Assessing the economic impact of Brexit: background report
- In this Special we assess the impact of Brexit on the UK’s economy in three scenarios: a ‘soft’ Brexit, a ‘hard’ Brexit, and a new free-trade agreement (FTA)
- A ‘hard’ Brexit would cost the UK 18% of GDP growth until 2030 compared to a situation where the UK would continue to be part of the EU. In absolute terms, this comes down to a cumulative amount of £400bn, which is equal to £11,500 per British worker
- The economic damage in our FTA and soft Brexit scenarios is less severe than in our hard Brexit scenario, although it will still cost the UK economy roughly 12.5% and 10% of GDP growth until 2030, respectively. This is equal to roughly £9,500 (FTA) and £7,500 (soft) per British labourer over this timeframe
- The Netherlands is an important trading partner of the UK, with export to the UK constituting 10% of total Dutch export. Consequently, a ‘hard’ Brexit will harm the Dutch economy and lead to GDP losses of between -3.5% and -4.25% in the long run. This is equivalent to roughly €25bn - €35bn or €3250 - €4000 per Dutch worker until 2030
- The impact on the euro area will be less severe than for the Dutch economy. We expect a cumulative impact on GDP growth of the euro area of roughly -2% in 2024 in all three Brexit scenarios
- We find much larger negative effects than most existing studies that use macro-econometric modelling to assess the effects of Brexit. This can be attributed to our methodology
- We use an improved tariff version of the macro-econometric model NiGEM, which enables us to better assess the negative impact of cost-push inflation resulting from imposed trade barriers between the UK and the EU
- We estimate a unique productivity model for the UK, which allows us to adequately gauge the UK-specific effects on productivity caused by Brexit
Sluggish British growth in anticipation of Brexit
The negotiations between the UK government and the European Union have finally begun this summer. The UK will probably leave the EU at the end of the first quarter of 2019, although it is also probable that a transition period will be part of the agreement between the UK and the EU. Despite the fact that it will take more than a year before Brexit actually takes place, the British economy is already starting to feel the impact. GDP growth in the first two quarters of 2017 was a meagre 0.2% (q-o-q) and 0.3%, respectively (Figure 1).
Household purchasing power is affected negatively, as the strong depreciation of the British pound is pushing consumer prices up, while nominal wages grow at a slower pace. Since the referendum in June 2016 the pound has fallen by more than 15% vis-à-vis the euro (Figure 1), although there has been some recovery since September due to the suggestion of an upcoming rate hike by the Bank of England in the not-too-distant future. In addition to a loss of purchasing power, corporate profits have deteriorated, as the purchase of intermediate inputs from abroad has become more expensive. Combined with the large uncertainty surrounding the negotiations on the Brexit deal, this has a pervasive effect on UK business investments. Underlining the worries of many corporates active in the UK, on 6 September more than 100 firms signed a letter and sent it to the Brexit negotiators to speed up the progress. In their opinion, the risk of ‘no deal’ remains real, and so businesses have to prepare for the worst, with inevitable consequences for jobs and growth.
How will the British economy be affected in the longer run?
The big question is what the permanent negative effects on the British economy will be once Brexit has actually materialised. The channels through which a Brexit can hit the British and European economies have been extensively described by Irwin (2015).
Trade and foreign investment
A first important channel is trade, as the EU is the UK’s most important trading partner (Figure 2). The introduction of tariffs on goods and non-tariff barriers on goods and services, such as customs controls, raises trade costs on UK exports for the EU and vice versa. Higher import inflation due to these increased trade costs also results in a lower real disposable income of households, which will squeeze their purchasing power. In addition, the UK purchases about half of its imported intermediate products from the EU (see García-Herrero and Xu, 2016). Imports of intermediates would become more expensive due to trade barriers, which means that British companies would face a deterioration of competitiveness, higher export prices and a lower global market share.
An exit from the EU also means that the UK is at risk of losing its position as a gateway to Europe, which will come at the expense of foreign direct investment (FDI). Currently, many firms are already putting investment plans on hold, while others will even decide to cease business activities in the UK. A recent survey of 1,200 major European companies, of which 80% are active in the UK, shows that half these companies are planning to invest less in the UK after Brexit. In addition, 28% indicate that they plan to move a large part of their capacity and 15% say that they intend to stop all activity in the UK.
Moreover, the UK runs the risk of losing its ‘financial services passport’ for UK financial institutions. These passport rights ensure that European and non-European financial institutions located in the UK can serve the entire European Single Market from one location. If the UK loses these rights, London will most likely have to give up its current position as the financial centre of Europe, which would not only have implications for foreign direct investment but would also have negative implications for all kinds of economic activity related to financial intermediation, such as legal advice and accountancy. It is still questionable whether the UK will retain these passport rights after Brexit. Even Switzerland, which is integrated in the EU internal market to a large extent, does not have these financial service passport rights.
Although the strong depreciation of the pound gives some relief to British exporters, only 40% of exchange rate fluctuations usually feed into export prices (IMF, 2015). Thus, the deteriorating effect of imported cost-push inflation on relative export prices outweighs the effects of the depreciation of the pound.
Lower foreign direct investment will have negative implications for growth of the capital stock, which will hold back productivity growth directly. However, labour productivity developments in the UK can be affected by Brexit via other channels as well. Lower foreign direct investment in Research & Development (R&D) could have negative implications for domestic innovative capacity. Moreover, it is well known that international knowledge developed abroad has a larger impact on domestic productivity if a country is more open to either foreign trade (Coe and Helpman, 1995; Grossman and Helpman, 1991) or foreign direct investments (Branstetter, 2006). The uncertainty surrounding the post-Brexit world has a major impact on the scientific community, which depends on long-term funding, cross-border mobility and international collaboration (New York Times, 2016). Stricter migration policies and a deterioration of the UK business climate could result in an exodus of high-skilled immigrants as well, or at least a lower net inflow of migrants. Results from a survey conducted by KPMG (2017) among 2000 EU immigrants working in the UK shows that 8% of the respondents are planning to leave and 35% are considering leaving. Especially younger, higher-paid and better qualified people are considering an exit from the UK, which increases the risk of a brain drain. We will return to the most important productivity conduits more extensively later in this Special.
The British government does have policy options that may mitigate the negative effects of Brexit on the British investment climate. A reduction of corporate taxes could generate positive effects on private investment and could prevent companies from relocating their activities. The downside is that this reduction has to be compensated for somehow by raising taxes on other fronts, such as income tax. Without compensation, public debt ratios would rise even further than the worrisome levels seen since the Great Recession of 2008. Second, deregulation could benefit firms. Then again, the OECD (2016) states that regulation of both network industries and the labour market has already been the least restrictive among OECD countries, which limits the scope for improvement.
Another benefit for the UK is a reduced contribution to the EU budget. The UK contributes roughly 8bn pounds on an annual basis and this is the largest net payment to the EU budget after Germany. In the case of a hard Brexit, the government will have GBP 8bn net budgetary savings. However, if the UK wants to keep access to the EU internal market, it will most likely have to keep contributing to the EU budget as well. Norway, for instance, is a member of the European Economic Area but still contributes marginally less than the UK (83% of the UK contribution). Moreover, the UK also has to pay a financial settlement, the ‘Brexit bill’, which could be as large as 60bn euro.
