Technological Advancement and Unemployment in 21st Century Developed Countries

Technological Advancement and Unemployment in 21^st Century Developed Countries

PSC 2012 – Research Methods in Political Science

Amy Schmidt ’15

Winter 2013

Topic of Interest

            When examining the causes of recent global recessions like the Great Recession of 2008, many view partisan gridlock in the United States and the debt crisis in Europe as the major causes (Condon and Wiseman 2013), while overlooking the possibility that high levels of technology may also have contributed to downturn in the economy through their effect on unemployment rate. As technology in industrialized states begins to advance at a higher rate than ever before, machines, computers, software programs, and other devices foster the capacity to perform the occupational tasks of the average middle class worker (Condon and Wiseman 2013). Historically, advancements in technology have been correlated with economic growth, and have provided more occupational opportunities in the long run (Frank and Bernanke 2013). In more recent times, though, the average citizens in developed countries may be suffering economically due to technological progress that has taken away millions of jobs from all sectors of developed countries’ economies without providing more jobs in the long run. As the industrialized countries of the global economy are striving to recover from the Great Recession five years after its commencement, it its crucial for such countries to assess all possible causes of economic downturn in modern times, including unemployment rates that may show a statistically significant relationship with technological advancement, in order to avoid future recessions or to recover more quickly.

Research Question

            The repercussions of the recent economic hardship brought upon by the global recession of 2008 impacted average middle-class citizens in developed countries largely through rising unemployment rates and struggles in the job market (Condon and Wiseman 2013). It is only natural to assume that high unemployment rates exist as both causes and byproducts of economic recessions. To address the economic hardship of middle-class residents, it is crucial to investigate the main causes of unemployment in developed countries. According to numerous economists, such as Frank and Bernanke (2013), Partridge and Partridge (1999), Izyumov and Vahaly (2003), and Nickell, Nunziata and Ochel (2005), factors such as labor market rigidities, large output gaps, outsourcing, lowered consumer demand, and minimum wage have enhanced the prospect of increasing unemployment rates; has technology played a statistically significant role in the unemployment rates of the 21^st century’s developed countries as well? This question arises due to the idea that highly advanced, widespread technologies today hold the capacity to surpass merely providing efficiency, by accomplishing tasks that have for so long been carried out by average workers. According to Condon and Wiseman (2013), “the global economy is being reshaped by machines that generate and analyze vast amounts of data,” such as smartphones, computers, and other efficient robot-like machines. Thus, more specifically, though technological advancement and usage tends to be associated with economic growth, have more modern technologies such as computers, cellular devices, and others derived from research and development expenditure caused unemployment rates in developed countries to increase, leading to economic hardship for middle-class citizens?

Literature Review

             When measuring unemployment, economists tend to distinguish between the natural rate of unemployment, consisting of structural and frictional unemployment, and cyclical unemployment, which is associated with recession in the economy. According to Frank and Bernanke (2013), structural unemployment is “the long-term and chronic unemployment that exists even when the economy is producing at a normal rate,” and frictional unemployment is “the short-term unemployment associated with the process of matching workers with jobs.” Though the different categories of unemployment may appear to hold significance when measuring the effects of specific variables on unemployment, Frank and Bernanke (2013) suggest otherwise. Theoretically, each individual category of unemployment has different causes and plays a substantial individual role in the overall unemployment rate; however, in practice, it is exceedingly difficult to distinguish between the categories precisely (Frank and Bernanke 2013). For this reason, the classification of unemployment rate into three types of unemployment is subjective and imprecise, rather than an approximate factor in measuring overall unemployment rate. Thereby, when investigating the main causes of unemployment, specifying between the categories of unemployment would be trivial rather than beneficial.

