(TIỂU LUẬN) FACTORS THAT AFFECTS THE CHANGES OF INFLATION IN THE US DURING THE PERIOD OF 1966 – 2012

FOREIGN TRADE UNIVERSITY FACULITY OF INTERNATIONAL ECONOMICS -*** - MID-TERM ECONOMETRICS REPORT TOPIC FACTORS THAT AFFECTS THE CHANGES OF INFLATION IN THE US DURING THE PERIOD OF 1966 – 2012 INSTRUCTOR : Dr Chu Thị Mai Phương Dr Vũ Thị Phương Mai CLASS : KTEE310 (2.1/2021).1 MEMBER : Hoàng Ngọc Anh – 1913340005 Nguyễn Đức Cường – 1913340012 Vũ Trường Sơn – 1913340034 Hanoi, December 2020 Table of Contents THEOREOTICAL BASIS OF THE RESEARCH TOPIC 1.1 1.1.1 1.1.2 1.1.3 1.2 The relationship between inflation, wages and unemployment In short-term Long term Conclusion The relationship between inflation and CPI MODEL SPECIFICATION TESTING THE INFLUENCE OF UNEMPLOYMENT, CPI, AVERAGE WEEKY EARNINGS ON INFLATION OF UNITED STATE DURING 1966-2012 2.1 2.1.1 2.1.2 Method to derive the model Method to collect and analyze the data 2.2 Population Regression Model 2.3 Explanation of variables 2.4 Description of the data 10 2.4.1 2.4.2 2.4.3 Methodology in the study Data sources 10 Statistical description of the variables 10 Correlation matrix between variables 11 ESTIMATED MODEL AND STATISTICAL INFERENCE 13 3.1 Estimated model and estimated result 13 3.2 Testing the level of relevance of the model 14 3.3 Testing the hypothetical violations 15 3.3.1 3.3.2 3.3.3 3.3.4 3.3.5 3.4 3.4.1 3.4.2 3.4.3 3.5 3.5.1 3.5.2 Testing Omit variable Ramsey Reset 15 Testing Multicollinearity 15 Heteroscedasticity 17 Testing autocorrelation 17 Testing normality of residual 19 Testing an individual regression coefficient 20 Testing the CPI 20 Testing the UNEMP 20 Testing the WGGR 20 Final result and Discussion 22 Final regression model 22 Discussion 22 CONCLUSION AND POLICY IMPLICATION 24 REFERENCES 25 ABSTRACT In recent years, on social media, there has been a country that are “famous for” the word Inflation – Venezuela From one of the wealthiest country on earth, with inflation, Venezuela is immersed in poverty, violence and illness So is inflation really a problem for the economy? According to economists and the prestigious website Investopedia.com, Inflation is the decline of purchasing power of a given currency over time More specifically, a quantitative estimate of the rate at which the decline in purchasing power occurs can be reflected in the increase of an average price level of a basket of selected goods and services in an economy over some period of time We need to clearly understand that the increase of the average price level here is the general increase of goods and services, not just a particular kind of good When the value of goods increases, it means that the purchasing power of money decline So that with the same amount of money, customers now can buy less than before Moreover, in case of relationship with other economies, inflation can be defined as the devaluation of a currency compared to others However, the devaluation of the money doesn’t always mean that inflation has negative impacts on the economy Otherwise, the negative influences of inflation only occur when the governments are unable to control the increase intensity of it In this case, when inflation is not measured and adjusted, it can lead to many negative impacts on the national economy and even affect other economies For example: trade balance instability, push costs, shoe-leather costs or hoarding wealth Conversely, if inflation is only moderate, it will cause the labor market to reach equilibrium faster, ensuring discount rates and rediscounts in the money and money markets main At the same time, moderate inflation also helps goods and services markets avoid the jagged pattern of price fluctuations In fact, in addition to the above three factors, there are also many factors affecting the inflation rate (based on Consumer price index - CPI) According to Keynes, besides the Consumer price index - CPI, cost-push or adaptive expectation are also two factors leading to inflation As for push costs, it can be simply understood that: When a government cuts taxes or increases recurrent consumption spending, budget deficits and currency devaluation that incur inflation taxes will increase the cost of raw materials Then, the increasing cost of raw materials leads to the bankruptcy of businesses, reducing total supply (potential output) Along with adaptive expectations, inflation is understood as the inherent state of the economy If workers try to keep their wages at the price (above the inflation rate), firms will pass on this higher cost of labor to the customer - through an increase in the prices of goods and services This leads to a loop of repeating, causing inflation However, when building the model, we find that three factors: Consumer price index, Unemployment