Another potential benefit for the UK from exiting the EU is that it can make separate trade agreements with countries outside the EU, such as the US and China. It is, however, quite uncertain whether the UK will be able to adopt the FTAs that the EU currently has with third parties or whether the country will have to build a completely new trade framework. In the latter case, the UK will have to enter into lengthy and tough negotiations. This also implies before the instalment of new FTAs, trade costs with these ‘third countries’ will rise. UK trade with third countries that have a FTA with the EU represents 14% of UK exports (Baldwin, 2016). Besides, it is questionable whether the UK will be able to get a better trading deal with the US and China than an entire trading block like the EU. The US has recently imposed a tariff of 219% on British aircraft maker Bombardier. In addition, the US and, amongst others, Brazil and New Zealand have filed a formal complaint at the WTO against a deal between the UK and EU to divide the agricultural quotas post-Brexit. When the WTO would validate this objection the UK will be forced to open up the market for agricultural products to foreign competition. These two examples show that the US will not hesitate to tighten trade screws in order to protect national interests, even at the expense of close allies. More importantly, it illustrates that the UK will come off worst in a dispute with bigger trading partners, as it misses the critical mass to retaliate.
Three exit scenarios plus ‘Bremain’
Broadly speaking, there are three scenarios conceivable for the future trade relationship between the EU and the UK when outside the EU.
- ‘Soft’ Brexit: The UK remains fully part of the European Single Market and trade costs will only result from non-tariff barriers, since the UK does leave the EU Customs Union.
- Bilateral Free Trade Agreement (FTA): the tariffs on products are expected to remain zero, but the increase in non-tariff barriers will be larger than in the soft scenario. Services will no longer be able to move freely.
- ‘Hard’ Brexit: negotiations between the EU and the UK break down in this scenario and the UK leaves the EU without any trade agreement. The WTO agreements will form the basis of the hard Brexit scenario. It is assumed that the UK will be able to copy the EU’s WTO schemes for tariffs and quotas.
We compare these scenarios against the alternative of ‘Bremain’: If the weakness in the British economy continues and the purchasing power of British households is further eroded, the British population may still rethink its decision to leave the EU. Via a second referendum or new elections a Brexit could be averted. We use this ‘Bremain’ scenario as our benchmark scenario, as it lends itself better to compare the damage of the Brexit scenarios. It is important to stress that a Bremain is not our official RaboResearch baseline scenario. In the Rabobank baseline, we assume that the UK will leave the EU, but will sign a trade deal with the EU. However, to assess the economic damage of a Brexit it is preferred to compare outcomes with a Bremain scenario.
Some British politicians, including Prime Minister Theresa May, have suggested the option of a transition period in which the UK’s EU membership would be more or less extended (particularly in her recent speech in Florence). Because it is still uncertain whether Brexit will come with such a period, and if so, what the trading conditions and the length would be, we did not include this in the scenarios.
Literature on permanent damage of Brexit
There are several studies that have estimated the impact of Brexit on the British economy; most studies find that there will be a negative impact on the British economy. We have found two exceptions. First, a study conducted by Capital Economics (2016), commissioned by Woodford Investment Management, which stresses that the negative impact on productivity will most likely be limited. This study, however, does not quantify the effects due to Brexit but only gives a descriptive analysis of the impact. A report by the Economists for Brexit (2016) even predicts a boost for UK productivity and GDP caused by an exit from the EU.
Most other studies, by contrast, find that Brexit will have a negative impact, e.g. OECD (2016), NIESR (2016), Oxford Economics (2015), PWC (2016), LSE (2017), HM Treasury (2016) and CPB Netherlands Bureau for Economic Policy Analysis (2016). In a hallmark overview, Van Reenen (2016) concludes that negative trade and investment effects will result in welfare losses between 1.3% and 2.6%. However, when taking into account dynamic productivity effects, the potential damage is far higher, estimated between 6.3% and 9.5%. The Bank of England (2016) acknowledged in its August 2016 Inflation report that there are a number of factors which are affected by Brexit and may lead to a reduction of the UK’s long-term level of potential output, but it does not provide an estimate of the impact.
To our knowledge, there are four studies that use general equilibrium models to assess the effects of Brexit and take into account dynamic productivity effects as well (see HM Treasury (2016), OECD (2016), CPB (2016) and NIESR (2016)). The impact on British GDP found in these studies is illustrated in Table 1. Broadly speaking, the impact of a soft Brexit on GDP is expected to be somewhere around -3%, a FTA would generate GDP losses of -5% and a hard Brexit would result in GDP damage between -7.5% and -9.5%. These effects are in line with effects found by Van Reenen (2016). CPB (2016) similarly assesses the impact on the Dutch economy, which in the case of a hard Brexit would face losses of -2% GDP (vis-à-vis the Bremain benchmark in 2030) and -1.5% in the case of a FTA. The EU27 is expected to see less permanent damage: -1.5% of GDP in the case of a Hard Brexit and -1.1% of GDP in the case of a FTA regime between the UK and the EU.
Methodology: a two-step approach
In this study, we use a two-step approach to assess the economic impact in different Brexit scenarios. We combine calculations using NiGEM together with calculations using a UK productivity model developed by RaboResearch.
Using the expanded tariff model of NiGEM for scenario analyses has three main benefits. First, the model allows us to assess the impact of several key variables in the short to medium term, such as exchange rate fluctuations, trade flows, foreign direct investment and the labour market. Second, NiGEM ensures that the global trade flows are viewed within a closed accounting setting. Thus, trade flows between countries add up to global trade and possible trade or economic shocks, such as a Brexit, are accounted for via the global world trade matrix. This also means that we can assess what the impact of Brexit has on other countries, e.g. the Netherlands, for which the UK is an important trading partner. Third, NiGEM is an error correction model (ECM), which ensures that short-term deviations of GDP from a country’s growth potential are made up eventually. So, ultimately, growth is driven by structural factors, such as capital formation, structural employment and labour-augmented technological change. Hence, in the short term, GDP is driven by demand:
where C is private consumption, I is investment, G is government consumption, XVOL are exports and MVOL are imports. Whereas, in the longer term, GDP is driven by supply factors:
where YCAP is potential output, K is total capital stock, L total hours worked, M represents oil input, and TECHL labour-augmented technological change. The parameters are either production function parameters or scale parameters. Equation (2) is based on a CES relationship between capital and labour, embedded in a Cobb-Douglas framework. Any deviations of Y from YCAP feeds into the price system, which brings demand in line with supply.
The use of macro-econometric models to assess the effects of Brexit is not new. HM Treasury (2016) and OECD (2016) also use NiGEM, whereas CPB (2016) uses WorldScan. CPB (2016) criticises the use of NiGEM for assessing Brexit effects, as it is not capable of dealing with trade policy shocks unless short-term effects and other non-trade effects are analysed. As such, the UK can only be shocked by lowering export and import volumes directly, while it is impossible to take into account import inflation as a result of higher trade barriers or changes in bilateral trade volumes. To obviate this problem, we use an integral approach that includes the majority of these short-term and non-trade effects. Furthermore, we use an adjusted tariff version of NiGEM, which has been released by NIESR this year. We adjust this model to be able to include changes in trade flows between the EU and UK. This model has the major advantage that the introduction of tariffs is embedded in the model and as such, import inflation is properly taken into account, which consequently affects real wages, real disposable income and consequently lowers private consumption. At the same time, relative competitiveness is hurt by cost-push inflation as intermediates become relatively more expensive, which raises production costs, export prices and lowers UK competitiveness.
