This document gives a desk no basis for trading Turkish base-load futures on moves in the lira, Brent or interest rates. The abstract promises that basis. Karataş Elçiçek describes an ARDL bounds test on monthly base-load electricity futures prices from January 2016 to July 2025. Its reported long-run effects are positive for Brent, electricity consumption and the exchange rate, and negative for interest rates.
The body we received belongs to a different paper: Abdurrahman Nazif Çatık's time-varying VAR of quarterly Turkish residential electricity prices. Its sample covers 1988Q1 to 2016Q1, and it appeared in the Romanian Journal of Economic Forecasting in 2020. The samples share one quarter.
What trade does the abstract promise?
An ARDL model relates the price level to its own lags and lagged regressors. The bounds test checks for a long-run relationship in levels, allowing a mixture of stationary and nonstationary regressors. An error-correction term measures the speed at which the price returns to that long-run anchor.
A desk could use this as a fair-value model. A weaker lira or a Brent rally raises the futures price anchor. The adjustment speed then tells the trader how long a deviation should last before fading it. Yet the abstract supplies only four signs. We found no futures-regression coefficient, bounds statistic or adjustment speed in the body. It contains no futures series either.
We could not test the claim ourselves. Our data contains neither Turkish electricity futures settlements nor Turkish consumption, inflation or interest-rate feeds. Using US energy futures would substitute an American pricing mechanism for the Turkish one. We therefore made no such substitution.
Inside the attached study
The body uses five series in log first differences: the US/TL exchange rate, high-sulphur fuel oil, natural gas, steam coal and residential electricity prices. The IEA supplies the fuel prices. Every price, including electricity, is expressed in national currency per tonne of oil equivalent. Coefficients and shock volatilities follow random walks, allowing both to change over time. Estimation uses MCMC with 100,000 draws and discards the first 10,000. Cholesky ordering identifies the shocks, placing the exchange rate first and electricity last.
Generation mix supplies the economic argument. Natural gas accounted for 17.7% of 57,543 GWh in 1990, behind hydro's 40.2% and coal's 35.1%. Its share rose to 37% in 2000 and 48.1% in 2015. About 99% of the gas is imported. If fuel costs determine electricity prices, growing reliance on gas should make its price increasingly influential.
Gas displaced the lira in household power prices
The impulse responses follow that pattern. A positive exchange-rate shock early in the sample raises electricity prices by 2.71% on impact and 11.33% cumulatively over twelve quarters. The impact response reaches its high of 5.7% in 2001Q2, alongside a cumulative response of 13.66%. By 2016Q1, those figures have fallen to 1.06% on impact and 3.10% cumulatively. The paper attributes the decline to inflation targeting.
Gas moves in the opposite direction. Its response rises from 1.36% in 2006Q1 to 5.69% at the sample's end, passing the exchange rate. Coal initially leads gas by 5.29% to 0.93%, then finishes at 2.64% accumulated.
Variance decompositions give a similar account. At a one-quarter horizon, the exchange rate accounts for 57.56% of electricity price variation around the 1994 crisis and 64.59% around 2001. By end-2007, gas accounts for 9.04%, compared with 4.28% for coal and 84.06% for the own shock. Gas subsequently reaches 29.60%. At twelve quarters, the final shares are 46.45% for the own shock, 30.17% for gas and 14.90% for the exchange rate. Oil's one-quarter share stays below 5% throughout. Its impulse response is significant only near the 1994 and 2001 crises, reaching 4.77% in 2001Q2.
The abstract omits natural gas from its regressors.
The body's evidence therefore makes gas the late-sample driver and assigns oil a marginal role. Karataş Elçiçek's abstract includes Brent among the long-run drivers and leaves gas out. Even the oil variables differ: the body uses high-sulphur fuel oil. Different variables and periods allow both findings to be true. They also prevent the body from supporting the abstract's claims.
A variance share cannot size this hedge
The forecast-error shares cannot supply a power hedge ratio in this setup. Each share assigns a fraction of variance to an orthogonalized shock, with the assignment dependent on ordering. Putting the exchange rate first credits it with any same-quarter movement shared by the lira and fuel prices. Because every price is in nominal lira, the fuel series already contain the exchange rate by construction. We found no results using alternative orderings.
The significance claims present another difficulty. The paper constructs its two-standard-error bands using the standard deviation of each response across the sample. Their width consequently measures how far responses move over time rather than posterior uncertainty at a particular date. Those bands support the verdict that oil is insignificant.
The dependent variable creates a deeper problem. The Energy Market Regulatory Authority (EMRA), established under the 2001 market law, regulates the market. We interpret the residential series as a regulated tariff rather than a clearing price. Private generators' output share rose from 18.23% in 2004 to 55.83% in 2015. A household tariff remains a poor proxy for a monthly base-load contract priced on expectations.
Sizing a hedge requires a coefficient estimated on the traded price itself. The attached text supplies none. It also fits five equations with drifting coefficients and volatilities to roughly 113 quarterly observations. Inefficiency factors reach 83.910. The authors describe those factors as low and the iterations as sufficient.
Keep the fuel-cost evidence; withhold the futures claims
The fuel-cost account is useful background. Turkish household electricity shifted from a currency-crisis price, with the lira accounting for 64.59% of short-horizon variance in 2001, toward a gas-cost price, with gas contributing 29.60% by 2016. For household tariffs through 2016, this evidence ranks gas first, the lira second and fuel oil a distant third. The 1.06% impact response measures an average-sized lira shock at 2016Q1. It says nothing about the abstract's January 2016 to July 2025 window.
The futures claims remain withheld. Brent, consumption, inflation and interest rates appear in the abstract but have no place in the body's model. The actual manuscript, including its long-run coefficients and error-correction speed, would change that assessment. Until those figures appear, the four signs in Karataş Elçiçek's abstract remain assertions. A desk should size nothing on them.