Effects of Immigration on the UK Economy
Disclaimer: This dissertation has been submitted by a student. This is not an example of the work written by our professional dissertation writers. You can view samples of our professional work here.
Any opinions, findings, conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of UK Essays.
How Immigration Affects the UK Economy
The main debate is to detect whether immigration has any adverse effects on the opportunities of natives in the United Kingdom, with many papers concerned with the subject of how immigration affects the economy looking into any impending effects.
2.1 Theoretical literature
When looking at immigration there are several factors we need to take into account which effect wages and unemployment in the UK.
As dictated by Friedberg and Hunt (1995), “most attention has been paid to the potential adverse effect on the labour market outcomes of native-born workers” stating that competition on native-born workers of a country from immigration could potentially lower wages and could result in dismissal of work. 
Going into more detail, this view is shared by Orrenius and Zavodny (2007) stating that increased immigration on labour supply will cause wage decreases if immigrants and natives are substitutes and wage increases if immigrants are a complement for natives. In industries with less skilled workers as opposed to skilled workers replacement is more probable because “employees are more interchangeable and training costs are lower.”
This view is also shared by Borjas, Freeman and Katz (1997) who argue that less skilled immigrants in contrast to natives will swing income-distribution towards the skilled and vice versa. However, if there is equilibrium in skills there is no negative affect on “supply of skills.”  Substituting native-born workers with immigrants may become difficult with regards to needing English as a language.
There has been some talk that there are equilibrium changes in the labour market according to Dustmann, Fabbri and Preston (2005). Using this labour market equilibrium model as opposed to Lalonde and Topel's (1991) model, they say that if immigrants' skills vary than those to the natives then this can cause “disequilibrium between supply of and cost-minimising demand for different labour types at existing wages and output levels.”  So effectively there could be wage increases if immigrants are complements to natives and vice versa. As a result restoration is expected in the short run in both wages and unemployment at different skill types and may run into the long run.
According to Borjas (1999) the concept of a general equilibrium model would result in the mean income distributions being depended upon the flows of immigrants. Here the Roy model (1951) distinguishes three ways into how immigrants and natives have different skill levels.
Above-average earnings immigrants in source and host countries results in positive correlation
Below-average earnings immigrants in both countries results in negative correlation.
For either selection to occur across countries; there needs to be positively correlated skills.
However, Borjas, Freeman and Katz (1991) say that equilibrium models are likely to change (i.e. increasing employment) in contrast to the flow of immigrants into a country. However, there are numerous factors to take into account the size of this effect (i.e. Share of output, skill composition) where Altonji and Card (1991) goes on to suspect short-run employment effects of immigration of less skilled natives will be greater than the long-run effects.
Going back to Friedberg and Hunt (1995), the model used defines the immigration impact on wages upon natives. Whether the host economy be open or closed to international trade is one of the modelling decisions along with the substitution effect between natives and immigrants. If the model is a closed economy this will result in perfect substitution with which immigrants lower the price of factors having an unclear picture with which they are imperfect substitutes and factor price rises with which they are complements. For example, examine an economy with capital and skilled labour being complementary and unskilled labour being the substitute of these. If there is an occurrence of unskilled immigrant workers then knowingly the wage of these workers will drop resulting in the return to capital and skilled wages being ambiguous. This results in a substitution effect where employers move away from skilled to unskilled labour. As a result this leads to increased supply and therefore higher optimal output.  This is turn has the role-reversal effect where skilled immigrants will now lower the skilled wage barrier resulting in unskilled wage ambiguous effects. These results in an increase in the complementary capital factor in return to the fall in skilled wage and rise in skilled employment.
According to the Hecksher-Ohlin theorem open economy model the results differ. Trade is driven by factor endowments where technology is assumed the same across countries and where these factor endowments do not differ much; factor price equalisation occurs even if migration is not free.
However, according to the open economy model labour embodied in traded goods seems to be thought of as the occurring adjustment: “immigration will cause the country to compensate by exporting more (or importing less) labour as embodied in goods.” 