Using NiGEM to assess the impact of Brexit has its merits but a major impediment is that long-term effects via an important channel of productivity are disregarded, being labour-augmented technological change (variable TECHL in equation (2)). This variable is more or less exogenous in NiGEM. This impediment applies to many macro-econometric models, which explains why HM Treasury (2016), OECD (2016) and CPB (2016) adopt exogenously-imposed TFP effects.
We are also forced to use this workaround to assess dynamic productivity effects in the tariff model of NiGEM. To assess these productivity effects, RaboResearch has developed a dynamic productivity model for the UK, based on almost half a century of macro data. In this model, we fully endogenise total factor productivity (TFP) and translate these effects to labour-augmented technological change in NiGEM.
We believe our approach has several benefits over the other Brexit studies discussed, which all adopt different methodologies to assess the impact of dynamic productivity effects. CPB (2016) only creates an exogenous link between trade volumes to productivity and uses a crude elasticity of 0.1. Based on the literature, HM Treasury (2016) adopts an elasticity of trade to productivity between 0.2 and 0.3. The impact of FDI on productivity is estimated using an industry-specific fixed effects model for the UK. NIESR (2016) calibrates productivity losses from declines in trade in their optimistic WTO scenario. The shock is based on effects assumed by the study of HM Treasury. NIESR calibrates the shock by adjusting labour-augmenting technology, which is the variable that we also adjust in this study. When taking into account productivity effects, GDP declines by 7.8% rather than 2.7%.
Both CPB (2016), HM Treasury (2016) and NIESR (2016) examine a limited amount of conduits by which productivity might be affected and possibly miss out on other factors that could be affected by Brexit and have a direct link to productivity, such as human capital and R&D.
The OECD (2016) does consider many determinants that might be affected by Brexit and at the same time are important for productivity, such as openness of the economy, anti-competitive product market regulations, business R&D and decline in management quality. But the OECD (2016) uses elasticities taken from panel data estimates by Égert and Gal (2016). The drawback of this approach is that country-specific effects might differ substantially from the average pooled effect. To illustrate this drawback, our UK-specific estimates show that the impact of R&D capital for the UK is much larger (elasticity of 0.6) than the impact found (elasticity between 0.1 and 0.14) in dynamic panel estimates for 20 OECD countries by Erken, Donselaar and Thurik (2016). Obviously, the higher return on domestic R&D activity in the UK compared to the OECD average might reflect a better dynamic innovation system, a better regulatory framework, higher dynamic spillovers of human capital etc.
Using an integrated productivity model for the UK provides insight into which factors have a decisive role in future productivity development, but also the magnitude of the country-specific effects. It also allows us to calculate a solid benchmark scenario, as we are able to fully decompose UK’s potential growth, based on assumptions of its underlying drivers. This adds to the robustness of our base case scenario, in our case a Bremain scenario.
It is important to properly take stock of productivity effects, as labour productivity has been the key pillar of increasing wealth in the UK over almost half a century and has especially been driven by labour productivity growth (Figure 4). In fact, due to higher prosperity levels, employed persons have been working fewer and fewer hours, which is a well-known phenomenon in many industrialized countries. Employment has been a major cause of GDP losses during periods of recession (i.e. the early 80s and 90s and 2009), due to the large amount of layoffs in these times. In short, assessing the permanent Brexit effects is not so much an assessment of the effects on employment, but rather permanent damage on productivity growth. Indeed, McKinsey draws the same conclusion: ‘economic growth will depend heavily on addressing long-standing productivity challenges’.
The major driver behind labour productivity has been the total factor productivity (TFP) component (Figure 5), although it is striking that TFP growth has slowed down markedly after the global economic meltdown in 2009 (Figure 6). Before the Great Recession, the average annual growth rate of UK TFP was 1.7%, whereas after the crisis this slowed to a meagre 0.5% annually.
A TFP model for the UK
TFP by itself is nothing more than a residual in the standard growth accounting (see, for instance, Total Economy Database), but at the same time it is the purest measure of technological progress, given that it captures the portion of economic growth that cannot be explained directly by an increase in labour inputs (due to demographic changes) or capital inputs (due to investment). As far as we know, there are no studies that look into the potential impact of Brexit on British TFP growth. We have developed a UK TFP model which is based on the methodology and very rich dataset developed by Erken, Donselaar and Thurik (2016).
Mechanisms behind TFP
Innovation is considered to be one of the most important drivers of productivity growth (Baumol, 2002; Jones, 2002). Erken et al. (2016), Griffith et al. (2004), Engelbrecht (1997), Bassanini and Scarpetta (2002) and Cameron et al. (2005) all find that there is a positive relationship between domestic technological capital on the one hand and productivity on the other. Next to domestic knowledge development, knowledge developed abroad is an important factor for productivity development (e.g. Coe et al., 2009; Coe and Helpman, 1995; Erken et al., 2016). The idea is that countries with a relatively low level of technological development can benefit from knowledge developed elsewhere by using it in their own products or production processes. Through a process of ‘catching-up’, low technological countries can narrow the gap in productivity with the technologically more advanced countries. Of course, knowledge developed abroad needs a conduit to be able to feed into the domestic innovation system, which can be either trade (Coe and Helpman, 1995; Grossman and Helpman, 1991; Lee, 2005) or foreign direct investment (Branstetter, 2006). In addition, in order to be able to internalise knowledge developed abroad, a country needs to build up ‘absorptive capacity’ (Cohen and Levinthal, 1989). In addition to being a conduit for foreign knowledge spillovers, a higher degree of openness to international trade will also result in a more effective distribution of labour and, consequently, a more productive sector composition, which will be prosperous for productivity. Furthermore, openness to foreign trade fosters market competition, which stimulates firms to reduce their X-inefficiencies and increase efforts to innovate. Both Edwards (1998) and Alcalá and Ciccone (2004) find that international trade has a robust positive effect on productivity. Labour input also generates adverse TFP effects (Belorgey et al., 2006; Bourlès and Cette, 2007; Erken et al., 2016). High labour participation is often characterised by increased deployment of less-productive labour, which lowers labour productivity. Working fewer hours may have a positive impact on productivity if less fatigue occurs among workers or if employees work harder during the shorter number of active hours. Three other factors that affect total factor productivity are business cycle effects, entrepreneurship and taxation (Erken et al., 2016). Business cycle effects are important, as labour and capital endowments are not immediately adjusted to business cycle volatility, which makes TFP susceptible to fluctuations of the business cycle. Entrepreneurship is important, as investments in knowledge and research alone will not automatically advance productivity, because not all developed knowledge is economically relevant. Schumpeter (1947) noted that entrepreneurship is an important mechanism for the creation of added value: the inventor creates ideas, the entrepreneur ‘gets things done’. Taxation could have a negative impact on productivity: a higher rate of taxation implies negative incentives in certain markets, which consequently could result in a less efficient economy. For instance, a higher taxation rate reduces revenues acquired through innovation, which could reduce incentives to innovate.