            A myriad of views have been formed regarding factors in the economy that cause unemployment rates to increase.  One factor that economists look to is the structure of the labor market. In particular, Nickell, Nunziata and Ochel (2005) hold that, “it is widely accepted that labor market rigidities are an important part of the explanation for the high levels of unemployment  . . . in OECD countries.” However, this group of economists has located errors in measuring unemployment solely on institutions in the labor market. In an empirical analysis of unemployment patterns in the OECD countries from the 1960s to the 1990s, Nickell, Nunziata and Ochel found that labor market rigidities in institutions of the OECD existed even when unemployment was low, leading them to conclude that labor market institutions can potentially explain cross-country differences in unemployment, but are most likely insufficient in measuring changes in unemployment over time (2005). Further, similar studies suggest that unemployment does not appear to be primarily the result of factors such as “overly generous benefits, trade union power, taxes, or wage ‘inflexibility’” (Nickell, Nunziata and Ochel 2005). Due to such claims, it is necessary to consider the combined affects of such economic factors and others in measuring unemployment, rather than citing one or two factors as the sole causes.

            Frank and Bernanke (2013) concur, noting that “the labor market is heterogeneous and dynamic” which provides for difficulty in matching jobs with workers. Further, they cite structural features of the labor market, such as labor unions and minimum wage laws as elements that act as barriers to employment. Both labor unions and minimum wage laws may keep workers’ wages above the market-clearing level, making it too expensive for employers to hire workers, and thus, leading to increased unemployment (Frank and Bernanke 2013).

            Falling under labor market rigidities, government-controlled minimum wage laws have gained leverage as a possible cause of unemployment. In 1999, Partridge and Partridge conducted an empirical analysis of 1980s’ data from the states of developed country, the United States, aspiring to determine the relationship between minimum wage and unemployment. These researchers collected data from a decade in which most states increased their minimum wage laws above the federal level (1999). Their results suggest, “a greater minimum wage increases long term unemployment rates” (Partridge and Partridge 1999). A study conducted by Campbell and Campbell supports this idea. Campbell and Campbell (1969) collected data from the Bureau of Employment Security on the unemployment rates in major labor market areas in states with and without a minimum wage. Though this empirical analysis provided a wide variation among individual states in the relationship between minimum wage and unemployment, Campbell and Campbell were able to conclude that, “on average, states with minimum wage regulation have higher rates of unemployment than states without such laws over a period of 16 years” (Campbell and Campbell 1969).

            Nevertheless, some economists claim that minimum wage does not have as significant of an effect on unemployment as many believe. For instance, White and Jones (1971) directly criticize the analysis published by Campbell and Campbell, claiming that their study is “certainly not persuasive that minimum wage laws are a determining reason for the difference” between states with minimum wage regulation and those without. White and Jones focus their criticism on the validity of the Campbell study and ultimately turn to the relationship between unemployment insurance coverage and unemployment rate, as well as the influence of labor migrations on unemployment differences (1971). Indeed, there are other causes of unemployment to consider, but despite the controversial views over the effects of minimum wage laws on unemployment rates, the existence of minimum wage may hold substantial ground in causing unemployment and thus, must not be ignored.

            Many economists tend to associate the existence of a welfare state with rising unemployment. Krugman (1994) suggests that a welfare state causes unemployment through taxes, which require employers to contribute to social insurance funds, and other regulations that raise the cost to businesses of offering jobs, and further, cause firms to lower the wages that they are willing to pay to workers. He also contends that unemployment benefits provided by government actively reduce the incentive for workers to accept or search for jobs, and thus, raise the wages they demand (Krugman 1994).

            Similarly, Frank and Bernanke identify “unemployment insurance” or government transfer payments as structural features of the labor market that affect unemployment (2013). These scholars argue that unemployment insurance helps unemployed citizens to maintain a decent standard of living while searching for a job; availability of such aid may prolong the average amount of time in which the typical unemployed worker is out of a job or searching for a job, according to Frank and Bernanke (2013). In other words, unemployment insurance in many countries may be too generous, and may also “remove the incentive to actively seek work” (Frank and Bernanke 2013).

            Economists frequently rely on the relationship between the output gap between potential output and real output in a country’s economy to evaluate the size of unemployment. This relationship can be demonstrated by Okun’s Law which states that on average, each additional percentage point in the unemployment rate above four percent is associated with about a three percent determinant in real GNP (Izyumov and Vahaly 2003). Relating to the developed countries, Izyumov and Vahaly state, “since Okun’s original publication, the existence of a trade-off between unemployment and output in mature market economies has been studied extensively and largely confirmed” (2003). Though Izyumov and Vahaly conducted a study testing the validity of Okun’s Law in transition economies, their cross-tabulation test provides significant information about the unemployment rates in more mature economies as they hypothesize that “as market reforms progress, a measureable link between growth and unemployment should emerge” (2003).

