New EU members have no chance to catch up the most developed countries

 In 2005, we published a paper quantitatively describing the transition of the former socialist countries to a new state similar to capitalism.  In 2009, we extended our model and predicted the evolution of real GDP per capita in 27 countries, including two virtual countries: the U.S.S.R and Czechoslovakia. Briefly, despite different social, cultural, ethnical, racial, demographical, religious, and technological histories our model demonstrates general success in the overall prediction between the start point (1989 to 1991) and the year of 2006(7). In all cases, the most accurate prediction is associated with the initial segment of the transition, when socialism was dynamically replaced by capitalism. The initial stage is characterized by the largest changes in the GDP per capita, and thus, provides a wide dynamic range as a crucial condition for accurate modelling. 

At later stages, some exogenous forces, such as economic recession, might disturb the agreement between measured and predicted GDP, and introduce some bias in the estimation of the defining parameters. This later stage, however, is of lower relevance to the transition itself and rather demonstrates the behavior of a regular capitalist economy.

Here, we revisit our old predictions and compare them with actual observations with the real GDP per capita estimates borrowed from the Maddison Project Database. In 2009, we concluded that the transition process has effectively finished in Central and East European countries. Thus, the long-term growth rate of GDP per capita in these countries is limited by the attained level of the GDP per capita, as in developed countries (see this post). Any deviations from the long-term rate can be explained only by inefficient economic performance. In fact, many developed countries are characterized by lower values of an economic trend than that in the U.S. at the same level of GDP per capita. We assume that the future economic growth in the FSC will follow either the potential or current value of the annual GDPpc increment. When no inefficiency is allowed, the best-case scenario of economic growth is realized.

Figure 1 displays Figure 10 from the 2009 paper - the past and future evolution in Hungary and Poland compared to that the U.S. All other FSC have time histories similar to those in Hungary and Poland, but with different initial values and actual slopes. In the current study, all theoretical slopes are equal: A=$370, i.e. annual increment in GDP per capita is $370 (1990 $).

The U.S. theoretical curve with A=$370 is much higher. Before 2008, it had no prominent horizontal or downward segments, as observed in Hungary and Poland. In 2008, the recession moved the US GDPpc curve down and it was below the theoretical curve as Figure 2 shows. In Poland, the stagnation process started in the late 1970s. Hungary is characterized by a shorter period of weak performance. Apparently, the years between 1950 and 1989 are characterized by a lower rate of economic growth with an increasing gap between the FSC and the U.S. The best-case scenario would allow maintaining a constant lag behind the U.S., X. In relative terms, the lag will decrease as a function G_FSC/(X+G_FSC), where G_FSC is the GDP per capita in a given FSC.

Figure 2 compares the predicted and observed GDPpc curves in Hungary. The MPD data (2011 prices) are used instead of the TED. All MPD readings are reduced to 1990$. The drop in the US curve was observed in 2008. The upper panel of Figure 2 shows the 2016 revision with TED data and the lower panel uses the MPD data. There is some discrepancy in the GDP per capita estimates in 2016 and 2020. The observations show that the economic performance in Hungary has been approaching its perfect trajectory with A=$370 (1990$) since 2012. Figure 3 presents the case of Poland, which shows a better performance likely related to the EU financial support. Figure 4 is for Bulgaria.

Overall, our prediction is still valid – the new EU countries lag behind the US and this lag will not be closed any time soon.

  

Figure 1. Upper panel: The evolution of per capita GDP in Hungary compared to that in the U.S. The gap between countries has been increasing since 1950. In the ideal case, when Hungary performs at the theoretical level (A=$370), it will be able to retain the gap. The absolute gap between the U.S. and Hungary can theoretically be constant at the level of $22,000 in the future. Lower panel:  Same as in the upper panel for Poland. The absolute gap between the U.S. and Poland is currently $22,300.

Figure 2. Comparison of predicted and observed GDPpc curves in Hungary. The MPD data (2011 prices) are reduced to 1990$. The drop in the US curve was observed in 2008. The upper panel shows the 2016 revision with TED data and the lower panel uses the MPD data. There is some discrepancy in the GDP per capita estimates in 2016 and 2020. The observations show that the economic performance in Hungary has been approaching its perfect trajectory since 2012.