The model: a framework for empirical TFP analysis
In order to model all mechanisms addressed above, we have adopted the framework used by Erken et al. (2016), where we adopting a flexible output elasticity of capital in the traditional growth accounting methodology developed by Robert Solow (1957). Under the neoclassical conditions of perfect competition in product markets and constant returns to scale in the production factors of capital and labour, the marginal products of capital and labour are equal to the return on capital and the wage rate, respectively. It can be derived that, in that case, the output elasticities of capital and labour are equal to the shares of capital income and labour income in total factor income. The annual growth of TFP can then be calculated as follows:
where Y denotes gross domestic product, K and L denote (physical) capital input and labour (measured in physical units such as hours worked) andis the share of capital income in total factor income, or stated differently, the share of capital income in the gross domestic product. From (3), we can define our TFP model for the UK:
where i is country, t is year and log represents the natural log. In the equation, a1 measures the effect of growth of domestic R&D capital (S) on TFP growth. Domestic R&D capital is calculated by using the perpetual inventory method on intramural R&D expenditure and taking into account a depreciation of capital of 15% due to the obsolescence of knowledge. This depreciation rate is often used to calculate R&D capital, based on Griliches (2000, p. 54), who refers to this percentage as the ‘‘conventional’ 15 percent figure for the depreciation of R&D-capital’. The term a2 is our knowledge catching-up variable (CU), which picks up the effect of the technological distance of the UK from the technological leader (i.e. the US). The technological distance is determined by looking at the distance of the patent stock. Moreover, we interact the technological distance of the UK with the domestic R&D intensity. The idea behind using this interaction term is that higher domestic R&D investment enables a country to close the technological gap with the technological global leader faster. This is in line with the idea of technological absorption capacity. The term a3 measures the impact of human capital and a4 is our business cycle indicator. The average years of (tertiary) education are usually used as an indicator to measure the amount of human capital in a country (see Barro and Lee, 2013).
This indicator is useful, although a disadvantage is that it does not take into account the quality level of education. Cohen and Soto (2007) claim to produce better data than Barro and Lee, as they use information from surveys based on uniform classification systems of education over time, and an intensified use of information by age groups. Moreover, Cohen and Soto claim that their data is more suitable for models in first differences, which is exactly how our UK TFP model is estimated. This measured as the unemployment rate (U) minus the non-accelerating inflation rate of unemployment (U*). Term a3 measures openness of the economy, which is based on trade exposure encompassing a weighted average of export intensity and import penetration, see (Bassanini et al., 2001, footnote 37). We use calculations by Donselaar (2011) to adjust trade exposure for country size, as small economies are by definition more exposed to foreign trade, regardless of their trade policy or competitiveness. Term a4 measures the corporate tax level as a % of total business capital income. Term a4 measures labour input control variables, i.e. the participation rate measured as the number of persons employed (N) as a share of total population (P), and the amount of hours worked (L) per person employed (N). These control variables are lagged one year. Finally, a9 measures entrepreneurship and a10 is a dummy to take into account the extreme effects of the Great Recession in 2008 and 2009. Table 2 sums up the description of all individual variables in equations (3) and (4) and their data sources. For technicalities on variable construction, we refer to Erken (2008), Donselaar (2011), Erken et al. (2016).
Econometrics and estimation results
Taking first differences of variables is a safe option to prevent the danger of spurious regression results when estimating relations between trended variables (Wooldridge 2003, p. 615). In our model, there are several variables trended, including TFP. We use several indicators to obtain the optimal lag structure to enhance the quality of our models: the Akaike info criterion, the Schwarz criterion and Hannan-Quinn criterion.
Our base estimation is illustrated in column (1) of Table 3 and consists of domestic R&D capital, technological catching-up and human capital. Domestic R&D capital shows a higher coefficient than is generally found in the literature, albeit its effect is statistically insignificant. The higher coefficient is easily explained as it is possible that the British innovation system is more efficient than the average OECD country. Technological catching-up and human capital have a significant impact on TFP growth. Moreover, the explanatory power is quite low, which indicates that we miss out on many important drivers in the base estimation.
If we add labour input variables in column (2), domestic R&D capital becomes statistically significant, but for the catching-up this is no longer the case. The labour input variables show the expected negative signs and coefficients in line with general findings in the literature. Moreover, the business cycle variable also seems to be important in explaining TFP growth. All in all, the explanatory power of our second model improves drastically: from 11% to 43% and the magnitude of the coefficients from the base model remain fairly stable.
As our catching-up variable is not significant in the second model, we add our openness variables in column (3), which might serve as a better mechanism to pick up the effect of international knowledge spillovers on TFP growth. The openness variable does indeed seem to be important and even eats away some of the already low explanatory power of our catching-up variable. We choose to lag the variable hours worked by two years, as this improved the fit of our model markedly. In column (4), we estimate the most complete model. Although the fit improves even more to 72%, catching-up and entrepreneurship have counter-intuitive signs, but are statistically insignificant anyway. The other variables have very significant and correct signs. If we remove the insignificant effects from our model in column (5), the TFP model again shows robust stable effects for all variables included. One problem is that the magnitude of the human capital variable seems to drop somewhat. The model in column (5) might not fully incorporate foreign knowledge spillovers and, consequently, fails to capture dynamic externalities which are generally important in the interplay between knowledge and human capital (see Griliches, 2000). In column (6), we interact our catching-up variable with our openness variable, lagged one year. Although this model still does not produce a statistically significant effect of the catching-up, the sign is correct and the magnitude of the human capital variable is again in line with the other estimated models.
As a quick and dirty robustness check, we use TFP growth figures produced by the Bank of England as a substitute for our own TFP growth figures. The estimation of this model is illustrated in column (7). The BoE model also shows convincing and significant effects of domestic R&D capital, hours worked and openness. The human capital variable is also significant but the coefficient is lower than our estimated models. Similarly, the BoE model refuses to show a significant effect of catching up. Moreover, the business cycle effect and labour participation are insignificant as well. All in all, as a robustness check, the BoE model produces quite robust results compared to our model estimates. Ultimately we chose to use the model depicted in column (6) to run our Brexit scenarios, which has a solid fit (Figure 7).
This section presents the results of our scenario analyses. All Brexit scenarios are compared to our benchmark scenario, which is the Bremain scenario, in which the UK decides to remain part of the EU. We will first briefly describe our Bremain scenario, before elaborating on the key economic variables in our Brexit scenarios.
An overview of all assumptions and extrapolations of our Bremain scenario can be found in Table A.1 in the appendix. Moreover, we provide an explanation of why we chose to calibrate parameters the way we did.
The British economy has quite an upbeat growth potential. In a Bremain scenario, we expect the British pound to appreciate to pre-referendum levels vis-à-vis the euro and USD. This will slightly hurt export in the short term but will boost private consumption due to improved terms of trade and lower cost-push inflation. Ultimately, growth will end up somewhere around 2% in 2025 and beyond (Figure 8). TFP growth is the most important driver of potential growth, contributing 1.1ppts. Due to positive net migration, the impact of ageing on labour supply growth is far less severe than in many countries on the European mainland, such as Italy and Germany. We expect structural employment to contribute 0.25ppts. Finally, the highly competitive business climate in the UK has historically attracted a solid amount of investment, both domestic and foreign. We expect investment to contribute 0.7ppts to growth from 2025 onwards.
Brexit scenarios: GDP
In all three Brexit scenarios, the UK ends up in a two-year recession right after Brexit has materialised in 2019 (Figure 9). The magnitude of the recession varies considerably in the scenarios, with a GDP decline over two years of -2.4% in a hard Brexit scenario, -1.1% in the FTA scenario and -0.3% in a soft Brexit. Although recovery sets in after the initial shock, growth remains below potential growth over a long period of time. In the hard Brexit and FTA scenarios there is still a substantial output gap even in 2030. Ultimately, a hard Brexit will cost the UK 18% of growth in 2030, compared to a situation where the UK would continue its EU membership. A FTA and soft Brexit will cause less harm but still cost the UK economy roughly 12.5% and 10% of GDP growth, respectively.