            Frank and Bernanke also identify Okun’s Law in that it “relates cyclical unemployment and the output gap” (2013). As previously mentioned, cyclical unemployment occurs due to recession in the economy. In their book, Frank and Bernanke (2013) explain that a recessionary gap may occur in the economy of a country if its actual output is less than its potential output. Though Okun’s Law distinguishes between cyclical and natural rates of unemployment, it is important to consider its parameters when measuring overall unemployment rate as it provides crucial information on whether an economy is in a recession or not. Evidently, negative output gaps or recessions tend to increase overall unemployment rates due to their effect on cyclical unemployment through Okun’s Law.

            The abundance of research and empirical analyses on the preceding plausible causes of unemployment render them sufficient enough to consider as control variables when evaluating unemployment rates and testing for further causes. In addition to the labor market rigidities existing though welfare status and minimum wage laws, and the output gap, more recent economist have begun to explore the possibility that technological advancements could also play a role in increasing modern unemployment rates. However, no statistically notable publication exists demonstrating the direct relationship between technology and unemployment rates in 21^st century developed country. However, “the investigation of whether or not the development of technology has become the enemy of employment is a boundless topic” (Huang 2009).

            In many cases, “a country’s ability to develop and apply new, more productive technologies helps to determine its productivity” (Frank and Bernanke 2013). Economists like Frank and Bernanke use the 18^th century example of increased transportation and refrigerating technologies, which increased economic productivity in many areas including that of American farmers (2013). Further, technology has allowed for numerous developments in communication, medicine, and computer technology, which have led to increased productivity (Frank and Bernanke 2013). Frank and Bernanke (2013) also argue that, “most economists would probably agree that new technologies are the single most important source of productivity improvement.” Concurrently, in general, it would seem as though technology has played a major role in economic advancement rather than hardship.

            Belgian sociologist, Paye, agrees that technology affects the economy positively and in fact, can help to improve long-run employment (1995). His studies in the late 20^th century involve developed countries who are members of the Organization for Economic Cooperation and Development (OECD), and he concludes that in most larger OECD countries and various smaller ones, “long-run economic and employment performance can be improved if there is a deeper involvement in technology development” (Paye 1995). Paye furthers his argument by placing responsibility on OECD countries to spread an understanding of the function of technology in economic growth. However, he does mention the existence of possible costs of technological change.

            Also in the late 20^th century, various economists have surmised a relationship between such technological change and unemployment rate. For example, Goldberg, Highfill and McAsey (1998) contend, “high-tech investment maximizes output but at the expense of unemployment of unskilled workers” or middle-class, average workers in our case. Their research examines the tradeoff between higher technology and more output in exchange for higher unemployment rates and lower technology and less output in exchange for jobs, though it does not statistically significantly prove such relationships. Their objective focuses on policy solutions to this tradeoff, but for the purposes of this study, their consideration of higher levels of technology as causes of job loss for average workers contributes to the main idea that indeed, higher technological advancement may cause unemployment.

            Brand (2002) also recognizes the issue of “technological unemployment” in his review of Amy Sue Bix’ book Inventing Ourselves Out of Jobs? America's Debate over Technological Unemployment. He recognizes that “the displacement of workers by advancing technology has been a concern of thinking people since the beginning of the industrial revolution,” and further notes that this concern has increased throughout the 19^th century (Brand 2002). Ultimately, Brand acknowledges the possible existence of displaced workers due to technological advancements.

            Others, like American journalist Lohr, do more than merely acknowledge technological unemployment, but address it as an urgent issue. Lohr identifies two researchers at the Massachusetts Institute of Technology who assert that although “a faltering economy explains much of the job shortage in America,” advancements in technology have “sharply magnified the effect” (2011). Lohr also investigates economists Erik Brynjolfsson and Andrew P. McAfee and their book Race Against the Machine, which contends that citizens are losing the race against the machines or technology (2011). He also cites W. Brian Arthur, an external professor at the Santa Fe Institute, and his emphasis on “job fallout from technology” (Lohr 2011). Lohr summarizes Arthur’s warning of such fallout in which “technology is quickly taking over service jobs, following the waves of automation of farm and factory work” (Lohr 2011). Ultimately, such research depicting economists’ urgency to address the issue of technological advancement increasing unemployment levels leads one to ponder whether such a relationship is truly statistically significant.