Schmidt et al. (1994) follows a similar yet different approach discussing migration impact where low skilled wages are set by a monopoly union where the outcomes are dependant on the union's constituents and instruments. Their models talks about when skilled and unskilled workers are complements then this leads to falling unemployment due to increases in immigration. Further there is likely to be a fall in unemployment when skilled workers are cared about by the union. Segmented labour market is another alternative between natives and foreign workers where in the secondary industry, foreign workers are likely to enter compared to the primary sector where natives tend to stay. There is the effect that there could be direct labour market pressure from immigration increases if there is a limited spill over effect between sectors. This direct effect of foreigners on natives leaves the labour market outcome of natives theoretically undetermined.
There has been some study that indicates that during the nineties both theoretically and empirically immigration has a beneficial effect on U.S born workers wages. Ottaviano and Peri (2005) states that increases in total employment by 10% due to flow of migrants and skill distributions to those in the nineties results in a 3-4% increase in U.S born workers wages. This is because the two (U.S and foreign-born workers) are not perfect substitutes. “Workers born, raised and partly educated in foreign environments are not identical to workers born and raised in the U.S.”  This difference is the gains from immigration that fall upon U.S born workers which we may call ‘diversity'.
The change in equilibrium employment and wages was further applied by Dustmann, Fabbri, Preston and Wadsworth (2003). There was more focus on the skill composition of UK-born and immigrants. They talk about labour market productivity differing from UK-born individuals when immigrants arrive into the host country. Accordingly this may be due to several different factors such as different levels of education, different demographic composition. So comparing an immigrant and a UK born worker with the same education level and of similar age, there is still the distinction that the immigrant may differ in areas such as wages and participation. Further there is reasoning that the skills acquired by immigrants in their home country may not be sustainable for the skills needed in the host country. For example a mechanic in India may have acquired skills necessary for performing well in his job  at home; however, they may lack the needed skills in performing this job  in the UK. So there are likely to be effects on skill composition upon wages and employment due to immigrants. However there are common skills which are more needed such as fluency in the host country language where this view like others is shared by Orrenius and Zavodny (2007).
Looking back we can see that most papers look at the key factor being the skills possessed by immigrants leading to their outcomes as an individual whereby education levels play a part too: higher education levels leading to better job prospects. To a certain extent Drinkwater, Eade and Garapich (2006) agrees with Dustmann, Fabbri, Preston and Wadsworth (2003) in that especially from the A8, qualifications overseas do not match that to UK qualifications specified by the Labour force survey  ; instead they are categorised as ‘other'. This further explains the differing skill composition of immigrants upon natives leading to changing wages and employment.
However, this paper has demonstrated that from A8 countries that even though migrants are low skilled and have high levels of education, the majority earn low paying jobs. As a result Drinkwater, Eade and Garapich's (2006) findings are that immigrants' opportunities are limited regardless of the level of skill and education they may possess.
Researching further not many papers talk into much depth about English Language proficiency being one of the characteristics which could affect the potential economic outcomes of migrants. Looking at Bauer et al. (2002) the paper concluded  that destinations with large ethnic enclaves  is where migrants migrate to especially those that speak very little English. The reasoning behind this is increases in size enclave results in decreases in using English. So those who have English proficiency will not be “as dependant on network externalities provided by large enclaves.”  So in effect the study finds that immigrants with good English proficiency levels are likely to migrate to areas of low immigrant concentration levels with similar English proficiency levels. The better your English proficiency levels, the better your opportunities links back to the study by Borjas (1994); workers able to speak English earn more and Grenier et al. (1983-1984); immigrants in the U.S gained higher earnings than those who are not simply able to speak the English language. As a result there seems to be a link between English proficiency and wages of both natives and immigrants.
2.2 Empirical Literature
During 2009 in the UK net immigration was 196,000  . However, unexpected fluctuations are likely to occur.
