, with output and employment both significantly lower ten years after the shock. Our estimates of the long-run impact are broadly in line with previous work for Ireland by Reference Bergin, Garcia-Rodriguez, Morgenroth and SmithBergin et al. (2017) and Central Bank of Ireland (2016). It is important to note that any assessment of the implications of a hard Brexit for both the UK and Ireland are highly uncertain and the models capture only some of the likely consequences for key variables such as trade, investment and productivity. For example, our modelling in COSMO does not take into account the impact of a hard Brexit on Ireland through the FDI channel. If the UK loses access to the EU single market for services, there could potentially be some diversion of FDI from the UK to Ireland. It is possible that this could partially offset some of the negative impact of Brexit through the other channels discussed above (Reference Bergin, Morgenroth and McQuinnBergin et al., 2016).
The uncertainty around the likely effects of hard Brexit is illustrated by the fact that OECD (2016) and HM Treasury (2016) also use NiGEM to estimate the effect of a WTO scenario on the UK and arrive at estimated impacts on GDP in the long run that are considerably larger than those in Reference Ebell, Hurst and WarrenEbell et al. (2016) (used in this paper). The OECD and HM Treasury studies assume a negative effect on UK productivity due to reduced openness and this is the main reason why the long-run GDP impact is substantially more negative in these studies. Reference Hantzsche and KaraHantzsche and Kara (2018) present a WTO scenario similar to Reference Ebell, Hurst and WarrenEbell et al. (2016), but including an additional negative productivity shock. The results suggest that the impact on long-run UK output would be more severe than estimated in the earlier Reference Ebell, Hurst and WarrenEbell et al. (2016) analysis. Using this scenario for the impact on the UK economy would in turn result in a larger estimate of the decline in Irish output due to a hard Brexit than reported in table 1.
3.2 An increase in interest rates
Interest rates in the Euro Area are currently at historically low levels and are expected to remain low over the medium term as inflation is projected to rise only gradually. At the same time, recent growth momentum in Europe has been strong and it is likely that interest rates will eventually begin to rise, even though there is uncertainty over the precise timing. The scale of the increase in mortgage lending in Ireland during the property bubble in the early and mid-2000s meant that by 2007, Irish households were highly leveraged. Household debt to disposable income peaked at over 210 per cent in 2010. Although significant deleveraging has taken place since then, household indebtedness remains high by international standards.Footnote 4
In this exercise, we consider a simulation in which there is a 1 percentage point increase in the ECB policy interest rate. The increase has a duration of five years. In terms of expectations formation, we assume that agents have rational expectations. It is important to note that this simulation examines a stand-alone exogenous shock to the interest rate. In the event that interest rates begin to rise, this could be due to changes in Euro Area economy or in response to shocks. This simulation does not consider the impact of other possible changes in the macroeconomic environment, beyond the effects of the direct shock to interest rates.
Table 2 shows the effect on key Euro Area aggregates from the shock to interest rates. The increase in the main policy rate raises the cost of borrowing and this reduces both consumption and investment. Higher interest rates reduce the expected real wealth of consumers, while firms face a higher user cost of capital. Changes to interest rates affect exchange rates through the forward-looking Uncovered Interest Parity (UIP) condition, so that the increase to the Euro Area interest rate causes the euro to appreciate, which in turn reduces exports. Taken together, this results in a fall in Euro Area output of over 1 per cent after three years, as shown in table 2.
The overall effect of the shock on the external environment for Ireland is shown in the bottom panel of table 2. Driven by lower Euro Area output, overall foreign demand for Irish exports would be around 1 per cent lower after three years. This is the main channel through which the shock is transmitted to the economy.
In COSMO, we assume that increases in policy rates pass through one-for-one to lending rates for Irish households and firms. The implications of this scenario for the Irish economy are shown in figure 2 and figure 3. As a result of the fall in Euro Area GDP and imports, external demand for Irish exports declines. Moreover, the appreciation in the euro results in a deterioration in Ireland's relative competitiveness. These two variables drive traded sector output in Ireland and the adverse movements in both result in a fall in output in the sector of around 1 per cent after five years. Investment falls, reflecting the direct effect of the rise in interest rates on the user cost of capital as well as the indirect impact due to the decline in exports.
Ireland: deviation of output from baseline: interest rate shock
Source: COSMO, authors' calculations.
Ireland : deviation of domestic demand from baseline: interest rate shock
Source: COSMO, authors' calculations.