 

Figure 3. Same as in Figure 2 for Poland

Figure 4. Same as in Figure 2 for Bulgaria

Economic comparison: Ukraine vs Russia. Real GDP per head in Ukraine is still below the 1989 level. Russia doubled the GDP per head

 In 2014, we compared the evolution of real GDP per head in Russia and Ukraine since 1980. The Total Economy Database (TED) of the Conference Board was used. In this post, we revisit our estimates from 2014 with a new dataset – now TED is replaced by the Maddison Project Database (MPD). Three figures below present different sides of the relative evolution. The upper figure shows the absolute GDP per capita values since 1980, and the Russia curve is still much above Ukraine. In 2018, Russia had $24,669 (2011 prices) and Ukraine only $9,813. The growth is absent in Ukraine since 2008 and the level observed in 1989 ($10,082) has to be reached in the future.  The middle (both curves normalized to 1980) and lower (both curves normalized to 1991) figures show that the real GDP per capita in Russia is by a factor of 2.05 larger in 2018 than in 1991 and by a factor of 2.01 compared to 1980. The actual problem is that this lag is forever. 

 

Real GDP is likely not correct as the inflation estimates are weird

 We revisited several times our original post published in 2011 on biased metrology of macroeconomic measurements. New data are needed to validate the original hypothesis or to reject it. It is important to retain in mind that economics as science often fails and gives counterproductive results due to inaccurate or biased measurements of the most basic macroeconomics variables – price inflation, labor force, unemployment, nominal GDP. In one of our previous posts, we focused on the estimates of real GDP in the USA before and after 1979 and here we extend the US case by several other cases. As mentioned before, we have devoted enough efforts to reveal and recover many trivial cases in our book “mecħanomics. Economic as Classical Mechanics”.

Real GDP (see Concepts and Methods of the U.S. NIPA for details) is the difference between nominal GDP and GDP deflator (price index). The latter is not easy to calculate or even evaluate.  In this post, we showed that it is so much a sophisticated problem that before 1980 there was no practical difference between the cumulative inflation values of the CPI and the GDP deflator in the US, as originally was demonstrated in Figure 1 of the 2011 post. (The cumulative inflation, i.e. the cumulative sum of inflation rates, is different from the price index when differently calibrated in the beginning.)

In this post, we are trying to find significant breaks in the linear dependence between the CPI and dGDP (i.e. between two measured time series) as related to new definitions of inflation. It is important that our observation of a linear relationship between CPI and dGDP in the USA is also valid for other studied developed countries. For our model of the linear and lagged relationship between labor force, unemployment, and inflation, such definitional breaks in economic parameters are equivalent to the breaks in the statistically estimated relationship. In other words, the breaks in the general relationship are not related to the change in the economic behavior of the involved parameters. These breaks are fully artificial and induced by major economic agencies (BLS, BEA, etc.) on their eternal way to perfect definitions of economic variables.

Figure 1 presents the case of the USA (here we use the OECD data) : upper panel - CPI and dGDP inflation rates since 1971; middle panel – price change with time relative to 1970 (both variables are normalized to their respective 1970 levels); lower panel – the difference between the curves in the middle panel.  We have changed the reference year for the USA from 1929 to 1970. Effectively, the CPI and dGDP curves in Figure 1 diverge from 1978. As we have already mentioned many times, before 1978 the CPI was used to estimate of the overall price inflation. Since 1978, the GDP deflator has been used. The difference between these two variables cannot be neglected: the cumulative price change between 1978 and 2018 is 1.3 units or 25%. The lower panel of Figure 1 demonstrates that the cumulative CPI and dGDP price change differences is very close to linear dependence. As we reported before, the coefficient of linear dependence is 1.2 for the USA data (BLS and BEA). For the OECD data, the coefficient is 1.26, i.e. another set of definitions is used by the OECD.

Figure 2 presents the same curves as in the middle panel of Figure 1, but the dGDP time-series is multiplied by a factor of 1.26. Now, the fit between the two curves is almost perfect. There is a small deviation from this linear link after 2008. It might be related to the formation of a new link after some change in dGDP definition in 2008. The potential change in the regression coefficient related to the period after 2008 is small and one has to wait for new data to estimate this coefficient statistically. In Figure 3, we present a tentative model with the dGDP additionally (to 1.26) multiplied by 0.8 after 2008. The fit is much better and the total multiplication factor compared to the pre-1978 definition is 1.008, i.e. practically 1. Are we in the pre-1978 definition era again? When analyzing the linear and lagged dependence between labor force, unemployment rate, and the inflation we have to introduce “non-structural” or artificial breaks (structural breaks are equivalent to the change in economic behavior) in 1978 and 2008. 