According to Borjas et al. (1996) the regressions run have concluded that the immigration effect upon wages is determined by crucial factors such as controlling restricted labour markets. However, this is not the only dependant as it also depends on the geographical area covered. Friedberg and Hunt (1995) assess the empirical effects looking into whether immigrants decide to settle down in the host (receiving) country. They said that immigrants are likely to move to regions with the highest wage demands as they are the most mobile workers. Empirical analysis showed that relating mainly to the US and other countries a 10% increase in immigrants upon the population will at most drop wages of natives by 1%. A graph showing the 1990 mean wage and salary income  against the foreign born of those cities indicated that higher mean densities has the same desired effect on wages in the cities with a positive correlation of 0.37 between the two variables. However, Friedberg and Hunt (1995) stated that if immigrants look into wages based upon level and not increases, “the endogeneity problem may be circumvented” by two or more time period data. Therefore there will be no affect on changes in immigrant density by changes in local wage but rather on the change in the wage. This is explained by Goldin (1994) data from 1890-1923 who examines the effect on changes in wages with respect to differing jobs and industries. The common result found a 1 percent increase in the foreign born population fraction lead to a 1-1.6 percent drop in wages. However saying this they may have been a “composition” problem which lead to some of Goldin's results being affected: “city-level wages are a composite of the wages of immigrants and natives in that location”. This being that, cities that have higher proportions of immigrants who earn less than natives results in lower average wages regardless of any impact (negative or positive) immigrants have on natives' wage.
About 60 percent of immigrants during the 1960's lived in the US six main states  and this increased by 15 per cent by 1990 according to Borjas et al. (1997) paper. As a result they look at the percentage of the adult population that were foreign born between the years 1950-1990. Between the 1960s the percentage of foreign born was stable if so declining, however, by the 1970 this almost tripled until 1990: this being alone in California (rose by 16.5 percent). Nearly doubled in the remaining states (rose by 6.2 percent) and rose by 1.2 percent in the rest of the country. Saying this effect of immigration on natives was dependant upon skill distribution between the two. Borjas et al. says that if the distribution of skills is equally likely between immigrants and natives then supply of skills is not affected by immigrants and therefore structural wages will remain unchanged. However, looking into Borjas et al. (1991) paper between the years 1980 to 1988 the share of the U.S work force rose by almost a third (2.4 percent) and this is backed by Orrenius and Zavodny (2007) finding of there being a positive relationship between higher immigrant shares and wages of skilled natives.
Altonji and Card (1991) goes into the skill levels that assess any effects on immigration. He looks at the elasticity of the skilled and unskilled labour indicating that a rise of 1 percent due to immigrants into a population reduces wages by 0.3-0.5 percent and this is further backed-up; if immigrants have the same skill level as the existing labour force that this may be escalated if they worked in the same industry as opposed to separately. Further into this Bartel's (1989) findings show that large concentration cities with previous immigrants are what immigrants are attracted to. Hindsight there is the assumption that the existing labour force formerly the natives are immobile and that unemployment could be dependant upon wages relative to benefits. Employment could be better-off for natives “if employers of immigrants are less likely to comply with minimum wage laws.”  As a result the data used between the years 1970 -1980 does not provide enough detail and breakdown to the nature of immigrants.
Dustmann et al (2005) found by computing a test that educational backgrounds and skill distribution between immigrants to Britain  are fairly similar to native born workers, indicating that there is no direct significant effects of immigration on the labour market as by the labour market equilibrium. This was true for Britain as opposed to empirical evidence. The empirical analysis did not focus on the outcomes had there been no occurrence of immigration on native born workers. The approach followed, was to use “variation in immigration to different spatial areas and to instrument this by variation in historical settlement patterns.” However, little evidence was found on the effects of immigration upon natives.
Looking into Pinschke and Velling (1997) they exploited the short-term disequilibria from immigration as opposed to Altonji and Card (1991) running immigration into the long term, as they felt less confident about immigration impact into the long run due to the data used. Pinschke and Velling (1997) ran two models: the first being foreigners divided by the total population of which they are in the economically active age group 15 – 64 and secondly the change in foreign share; the variables consisting of all registered foreigners. The regression consisted of the change in employment to population ratio from the years 1985-1989 giving a coefficient of 0.07 indicating employment growth in areas of high employment. However, these models did not analyse wages as they felt German unions are successful enough to implement consistent wages across the country which contradicts Schmidt et al. saying that outcomes are dependant on the union's constituents and instruments. Secondly, the data on wages will provide little evidence on the potential impacts of immigration. The results found were that increased immigration had no effect on the unemployment rate. On the other side between the years 1985-1989 they found unemployment fall the least in lower unemployment areas where foreigners are at the most.
Dustmann et al. (2005) says that immigrants have no effect on the labour market which is a point Pinschke and Velling (1997) doesn't contradict saying that there is “no direct evidence that foreign immigration influences the migration pattern of natives.”