Annual potential output is affected by Brexit as well (Table 1). In the hard Brexit scenario, potential output in 2030 amounts to 1.3%, compared to a potential growth of 2.1% in our Bremain scenario. The FTA and soft Brexit scenarios both show a potential growth of 1.6%. In all three Brexit scenarios, the factor holding back potential growth the most is lower productivity growth (Figure 10). In the Bremain scenario labour productivity in 2030 is £56 per hour, whereas in our hard Brexit this is roughly ten pounds lower (£46 per hour).
The slowdown in productivity is especially caused by lower TFP. Compared to the Bremain scenario, TFP drops by more than a factor 2 from 1.1% to 0.5% in the hard Brexit scenario (Table 4). The slowdown in TFP is caused by a slowdown in growth of domestic R&D capital, less openness of the economy which is an impediment to benefitting from knowledge spillovers from abroad and at the same time lower competitive pressure domestically. Finally, TFP in all three Brexit scenarios slows due to all kinds of labour market metrics. In the FTA and hard Brexit scenarios, the amount of hours worked by people decreases less than in our Bremain scenario, as people try to compensate for a loss of wealth by working more hours. The downside of working more hours is that it holds back productivity gains as well. This explains why the contribution of structural labour input is higher in our most pessimistic scenarios (i.e. FTA and hard Brexit). Finally, the contribution of growth of the capital stock per unit of labour is lower in the Brexit scenario, due to lower foreign direct investment.
Brexit scenarios: trade and prices
Trade volumes deviate substantially in our three Brexit scenarios, with hard Brexit export volumes about 30% lower than in our Bremain scenario and 15% and 10% lower in our FTA and soft Brexit scenarios, respectively (Figure 11). Import volumes are 27% (hard), 23% (FTA) and 16% (soft) lower compared to our Bremain scenario.
The slowdown in trade in all three scenarios is the result of higher trade barriers between the EU and the UK in the post-Brexit era. Due to imposed tariffs and non-tariff barriers between the EU and the UK, export and import prices in the UK rise steeply. Export and import prices in the hard Brexit scenario are 25% higher than in our Bremain scenario. Export prices in FTA and soft Brexit are roughly 20% higher, whereas import prices are 14% higher. The fact that trade prices in both FTA and soft Brexit are roughly equal immediately shows that non-tariff barriers are far more important in determining the prices of future trade with Europe than direct tariffs. The increase of import and export prices is higher than the tariffs imposed, as production costs rise due to overall higher inflation, more expensive imported intermediates and commodities, higher wage costs, and a less efficient production.
The consequence of the higher import prices is that it generates cost-push inflation, which will prop up inflation (Figure 12) and weigh on domestic consumption and GDP. Moreover, higher inflation will result in nominal wage increases, which will boost unit total costs of manufacturing firms operating in the UK. This is detrimental to firm competitiveness, which will result in lower export market shares of British firms.
Brexit scenarios: labour market
In our scenarios the labour damage will be limited in the UK, due to very flexible labour market institutions. Hence, there is no indication that a Brexit in any form will result in higher structural unemployment. Cyclical unemployment will rise in our scenarios, with a hard Brexit causing a jump from 4.6% in 2018 to 6.2% in 2020, whereas in our Bremain scenario, unemployment is stable, hovering just above 4% (Figure 13).
Unemployment rates decline quite rapidly in all three scenarios. As stated, the UK labour market is quite flexible, which means deviations from structural unemployment are not persistent. Furthermore, labour-augmented technological change is growing at a slower pace in all three of our Brexit scenarios, which implies that technology will shred less jobs compared to our Bremain scenario. Finally, real wages grow less rapidly than in our Bremain scenario (and even decline slightly in the hard Brexit scenario), which will be beneficial to employment growth but weighs on private consumption (Figure 14).
Effects on the Dutch economy and the euro area
The Dutch economy has a strong trade relation with the UK. After Germany, the UK is the Netherlands’ second largest trading partner in the EU. Around 10% of total Dutch exports are shipped to the UK and therefore, we expect larger negative effects of Brexit on the Netherlands than for the EU average. Figure 15 shows the average damage in the case of a hard Brexit, where growth in the Netherlands will slow from 1.5% to 0.2% in 2020. In the FTA and soft Brexit scenarios, the Netherlands will see a slowdown to 0.5% and 0.7% in 2020, respectively. The negative effects are not only the result of less direct trade with the UK but are also the result of higher import inflation which weighs on private consumption and investment in the Netherlands.
According to our calculations, the cumulative losses in the long run will be somewhere between 3.5% and 4¼% in a hard Brexit scenario until 2030, which comes down to somewhere between €25 and €35bn, which equals €3250 to €4000 per Dutch worker. The damage to the Dutch economy in our hard Brexit scenario is almost twice as high as the effects found by CPB (2016), which reports cumulative GDP losses of roughly 2% in their most severe Brexit scenario. The deviation in results stems from the more severe economic impact on the British economy in our scenario compared to effects found by the CPB (2016). In the FTA and soft Brexit scenario, the cumulative losses are in the range of 3% and 3.5%.
The impact on the euro area will be less severe than for the Dutch economy. We expect a cumulative impact on euro area GDP growth of roughly -2% in 2024 in all three Brexit scenarios (Figure 16).
In this study, we evaluate the effects of a Brexit in three different scenarios: 1) a hard Brexit scenario in which negotiations between the UK and EU fail and do not lead to a new trade agreement, 2) a free-trade agreement equivalent with the agreement that for instance Switzerland has with the EU, and 3) a ‘soft’ Brexit scenario where the UK remains part of the European internal market, but exits the Customs Union. Our results show that the economic costs of a Brexit in 2030 are expected to range between GBP 400bn (hard Brexit) and GBP 260bn (soft Brexit), compared to a scenario where the UK would continue to be a member of the EU (Bremain). These costs translate to a whopping £11,500 - £7,500 per British worker in 2030.
We deviate strongly from previous studies in our approach and assumptions and find much higher costs associated with the Brexit. This can be attributed to differences in methodology. First, we use an improved tariff version of macro-econometric model NiGEM, which enables us to better assess the negative impact of cost-push inflation resulting from imposed trade barriers between the UK and the EU. Second, we estimate a unique productivity model for the UK, which allows us to adequately gauge the UK-specific effects on productivity caused by Brexit. In this sense, we follow-up on criticism by Dhingra et al. (2016), who state that the HM Treasury has been too careful in their assumptions with respect to trade, FDI and productivity.
Lastly, we want to note that scenario analyses are always subject to a large degree of uncertainty. Despite this, we believe that the estimated economic impact of a Brexit will be severe in any case, as even in a soft Brexit scenario the associated costs for the UK are substantial.
In our Bremain scenario, the UK decides to stay in the European Union, after for example a second referendum or new elections. This Bremain scenario is the benchmark scenario and is used to compare the damage of the Brexit scenarios. In this scenario we extrapolate many variables based on pre-referendum trends. In Table A.1 we report the assumptions for each variable, the extrapolation technique, the computed annual growth rate (CAGR) over the period 2016-2030, or if opportune, the level of the variable in 2030. In some cases, variables are generated endogenously using NiGEM, such as the capital stock, trade flows, unemployment and employment levels. In other cases, we choose to make our own extrapolation based on Autoregressive Integrated Moving Average (ARIMA) models.
Brexit scenario assumptions
Below we discuss all assumptions of our Brexit scenarios for a set of key factors determining the economic impact.
Trade costs: tariffs and non-tariff barriers
In all three Brexit scenarios, we assume that trade costs will rise due to increased trade barriers after Brexit. In the soft Brexit scenario, the UK still has access to the European Single Market, which implies that no tariffs on goods are imposed. In the FTA scenario, we assume the EU and UK will replace the current trade regime directly with a new FTA with zero tariffs on goods. We take a different approach than, for instance, the CPB (2016), where the UK manages to successfully negotiate a FTA only after 10 years, and in the meantime falls back on WTO agreements. The hard Brexit scenario is a different story, as we assume that UK-EU trade relies on WTO agreements and the UK is able to copy the EU’s WTO tariff and quota schedules. This means that UK goods exports to the EU are subject to the EU’s WTO external tariff and vice versa. Based on a weighted average between 2012 and 2015, we adopt an import tariff of 3.17% on UK exports to the EU and a tariff of 3.87% on EU goods exports to the UK.
Trade costs will also rise due to non-tariff barriers, like customs procedures, rules of origin declarations and different product standards. This is partly based on the assumption that the UK will leave the Customs Union in all three scenarios, since the UK has indicated that it prefers to be able to make free trade agreements with non-EU countries after Brexit. Estimates from the literature show that non-tariff barriers between the EU and the US are equivalent to a tariff of about 13% (Egger et al., 2015 and CPB, 2016). However, we expect that trade between the EU and the UK would be less subject to non-tariff related barriers after Brexit than trade between the EU and the US, because compared to the US, the UK enforces more similar standards, legislation and procedures. We therefore assume that only two thirds of these non-tariff barriers apply to trade between the UK and the EU in the hard Brexit scenario. In the soft Brexit scenario, the UK would continue to follow EU policy on, for example, product standards which would limit the rise of non-tariff barriers. However, membership of the European Economic Area (EEA) does not cover the Customs Union, which indicates that a physical border would emerge between the EU and the UK, and non-tariff barriers like customs procedures will come into place in the soft Brexit scenario. Non-tariff barriers will rise by a quarter of the level found for EU-US trade. In the FTA scenario, we assume that non-tariff barriers will rise to a lesser extent than in the hard Brexit scenario, because the EU and the UK will make some agreements to lower customs procedures, but not as much as in the soft Brexit scenario. We assume that non-tariff barriers will raise trade costs by 45% of the rates found for EU-US trade in the FTA scenario.
In short, in the soft Brexit, FTA and hard Brexit scenarios, UK exports to the EU and vice versa will be subject to a price rise due to tariff and non-tariff barriers of 3%, 6% and 11%, respectively.
Since we assume that the UK will leave the EU Customs Union in all three scenarios, the EU FTAs with third countries will cease to apply to the UK. We assume that the UK will replace 40% of these FTAs, but for the remaining 60% of the FTAs we expect trade costs to rise. UK trade with third countries that have a FTA with the EU represents 14% of UK exports (Baldwin, 2016). Although many of these third countries are a WTO member, every WTO member has its own tariff and quota schedules in place. This makes it very difficult to estimate the precise increase in the UK’s trade costs with the rest of the world. We therefore choose a conservative approach to calculate the effect. In all three scenarios, we assume that trade costs with non-EU countries rise by 0.8%. The rationale behind this effect is as follows. Roughly 50% of all UK trade is with the EU and 50% is with non-EU countries. 28% of these non-EU countries have a FTA with the EU, which means that 28% of the UK’s trade with the rest of the world will face an increase in tariffs. We assumed a higher tariff for these countries than the average WTO tariff of 3% since the dispersion between individual country tariffs can be high, and varies from under 5% (Israel), to almost 30% (Egypt). Consequently, we assume tariffs of 5% rather than 3%, which means that UK tariffs with the rest of the world will increase on average by 0.8% (60%´28%´5% = 0.8%).
The exit bill and UK contributions to the EU budget
The EU has made it clear that it expects a financial settlement for the assumed obligations of the UK towards the EU, such as outstanding budget commitments, pension obligations of EU civil servants and contingent liabilities. The magnitude of the exit bill is still uncertain at the moment, but we assume that in the soft Brexit scenario the UK will pay the current estimates of EUR 60bn, which is an amount that has been circulating in the media. This bill would be a price to gain full membership of the Single Market. In the FTA scenario, the UK will pay three quarters of the estimated EUR 60bn, which comes down to EUR 45bn. This is to keep partial access to the Single Market. In both the soft Brexit scenario and FTA scenario, the Brexit bill will be paid on a quarterly basis with equal contributions, spread over 11 years. In the hard Brexit scenario, however, we assume that the UK will not be willing to pay a high amount and therefore assume an exit bill of EUR 15bn, which is 25% of the current estimated bill of EUR 60bn. The EUR 15bn bill will not be paid at once but be spread out over sixteen quarters, assuming that a hard Brexit leads to deteriorating ties between the UK and EU and the UK is not willing to commit to a long-term payment to the EU.
Although eliminating contributions to the EU budget was one of the reasons for the British people to vote in favour of Brexit, the UK probably still has to pay a proportion or the entire amount, if it wants to negotiate a trade agreement with the EU. Norway and Switzerland both contribute to the EU budget, albeit a smaller amount than EU members. As a member of the EEA, Norway’s financial contribution to the EU is 83% as large as the UK’s contribution (House of Commons, 2013). Therefore, we assume that the UK’s contribution to the EU budget falls by 17% in the soft Brexit scenario. Between 2009 and 2014, the average UK net contributions amounted to GBP 8bn. We used this average to calculate the annual net savings for the UK government. In the soft Brexit scenario this consequently amounts to GBP 1.36bn. In the FTA scenario, we assume that the UK’s contribution falls by 50% as the UK maintains only partial access to the EU Single Market, resulting in annual net savings of GBP 4bn. In the hard Brexit scenario, we assume that the UK will not make any contributions and therefore will save GBP 8bn on the government budget. The UK government starts saving net contributions from 2019Q2 onwards.
The net amount of the contribution savings and the Brexit bill payments (thus the new government expenses) will be allocated to government investments and consumption according to their respective weights of total government expenditure in the past. Government consumption has (in the past 7 years) been about 87% of total expenditure and we allocate the rest to government investment.
FX and interest rates
A large part of a potentially hard Brexit has already been priced in the exchange rate of the GBP. However, given the uncertainty of what a hard Brexit will exactly mean for the British economy, the GBP might weaken a bit further if a hard Brexit is ‘harder’ than currently expected by financial markets. Accordingly, we assume that the GBP will depreciate slightly further against the EUR in a hard Brexit scenario in the short term (about 1 year). Subsequently, we expect the GBP to start appreciating gradually, as uncertainty about the effects clears and the Bank of England might be prompted to increase rates due to higher inflation, caused for instance by the initial depreciation of the GBP and higher tariffs between the EU and the UK.