Hypothesis

            Many economists have pondered the issue of technology affecting unemployment without providing statistical evidence. It remains a controversial topic due to the efficiency and long-run employment that technology has provided economies with in the past, and possible negative impacts it may have had in the 21^st century thus far. By statistically testing whether a relationship between higher usage of technology and unemployment rates in the 21^st century exists, I hope to draw nearer to a conclusion that contributes to settling the debate on this topic. I also plan to conduct a study with more recent data than the late 20^th century, the time period where many economists researching this issue gathered their data. After controlling for the variables that are widely accepted as causes of unemployment, including a country’s welfare status and minimum wage laws that construct the labor market, as well as the output gap to determine either the recessionary or expansionary nature of a state, I hypothesize that technological advancement will have a statistically significant relationship with unemployment in which increased technology levels and usage result in higher unemployment rates in developed countries (OECD countries). More specifically, I hypothesize

1.      There will be a statistically significant positive relationship between the number of personal computers in a country and the unemployment rate three years later in which higher numbers of personal computers in countries will cause the unemployment rates in those countries to increase.

2.      There will be a statistically significant positive relationship between the number of mobile cellular subscriptions in a country and the unemployment rate three years later in which higher numbers of mobile cellular subscriptions in countries will increase the unemployment rate in those countries.

3.      There will be a statistically significant positive relationship between a country’s research and development expenditure and the unemployment rate three years later where higher research and development expenditure will cause unemployment rates to rise.

Research Design

            In order to test my hypotheses, I will run a multiple regression test that will determine the relationship between higher technology level and usage and the unemployment rates in twenty different developed countries (n=20). To ensure that the data collected are from developed states, I will focus on countries that are members of the Organisation for Economic Co-operation and Development (OECD) who are considered economically advanced.  The factors that determine higher technology level and usage—number of computers, number of mobile cellular subscriptions, and research and development expenditure in each country—will be taken from the OECD List of Key Indicators as well as the United Nations Millennium Development Goals Indicators. To ensure legitimacy while testing the relationship between technology and unemployment, I will control for the following independent variables that are widely accepted as causes of higher unemployment rates: real hourly minimum wage, welfare cash benefits and in-kind benefits, and output gap. These control variables along with the dependent variable of unemployment rate, will be taken from the OECD List of Key Indicators. To account for lag time in the effect that technology has on unemployment rates, all independent variable data will be measured from the year 2000 until 2005, while the dependent variable data will be collected from 2003 until 2008. 

            The regression model I will use will be a linear multiple regression in the following form: , where represents the progression of years from 2000 to 2005,  represents the personal computers per 100 people from 2000 to 2005,  represents the mobile cellular subscriptions per 100 people from 2000 to 2005,  represents the research and development expenditure as a percentage of GDP from 2000 to 2005,  represents the real hourly minimum wage in U.S. Dollars,  represents the welfare provided as cash benefits and in-kind benefits as a percentage of GDP,  represents the output gap in a country as the deviation of actual GDP from potential GDP as a percentage of potential GDP, and  is the predicted value of Y which represents the unemployment rate from 2003 to 2008.

            To test the significance of the relationship between the variables representing technology levels and the unemployment rates, I will conduct a two-tailed test with a 95% significance level and critical t-value at t = ±2.160. That is, any value with a t-score that is greater than +2.160 or less than -2.160 will be considered within the critical region, and thus, will be considered statistically significant. In the multiple regression test, p = .05 represents the significance level where a p-value less than p = .05 signifies a statistically significant relationship, while a p-value greater than p = .05 denotes that a value is within the average region, and thus, is not statistically significantly related to the dependent variable. Ultimately, this test will provide me with the proper information to determine whether a statistically significant relationship exists between higher technology levels/usage and unemployment rates in developed countries.