Borjas et al. (1996) argues that in order to investigate the effect of immigration on wages we have to study how natives will respond in terms of internal migration. For example if native women migrate to another city the labour supply will not increase due to migration so wages will not be affected. Here two methods; area and factor proportion approach are used. The area approach studied the effect of immigration in gateway cities and showed varying results in 1980 and 1990. This looked likely because in the 1980s there was the economic boom period. Evidence in this paper explains that induced immigration effects the migration of natives and this seemed primarily down to the less educated natives; “if native migration responses are sufficiently large over the relevant period, comparisons of small areas will mask the true effect of immigrants on native wages.” The factor approach focuses on the changes in immigrants' different level of skill and labour supply. Overall the evidence suggested that immigrants have played a role in reducing the pay of high school dropouts as has immigration and trade to falling pay of these workers. Looking into this there have been several reasons with the analysis of the models such that there is the failure that other regions have not been taken into account and that the models assume education of natives and immigrants to be the same which may not be entirely true.
Besides this other papers such as Dustmann et al. (2003) look into the varying areas focusing entirely on the U.K. From the estimates provided a 1 percent increase in immigration in the U.K leads to a 0.71 percent increase in unemployment showing a positive effect. In the two decades 1981 and 1991 evidence suggests that the high influx of immigrants was situated mainly in Greater London and this was simply because immigrants tend to settle in areas of higher wages. Data from the LFS seemed advantageous being able to individually analyse different skill groups (i.e. plot regressions for the skilled, unskilled and semi-skilled) providing evidence that these are all positive. However, several factors were not taken into consideration such as within London itself the flow of immigrants may vary according to area and that immigrants have had positive economic effects in these areas proven by Pischke and Velling (1997).
All in all there seems to be several papers which as a whole determine that the effects on immigration upon wages and unemployment at a minimal with many papers using different empirical methodology. Further to this different data in the papers have been used in assessing these effects where most of the effects remain to be the same. Many papers show the effects of immigration running into the short-run as opposed to the long run. However, this differs adversely when individually studying certain parts of the papers.
SECTION 3: METHODOLOGY
Here raw data from various related sources will be used to make a comment on the effects of immigration, identifying the variables to be used assessing reasoning behind them as well as looking into the limitations. Statistical methods will also be considered in order to analyse the data.
3.1 Sources of data
The time series approach data was gathered from various sources. The first variable being immigration was obtained form the International Passenger Survey and the remaining variables were obtained from the Office for National Statistics under the sub category of Social Trends 2009 edition.
3.2 Theoretical Basis
When looking into the effects of immigration upon wages there are several variables that would be of interest in producing a concluding validation. The stated obvious variables would be wages and immigration, however looking at the discussed literature there are other variables that need to be taken into account.
Looking into the variable of wages, this could be gross wages or household income. Looking into gross wages this takes into account the money you earn based on your hourly or salary pay, before any taxes or other deductions have been taken out; as a result not being taken into account as it does not represent the true reflection of the effects on wages. However, looking into Household disposable income, this shows the average amount each person has available to spend or save summarising people's economic well being as opposed to the country.
Data for immigration has been extracted from the International Passenger Survey looking into long-term international migration into the UK. However, as they are based purely on estimates the outcome is possible that it will not give a true accountable picture of the effects and is left open for sampling errors. Looking into migration between EU countries; citizens can freely travel, work, retire, or just vacate without any problems giving me the solving factor as to why statistical data is not available. One of many ways citizenship can be obtained is through long term residency so basing this in the UK implies that the long term effects would outweigh the short term effects. This is supported by Friedberg and Hunt (1995) mentioning a key factor into whether immigrants decide to settle down in the host (receiving) country and for how long. However, there are some limitations to the data in the sense that it does not account for the number of illegal immigrants into the country. Dustmann, Fabbri, Preston and Wadsworth (2003) points out that there are still distinctions between immigrants and UK born workers such as wages and participation so may not help the process regarding the accountability of the data with regards to illegal immigrants.