Foreign direct investment
Due to an increase in trade barriers between the UK and the EU, and possibly also between the UK and the rest of the world, the UK is expected to become less attractive to foreign investors after Brexit. After all, the UK could lose its position as a gateway to Europe after Brexit, which will hamper the UK business climate. This will probably result in a reduction of foreign direct investment inflow and lower foreign business activity. According to Bruno et al. (2016), an exit from the EU will decrease FDI inflows by 12% to 28%. For our hard Brexit scenario we use the upper bound of these estimates (28%). Given ONS data on FDI inflows of GBP 21.6bn, we assume an annual decline of GBP 6bn in FDI in our hard Brexit scenario. In the soft Brexit scenario, we use the lower bound of the estimates (12%), which is equivalent to an annual decline of GBP 2.6bn. In the FTA scenario we assume a decline by 20%. These assumptions are in line with NIESR (2016) and HM Treasury (2016) but smaller than OECD (2016). We assume that the FDI shocks start in the second quarter of 2019.
Export markets shares
We assume that trade between the EU and the UK will decrease after Brexit, as both parties find trading on each other’s markets less attractive and more cumbersome. Effectively, we assume that the export share of the UK to the EU (UK to EU export divided by total UK export) and vice versa will decrease. It is important to note that this encompasses an additional negative effect on trade, next to the negative effect of higher tariffs on trade. Because the effect of higher tariffs also affects other factors (e.g. inflation), reducing exports shares affects trade more directly.
We gauge the reduction in trade shares between the UK and the EU by multiplying our estimated increase in export prices (assuming tariffs of 11%, while also including the effect of GBP depreciation) with the export elasticities of the UK and the EU. We adopt an export elasticity of 3.1. This means that UK export volumes will decrease by 3.1% for every 1% increase in export prices. The elasticity is based on a study by the European Commission (Imbs and Mejean, 2010). We use the average export elasticity for the UK in the different model estimates. Using the same methodology, we assume export elasticities for the EU to the UK to be 2.8 and for the UK to the world to be 3.1. Regarding export price increases, we use our previously mentioned increase in tariffs of 11%, while also taking into account the effect of the depreciation of the GBP of 15% against the EUR since the Brexit vote. For the FTA and soft scenarios we use the same elasticities, but our tariffs assumptions are lower, namely 6% (FTA) and 3% (soft).
An argument often mentioned by advocates of Brexit is that a weak pound will boost UK competitiveness, as it will make British exports relatively cheaper. However, we think that the weaker pound will have a limited positive effect on UK exports. First, exporters mainly assess the long to medium term when determining their export strategy and thus will also incorporate a partial reversal of the current pound weakness. Second, the literature shows that less than half of a depreciation of the pound passes through to export competitiveness (IMF, 2015 and Kirby et al., 2016). Exporters use the increased demand for their products to increase their profit margins. Hence, exports of countries experiencing a depreciation of their currency do not become a lot cheaper and exports do not increase as much as they would have in the case of a full exchange rate pass-through. We use pass-through estimations of 40%. If we assume a real GBP depreciation of 12% (given UK inflation of 3%), a pass-through of 40% and tariffs of 11%, we estimate that the export prices will increase by 6% (11% - 0.4´12%). Multiplying the price increase with the estimated elasticities implies that UK-EU exports will decline by 19% (6%´3.1). In a similar vein, we assume EU to UK exports to decline by 17% (6%´2.8). Using the same reasoning, we assume UK to EU exports and EU to UK exports to decline by 6% and 10% in the soft and FTA scenario respectively. We also assume that the UK will need a substantial amount of time to re-negotiate new trade agreements with the rest of the world. Thus, we assume no beneficial effect for UK exports to the rest of the world.
Migration and labour supply
We assume that after Brexit the UK government will impose a stricter migration policy and that EU migration will slow down in the hard Brexit and FTA scenarios. In the soft Brexit and Bremain scenarios, the UK still takes part in the Single Market and, consequently, free movement of persons still applies. According to ONS statistics, the average EU net migration in the period 2013-2015 was 145,000. In our hard Brexit scenario, only migrants with job guarantees are allowed as well as 50% of all migrants who applied for a study. Consequently, net migration will drop by 44%. In the FTA scenario, we assume that all migrants who want to study in the UK are allowed to enter the UK, which will mitigate the negative impact on net migration by 15ppts (i.e. net migration in FTA will still decline by 29%). Furthermore, according to a survey by KMPG (2017), 8% of the current EU migrants residing in the UK are planning to move and 35% are considering an exit. In the hard Brexit scenario we assume that the 8% who are planning to leave will actually move elsewhere. In addition, we assume that half of the 35% of migrants who are considering an exit will leave as well. This comes down to a decline in the stock of EU migrants of 555,000, which we will spread evenly over the period 2019-2023. In the FTA scenario, we assume that only the 8% who are actually planning to leave will actually move elsewhere, which equates to 174,000 EU citizens leaving until 2023. We assume all EU citizens leaving the UK fall within the working age. Consequently, our migration assumptions will negatively affect the working age population of the UK.
To mitigate the negative impact of Brexit on the investment climate in the hard Brexit scenario, we assume that the UK government will lower the corporate tax rate of 19% to the level in Ireland, which is 12.5%. We adopt a gradual decline in the corporate tax levels in 5 years, starting in 2017, with corporate taxes on a par with the Irish levels in 2022. This will marginally improve the UK’s business climate. But, in order to keep government finances healthy, we assume that taxes on income will have to rise slightly. The impact on growth will therefore be limited. In the soft Brexit and FTA scenarios we do not expect a change in the UK’s corporate tax rate.
In the UK, roughly 50% of all investments in R&D are made by foreign companies (Figure A.1). If foreign companies decide to put investments on hold, or relocate business activities away from the UK, we assume that a proportional part of these investment decisions will affect R&D spending. In our three Brexit scenarios, we assume that FDI will decrease by 28% in the hard Brexit scenario, 20% in the FTA scenario and 12% in our soft Brexit scenario. We use these figures to lower R&D investment of foreign firms accordingly. So, in our hard Brexit scenario, for instance, R&D spending is decreased by 14% (0.5´28%), which builds up to annual lower investments in total intramural private R&D of USD -4bn. Some companies will already anticipate a Brexit and cease certain research programmes in the UK in 2018. We assume that roughly half of the annual R&D investment losses in the post-Brexit period will be booked in 2018 (hard Brexit: -7%, FTA: -5%, soft Brexit: -3%).
As discussed, lower trade also affects productivity. We use the trade volume metrics from the NiGEM scenarios to adjust our openness variable. For the FTA and soft Brexit scenarios, we assume that trade patterns will return to their long-run trends. Moreover, we assume that this normalisation process takes longer in the FTA scenario (full recovery 2028) compared to the soft Brexit (full recovery in 2024), as it takes time for the UK to renegotiate FTAs with third countries. Figure A.2 shows the different adopted paths of our openness variable.
Our TFP model for the UK shows that human capital deepening, i.e. the amount of human capital per unit of labour, is an important determinant for productivity growth. Although migration has an impact on the amount of labour input, i.e. the amount of total hours worked in the British economy, we do not expect human capital to be affected by Brexit. The ONS produces statistics on skills levels by nationality. Figure A.3 shows that the average skill level of EU migrants is slightly below the skills level of domestic workers, and even more so compared to non-EU migrants. So, EU migrants leaving the UK will not weigh on the skills composition of the British labour force, or would even slightly improve this variable. However, the survey by KPMG (2016) questioning immigrants about plans to leave the UK shows that the respondents considering an exit are above all the more highly skilled ones. So this will counterbalance any downward pressure on human capital resulting from migration of EU immigrants.