Results

            The data in the regression was collected from 20 developed countries that are members of the OECD from years 2000 to 2005, including number of personal computers (per 100 people), number of mobile cellular subscriptions (per 100 people), research and development expenditure (% of GDP), real hourly minimum wage (U.S. Dollars), welfare benefits (% of GDP), and output gap (real GDP deviation from potential GDP as % of potential GDP). The unemployment rates in these countries from years 2003 to 2008 were regressed on these explanatory variables.

            The following regression equation resulted: Unemployment Rate = 114 – 0.052 Year – 0.0528 Personal Computers – 0.0267 Mobile Cellular Subscriptions – 0.468 R&D Expenditure + 0.0516 Real Hourly Minimum Wage + 0.463 Welfare Benefits – 0.334 Output Gap. In condensed numerical terms, the equation takes this form:  with an  value of 32.3%.

            According to this regression model, year as an independent variable, number of mobile cellular subscriptions, research and development expenditure, real hourly minimum wage, and welfare benefits did not have a statistically significant effect on unemployment rates. The progression of years did not have a statistically significant impact on unemployment rate three years later with a p-value of p = 0.813 and a t-score of t = -0.24. If the succession of years did have a statistically significant relationship with unemployment rate, the coefficient for year in the multiple regression equation (-0.052) shows that for every one additional year, the unemployment rate three years later would decrease by 0.052 percentage points. Though I controlled for year as an independent variable in the test, when tested in this particular data set, it was not a statistically relevant control variable.

            Though I had controlled for real hourly minimum wage as a common cause of higher unemployment rates, this variable failed to statistically significantly affect the unemployment rates with a p-value of p = 0.556 and a t-score of t = +0.59. If minimum wage did have a statistically significant relationship with unemployment, the real hourly minimum wage coefficient in the multiple regression equation (0.0516) suggests that for every one dollar increase in minimum wage, the unemployment rate three years later would increase by 0.0516 percentage points. Therefore, if this variable had been in the critical region, rather than statistically insignificant, it would support my hypothesis, which assumes that higher minimum wages have a positive relationship with higher unemployment rates. Since it is not statistically significant, though, this control variable was irrelevant in this particular test.

            Likewise, although I controlled for welfare benefits as a common cause of higher unemployment rates, this variable failed to affect the unemployment rates in a statistically significant relationship due to a p-value of p = 0.128 and a t-score of t = +1.53. If welfare benefits did have a statistically significant impact on unemployment rates, its coefficient in the multiple regression equation (0.463) denotes that for every one percent increase in welfare benefits (% GDP), the unemployment rate three years later would increase by 0.463 percentage points. Hence, if this variable had been in the critical region, rather than statistically insignificant, it would support my hypothesis, which assumes that the percentage of GDP spent on welfare benefits has a positive relationship with unemployment rates. This control variable was not relevant when tested in this data set, though, since its t-score and p-value rendered it statistically insignificant in impacting unemployment rates.

            Similarly, with a p-value of p = 0.084 and a t-score of t = -1.74, the number of mobile cellular subscriptions per 100 people did not statistically significantly impact the unemployment rate three years later. Thus, I must reject hypothesis 2, which predicts a statistically significant relationship between mobile cellular subscriptions and unemployment rates. If mobile cellular subscriptions were statistically significant in this test, the coefficient in the multiple regression equation (-0.0267) suggests that for every one additional mobile cellular subscription, the unemployment rate three years later would decrease by 0.0267 percentage points. Thus, even if this variable held statistical significance, it would lead me to reject hypothesis 2, which states that there will be a positive relationship between the number of mobile subscriptions per 100 people and the unemployment rates; there would actually be a negative relationship.

            Research and development expenditure also failed to have a statistically significant relationship with unemployment rate due to a p-value of p = 0.477 and a t-score of t = -0.71. Concurrently, I must reject hypothesis 3, which predicts that a statistically significant relationship exists between research and development expenditure an unemployment rates. If research and development expenditure did have a statistically significant impact, its coefficient in the multiple regression equation (-0.4678) denotes that for every one percent increase in research and development expenditure, unemployment rate would decrease by .468 percentage points. Again, even if this variable held statistical significance, it would lead me to reject hypothesis 3, which states that there will be a positive relationship between research and development expenditure and unemployment rates; there would actually be a negative relationship between the two variables.