rate, and Percent change in average weekly earnings greatly affect the changes of inflation rate According to William Philips, there is an inverse relationship between the unemployment rate and inflation, through the intermediate factor is the level of food That means if the unemployment rate is low, the economy must create more jobs, businesses expand production and total output increases That also means accepting high inflation and vice versa Therefore, we choose inflation and the factors affecting inflation (Consumer price index - CPI, Unemployment rate, Percent change in average weekly earnings) as the topic for our econometrics report THEOREOTICAL BASIS OF THE RESEARCH TOPIC 1.1 The relationship between inflation, wages and unemployment The relationship between these three quantities is a two-way relationship and the study focuses on the effects of wages on inflation and unemployment The author admits that it is impossible to have a perfect relationship between two quantities (one quantity is only affected by the other and vice versa) There are many factors other than wages that influence inflation but certainly the rate of change in wages has a strong effect on inflation In addition, the author does not deny the impact of unemployment on inflation, but it is a two-way relationship and both are affected by many other factors 1.1.1 In short-term A.W.Phillips was one of the first economists to seek to demonstrate a negative correlation between inflation and unemployment Philips has studied extensively on the relationship between the inflation rate and the UK unemployment rate for almost a century (from 1861 to 1957) Eventually he discovered that there was a trade-off between the two The trade-off of unemployment for wages His assumption is that the demand for resources increases, labor becomes scarce, businesses will quickly offer higher wages to attract workers However, when the demand for resources decreases and unemployment increases, workers will be reluctant to accept a salary lower than what they deserve So the rate of wage growth will gradually decrease The trade-off of wages to unemployment The second problem that affects the change in wage growth rate is the change in the unemployment rate When the economy thrives, businesses reap great profits, they are willing to pay generous wages in hiring labor This greatly increases the labor supply, and the unemployment rate then dropped rapidly On the contrary, when the business is not doing well, the salary of employees does not increase or increase very slowly, the demand for labor decreases, the unemployment rate becomes highly The above assumption by Phillips describes a non-linear relationship between unemployment and wage inflation The curves that form the relationship between the unemployment rate and the overall rate of price inflation (or rather wage inflation) have made the Phillips curve famous The wage inflation data Phillips uses can also be descriptive for general price inflation Because wages are also year in the production costs of the business Raising wages will result in the price of goods and services, and this is also the most basic definition of inflation 1.1.2 Long term In the late 1960s, a group of economists representing the money major, typically Milton Friedman and Edmund Phelps, gave sharp analysis and criticism that the Phillips curve could not be applied in the long run term In the long run, unemployment will return The natural adjustment mechanism of the market will return unemployment, this period is called by Paul Samuelson in the period of stagflation In the short term, wage increases will attract more workers At this point, the supply of labor will become plentiful, leading to the unemployment rate starting to decline However, workers will gradually find that their wage purchasing power is reduced due to inflation, and they will offer a higher salary to keep wages up to date The supply of labor thus began to narrow while wage inflation and general price inflation continued to rise, or even faster than before Increasing inflation to reduce unemployment is, in the long run, not conducive to the economy Similarly, reducing the inflation rate does not cause the unemployment rate to rise Since inflation does not affect the long-run unemployment rate, the Phillips curve becomes a vertical line when it intersects the horizontal axis at the value of the natural unemployment rate 1.1.3 Conclusion The negative correlation between inflation and unemployment in the Phillips curve can only describe the economy in the short run, especially when the inflation rate is in a steady state It cannot be applied in the long run, because the market mechanism will adjust the unemployment rate to its natural rate The Phillips curve is not a key for