The level of real Gross Domestic Product (GDP) hasn't been discussed much and seems an influencing factor in unemployment. GDP refers to the amount of goods and services produced in a year, in a country. For example, a change in real GDP will have a change on the productivity and output in the economy leading to changes in the level of earnings and impacting the level of unemployed workers. Here Altonji and Card (1991) relates this to a 1% percent rise in immigration leads to reducing wages of about 0.3-0.5 percent relating to Dustmann et al. (2003) 1 percent increase in immigration in the UK leading to a 0.71 percent increase in unemployment.
Another situation to look into is the differing level of skill to measure immigration. Dustmann, Fabbri, Preston and Wadsworth (2003) look at the skill composition, splitting up immigrants and UK born into different education groups as well as demographic composition where this coincides with Dustmann et al (2005) saying that the skill distribution between immigrants and British native born workers are fairly similar. So this may not entirely be one of many key issues to look into as well as there being limitations upon the Labour Force Survey (LFS) providing information on educational and skill levels.
Unemployment plays a part too which has a direct relationship with wages. According to Altonji and Card (1991) unemployment could be dependant upon wages establishing this as another factor to take into consideration. Here we look at the unemployment rate as anyone over the age of 16  looking for a paying job but not having one. However, with using this variable there comes the limitation being other related factors not being included such as benefits. Saying this information relating to benefits is not available for my time series approach.
Economically inactive is another variable where this looks at those people either not seeking work or is unavailable to start work. This may impact unemployment directly as the working population could move into this at some point in the future so further tests will be conducted to see any effects.
Household expenditure is closely followed upon GDP and Household Income so any changes in these may see an impact upon expenditure. So as a result this variable will be used to see any closely related changes.
Government expenditure has been taken as a percentage of GDP so any changes in one may see an effect upon the other, so this variable will be taken into account to identify any reasoning changes.
The variables chosen are of particular reasoning as the analysis will be in a time series approach to investigate the effects of immigration on the UK. We have opted for this because it allows me to analyse immigration effects on an aggregate level as opposed to the cross-sectional approach. The cross-sectional approach was not feasible to follow to allow me to obtain accurate results; this is due to wages and flows of immigrants being an unconditioned link (Borjas et al. 1996). Data will be ranging from 1975 to 2007  obtaining a total of 32 observations in this time period which will provide me with the reasoning behind the underlying effects of immigration in the UK. Further the variables have also been logged to provide a more accurate reading in the testing procedure as opposed to using the larger data obtained.
3.3 Raw Data Description
The graphs presented show the relation between the variables. Looking at the variables inflow of immigrants, income, household expenditure and GDP we can see that there is a positive relationship. The only distinction showing is that the remaining variables do not follow a similar trend.
However according to Friedberg and Hunt (1995) inflow of immigrants could result in lower employment and looking at the graph it does follow economic theory in certain periods.
Further to this, recessions in Britain in the 1980s and 1990s took its toll showing high levels of unemployment during these years as well as real GDP being constant yet having a slight downturn curve. Saying this it took five years for unemployment to fall back to its original levels.
3.4 Model Specification
As more than two variables are being used, a multivariate model is to be conducted as follows:
In = β1 + β2X i2 + β3X i3 + êi
This model was derived from various journals such as Borjas (1994) where a similar equation was used with the wage rate being the dependant variable, however, differing independent variables were used to see any alternative effects. Below shows the following;
Inc = α + β1M + e (Model 1)
Inc = α + β1U + β2GDP + β3HE + β4GE + β5EI + β6M + e
α = Alpha constant
Inc = Income
U = Unemployment
G = Real GDP
HE = Household Expenditure
GE = Government Expenditure
EI = Economically Inactive
M = Inflow of immigrants
e = Error term
U = α + β1M + e (Model 3)
U = α + β1Inc + β2GDP + β3HE + β4GE + β5EI + β6M + e
Model 1 specification indicates that the Income will be our dependant variable as opposed to Immigration being our explanatory variables. This model look into whether immigration has any effect upon income directly; being compared to model 2 taking into account the additional explanatory variables, so to look at any altering effects upon income to spot any changes. The same models will be run again switching the dependent variable income with unemployment to see the effects upon unemployment as well.
Testing for stationarity will involve using the Augmented Dickey-Fuller (ADF) test. Non-stationary series are unpredictable and should not be used as it may result in spurious and unreliable statistical inferences. For example in spurious regression, X and Y are not related at all however regressing Y on X shows it is statistically significant even though this is not the case.