            Oppositely, the output gap in the economy, measured as a deviation of real GDP from potential GDP and a percent of potential GDP, did have a statistically significant impact on unemployment rates three years later with a p-value of p = 0.002 and a t-score of t = -3.12. The output gap coefficient in the multiple regression equation (-0.334) shows that for every one percent increase in output gap, there is a decrease of 0.334 percentage points in the unemployment rate three years later. This control variable does support my hypothesis, which assumes that larger positive output gaps cause lower unemployment rates through a negative relationship. This relationship makes sense due to the idea that higher positive output gaps cause an expansion in the economy when employment resources are applied in full or overextended. Thus, output gap is a relevant control variable, with a critical p-value and t-score, and must be accounted for as a control variable when testing for the relationship between technology and unemployment rates.

            Finally, the regression results show that the number of personal computers per 100 people did have a statistically significant affect on unemployment rate three years later by a p-value of p = 0.002 and a t-score of t = -3.12. Its coefficient in the multiple regression equation (-0.0528) indicates that for every one additional personal computer, the unemployment rate three years later decreases by 0.0528 percentage points. This relationship leads me to reject hypothesis 1, which states that there will be a statistically significant positive relationship between the number of personal computers and unemployment rate three years after. Oppositely, the negative coefficient in the regression equation denotes a negative relationship in which the unemployment rate decreases as the number of personal computers increase. Though I was correct in hypothesizing a statistically significant relationship, higher amounts of personal computers affected unemployment rate differently than I expected.

            In sum, the results of the multiple regression show that three out of four control variables in this test failed to have a statistically significant impact on unemployment rates. Though a larger output gap seems to decreases unemployment rate as expected, hourly minimum wage, welfare benefits, and succession of years did not statistically significantly impact unemployment rates. With this particular data set, these control variables appear irrelevant; however, in a data set with a larger sample size of developed countries or a different selection of control variables, these three may become more statistically significant when examining unemployment rates. Also, it is important to consider the possibility that other variables exist that impact unemployment rates. In this particular regression, I chose to control for the three variables that I perceived as most important; with further research, I may have found other crucial factors in determining unemployment in developed countries.

            Ultimately, though, the focus of this multiple regression test is to determine whether technology advancement and utilization—represented by personal computers, mobile cellular subscriptions, and research and development expenditures—has a statistically significant affect on unemployment rates in developed countries in the 21^st century. Only one out of these three variables, personal computers, tested statistically significant, leading me to reject hypothesis 2 and hypothesis 3 immediately. Further, since the test results indicate that the statistically significant relationship between personal computers and unemployment rate is negative rather than positive, I must also reject hypothesis 1. Thus, I can tentatively conclude that technology advancement and usage by way of personal computers will decrease unemployment rates in developed countries in the 21^st century. I can also conclude tentatively that in general, certain aspects of technology like mobile cellular subscriptions and research and development expenditure do not have a statistically significant relationship with unemployment rates and overall, higher levels of technology do not cause unemployment rates to rise in 21^st century developed countries.

            Clearly, my hypotheses and multiple regression test did leave room for some error. For example, the particular year-span in which the data were collected may have skewed the results; I collected data where each independent variable in each year from 2000 to 2005 was tested for its impact on unemployment rate exactly three years later. Although it is logical to conclude that any factor affecting unemployment will lag in its impact on unemployment rates, I retained no evidence to support that a three-year lag-time is sufficient. If this lag time were longer or shorter, the regression results may have turned out differently. Thus, I can only tentatively reject my hypotheses and accept the results with a three-year lag time for the time being. A follow-up research study may focus more on studying the particular amount of time it takes for particular variables to have a significant impact on unemployment rates.

            Furthermore, technological advancement and usage in developed countries depends on many different factors beyond number of personal computers, mobile cellular subscription, and research and development expenditure. Though these three components do measure different aspects of technology in developed countries, many more technological components exist that are much too wide-ranged for one to collect data on. In a follow-up research study, if I were to test for additional sources of technology usage and advancement such as software programs used in businesses, the regression results may have been slightly more accurate. In another study it may be beneficial to attempt to measure the correlation between number of workers laid-off with the number of machines and computers within businesses or factories in developed countries. Though this information may be difficult to locate, it would greatly enhance the legitimacy of a study seeking to explain unemployment rates in terms of technology. Until further variables and their correlations are tested, I must tentatively reject my hypotheses.