 

Figure 1. Upper panel: The evolution of the CPI and GDP deflator in the USA since 1970 according to OECD data.  The CPI inflation curve is higher than the GDP inflation one. Middle panel: Cumulative price change according to price indices: CPI and GDP deflator. Lower panel: The difference between curves in the upper (running) and middle (cumulative) panels.

 

Figure 2. The evolution of the cumulative inflation (the sum of annual inflation rates) for the CPI and GDP deflator in the USA since 1970. The dGDP is multiplied by a factor of 1.26.  

 

Figure 3. Same as in Figure 2, but the dGDP after 2008 is additionally multiplied by a factor of 0.8.  In fact, the total multiplication coefficient (1.26*0.8) = 1.0, i.e. the same as before 1978. 

Figure 4 presents the case of the UK. Panel a) repeats the plot published in 2011. We reported the deviation between the CPI and dGDP since 1978 as in the USA. Panel b) presents the same case with the OECD data for the period between 1955 and 2018. One can see a clear kink in the dGDP curve between 1994 and 1996, which is absent in panel a). This is likely a later OECD revision. Interestingly, the dGDP curve is above the CPI one. In the USA, the order is the opposite. The difference between the CPI can be overcome by two corrections: the dGDP is multiplied by a factor of 0.98 since 1978, as shown in panel c), and the kink is corrected by a constant  -0.055 – panel d). As a result, the overall fit since 1955 is excellent and we have to take the break in 1978 into account when assessing our model statistically.

a)

b)

c) 

d)

Figure 4. The case of the UK. Described in the text.

Figure 5 presents the case of Austria. The cumulative price change (CPI and dGDP) curves in panel a) demonstrate the start of deviation around 1990. Panel b) shown the running (inflation) and the cumulative difference between the curves in panel a). The pivot point is 1991. We have corrected both time periods – before and after 1991. The pre-1991 period has a very small multiplication factor of 1.005 and the after-1991 period is the best fit with coefficient 1.17 – panel c). The model linking the dGDP inflation with labor characteristics definitely needs a data-driven brake in 1991.

 a)

 

b) 

c)

Figure 4. The case of Austria. There is a break in 1991.

 Figure 5 presents the Kingdom of the Netherlands. The CPI curve is above the dGDP one. The relatively good fit between the curves needs three breakpoints – 1982 (coefficient 1.2), 1997 (0.8) (both changes are shown in panel c), and 2009 (2.0)  - panel d).

 a)

b)

c)

d)

Figure 5. The Kingdom of the Netherlands

In Figure 6, we present Japan. The CPI is above the dGDP – panel a). The cumulative difference in panel b) is an almost linear function of time. The multiplication factor between 1974 and 2018 is 1.25, but the CPI is constant after 1996 – panel b). This constant can be fit only with a 0 linear coefficient or a constant added to all dGDP or CPI inflation readings after 1996. In panel d) we added -0.011 to each reading of the CPI inflation since 1997 and obtained a good fit between the CPI and dGDP. We are sure that the Japan Statistics did not add 0.011 to the CPI inflation estimates in order to keep it above 0. It would be a good trick to avoid formal deflation, which is observed in the dGDP since the mid-1990s, but it would be a bad trick for the economic models we have been developing for Japan.

 a) 

b)

c) 

d)

Figure 6. Japan 

Italy is a perfect case with one constant describing the whole difference between 1970 and 2018. Figure 7 presents the original CPI and dGDP time-series – panel a), the difference – panel b) and the overall fir when the dGDP is corrected by a factor of 0.94. The overall fit is excellent through the studied interval and no breaks are needed in the model. For Italy, the CPI is below the dGDP curve. 

a)

 

b)

 

c)

Figure 7. Italy

 

 

 

On the potential non-equivalent exchange between developed countries

 I am in the middle of modeling the relationship between the labor force, unemployment, and price inflation.  Twelve years ago we reported several statistical models revealing a linear lagged relationship between these three parameters in developed (e.g., USA, UK, Japan, Germany, France, Austria) countries. It is time to revisit them and validate these models with new data published since 2010. There are several problems we have to overcome before the published data can be used in statistical estimates. The most important problem is the change in definitions of all three parameters. For example, for the USA we found that the CPI (consumer price index) and dGDP (GDP price deflator) change their relative behavior due to changes in definitions (e.g., imputed rent). Moreover, these changes are well described by a linear relationship.

On the way to the final statistical analysis of the model, I formulated an interesting idea, the first time in some implicit form in the post “Growth rate of the GDP per capita revisited. 4. Developed countries – cntd”. The statement was more qualitative, but I assumed that Germany is the principal beneficiary of the EU. It is difficult to prove a non-equivalent exchange. The annual increment in real GDP per capita is higher in Germany than in France, Italy, Spain, and the UK. This is a good argument. 

