The impact of cryptocurrencies on selected macroeconomic variables of Turkey: Bitcoin case

In this study, the relationship between Bitcoin and other variables in the model was investigated with the data obtained from the monthly percentage changes of Bitcoin, USD-TRY (Dollar - TL exchange rate), GAU-TRY (Gold - TL exchange rate), BIST 100, Brent Oil, unemployment rates, one-week repo rates, consumer price index, producer price index and industrial production index variables between January 2011 and December 2023 in Turkey. In this context, the stationarity tests of the variables obtained in this context were carried out by Augmented Dickey Fuller, Phillips-Perron and KPSS Unit Root Test. After determining the appropriate lag length for the VAR model, which is planned to be constructed with the variables that are determined to be stationary at 99% confidence level at the level values, the VAR model was constructed. The unit root test of the VAR model created in the following bet was performed and it was concluded that the model was stationary. In the study, it was observed that the interactions of the Bitcoin variable with other variables in the model differed in the positions of being an explained variable and being an explanatory variable within the relevant period. As a result of the VAR model in which BTC variable is considered as the dependent variable and 1 lagged values of other variables in the model are considered as independent variables; With monthly data for the period January 2011 - December 2023, a positive relationship was found between the BTC dependent variable and USD-TRY(-1), BRENT(-1), REPO(-1), CPI(-1) independent variables, and a negative relationship was found between GAU-TRY(-1), BIST100(-1), UNEMP(-1), D-PPI(-1) and PPI(-1) independent variables at 1 percent significance level. In addition, it has been concluded that the most important variables that have a positive effect on the change in Bitcoin prices in Turkey between January 2011 - December 2023 are the CPI variable with a coefficient value of 1.86 and the USD-TL exchange rate variable with a coefficient value of 0.77. This situation shows that with the inflation and exchange rate increases that will take place in our country in the following periods, serious increases will occur in Bitcoin prices due to the direct relationship; This increase in Bitcoin prices can be suppressed with successful monetary and fiscal policies to be implemented within the scope of combating inflation. Therefore, evaluating Bitcoin together with other investment instruments in periods of high expected inflation rates may allow the decrease in purchasing power due to inflation to be absorbed. In this context, when the results of the analysis in the relevant model are analysed, it is seen that the effect of the CPI variable on BTC has a higher value compared to other investment instruments. Therefore, it can be predicted that possible investments in Bitcoin during periods of CPI increase in our country may be more profitable than the investment instruments in the model (Dollar, Gold, BIST100, REPO). However, the fact that other variables were not included in the model in the study and Bitcoin has been in a continuous upward trend since the first day it emerged, and the continuous increase in inflation rates during this time period may have paved the way for the emergence of such a relationship between the variables. A one unit increase/decrease in the CPI(-1) variable causes a 1.861981 unit increase/decrease in the BTC variable. This means that an increase in inflation also increases the price of Bitcoin. Some investors may turn to alternative assets such as Bitcoin in order to prevent their money from losing value during inflation periods. Therefore, as inflation increases, the increase in demand for Bitcoin may contribute to the increase in Bitcoin prices, in other words, to the increase in Bitcoin prices. However, although the model shows the relationship between the CPI(-1) variable and BTC, this does not necessarily imply causality. The result is a finding obtained only within the scope of the data within the scope of the study and the model applied. In this context, it would not be a very correct statement to say that the increase in CPI directly increases the price of Bitcoin, but the findings obtained are expressed in this way. A one unit increase/decrease in the USD-TRY(-1) variable will lead to a 0.777810 unit increase/decrease in the BTC variable. Therefore, it is concluded that the appreciation of the dollar against TL will generally lead to an increase in BTC prices. Since the appreciation of the dollar is generally interpreted as a decrease in risk appetite in global markets, investors can increase the price of Bitcoin by turning to Bitcoin during uncertain periods. In addition, since the dollar is used as a reserve currency worldwide, movements in the dollar exchange rate can affect Bitcoin prices as well as the prices of many financial assets. In this context, it may be a rational decision for Turkish investors to turn to alternative investment instruments such as the dollar or Bitcoin in the face of the depreciation of the TL. However, it should not be forgotten that there are many factors affecting the price of Bitcoin in real life (US interest rate decisions, global economic developments, speculative movements and manipulative news, legal regulations on cryptocurrency, etc.). In line with the variance decomposition analysis results, a deviation of 0.192528 in the BTC variable as of the tenth period is explained by the BTC variable itself by 90%, while 2. 41% is explained by D-PPI, 1.839384% by GAU-TRY, 1.838942% by BRENT, 1.544308% by REPO, 1.230593% by USD-TRY, 0.52% by UNEMP, 0.11% by CPI, 0.027241% by BIST100, 0.021936% by IPI variables. Therefore, it is seen that the BTC variable is mostly under the influence of its own lagged values and the effects of other variables on the changes in the BTC variable as of the periods have very close values to each other. When the variance decomposition results of other variables for the VAR model are analysed in the following section, it is concluded that the BTC variable is the most exogenous variable in the VAR model since 100% of the change is covered by the BTC variable. In this context, the fact that the entire variance of the BTC variable in the first period is explained by itself shows that the most important factor affecting the price movements of BTC in that period is its own movements in the past, that the other variables in the model do not play an important role in the price movements of BTC in that period, and that BTC acts independently of other selected variables in the relevant period or is affected very little. The fact that the variance of the Bitcoin variable decreased to 0.90 levels as of the periods reveals that it is in a relationship with other selected macroeconomic variables and that it is not sufficient to use only the historical price data of Bitcoin in order to make long-term BTC price forecasts, the effects of other factors should also be taken into consideration. Although it may be sufficient to use Bitcoin's historical price data to predict BTC price movements in the short term, other variables should not be ignored in realising long-term forecasts. While the reaction of the Bitcoin variable to a one standard error shock in the Bitcoin variable was positive in the first four periods, it followed a negative course in the fifth period. However, it is concluded that the BTC variable exhibited the largest positive reaction in the first period and the largest negative reaction in the seventh period against a one standard error shock in the BTC variable series. In line with all this information, when the shocks within Bitcoin are analysed, it is observed that there is a positive reaction in the short term and a negative reaction in the long term. This situation may reflect the volatility of the Bitcoin market and the short-term earnings expectations of investors. Since the largest positive reaction occurs in the first period, while the largest negative reaction is observed in later periods, it is obvious that Bitcoin prices show different sensitivities to shocks at different times. When the analysis continues, it is seen that the shocks in all variables do not affect BTC in the first period, indicating that the Bitcoin market can act independently of other markets in the short term. In addition, the fact that shocks in other variables affect BTC in different directions in different periods shows that economic conditions and market sentiment are constantly changing and this situation affects Bitcoin prices. The fact that shocks in all variables except Bitcoin affect Bitcoin prices negatively in the long run may reflect the correlation of Bitcoin with other assets and its perception as a risky financial asset instrument. According to the results of VAR Granger Causality / Block Exogeneity Wald Test and Pairwise Granger Causality analyses, it is concluded that there is unidirectional causality from the Industrial Production Index variable to the unemployment variable, unidirectional causality from the USD-TRY variable to the REPO variable, and unidirectional causality from the BIST100 variable to the industrial production index variable. In addition, as a result of VAR Granger Causality / Block Exogeneity Wald test, a unidirectional causality relationship was found from the Industrial Production Index to the CPI variable, from the USD-TRY variable to the CPI and D-PPI variables, from the BIST100 variable to the CPI variable, from the D-PPI variable to the CPI variable, from the BRENT variable to the IPI variable, while a bidirectional causality relationship was found between the IPI and D-PPI variables. With the Pairwise Granger Causality Test, a unidirectional causality relationship was detected from USD-TRY and GAU-TRY variables to BRENT variable, from BIST100 variable to CPI variable, from GAU-TRY variable to BIST100 and REPO variables, and from D-PPI variable to BIST100 variable Key Words: Cryptography, Cryptocurrency, Bitcoin, Altcoin, Blockchain Technology, Inflation and Bitcoin, Financial Investment Preference, Time Series Analysis, VAR model, ANOVA, Impulse-Response Analysis, VAR Granger Causality / Block Externality Wald Test, Pairwise Granger Causality Test