 The report of KPMG (2017) shows that half of the respondents with PhDs and 39% with postgraduate degrees are thinking about leaving. Furthermore, over half of those earning £50,000-£100,000 said they would leave or were thinking about it.
 Although PWC (2016) uses a CGE model to estimate the impact of Brexit, they seem to abstract from any dynamic productivity effects. Or as PWC (2016) puts it: ‘Our study also does not cover potential structural changes to the economy.’ This complicates a comparison with the other CGE studies that do take into account these dynamic productivity effects, including ours.
 NiGEM stands for National Institute Global Econometric Model.
 The term X-inefficiencies refers to slack in the production process and higher production costs than necessary, which are the result of a lack of competitiveness pressure in the market.
 Trade exposure is defined as: TRADE = X + (1-X) ´ M, where X represents the ratio of exports in relation to GDP, M is the ratio of imports in relation to domestic demand. The domestic demand is calculated by domestic production minus exports plus imports.
 Export prices are shocked in 2019Q2 and gradually decline over time.
 This approach is in line with NIESR (2016) and LSE (2016).
 In determining the export elasticities for the EU we simply used the average of the export elasticities of France and Germany, since export elasticities were not estimated by Imbs and Mejean for all EU countries. Given that a large part of UK exports to the EU are to Germany and France, this seems to be a reasonable assumption.
 Note that we spread out the changes in trade shares over several quarters, assuming that trade between the UK and the EU will decline gradually rather than abruptly. In addition, our assumptions are more conservative than the assumptions of NIESR (2016), but we believe that our method is more valid as it is closer to reality. For example, NIESR assumes an exchange rate pass through of 50%, while the IMF (2015) finds an exchange rate pass through of 40%.
Alcalá, F. and A. Ciccone (2004). Trade and productivity. The Quarterly journal of economics, 119(2), 613-646.
Baldwin, R.E. (2016). Brexit Beckons: Thinking ahead by leading economists. CEPR Press.
Barro, R.J. and J.W. Lee (2013). A new data set of educational attainment in the world, 1950–2010. Journal of development economics, 104, 184-198.
Bassanini, A., S. Scarpetta and P. Hemmings (2001). Economic growth: the role of policies and institutions. Panel data evidence from OECD countries, OECD, Economics Department Working Papers, no. 283, Paris.
Bassanini, A. and S. Scarpetta, (2002). Does human capital matter for growth in OECD countries? A pooled mean-group approach. Economics Letters, 74(3), 399-407.
Baumol, W.J. (2002). The Free Market Innovation Machine: Analyzing the Growth Miracle of Capitalism, Princeton University Press, Princeton/Oxford.
Belorgey, N., Lecat, R. and T.P. Maury (2006). Determinants of productivity per employee: An empirical estimation using panel data. Economics Letters, 91(2), 153-157.
Bourlès, R., and G. Cette (2007). Trends in ‘‘structural’’ productivity levels in the major industrialized countries. Economics Letters, 95(1), 151-156.
Branstetter, L. (2006). Is foreign direct investment a channel of knowledge spillovers? Evidence from Japan’s FDI in the United States. Journal of International Economics, 68(2), 325-344.
Bruno, R., Campos, N. and S. Estrin (2016). Gravitating towards Europe: an econometric analysis of the FDI effects of EU membership, mimeo, London School of Economics.
Cameron, G., Proudman, J., and S. Redding (2005). Technological convergence, R&D, trade and productivity growth. European Economic Review, 49(3), 775-807.
Coe, D.T. and E. Helpman (1995). International R&D spillovers. European Economic Review, 39(5), 859-887.
Coe, D.T., E. Helpman and A.W. Hoffmaister (2009). International R&D spillovers and institutions. European Economic Review, 53(7), 723–741.
Cohen, W.M. and D.A. Levinthal (1989). Innovation and learning: the two faces of R&D. The Economic Journal, 99(397), 569-596.
Cohen, D. and M. Soto (2007). Growth and human capital: good data, good results. Journal of Economic Growth, 12(1), 51–76.
CPB (2016). Trade effects of Brexit for the Netherlands. CPB Background Document June 2016.
Dhingra, S., G. Ottoviano, T. Sampson and J. van Reenen (2016), The UK Treasury analysis of 'The long-term economic impact of EU membership and the alternatives': CEP Commentary. London School of Economics and Political Science
Donselaar, P. (2011). Innovatie en productiviteit: het Solow-residu ontrafeld. Rotterdam.
Edwards, S. (1998). Openness, productivity and growth: what do we really know? The Economic Journal, 108(447), pp. 383-398.
Engelbrecht, H.J. (1997). International R&D spillovers, human capital and productivity in OECD economies: An empirical investigation. European Economic Review, 41(8), 1479–1488.
Egger, P., J. Francois, M. Manchin and D. Nelson (2015). Non-tariff Barriers, Integration, and the Transatlantic Economy. Economic Policy, 30(83), 539-584.
Erken, H.P.G., P. Donselaar and A.R. Thurik (2016). Total factor productivity and the role of entrepreneurship. Journal of Technology Transfer, 1-29.
Griffith, R., Redding, S. and J. Van Reenen (2004). Mapping the two faces of R&D. Productivity growth in a panel of OECD industries. Review of Economics and Statistics, 86(4), 883-895.
Griliches, Z. (2000). R&D, education, and productivity. A retrospective. Cambridge (MA): Harvard.
Grossman, G.M. and E. Helpman (1991). Innovation and growth in the global economy. Cambridge, University Press.
HM Treasury (2016). HM Treasury analysis: The long-term economic impact of EU membership and the alternatives, HM Government.
House of Commons. (2013). Leaving the EU. Research Paper 13/42, 1st July.
Imbs, J. and I. Mejean (2010). Trade elasticities: A final report for the European Commission, Report written for the European Commission.
IMF (2015). Adjusting to lower commodity prices. World Economic Outlook.
Irwin, G. (2015). BREXIT: The Impact on the UK and the EU, London.
Jones, C.I. (2002). Sources of U.S. economic growth in a world of ideas. American Economic Review, 92(1), pp. 220-239.
Kirby, S., O. Carreras, J. Meaning, R. Piggott and J. Warren (2016). Prospects for the UK economy. National Institute of Economic and Social Research, no. 235.
KPMG (2017). The Brexit effect on EU nationals – A survey on what European workers will do now.
Lee, G. (2005). International R&D spillovers revisited. Open Economies Review, 16(3), 249-262.
LSE (2016). The costs and benefits of leaving the EU: Trade effects, London School of Economics, CEP Discussion Paper No 1478.
NIESR (2016). The long-term economic impact of leaving the EU. National Institute of Economic and Social Research, 121-238
OECD (2016). The economic consequences of Brexit: A taxing decision. OECD Economic Policy Paper, 16, 1-36.
Schumpeter, J.A. (1947). The creative response in economic history. Journal of Economic History, 7(2), 149–159.
Solow, R.M. (1957). Technical change and the aggregate production function. Review of Economics and Statistics, 39(3), 312–320.
Van Reenen, J. (2016). Brexit's Long-Run Effects on the UK Economy. Brookings Papers on Economic Activity, 2016(2), 367-383.
Wooldridge, J.M (2003). Introductory Econometrics. A Modern Approach. Ohio: Thomson/South-Western.