Reduced activity in the traded and non-traded sectors would result in a fall in employment and the unemployment rate would increase by around ½ a percentage point. Lower incomes would reduce consumption and, together with the fall in investment, would cause output in the non-traded sector to fall by over 1 per cent by year 5. The effects on overall output (figure 2) are in line with the estimates in Department of Finance (2017). The direct effect of higher interest rates, along with the fall in output and employment, gives rise to a deterioration in the fiscal position, with the government debt-to-GDP ratio projected to be around 4.6 percentage points higher by the end of the period. In the financial sector, the rise in rates would directly reduce the demand for credit so that lending to households and firms would decline.
3.3 A depreciation of the sterling exchange rate
Since the result of the Brexit referendum on 23 June 2016, the value of sterling vis-à-vis the euro and the US dollar has fluctuated significantly. In the week immediately after the Brexit referendum on 23 June 2016, the bilateral euro-pound sterling exchange rate moved from 0.77 pence sterling to 0.82 pence.Footnote 5 This abrupt change of 6.4 per cent was the largest daily change in the history of the exchange rate. Since then, the value of the euro strengthened further, reaching up to 93 pence sterling in August 2017. Compared to its pre-referendum value, this represents a 21 per cent depreciation of pound against the euro. These changes in the sterling exchange rate have important macroeconomic implications by changing the relative competitive position of the UK versus other trading blocks. In the context of ongoing uncertainty over the UK's future trading relationship with the EU, there is likely to be further volatility in the months ahead. For our third shock, we consider the impact of a 10 per cent depreciation of sterling on the Irish economy.
A sterling depreciation shock has two main effects on the UK that are relevant for Ireland: UK imports decline as they are now more expensive following the shock while UK GDP increases as tradable sector output and export volumes increase. As shown in figure 4, we can distil the impact of these changes in the external environment into the effect on the two key variables in COSMO: external demand and competitiveness. In terms of competitiveness, the sterling depreciation results in a deterioration in the relative competitiveness position of the Irish economy. This negatively impacts the tradable sector where Irish firms are price takers on international markets. Regarding foreign demand, the increase in UK growth would help raise UK demand for Irish exports but this positive effect is counterbalanced by the direct reduction in UK imports from Ireland due to weaker sterling. As shown in figure 4, foreign demand for Irish exports declines in the first three years before turning positive after year 4 when the positive impact of the sterling shock on UK GDP reaches its peak.
Ireland : deviation of competitiveness and foreign demand from baseline: sterling shock
Source: NiGEM, authors' calculations.
Taking the results for foreign demand and competitor export prices from NiGEM, figure 5 shows the effect of the shock on Irish output. Overall, the 10 per cent sterling depreciation results in a fall in output of about 0.3 per cent after 4 years. The boost to foreign demand from stronger UK growth helps to lessen the negative impact of the shock but the overall impact of the shock is to reduce the level of Irish output as the negative competitiveness effect dominates. The shock is mainly transmitted to the Irish economy through the traded sector where the deterioration in competitiveness reduces output by almost 0.5 per cent and this in turn lowers investment. The fall in domestic demand and net exports would reduce the level of employment and result in a small rise in the unemployment rate.
Ireland: deviation of output by sector from baseline: sterling shock
Source: COSMO, authors' calculations.
4. Conclusion
In this paper we examine the results of a series of shocks to key external variables that influence growth in the Irish economy. We augment the modelling work in COSMO by first simulating the effect of the shocks on the external environment in NiGEM and then feeding the values of the key exogenous variables into the COSMO model. The results provide a benchmark against which to evaluate the long-run properties of the model as well as illustrating how the model can shed light on the key transmission channels in the economy.
The results indicate the potential negative impact on the economy in the event of a hard Brexit. In this scenario, the level of Irish output would be reduced by over 3 per cent after ten years and the unemployment rate would be close to 1 percentage point higher than in a no-Brexit case. An increase in interest rates – by damaging competitiveness and reducing investment – would lower economic activity and increase the level of debt. In the context of the recent volatility, our simulation results also quantify the possible effect of a depreciation in the sterling euro exchange rate. The model does not take into account possible non-linearities in the responses of firms and households to changes in the exchange rate. Such non-linear effects could be important given scale of the drop in sterling already observed since the Brexit referendum.
These simulations help to shed light on the effects of major changes in key exogenous variables such as foreign demand. The simulations are partial in that they consider the impact of changes in a limited number of key external variables. However, in the event of a major external risk materialising, a range of important variables or policy parameters are likely to change simultaneously. In this case, it would be important to undertake a full simulation to take account of interaction effects between the simultaneous changes in many variables or instruments.