There is another possible indicator related to dGDP and CPI, however. The former is defined only by the prices of domestic goods and the latter includes prices of imported goods and services. I do not understand how the mechanism of the non-equivalent exchange works but the only big country in the EU has the CPI curve higher than the dGDP - Germany. The same configuration is observed in the USA and Japan. In France, Italy, the UK, Canada, and Australia, the dGDP curve is above the CPI one.  

 

Our prediction from 2008 on the evolution of unemployment rate in Italy at an 11-year horizon is still valid

In this blog, we regularly revisit our prediction of the rate of unemployment in Italy, which had been made in our 2008 paper. (The previous revision was in 2017.) Five years after this publication, we found that the accuracy of prediction was excellent. We decided that our model works well. Since the model has a natural 11-year horizon, we were able to check our original (2008!) prediction for 2013 and 2016 and now for 2019 using new estimates of the unemployment rate in Italy (here we use the OECD database).  According to the OECD, the unemployment rate in 2019 was 9.95%. For 2017, the rate was 11.2%.   

Our model of unemployment as a function of the change in labour force predicts two pivot points in the unemployment rate – in 2008 and 2014. We introduced the model of unemployment in Italy in 2008 with data available only for 2006. The rate of unemployment was near its bottom at the level of 6%. The model predicted long-term growth in the rate of unemployment to the level of 11% in 2013-2014. The next pivot point is expected in 2023 and the rate will start to grow again. 

The overall agreement between the measured and predicted unemployment estimates in Italy validates our concept, which states that there exists a long-term equilibrium link between unemployment, u_t, and the rate of change of labour force, l_t=dLF/LFdt. Italy is a unique economy to validate this link because the time lag of unemployment behind l_t  is eleven (!) years.  The estimation method is standard – we seek for the best overall fit between observed and predicted curves by the LSQR method. The best-fit equation for the original data obtained from the national account is as follows:

 

u_t = 5.0l_t-11  + 0.07         (1)

 

As mentioned above, the lead of l_t is eleven years. This defines the rate of unemployment many years ahead of the current change in the labour force. When the OECD data are used, the model has different coefficients:

 

u_t = 3.0l_t-11  + 0.09         (2)

 

The difference between national and international estimates of the labour force and unemployment is thoroughly discussed in our papers. We have also reported the change in definitions of GDP and CPI in several posts published this December. 

Figure 1 repeats the picture from the 2017 post and displays the observed unemployment curve and that predicted using the rate of labour force change 11 years ago and equations (1). Since the estimates of the labour force in Italy are very noisy we have smoothed the annual predicted curve with MA(5). All in all, the predictive power of the model is excellent and timely fits major peaks and troughs after 1988. The period between 2006 and 2016 was predicted almost exactly. 

In Figure 2 we add the most recent period and use the most recent version of the OECD data. The model (2) is presented in its annual and MA(5) versions. The prediction still works well and the next pivot point to increasing unemployment in Italy is expected in 2023. 

The fit between predictions and observations is the best validation of any quantitative model. We do not know any other macroeconomic model capable to describe such dramatic turns many years ahead. The evolution of the rate of unemployment in Italy is completely defined ten years ahead.  Since the linear coefficient in (2) is positive one needs to reduce the growth in the labour force in order to decrease the rate of unemployment.

 

Figure 1. Borrowed from the 2017 post. The observed and predicted rate of unemployment in Italy.

 

Figure 2.  Observed and predicted rate of unemployment in Italy updated using the OECD data between 1965 and 2019.

Information technology is still the leader of deflation

Eight years ago, I presented the absolute leader of deflation in the US – the consumer price index of information technology, hardware, and software (see Figure 1). It has been falling further since 2012 and now is reaching 7 points relative to 256 points of the headline CPI as Figure 2 demonstrates. Figure 3 depicts inflation for both indices. The information technology price index demonstrates negative inflation from the start.   This inflation approaches the zero line, however. It is important that the index itself will never be negative. it is not excluded that information technology becomes a price setter.  

Figure 1. The evolution of the headline CPI and the index of information technology, hardware, and 

software as presented in 2012.

Figure 2. The evolution of the headline CPI and the index of information technology, hardware, and software since 1988. data between 2012 and 2020 are added

Figure 3. Inflation: Headline CPI and the index of information technology, hardware, and software. The latter is always positive.