Bauer et al. (2002) is a paper that looks into the surrounding issues of immigration. Here the paper talks about large ethnic enclaves driving immigration as opposed to immigration causing this. This lies simply on the fact that the larger the ethnic enclaves, the less proficient you need to be in English. As a result we must run the Granger causality test to know whether changes in a variable have an impact on changes on the remaining variables. For example if X causes Y, then, changes of X happened first followed by changes in Y.
When examining the regression outputs, we will be looking at the signs of the coefficients in each regression model confidently predicting whether they are positive or negative (i.e. statistically significant or not). For example, it is satisfactory to predict that an increase in unemployment is likely to result in a decrease in income as competition is at a low, expecting a negative coefficient sign.
Another aspect of the coefficient is the p-value, referring to the estimated probability of rejecting the null hypothesis. A result is said to be statistically significant when the P-value is less than a preset threshold value being at a 1%, 5% and 10% level. So for example, performing a test at a 10% significance level (i.e. 0.1), a p-value lower than 0.1 implies that you reject the null hypothesis indicating its statistical significance hence accepting the sign of the coefficient. This in turn has a reversal effect with the p-value being greater than the threshold (e.g. greater than 0.1). Focus will be set mainly on the 5% critical level.
The adjusted R-square measures the proportion of the variation in the dependent variable accounted for by the explanatory variables. So this being incorporated into our model will tell us that the higher the value the better the goodness of fit. For example, a value of 0.3 implies that 30% of the explanatory variables accounts for the total variation in the dependant variable.
The t statistic is a measure of how extreme a statistical estimate is and the greater the value the lower the standard error.
Further keeping the Ordinary Least Square (OLS) model as unbiased is important. A set of assumptions known as the Gauss-Markov assumptions will determine whether the OLS model is the best estimator.
Assuming that the errors have an expected value of zero
Assuming that the independent variables are non-random
Assuming that the independent variables are linearly independent.
Error terms have constant variance
High multicollinearity not present in the independent variables
Serial correlation not present
Here the above will be undergoing a series of tests to determine if the OLS model is BLUE. 
Moreover, looking at the identification of multicollinearity will be determined to check if any is present. Multicollinearity occurs when some independent variables in the model are correlated with other independent variables. When multicollinearity is present, p values can be misleading and the regression coefficients' confidence intervals will be very wide. However saying this if multicollinearity exists, the models are still computed with great accuracy but in order to rectify the problem removing a non-logical variable may reduce or eliminate multicollinearity.
On top of this we will be looking at serial correlation. This is when error terms from different time periods are correlated; we say that the error term is serially correlated. However, serial correlation will not affect the unbiasedness or consistency of OLS estimators, but it fact their efficiency. Also it provides wrong hypothesis tests based upon standard errors, for example, with positive serial correlation, the OLS estimates of the standard errors will be smaller than the true standard errors. A way to test for serial correlation is to look at the Durbin Watson Test; provided below are the following guidelines to which the results can be assessed: -
dw < dl: Statistical evidence that the error terms are positively auto correlated
dw > dl: Statistical evidence that the error terms are negatively auto correlated
du < dw < 4-du: No positive or negative serial correlation
dl < dw < du: The test is inconclusive
The Durbin Watson statistic results will lie in the 0-4 range, with a value near two indicating no serial correlation.
Further into this, detection of heteroscedasticity will be determined. This is where the error terms do not have a constant variance. Here what is known as the White's test will be applied, where if the p-value is seen to be less than 0.05 we can accept the hypothesis that there is heteroscedasticity.
The Breusch–Godfrey Serial Correlation LM Test is a more robust test for autocorrelation in the residuals from a regression analysis taking into account lags. The adding of the lags will allow me to see the immigration effects from previous years to the current years. This plays a vital process in Friedberg and Hunt (1995) journal mentioning a key factor into whether immigrants decide to settle down in the host (receiving) country and for how long. As a result the lags will play a part in assisting me into their evaluation and whether this has a similar effect in the UK. Constructing the regression model with lags and identifying the sign of the coefficient will determine if there are any influencing factors as well as the F-statistic and P-value determining the rejection of the null hypothesis at a 5% significant level.
Cite This Dissertation
To export a reference to this article please select a referencing stye below: