You are incorrect in your titled assertion. But I can guess where it comes from.

Most people use the simplest concepts they need to get the job done. They are considered about the forward voltage, \$V_{BE}\$, which is somewhat impacted by the collector current and very much impacted by temperature... so that's important... and \$V_{CE}\$ is immediately related to whether or not the BJT is saturated or not and this impacts very basic questions about available \$\beta\$, likely dissipation and temperature of operation which is pretty important. Besides, if you know \$V_{BE}\$ and \$V_{CE}\$ then you know \$V_{BC}\$. You might care about that, too. For example, the Early effect... But it's of secondary importance.

But you are wrong, anyway. The first model of the transistor to learn about is the Ebers-Moll model. It's level 1 model includes three distinct ways of looking at the BJT: transport, injection, and hybrid-pi. They are equivalent views, but they have different areas where they are easier to apply.

Let's look at the injection model first (addressing itself to _diode_ currents):

 - \$I_F = I_{ES} \cdot \left[ e^{\frac{q\cdot V_{BE}}{k\cdot T}} - 1 \right] \$
 - \$I_R = I_{CS} \cdot \left[ e^{\frac{q\cdot V_{BC}}{k\cdot T}} - 1 \right] \$
 - \$ I_C = \alpha_F \cdot I_F - I_R \$
 - \$ I_B = \left( 1 - \alpha_F \right) \cdot I_F + \left( 1 - \alpha_R \right) \cdot I_R \$
 - \$ I_E = -I_F + \alpha_R \cdot I_R \$

Now, the transport version (addressing itself to _collected_ currents):

 - \$I_{CC} = I_S \cdot \left[ e^{\frac{q\cdot V_{BE}}{k\cdot T}} - 1 \right] \$
 - \$I_{EC} = I_S \cdot \left[ e^{\frac{q\cdot V_{BC}}{k\cdot T}} - 1 \right] \$
 - \$ I_C = I_{CC} + \left[ -\frac{1}{\alpha_R} \right] \cdot I_{EC} \$
 - \$ I_B = \left[ \frac{1}{\alpha_F} - 1 \right] \cdot I_{CC} + \left[ \frac{1}{\alpha_R} - 1 \right] \cdot I_{EC} \$
 - \$ I_E = \left[ -\frac{1}{\alpha_F} \right] \cdot I_{CC} + I_{EC} \$

Finally, the non-linear hybrid-\$\pi\$ (nice, because linearizing it in the small-signal case leads directly to the well-known linear small-signal hybrid-\$\pi\$ model):

 - \$\frac{I_{CC}}{\beta_F} = \frac{I_S}{\beta_F} \cdot \left[ e^{\frac{q\cdot V_{BE}}{k\cdot T}} - 1 \right] \$
 - \$\frac{I_{EC}}{\beta_R} = \frac{I_S}{\beta_R} \cdot \left[ e^{\frac{q\cdot V_{BC}}{k\cdot T}} - 1 \right] \$
 - \$I_{CT} = I_{CC} - I_{EC}, \rm{(generator \,\, current)}\$
 - \$ I_C = \left( I_{CC} - I_{EC} \right) - \frac{I_{EC}}{\beta_R} \$
 - \$ I_B = \frac{I_{CC}}{\beta_F} + \frac{I_{EC}}{\beta_R} \$
 - \$ I_E = -\frac{I_{CC}}{\beta_F} - \left( I_{CC} - I_{EC} \right) \$

As you can easily see now, \$V_{BC}\$ figures quite prominently in the most basic and first level BJT modeling. And it doesn't stop there. It's present in EM1, EM2, EM3, Gummel-Poon, modified versions of those, and even into the latest models. You just haven't been exposed to even the first level of BJT modeling, yet. That's all. That's because for many (if not most) needs, you can simplify the basic BJT EM1 model still further and ignore quite a bit and still get by, okay.