the economy, nor can it be applied to “fight against” the unemployment of a country, it can even harm to the economy 1.2 The relationship between inflation and CPI Economists often use two indicators to evaluate the inflation rate of the economy: the consumer price index CPI and the total domestic product deflator Here, the CPI represents the fluctuation of a general price level for a basket of consumer endconsumption goods and services However, according to Investopedia, CPI is not a perfect indicator to measure inflation because of the following factors: • The quality of goods increases, the price also increases, but the CPI sees that increase as inflation • The appearance of new goods makes it difficult to compare with old substitutes • Consumers have many alternatives to where they will buy, but the CPI surveys not take these into account MODEL SPECIFICATION TESTING THE INFLUENCE OF UNEMPLOYMENT, CPI, AVERAGE WEEKY EARNINGS ON INFLATION OF UNITED STATE DURING 1966-2012 2.1 Methodology in the study 2.1.1 Method to derive the model The process using in the research is called Multiple Linear Regression This is a linear approach to modeling the statistical relatonship of a dependent variable on one or more explanatory variables 2.1.2 Method to collect and analyze the data 2.1.2.1 Collect the data Collected data are secondary data, in form of Time Series, showing the numerical information of some factors of United State in 47 years from 1966 to 2012 The data was collected base on the data set 2-3 of the book “Introductory Econometrics With Applications by Ramanathan” from the source Economic Report of the President 1995 and 2012 govinfo.gov, which has a very high level of accuracy 2.1.2.2 Analyze the data Our group has used Stata to analyze the dataset and interpret the correlation matrix between variables INFL= f(CPI, UNEMP, WGGR) Where: • INFL: Percent change in CPI (inflation rate) • CPI: Consumer Price Index • UNEMP: Civilian unemployment rate (%) • WGGR: Percent change in average weekly earnings, in current dollars Thus, according to the economic theories, in order to analyze the factors influencing the Inflation rate, our group has discussed and decided to choose the regression analysis models 2.2 Population Regression Model INFL=  + 2*CPI +  3*UNEMP +  4*WGGR + Ui 2.3 Explanation of variables • Dependent variable INFL: Inflation rate (%) Inflation happens when the purchasing power of a given currency over time decreases The increase of an average price level of a basket of selected goods and services can reflect the decline in purchasing power and receive the table below: Table Summary Statistics, using the observations 1966 – 2012 Variable Obs Mean Std.Dev Min Max infl 47 4.353191 2.808748 -0.4 13.5 cpi 47 123.7377 62.32767 32.4 229.594 unemp 47 6.148936 1.669401 3.5 9.7 wggr 47 4.104255 1.852021 1.5 8.5 (Source: the team synthesized under the support of Stata software) According to the statistic table above, we can see that almost all the Standard Deviation (Std.Dev) is much smaller than the mean, the value of maximum and minimum is quite far, this means the statistic distribution is absolutely large It happens since the data is updated each year and collected in long period On the other hand, these data are sensitive to other factors such as society, economy, politics, Due to this, the reflection of changing in the dependent variable may not get high confident 2.4.3 Correlation matrix between variables The correlation coefficient measures the strength and directioin of a linear relationship between two variables on a scatter plot In STATA, the correlation with matrix is generated by using the command corr infl cpi unemp wggr to check the correlation between the variables, we have the result as the table below: Table Correlation matrix infl cpi unemp infl 1.0000 cpi -0.5597 1.0000 unemp 0.0740 0.1800 1.0000 wggr 0.7194 -0.6443 -0.0279 wggr 1.0000 (Source: the team synthesized under the support of Stata software) According to the matrix, it can be inferred 11 • The correlation between the independent and dependent variables: The correlation coefficient between INFL and CPI is -0.5597, which is negative and pretty high It means that cpi negatively affects infl, any changes in the cpi will lead to an inverted change in the inflation rate (percent change in cpi) The correlation coefficient between INFL and UNEMP is 0.0740, which is positive and relatively low It means that unemployment rate positively affects inflation rate, any changes in the unemployment rate will lead to a slightly covarieted change in the inflation rate The correlation coefficient between INFL and WGGR is 0.7194, which is positive and relatively high.This is also the highest coefficient, showing that WGGR has the strongest impact on INFL It means that percent change in average weekly earnings positively affects inflation rate, any changes in the percent change in average weekly earnings will lead to a significant covarieted change in the inflation rate • The correlation between the independent variables: The correlation coefficient between WGGR and UNEMP is -0.0279, which is negative and relatively low It means that unemployment rate negatively affects percent change in average weekly earnings, any changes in the unemployment rate will lead to a slightly inverted change in the percent change in average weekly earnings It is explained by A.W.Phillips Phillips hypothesized that when demand for labor is high and there are few unemployed workers, employers can be expected to bid wages up quite rapidly However, when demand for labor is low, and unemployment is high, workers are reluctant to accept lower wages than the prevailing rate, and as a result, wage rates fall very slowly A second factor that affects wage rate changes is the rate of change in unemployment If business is booming, employers will bid more vigorously for workers, which means that demand for labor is increasing at a fast pace (i.e., percentage unemployment is decreasing rapidly), than they would if the demand for labor were either not increasing (e.g., percentage unemployment is unchanging) or only increasing at a slow pace In addition, all correlation coefficients of all variables are smaller than 0.8 Therefore, this model does not have multicollinearity problem 12 ESTIMATED MODEL AND STATISTICAL INFERENCE 3.1 Estimated model and estimated result In Stata, using the command reg infl cpi unemp wggr we have the result as table below: Table Estimated result based on OLS method F(3,43) = 17.42 P – value (>F) = 0.0000 R2 = 54.86% Variables Coefficient Standard T P > |t| Confident interval (95%) Error cpi -0.0089905 0.0061778 -1.46 0.153 -0.0214492 0.0034682 unemp 0.2129014 0.1764592 1.21 0.234 -0.1429626 0.5687654 wggr 0.9014218 0.204589 4.41 0.000 0.4888288 1.314015 constant 0.4568732 1.67851 0.27 0.787 -2.928166 3.841912 (Source: the team synthesized under the support of Stata software) According to the estimated result from Stata using the Ordinary Least Squares (OLS) method, we obtained the Sample Regression Functioin (SRF) as below: INFL= 0.4568732 - 0.0089905*CPI + 0.2129014*UNEMP + 0.9014218 *WGGR 13 3.2 Testing the level of relevance of the model • We have: R2 = 0.5486 → The level of relevance of the model is 54.86%: The variations of the cpi, unemp, wggr variables explain 54.86% of the average variation of the infl dependent variable • The F-test Given the hypothesis is: {H0: 2 = 3 = 4 = H1 : 22 + 32 + 42  0} The critical F-value: F 4−1 𝑘−1 = F = 2.84 =F 47 − 43 𝑛−𝑘 The formula to find test statistic: Fs 𝑅 (𝑛−𝑘 ) Fs = (𝑘−1)(1−𝑅2)= 0.5486 x(47−4) (4−1)(1−0.5486 ) = 17.4197312 We can easily see that Fs = 17.419731 > F = 2.84 → Reject H0, accept H1 → The overall model is statistically significant at a significance level of 5% • The P-value approach We use the command test cpi unemp wggr and receive the table below Table Testing the relevance of the model using P-value approach (1) cpi = (2) unemp = (3) wggr = F (3, 43) = 17.42 Prob > F = 0.0000 (Source: the team synthesized under the support of Stata software) According to the result, we have the P-value = 0.0000 < 0.05 As a result, we can reject H0 Therefore, the overall model is statistically significant at a significance level of 5% 14 In conclusion, by comparing two methods, we can conclude that the overal model is statistically fitted at the significance level of 5% 3.3 Testing the hypothetical violations 3.3.1 Testing Omit variable Ramsey Reset With the significance level of 5%, we conduct the test: Given that the hypothesis is: {H0: The model does not omit variable H1: The model omits variable} Using the command “estat ovtest” in the Stata software, we receive the table below: Table Ramsey Reset test result Ramsey RESET test using powers of the fitted values of infl Ho: model has no omitted variables F(3, 40) = 0.89 Prob > F = 0.4554 (Source: the team synthesized under the support of Stata software) According to the result, we have the P-value (Fs) = 0.4554 >  = 0.05 → Not Reject H0 → At the 5% significant level, the model does not suffer from omitted variables bias 3.3.2 Testing Multicollinearity • First method: Use the variance increment factor VIF With the significance level of 5%, we conduct the test: Given that the hypothesis is: {H0: The model does not have multicollinearity H1: The model has multicollinearity} 15 Using the command “vif” in the Stata software, we receive the table below: Table Multicollinearity test result by VIF Variable VIF 1/VIF cpi 1.79 0.558567 wggr 1.73 0.576829 unemp 1.05 0.954318 Mean VIF 1.52 (Source: the team synthesized under the support of Stata software) According to the result, we have: VIFcpi = 1.79 VIFwggr = 1.73 VIFunemp = 1.05 Comparing, we easily see that VIFcpi = 1.79 < 10; VIFwggr = 1.73 < 10; VIFunemp = 1.05 1.96 → Not reject H0 → At the 5% significant level, the model does not have multicollinearity 16 3.3.3 Heteroscedasticity With the significance level of 5%, we conduct the test: Use White test to examine heteroscedasticity Hypothesis {H0: The model does not have heteroscedasticity H1: The model has heteroscedasticity } Use the command “imtest, white” in Stata, we receive the table below: Table Heteroscedasticity result by White test White's test for Ho: homoscedasticity against Ha: unrestricted heteroscedasticity chi2(9) = 14.44 Prob > chi2 = 0.1074 (Source: the team synthesized under the support of Stata software) We can see that Prob > chi2 = 0.1074 >  = 0.05 → Not reject H0 → Heteroscedasticity does not occur 3.3.4 Testing autocorrelation With the significance level of 5%, we conduct the test: Use the Breusch-Godfrey test to examine autocorrelation Given that the hypothesis is: {H0: The model does not have autocorrelation H1: The model has autocorrelation } Use the command “estat bgodfrey” in Stata, we receive the table below: Table Autocorrelation result by Breusch-Godfrey Breusch-Godfrey LM test for autocorrelation lags(p) chi2 df Prob > chi2 17 7.220 0.0072 (Source: the team synthesized under the support of Stata software) We can see that Prob > chi2 = 0.0072 <  = 0.05 → Reject H0 → At the 5% significant level, the model has autocorrelation • Correcting Autocorrelation To fix the problem, using Robust Use the command “reg infl cpi unemp wggr, robust”, we receive the table below: Table Regression model using robust standard errors F(3,43) = 10.04 P – value (>F) = 0.0000 R2 = 54.86% Variables Coefficient Standard T P > |t| Confident interval (95%) Error cpi -0.0089905 0.0040094 -2.24 0.030 -0.0170761 -0.0009049 unemp 0.2129014 0.1544124 1.38 0.175 -0.0985009 0.5243037 wggr 0.9014218 0.2386446 3.78 0.000 0.4201492 1.382694 constant 0.4568732 1.405421 0.33 0.747 -2.377428 3.291174 (Source: the team synthesized under the support of Stata software) Note that comparing the results with the earlier regression, none of the coefficient estimates changed, but the standard errors and hence the t values are different, which gives reasonable more accurate p values 18 3.3.5 Testing normality of residual In statistics, normality tests are used to determine if a data set is well-modeled by a normal distribution and to compute how likely it is for a random variable underlying the data set to be normally distributed Given that the hypothesis is: {H0: The model has normality H1: The model does not have normality} → May normally distributed Use the command “sktest r”, we receive the table below: Table 10 Testing normality of residual Variable Obs Pr(Skewness) Pr(Kurtosis) Adj chi292) Prob>chi2 r 47 0.0976 0.0103 8.11 0.0174 (Source: the team synthesized under the support of Stata software) We see: P-value = 0.0174 <  = 0.05 → Reject H0 19 → At the 5% significant level, the model has normality 3.4 Testing an individual regression coefficient 3.4.1 Testing the CPI With the significance level of 5%, we conduct the test: Given that the hypothesis is: {H0: 2 = H1 : 2  0} From Table 1, we have p-value = 0.030 <  = 0.05 → Reject H0 → The coefficient 2 is statistically significant at the 5% significance level → The Consumer Price Index has effect on the inflation rate 3.4.2 Testing the UNEMP With the significance level of 5%, we conduct the test: Given that the hypothesis is: {H0: 3 = H1 : 3  0} From Table 1, we have p-value = 0.234 >  = 0.05 → Not reject H0 → The coefficient 3 is not statistically significant at the 5% significance level → The Civillian Unemployment Rate has no effect on the inflation rate 3.4.3 Testing the WGGR With the significance level of 5%, we conduct the test: Given that the hypothesis is: {H0: 4 = H1 : 4  0} From Table 1, we have p-value = 0.000 <  = 0.05 → Reject H0 → The coefficient 4 is statistically significant at the 5% significance level → The Percent change in average weekly earnings has effect on the inflation rate 20 Meanings of estimated coefficients: Regression coefficient Sign of reression Meaning coefficient 2 = -0.0089905 0 Estimation of 3, on the condition that the other elements are constant, when UNEMP increases unit, INFL will increase 0.2129014 4 = 0.9014218 >0 Estimation of 4, on the condition that the other elements are constant, when WGGR increases unit, INFL will increase 0.9014218 21 3.5 Final result and Discussion 3.5.1 Final regression model INFL= 0.4568732 - 0.0089905*CPI + 0.2129014*UNEMP + 0.9014218 *WGGR 3.5.2 Discussion Based on the regression model and testing hypothesis, we can get a conclusion: • On the condition that the other elements are constant, when WGGR increases unit, INFL will increase 0.9014218 • On the condition that the other elements are constant, when CPI increases unit, INFL will increase 0.0089905 • The change in Unemployment rate does not affect the Inflation rate • The model explains 54,86% the change of Inflation rate The estimation and testing hypothesis shows that the unemployment rate is not meaningful in model or has no impact on the inflation rate This is not reasonable with the theory of the economist William Phillips with the argument that “inflation rate and unemployment have negative relationship” and be illustrated by the Phillips curve This contradictory happens due to further researches have proved that there is no tradeoff between inflation and unemployment in long-run, the arguments of Phillips just show the effect in short-run Specifically, In the last 1960s, a group of economists typical Milton Friedman and Edmund Phelps did some researches and separate The Long-run Phillips curve and The Short-run Phillips curve 22 According to this, in the short-run, raising wage will attract more labor, which make the labor force increases and more plentiful, since reduces the unemployment rate However, paying high wages means that the operation cost of company increase hence the price of product also goes up to ensure the profit During period, employees will recognize that their purchasing power decline and they will require employers to pay even higher salary, if not they are willing to be unemployed to look for a higher-paid job, so the unemployment rate increases In the long-run, the expected inflation rate will come near to the real one, at the same time the unemployment rate will back the natural one In conclusion, some policies which increase inflation rate to decline unemployment rate are just effective in shortrun In long-run, the unemployment rate is unchanged at the natural rate 23 CONCLUSION AND POLICY IMPLICATION Our team has just researched factors affecting inflation rate based on the theories of A.W.Phillips, Milton Friedman and Edmund Phelps After testing, we can see that the model which is built based on the data set 2-3 of Ramanathan has several hypothetical violations such as Heteroskedasticity and Autocorrelation They are fixed by robust and Prais-Winsten respectively However, we cannot conclude that this is the suitable model since the number of observations is really small (Obs: 47) Beside, we can realize that the unemployment rate does not affect much on inflation rate After the World War II, many countries deep into the condition that both inflation and unemployment rate are high However, unemployment rate is a signal of the recession or stagnation of the economy That is when the total product and demand for goods & services decreases, firms decide to reduce manufacturing Obviously, the stagnation period is not correlative to the theory about negative relationship between inflation and unemployment rate Since these disadvantages, our team conclude: This model still have some contraints and may not reflect the most objective evaluations about the effect of cpi, unemployment rate, average weekly wage on the inflation rate However, by using regression model and testing hypothesis, we partly understand the way how the politicians evaluate the inflation rate When inflation rate increases rapidly and lasts for a long time, it will make a lot of consequences, affect deeply to the life of households and the growth of economy Since that, to reduce and restrict inflation, we need not only immediate solution but also the long-term strategy 24 REFERENCES • Investopedia.com • The data set 2-3 of Ramanathan form the source of Economic Report for President govinfo.gov • The links between wages, inflation and unemployment (Nevile,1983) • The relationship between inflation and unemployment - A.W.Phillips • https://www.bbc.com/vietnamese/business-46823162 • The relationship the “Phillips curve" – Paul Samuelson and Robert Solow • The Natural Rate of Unemployment Reflections on 25 years of the Hypothesis by Rod Cross 25 ... inflation The curves that form the relationship between the unemployment rate and the overall rate of price inflation (or rather wage inflation) have made the Phillips curve famous The wage inflation. .. item in the predetermined basket of goods and averaging them Changes in the CPI are used to assess price changes associated with the cost of living The CPI is one of the most frequently used... versa) There are many factors other than wages that influence inflation but certainly the rate of change in wages has a strong effect on inflation In addition, the author does not deny the impact of