Significance

            The multiple regression results suggest that technology is not a leading cause of higher unemployment rates for developed countries in the beginning of the 21^st century. In terms of this research study, only computers statistically significantly impact unemployment rates, where higher numbers of personal computers per 100 people causes lower unemployment rates overall. This relationship reflects the theories of economists like Paye (1995) and Frank and Bernanke (2013) who attribute higher productivity and employment to higher levels of technology. Though the findings do not support the suspected idea that 21^st century developed countries suffer from technological unemployment, they do provide some assistance in closing the controversial gap over technology’s affect on unemployment. Through this study, one will recognize that computers, common pieces of technology, hold a statistically significant negative relationship with unemployment rates. If it were acceptable to consider computers alone as a representation of technology advancement, this study would support the view that technology actually increases employment rates, not unemployment rates.

            However, the regression also demonstrates that it is not only important to control for non-technology causes of higher unemployment like output gaps, which have a statistically significant relationship with unemployment rates, but also to consider the numerous different pieces of equipment that are considered advancements in technology. In addition to applying computers, mobile cellular subscriptions, and research and development expenditure as representatives of technology, researchers must also consider technological components like updated software, tablets, Internet usage, etc. Ultimately, though, the number of technological constituents is much too large, and it would be exceedingly difficult to pinpoint the data for each component. Data collection on this topic is quite difficult, as the definition of technology has become increasingly ambiguous, even for developed countries where new pieces of technology originate every day.

            As previously mentioned in the results, errors in the study may exist due to lack of sufficient matching of years by incorrect lag-time. Thus, this study suggests that research on lag-time for each independent variable’s effect on unemployment rate must be paid closer attention. Also regarding years, all data for this multiple regression was gathered between years 2000 and 2008 to represent the 21^st century thus far. However, most urgent concerns regarding technological unemployment in developed countries have arisen in the last three years (Condon and Wiseman in 2013 and Paye in 2011) and tend to allude to the future. This suggests that technology-caused unemployment may become more relevant in the future of the 21^st century rather than its initial eight years. Therefore, this study may be more relevant if conducted after the next decade of the 21^st century. If data is collected in the future, between years 2010 and 2020, one may find results more supportive of a positive relationship between technology advancement and unemployment rates than this particular study. Despite the unpredicted findings from this multiple regression analysis, its results and error still provide incentive for researchers to continue studying this issue in the future.

            Ultimately, since the regression results indicate that personal computers have a statistically significant impact on unemployment rates, components of technology, such as computers, cannot be ignored when measuring unemployment rates in developed countries. If nothing else was gained from this study, one’s understanding of technology’s role in the economy is certainly deepened. This research study provides the basic framework for economists in the future who look to measure technology’s effect on unemployment and the overall economy. Any follow-up tests will be enhanced by first, noting the results and possible errors in this particular study and then, researching more thoroughly to correct its inaccuracies. After the Great Recession of 2008, developed countries will continue to look for ways to mend their economies, and thus, must take into account causes of higher unemployment. As a more direct result from this particular test, developed countries may consider working to increase the number of personal computers utilized, since greater amounts of personal computers seem to have been correlated with lower unemployment rates in the 21^st century so far. Though developed economies are composed of an abundance of factors, making recovery a complicated, multidimensional task in the 21^st century, this research study provides a small piece to the puzzle of economic understanding and improvement through controlling unemployment rates.

 Appendix

Scatterplot Graphs with Regressions for Statistically Significant Variables

 Bibliography

Brand, Horst. 2002. “Technological Unemployment.” Monthly Labor Review 125.5 (May): 52-53. http://search.proquest.com (4 March 2013).

Campbell, Colin D and Rosemary G. Campbell. 1969. “State Minimum Wage Laws as a Cause of Unemployment” Southern Economic Journal 35 (April): 232. http://web.ebscohost.com (2 March 2013).

Condon, Bernard and Paul Wiseman. 2013. “Middle-class jobs being lost to technology.” January 27. Democrat and Chronicle. Print.

Frank, Robert and Ben Bernanke. 2013. Principles of Macroeconomics. 5^th ed. McGraw Hill/Irwin.

Goldberg, Kalman, Janett Highfill, and Michael McAsey. 1998. “Technology choice: The output and employment tradeoff.”  Economics and Sociology 57 (January): 27-46. http://search.proquest.com (4 March 2013).

Huang, Xiaowea. 2009. “The Risk of Information Technology on Employment.” Asian Culture and History 1.2 (July): 108-112. http://search.proquest.com (3 March 2013).

Izyumov, Alexei and John Vahaly. 2003. “The Unemployment-Output Tradeoff in Transition Economies: Does Okun’s Law Apply?” Economics of Planning 35 (February): 317-331. http://search.proquest.com (2 March 2013).

Krugman, Paul. 1994. “Past and Prospective Causes of High Unemployment.” Q  IV. Economic Review. https://www.kc.frb.org/PUBLICAT/SYMPOS/1994/s94krugm.pdf (2 March 2013).

Lohr, Steve. 2011. “More Jobs Predicted for Machines, Not People.” October 23. The New York Times. www.nytimes.com (4 March 2013).

Nickell, Stephen, Luca Nunziata and Woflgang Ochel. 2005. “Unemployment in the OECD Since the 1960s. What Do We Know?” The Economic Journal (January): 1-27. http://www.jstor.org/ (2 March 2013).

Organisation for Economic Co-operation and Development (OECD). 2012. “Economic Outlook No 92 - December 2012 - OECD Annual Projections: Output gaps: deviations of actual GDP from potential GDP as % of potential GDP.” December 2012. http://stats.oecd.org/Index.aspx?QueryId=40550# (26 March 2013).

Organisation for Economic Co-operation and Development (OECD). 2013. “Labor Force Statistics by sex and age – indicators.” http://stats.oecd.org/Index.aspx?DataSetCode=LFS_SEXAGE_I_R (26 March 2013).

Organisation for Economic Co-operation and Development (OECD). 2013. “List of OECD Member countries - Ratification of the Convention on the OECD.” http://www.oecd.org/general/listofoecdmembercountriesratificationoftheconventionontheoecd.htm# (21 March 2013).

Organisation for Economic Co-operation and Development (OECD). 2013. “Real hourly minimum wages http://stats.oecd.org/Index.aspx?DatasetCode=RHMW# (22 March 2013).

Organisation for Economic Co-operation and Development (OECD). 2013. “Social Expenditure – Aggregate data: Public expenditure on family by type of expenditure, in % GDP.” http://stats.oecd.org/Index.aspx?QueryId=33817# (26 March 2013).

Partridge, Mark D and Jamie S. Partridge. 1999. “Do minimum wage hikes raise US long term unemployment? Evidence using state minimum wage rates” Regional Studies 33 (November): 713-726. http://search.proquest.com (2 March 2013).

Paye, Jean-Claude. 1995. “Technology, unemployment, and structural change.” Organization for Economic Cooperation and Development. The OECD Observer 194 (June): 4. http://search.proquest.com (4 March 2013).

The World Bank. 2013. “Research and development expenditure (% of GDP).” http://data.worldbank.org/indicator/GB.XPD.RSDV.GD.ZS/countries?page=1&display=default (26 March 2013).

United Nations. 2012. “Millennium Development Goals Indicators: Mobile cellular subscriptions per 100 inhabitants.” July 2012. http://mdgs.un.org/unsd/mdg/SeriesDetail.aspx?srid=756&crid (26 March 2013).

United Nations. 2012. “Millennium Development Goals Indicators: Personal computers per 100 population.” June 2010 http://mdgs.un.org/unsd/mdg/SeriesDetail.aspx?srid=607 (26 March 2013).

White, Rudolph A and Ethel B. Jones. 1971. “State Minimum Wage Laws as a Cause of Unemployment: Comment.” Southern Economic Journal 37 (January): 366. http://web.esbscohost.com (2 March 2013).